Please note, this video is a talking point. Not a real practical choice for an audio preamp. Please watch ALL of the video before jumping in with a comment that I have probably already made within the video. If you don't believe wide bandwidth, and fast slew rate, is important, take a look at this video th-cam.com/video/q_HfTgN1kB4/w-d-xo.htmlsi=stvREkEbJjxS5NAK
MB, the "best" preamp we've found is based on Graham Cohen's "Double Balanced" topology. An OPA2891 would be an ideal "DC comparator" in this topology, especially if driving more than 4 bipolar transistors. We've used this for 35 years. Graham was from Australia.
I would like to add a few things to Mike's discussion points. First- The OPA2891 (also available as a slightly lower spec device- the OPA892) is a **video** amp primarily. However, TI's specification sheet for it quite clearly touts it as a device for "Professional" audio devices (of various sorts). Well, yes, sort of... maybe. The point that Mike and I were trying to make was NOT that it is the be-all and end-all device for audio, but rather that it could be used for that purpose **IF** the proper circuit was designed around it. However, quite honestly, NO ONE would be able to actually hear any potential improvements in slew rate and distortion as compared to a device such as the LME49720. So in the end it is just not worth the added time and effort. Stick with NE5532s, LM4562a, LME49720s, or similar family devices as you may like. Second- Unfortunately even if you did solder it to a DIP adapter and piggy-back the necessary bypass capacitors literally ON the device package (to keep it from breaking into oscillation) , it still cannot be a drop-in replacement for a 5532, 4562, 49720 or other similar spec devices. For example, its input bias current is about 1000 times (yes, 1000X..!) higher that of an LME49720. That alone means that the circuitry around it would have to be re-designed for it to even function properly. Finally- The real point of this video is to show just how slow 1970's vintage "audio" chips such as the LM358 were, and how very much the state of the art has improved, even as recently as with the LM4562/LME49720. In fact, if you have elderly hearing (such as ours), and assuming that the remainder of your hi-fi system is at all decent, you CAN hear the difference in audio played through an LM358, an NE5532 and an LME49720 using the very same amplifier circuit. You absolutely do NOT have to have magic ears or be a rich audiophile. More detail and clarity mainly. And those differences are primarily due to the faster slew rates. It is not that the detail was not there already, it was always there in the recordings. It is that slower devices and circuits mask it. The faster devices are not adding things, they are simply taking away less. So- please DON'T go buy OPA2891, OPA891 or OPA892 chips expecting to drop them in and have magical things happen (other than your money disappearing into thin air) because they just won't. 🤔
@@Mrhifitunes Nope. Just like in audio applications; it has no virtues that would be useful for a PS regulator and has too many issues that would make it almost impossible to work properly. Jut use it as intended; for video signals. A perceived "better" is not always better!
@ronschauer839 Absolutely right! Whatever the point of this YT is it's silly at best and harmful to the less experienced at worst. This "rolling" of opamps is often a bunch of BS with people claiming to hear differences between (hopefully) similarly speced devices. Between a LM358 and a LM4562? Sure. Between an NE5532 and a 4562, I don't think so. The one constant is the omission of double blind or ABX tests. If you're not evaluating the swaps that way you're just kidding yourself thinking you hear a difference.
@@Obsfucation If you think the audible differences between ICs are small (which I agree 100%) If you bring in differences in these so-called discrete devices a supper high price. Then the snake oil really starts to flow! Most do sound great but no better than the LM4562 but costs 100 times more money.
@@MichaelBeeny I'm not talking about using it as audio opamp but as power supply. It has the advantage of fast switching, low noise and low supply voltage. It can be used for 5v linear power supply, which cannot be with the lme.
The Texas Instrument datasheet says: 7.1.1 Driving a Capacitive Load The OPAx891 are internally compensated to maximize bandwidth and slew-rate performance. To maintain stability, take additional precautions when driving capacitive loads with a high-performance amplifier. As a result of the internal compensation, significant capacitive loading directly on the output node decreases the device phase margin, and potentially lead to high-frequency ringing or oscillations. Therefore, for capacitive loads greater than 10pF, place an isolation resistor in series with the output of the amplifier. Figure 7-1 shows this configuration. For most applications, a minimum resistance of 20Ω is recommended. In 75Ω transmission systems, setting the series resistor value to 75Ω is a beneficial choice because this value isolates any capacitance loading and provides source impedance matching. So if a 20 ohm resistor can solve the problem, we're good!
Spec sheet suggests that you can simply put a 20Ω resistor in series with the output if you need to drive a larger load. If you drive 100 pF, the RC time constant ends up being 2ns, which is still a small time constant compared to the intrinsic 100 V per µs, or 1 volt per 10 ns. So, yeah, you probably can use this part with an added output resistor and it will work just fine.
I made the preamp using Analog Devices AD815 ,which is a video amp with two channels, and I liked it immediately with its very broad image though some describe it as grainy. But to experience that level of image I liked it regardless. It was back in 2010 Carlos FM introduced the video ic AD815 on many forums. Thank you Michael and thank you Carlos!
Howdy michael, yet another great video. It will be very interesting to see where this eventually ends, and what solution/selection will become the «final draft» for the preamp. Best regards to you and your family. Take care.
Dear Mr Beeny thank you for your always excellent and very valuable analysis There is an old chip that was considered one of the best for a line stage preamp The ad797 from Analogue Devices Its noise and distortion are very low when it is used correctly
Hey Everybody! I just had a fascinating discussion with chat GPT to help understand something. Why on earth does it make such a big difference to put the sparkos ss3602 in these TPA3255 amps (fosi V3, V3 mono etc.). I think I found the answer and it’s more than “it’s a better opamp.” See the TPA3255 chip is highly capable but has a severe limitation. It requires a great deal of signal power and drive current to feed it. It’s not an amplifier that can be driven with .5V input to any satisfactory volume level without a huge boost from an opamp. This differs from most amplifier designs that simply do not need this boost and only use opamp as a buffer stage for volume or tone controls. In the Fosi V3, you will find two dual opamps. One is the preamp the other is the amplifier driver which boosts the signal after the preamp. The driver opamp is being pushed to deliver quite a lot of current similar to driving headphones at higher volume levels. This is the reason why the amplifier benefits so dramatically when you stick an SS3602 in that position because it has a 50ma output current capability which is really impressive and it can handle that gracefully. For the preamp socket, you could probably do just fine with something like the LME49720 which is about $4 and is about equally capable in situations where drive current isn’t needed. HF oscillation doesn’t seem to be an issue in these circuits because they are actually pretty well optimized. At least with the V3 I’m testing. I think it’s important to consider this because these amps are very cheap and you may not want to spend twice as much on opamps as you are spending on the amplifier and you probably don’t need to. Now what’s interesting is that the ne5532 has much higher drive capabilities as the LME49720 but the LME49720 has better sound quality. So this means if you just buy 2 LME49720s and populate both spots, it might be a downgrade at higher volume levels because it will run out of juice as a driver but will be great in the preamp circuit. Interesting right? I thought so. Also it’s freaky how much chatGPT understands about this stuff. Revision: Turns out chatGPT might have been wrong. The TPA3255 should only need 1-3ma of drive current so it seems like the LME should be just fine with its 26ma of drive current although ChatGPT still debates me on this saying that higher is better in this application and that it could be an audible difference.
I had enough with ChatGPT after I asked a simple question about a common base stage with bjt. Instead of answering my question "he" tried to convince me that common emitter is better.
@@jjcale2288 ai can listen that way. I am comfortable with discreet components. I have human ears. Amplifiers are not the problem. Microphones and speakers are.
My question is: if almost all recordings and mixes are produced using devices with "normal" slew rates (like a 5532), do playback devices using much higher slew rates "reclaim" detail and perceived speed/dynamics or do they just effect the signal in such a way that it adds it's own fast/clear character?
The object in a music **reproduction** system should be to not take anything more away, but rather to reproduce whatever actually made it to the recording as faithfully as possible. You should not be adding anything to the signal (unless that is your goal of course, such as bass boost). Regardless, you cannot re-create things that were lost in the studio because they are not there to be reclaimed, if that makes sense.
Not a chance. And there is no way that you are ever even going to get the potential bandwidth of even a middling good op-amp by the time the content gets to you. The bandwidth (and therefore the slew rate) of the playback from any source you care to name will be, by a wide, wide margin, the limiting factor.
Accuracy is illusion. Once many years ago i was for accuracy but now i realize that it's not important for enjoying music , accuracy is important for building bridges not for music ,it is impossible to know what is accurate, starting from chossing microphone for vocal or instruments in the studio and continues along the chain sources amplification and loudspeakers, every piece of equipment is different and have it's own flavor, accuracy is subjective. Once i had toshiba tv that was for me quite natural colors then i bought sony tv that the colors were much more vibrant and possibly more saturated. At first i tried too make sony looks like toshiba as used to but it wasn't possible. After some time i realized that picture is more enjoyable that way although not perfectly true to life as my eyes sees. No need to seek accuracy for music enjoyment.i have both valves and solid state amplifiers of any classes , although ss amplifier tend to be more accurate the valves brings more enjoyment and engagement, there is no true to life music and even the man who did the mastering in the studio don't remember after a year how it sounded then.
Many people think that audio hi-end is about an extreme precision, but in fact (aside of top notch connector, cables, switches, and a wow-design) it’s about a unique sound. And people with million$ are able to find their own one. For us an accuracy remains. We can chase zeroes in THD
There is a small firm in the US that have an interesting product on the market where they produce these miniature surface mount circuits which are the size and pinout of the classic 741 so they are a direct plug-in replacement for a single op amp. The idea is for about 50 quid you can take any commercial amp which uses a chip with that pinout and swap it in for an instant improvement in quality. They claim it is impossible to create a chip as good as their circuit because the integration of the transistors in the chip is inferior to a circuit where you use separate components. Two possible reasons are the noise problem of chips and the ability to accurately match transistors for the longtail pairs and current mirrors.
Very nice find! I'll be watching with interest your follow up videos with circuit diagrams etc. One very good chip is the Harris HA 5002/3 which has higher slew rate low noise and lower output impedance. I have been using these for years for both a preamp and many paralleled together to use as an audio amp direct.
Michael, this opamp is not perfect for audio (thats why its marketed for video😉) I glanced through the datasheet of this opamp and it has a few problems. First.. It has very low voltage noise but it has EXTREMELY HIGH current noise, it is highly dependent on low values for the feedback resistors and circuit topology. Secondly the frequency response is also highly dependent on the feedback and circuit topology. Michael, promise you slew rate and speed isn't the end all be all. Think of your analogy of using dual transformers for a preamp. As long as you have 'enough' slew rate with some margin is good enough. Monolithic IC designers spend thousands of hours and hundreds of thousands of dollars to come up with a solution for individual problem. Like everything in physics there may be an optimum solution but not one to fit every problem. If you're really looking for a solution for an audio preamp I would look at a composite amplifier do some research and feel free to message me or respond to this comment and I'll get you in the right direction
Exciting video! Would be good if you would state the bandwidth of the chip at 10 pF load, 100 pF and 1 nF. If it is so good at no load, then it must also be good at small loads and you get a long way with 1 nF.
For power amplifiers, the ADA4700-1 Opamp is really to be preferred. It can be used in the voltage amplification stage up to +/- 50 V, and also can set accurate bias.
There is a limit on the maximum slew rate you will see in a real world audio signal. The slew rate will be proportional to the frequency and maximum signal. Slew rate is in essence the first time derivative of the audio signal.
There is a section on the datasheet regarding capacitive loads, 7.1.1 Driving a Capacitive Load. I would add that, in principle, OPA2891 would be a very good solution for the first stage of amplification as the issue of noise is more pressing on a low level signal like one from a phono MC input, once the level is in the hundred milivolts, then potentially almost any modern op-amp can further amplify the signal for driving a power amp. But there is a potential issue with OPA2891, the voltage noise level below 100Hz becomes increasingly higher and the current noise is much higher than OPA1612 for example, which could be important depending on application. Important thing is to experiment and check what works best, on paper it could be perfect but in practical application it could sound worse of the same as a cheaper op-amp, it happened to me several times.
only slew rate dosn"t say annything about sound quality, the slewrate you need for audio is not that very high, even if you go up to 40.000 Hz , look at your scoop , see the flank, thats all you need, more is not needed.( a faster slewrate does not exist in 20 to 40.000 hz, a flank of a square will never be faster in the audio range)
Thank you Michael! As always, you've suggested quite a bit of gear. What do you think about the TL072? I was planning to use it in a guitar pedal it has high accuracy and sounds nice; sadly only available as SMD these days. However, I've never heard of the OPA2891, I'll give it a try
I agree that a LM358 is objectively terrible. Beyond demonstrating that point, I think the focus on slew rate here is misplaced. The things that make this op amp interesting for audio applications are its commendably low voltage noise spec, 0.95nV / sqrt(HZ), and its low distortion. The exceptional bandwidth only matters because it preserves a decent amount open loop gain clear out to 20 kHz, where it's still 80 dB. That's 10 dB better than an LM4562 or LME49720 at the same frequency.
You do of course realize, there is NO musical instrument or recording microphone that has a slew rate better than 1V/us? Perfectly fine for any preamp driving a power amp with 2V input sensitivity at 40kHz (>90% of all consumer products). Fantasy vs real world... it's not about speed; rather, the noise floor spectrum and out of band interference/intermodulation products and interface impedances! (that is where the DETAIL is hidden) And let's not forget those really big dollar bills always sound better, eh? Graph paper is wonderfully good at confusing audiophools. { short grain Egyptian papyrus only handled by virgins } ...and while I'm busy here screaming about nonsense - take a walk in the studio of your favorite recordings - YUP, daisey chained 5532's
Just chucking my two cents here: I use a 9U 104hp eurorack synthesizer which has a section of video modules and modules that use preamps that are specifically compatible with video rate signals as well as audio. And for either my audio or my control signals, running them through one of those modules enhances detail, and the breadth of nuances that change with a knob turn connected to one of those preamps has much more detail in a smaller section of a knob turn than what comes from audio preamps. I should stick to hem on the scope and sniff around, but just going by ear and a surface trace of things using them, I 100% agree that they introduce incredible detail because they’re over spec
It is best to make air wire connection or make pcb with Rogers which is usually for microwave pcb. Feedback wire should be air wire connected. I hade a gain peaking with feedback on pcb layout.
For me THD and noise in the audio pass band are more important than slew rate beyond a relatively low point, especially for a preamp, and then if you want to drive cable you have a 600R resistor in the output that will isolate the cable capacitance from the output and feedback stage. Then you're down to cable capacitance limiting your bandwidth only which should make it out to 20kHz give or take 15kHz is good enough for pop music which is somewhat band limited anyway.
But a faster opamp in a power amplifier, driving the power transistors, with benefit from a high slew rate to correct the cross-over distortion of the transistors, resulting in a much lower THD of the entire amplifier.
Hi, grateful for this interesting presentation. What would be your choice (and why) of opamp for performing I/V conversion after a DAC chip ? Thank you
Nothing wrong with the 5532 so far. As long 'audiophiles' (with a hearing range up to 12-15 kHz) listen to crappy bookshelf speakers randomly placed in untreated rooms, swapping OpAmps is just a fetish :) However, I like your videos. Is your new digital amp working fine? What about getting better coil filters with higher ratings for it? Guess this makes way more difference when power rises than any other parts on these boards.
It's hard to believe the 5532 is still as respected as it is. I should really say it's hard to believe something hasn't come along that would make its performance as outdated as the LM358 or the (gak!) 741. I remember upgrading our studio mixing consoles to 5532 opamps in the mid 1980's, so 40 years ago. It was a noticeable improvement if you knew what to listen for. Then about 10 or 15 years ago, OPA2134 opamps were the chip of choice. I couldn't hear much difference (I've been listening professionally for decades) but they had a positive placebo impact so that's what I've stuck with.
@MichaelBeeny I thought this chip was going to have an amazing slew rate, but at 105V/uS it's way behind the OPA861 that is already in some audio designs. Try 900V/uS
Although it cannot be used as a drop in replacement, PCB's from China are dirt cheap. The 10pF output is simple to overcome when designing a small dedicated board. Simply isolate the output of the op-amp with a 20Ω resistor. This isolation method is commonly used with all new op-amps.
That 20 ohm resistor only compensates for overshoot. It impedance matches the pcb trace to the opamp and will require experiment to match perfectly ie could be between 10-30 ohms.
Interesting, not that I have but a very basic understanding of electronics, but how is the OPA2891 actually used in a video circuit that the very low capacitance isn't a problem and why that isn't suitable to use in an audio circuit?
I am not a designer of video circuits, but I can guess that they use a discrete line driver buffer stage to decouple the OPA's output from downstream loads such as coax cables. There is even such a buffer technique shown in the LME49720 specification sheets (to boost its capabilities) from TI, and there are app notes for that sort of thing.
Michael and Ron are such teases. First they promise me a headphones amp, then they withhold Gerber files for a board accommodating a crazy fast chip. 👻
Well not exactly true. We did show the Accuphase clone preamp used as a great headphone amp, that works really well. We have no circuit using this chip. Not very practical, more a theoretical idea. We are working on a new preamp design, a bit far off yet. Measures well but still not done any listening tests. Might sound dreadful lol
@@el_arte There are two preamp circuits in the works. One is a Baxandall-style active volume control with gain (in other words, a preamp). The other is a simplified preamp based upon either the NE5532 or the LME49720 (your choice) but having a hybrid output buffer to decouple the effects of the load capacitance of successive stages from the op amp. So far so good in testing but we are not ready to divulge the circuit and results just yet. Sorry... Oh, and the "Accuphase C-3850" clone Mike reviewed works fine as a headphone amp with headphones as low as 32 ohms impedance. The final output transistors get a little warm, but nothing unmanageable.😀
Looks exciting but I agree probably no real time gains in audio, as for that piece of junk LME49720 well I'm not saying a word Lol! Now for serious audio with that dash of Burr Brown goodness I think that the OPA1656 is the leader of the pack ! ask your friends I would like to here yours and there's response to my statement and yes I love your videos but there isn't a button marked 'Love' so you will have to put up with a 'like'
I use these as well along with the OPA1642 etc from the OPA16XX 'Sound Plus' range in my pre-amps and they sound great for the price they charge for them. Even on DIP adaptors I prefer them now compared to LME497XX.
Go grab an OPA209 - lowest noise below 100Hz (however, many people prefer the 'warmth' and fuzziness of lo freq. noise like vinyl, cheap LF347 keeps them content)
Where is the input waveform and the 2891 waveform? The overshoot/undershoot looks like an impedance mismatch to the pcb trace, solved using series termination. Fast outputs have low drive capabilities and a suitable buffer could make this a brilliant (perfect) gain stage.
If speed is so inportant why not used LT1226. More expencive ($11.34 single vs $6.67). LY1226 have 1GHz bandwith. Another thing: there is leaded verion, so you could used socket.
What about potentiometers which can quickly have a negative effect on the sound. (PS Audio) user is a kind of digital potentiometer. not quite sure what the principle is
nice video. years ago, me and my friend experimented with the samples we got from Ti. we used the lme49990 and added the lme49600 as buffer just for fun.
If we need a speed (and if that means an accuracy) we should use a discrete pre-amp with local negative feedback only. In a current amplification mode. Some 300MHz bjt, and it will amplify PAL signal with a good linearity (0.01% this is what nobody hears). I remember when tape recorders switched to opamp driven pre-amplifiers, a sound became dead. If I still need an opamp for a sound these days, that would be muses02, burson v6, or sparkos 360x
I really don't understand why some people think IC don't sound good. Afterall it's full of transistors. The devices are very precision indeed. Much more accurate in component tolerances than any discreet circuit can be. And at a fraction of the cost. But here comes the snake oil. One is often in a shiny metal can, the other in a black plastic mold and looks very boring and just like a hundred other chips. So, they cannot be as good, can they?
@@MichaelBeeny For my opinion opamp is "worse" for two reasons. First, because it's a differential amplifier. DA affects a dynamic (however perhaps not all of them, as an internal structure of all those current mirrors can affect). Secondly, because it's a " devices are very precision indeed". That means that a stability is more important than a dynamics. If we measure a sine wave, we see nothing (because by ear we hear a difference between 5532 and 797, so it's about a processing of a complex wave). Third, like a bonus, most of opamps were designed for a measurement, not for a sound, when an input bias means more than a transient characteristics. I'm pretty confident that a linearity is not the most important for a sound (well, for a sound people like).
CMRR 90 dB min, Open Loop Gain 93 dB min, Voltage Noise 1.9 nV @ 40Hz, I would only use it with low gain (10 max), I haven't found where the 10 pF problem is hidden in the data sheet. At 10nF and 20KHz, the permissible output current is only used to about 10%. All my opinion😉
That's how it goes. We old farts we have the knowledge and wisdom, but the hearing accuracy is long gone. The younger may have golden ears (I doubt though) but not so much experience and hands-on for the practical part of hi-fi. They grew up with plastic PC loudspeakers and 5.1 home theater. They don't need high slew rate for Bluetooth speakers.
Only measurements will show which is "best"........but whether the improvement in measurements,slew rate etc over the rival would be audible is a different matter.
@@Pete.across.the.street No, it's the exact opposite. Humans are poor sensors and defining "best" with only the hearing experience as a metric only means not quality matters but everything that is subjective by nature, therefore opinion based. This is why measurements are the only way to make everyone objectively agree on facts. Listening will maybe tell you the one you *like* the most. But definitely not which one is "best".
@@nashaut7635 With any audio project, I always start with, in some cases many weeks of listening tests. Only then does it go in the bench. Often the measurements confirm what I've heard but as circuits and ICs get better, the differences are also harder, if not imposable hear OR measure.
It's not. Slew rate only matters when high frequencies are amplified to high voltages. A few volts per microsecond is more than fast enough, even in a thousand-watt (90VRMS at 8 ohms) power amp playing at 20kHz. For a preamp operating at only a couple of volts RMS, even less slew rate is necessary. The slew rate of a sinewave (in volts per second) is (2π * Vp * f) where Vp is the peak voltage (not peak to peak or RMS) and f is the frequency. For 90VRMS (1000 watts into an 8-ohm load), Vp is 127.3 (rounded to a tenth). At 20kHz, the slew rate is approximately 16 million volts per second, or 16V/uS. For 2VRMS (the output level of a CD player at 0dBFS), the slew rate is only 0.35V/uS. To avoid issues, the slew rate of a circuit should be at least a couple of times the slew rate of the signal being reproduced. The slew rate of a 4558 opamp (very common, very cheap audio opamp) is 2.2V/uS, which is six times the slew rate of a 20kHz signal at 2VRMS. That's plenty.
I agree that it's generally not an issue (anymore). For example, the LM358: if I did the math right, it can't produce a 10 V peak to peak signal over 16 kHz. So that is a realistic flaw but there's no need to use that IC.
I'm not 100% sure I understand the question! Are you talking preamps, integrated amplifiers or just power amps? Is the 1,500 dollars US of pounds UK? you gotta help me lol
The LME49721 is our favorite right now. The results are clearly shown in this video. It does need a capacitor mod, again shown in past videos, to avoid stability issues.
@@Pete.across.the.street Hi Pete, you will be shocked, like I was to hear, not everyone watches all my videos!! and one or two don't watch them all the way though. I know that's hard to believe. I was devastated.
When used as an output driver, the chip won't be connected directly to the cable. It will have a 75 ohm resistor in series with it to match the cables characteristic impedance. Of course, one must use careful design and board layout to keep a chip like this happy. Do that, and it's a monster capable of pushing 200 mA.
Why being so obsessed about slew rate? What is the bandwidth of your ears? You're not the youngest any more, so if you’re lucky, you might still hear 15 or 16 kHz. A large signal rise time of 1 µs results in a large signal bandwidth 350 kHz! Your oscilloscope already shows a slightly better rise time for the NE5532. So the large signal bandwidth of the NE5532, being >350 kHz, is already miles beyond the bandwidth of your ears. So why bother for more slew rate and bandwidth? Even if the rise time is only 5 µs, the bandwidth is still 70 kHz, way better than your ears. What about accuracy? What would be the accuracy of all the equipment between a musical instrument and your ears? That includes: microphones, equipment for recording, editing, mixing, ADCs for CD/DVD, your audio chain with DACs and speakers and the acoustical properties of your room. Any accuracy, if ever present, will be long gone in this long chain. Adjust the controls of your audio chain as you like it and enjoy the music!
Op-amps have a very high gain, that is reduced by feedback from end to start in the application. Many designers of amplifiers therefore don't like op-amps and insist on discrete transistors, where each stage can have its own feedback loop (emitter resistor) which is reacting immediately on this stage, not delayed over many stages. These designers avoid an over-all feedback loop from output back to input, or at least use only a small part of the total feedback over the long way. I never designed audio amplifiers, but I can understand the motivation to avoid or at least reduce over-all feedback. However - if the op-amp is extremely fast compared to the frequency band needed - that critic can be forgotten. The delay from output back to input is negligible.
your ears can hear a phase difference of microseconds. At 5kHz, 5 degrees of phase shift is 2.8us. So yeah you can't hear 20kHz, but you can hear the phase shift of an amp that rolls off at 20kHz. Or a speaker for that matter.
Whats with this obsession with slew rate? Listening to ultrasonics? 10-20v/us is plenty enough for anything audio. Capacitance doesn't matter as long as you push enough current.
I think the LM358 has never been used in any commercial piece of audio equipment. Regarding slew rate. A 20 V 20 kHZ sine wave would require a slew rate of only 2.5 V/µs so I fail to see why you would need over 9 V/µs for audio applications. For a 20 V sine wave, the NE5532 would work fine up to 70 kHZ until slew rate would start to be a problem. Audio processing in the analog domain is still done with TL072, NE5532, RC4560, RC4580... nothing exotic at all. The signal from a microphone in an old analog mixing console would pass trough maybe 50 op amps (the ones everyone seems to hate because they seem to sound bad) before it reaches the output of a mixing console. Even your most beloved audiophile recordings mixed on some world famous analog mixing consoles are all equipped with ordinary op amps and many of them. Schematics of those famous mixing consoles are out there. The trick in those consoles is not in the op amps but the reference structure of the mixing console. This structure you will probably never find in home audio equipment not even the expensive units. (at least not to my knowledge)
I've seen the LM358 used in the audio stages of some Tait Electronics base station transmitter designs from the 1980s. Perhaps not the sort of audio that some people would be interested in, but it certainly was audio. I seem to remember these were quite sensitive to strong RF fields which sometimes meant the transmitter could leak back into the audio stages with some undesirable side effects. One quick fix for this was to replace them with something like a TL062. The LM358 was cheap, but not my favourite Op Amp.
Kâğıt üzerinde opa 1612 cok iyi degerlere sahip,ama kör dinleme testinde muses02 ile yarışamaz, ayrica opa2134 vee lm4562 opa1612 ye göre kör testte kulağa daha iyi gelen sesi verir! Kör test nihai objektif dinleme testidir yalan söylemez ve muses02 yi ancak akliam lc5 geçer ve a sınıf çalışan bir op amp tir ve cok ısınır! Opa 1656 denemedim ama lm4562 seviyesinde olacagini tahmin ediyorum
Hi Michael. While I like some of your reviews of the various cheap audio kits available, I believe your assertions that high slew rates are required for good audio amplifiers is non sensical. The required response range for "hi-fi" is generally regarded as 20 Hz to 20 kHz. Human voice, musical instruments, microphones and loudspeakers don't respond much outside this range and certainly human hearing covers a smaller range which further decreases quickly with age. Maximum slew rate is a simple function of the peak voltage and the frequency being reproduced. Maximum slew rate i.e. rate of change of a sine wave occurs at the zero crossing and can be found from a straight forward differential equation: ∂/∂t [Vpeak * sin(w*t) where w = 2*π*f. Do the maths and you will see what slew rates are actually required which can be easily verified with a signal generator and an oscilloscope. For example, the output stage of a power amp specified to provide 100 W into 8 Ω (which corresponds to about 40 V peak) needs a slew rate of: 2*π*20000*40 = 5.02 V/µs to reproduce 20 kHz. ICs that might be used in preamplifier stages only ever need to produce output voltages that peak in the 100s of millivolts otherwise the power amp will be overdriven so have a maximum slew rate requirement of less than 1V/µs to be able to full reproduce signals in excess of 20 kHz. One of the first lessons I learnt as an electrical engineering undergraduate was do not design anything faster than it needs to be. An amplifier that is faster than necessary requires extra design effort to guarantee stability and minimise noise. If you study the work of some of the reputable practitioners of good audio design such as John Linsley-Hood, Douglas Self, Peter Walker etc you will see a realistic approach taken to design requirements.
You stick to pure theory, but when you hear the difference in sound clearly you have missed something. Many amplifiers measure the same but sound way different, sine-wave measurements tell nothing as it is the easiest form of signal to reproduce up to the levels of 0.001 THD and below, you forget about phase delay of low frequences relative to high frequences, settling time also in some circuits it can be included in a feedback loop and have an effect on damping the drivers, an increased delay in this case will be detrimental but will not be apparent on pure sine-wave as it will stay a pure sine-wave. Probably this chip has also other parameters giving him an edge over other audio op-amps.
My current discrete build preamp has a slewrate of 200Vu/s. Bandwidth is way beyond human hearing, yet it sounds so good and natural. Cutting the bandwidth would essentially reduce the slewrate. Since the preamp runs happily without problems, I'll keep my +1MHz range and enjoying the 20-20000hz music it produces with the enormous headroom that the slewrate provide.
@@glasslinger why? You can actually hear this even at low level music. Many DIY'ers (and often expensive consumer amplifiers) aim for a high slewrate, it's just a matter of design and parts selection. Often no extra cost if done right.
@@kenielsen The comment left out the :) I was making a funny, having all that power and not needing it. I'm 79 so my listening days are muted. I can use a 358!
You really don't seem to understand preamp design using opamps. It is NEVER good practice to use directly the output of an opamp to an output. NEVER EVER. Typical is the use of 100R from output of the opamp to the output. If you don't understand why, I can be bothered to explain it, it has something to do with the term "infinity" which is impossible..
You clearly have a fundamental misunderstanding of how opamps work if you think that dropping them into an otherwise identical circuit is going to give a fair comparison of how well each opamp can work. I suggest that anyone who's serious about getting the most out of actual audio opamps consults the works of a Mr Doug Self.
You clearly have not watched my video with care. In fact, I have stated you CANNOT do this. In this video alone I've said if you swap the 5532 with a LME49720 you need extra caps across the supply pins 4 & 8 to keep it stable. This whole video is about that very fact that in MANY chases you cannot just swap them over. Many people do just that. I'm fully aware of Doug's bible, I have a copy of a number of his books right here. You will note, I only swapped 3 explaining the need for extra decoupling. The OPA2891 was the main subject of the video, explaining you CANNOT JUST DROP IT IN. It looks like you watched 3 mins of my video and started typing!
Maalesef türkçe altyazi yok michael en ilgilendiğim sevdigim konular saf ses üzerine bu op amp ses için en iyi değerlere sahip ama pratikte iyi ses vermiyormu anlamadim?!
I'm so sorry, but subtitles are 100% out of my control. TH-cam just do it (sometimes) I have written to them about this. I hope I receive a reply soon, or better still, they fix it. Regarding the IC. It's not really practical to build for audio. Really just a talking point. Should it be possible to build a working circuit, I don't doubt it WILL sound good.
Just chucking my two cents here: I use a 9U 104hp eurorack synthesizer which has a section of video modules and modules that use preamps that are specifically compatible with video rate signals as well as audio. And for either my audio or my control signals, running them through one of those modules enhances detail, and the breadth of nuances that change with a knob turn connected to one of those preamps has much more detail in a smaller section of a knob turn than what comes from audio preamps. I should stick to hem on the scope and sniff around, but just going by ear and a surface trace of things using them, I 100% agree that they introduce incredible detail because they’re over spec
Please note, this video is a talking point. Not a real practical choice for an audio preamp. Please watch ALL of the video before jumping in with a comment that I have probably already made within the video.
If you don't believe wide bandwidth, and fast slew rate, is important, take a look at this video
th-cam.com/video/q_HfTgN1kB4/w-d-xo.htmlsi=stvREkEbJjxS5NAK
Michael. My apologies I did watch the video but did not see this comment. Have a great day!
If you don't LIKE my video, I might have to take up stamp collecting! so P L E A S E like it.
I collected stamps as a kid but I was saved by music. Understood, liked!
Mike, given the cost of postage to NZ you won't be getting any letters from me!
@@stevesmyth4982 Does anyone STILL collect stamps these days?
@@stevesmyth4982 But Steve, what if I don't get enough LIKES? I will need all the stamps I can get!
NO. If you didn't ask I might.
MB, the "best" preamp we've found is based on Graham Cohen's "Double Balanced" topology. An OPA2891 would be an ideal "DC comparator" in this topology, especially if driving more than 4 bipolar transistors. We've used this for 35 years. Graham was from Australia.
I would like to add a few things to Mike's discussion points.
First-
The OPA2891 (also available as a slightly lower spec device- the OPA892) is a **video** amp primarily.
However, TI's specification sheet for it quite clearly touts it as a device for "Professional" audio devices (of various sorts).
Well, yes, sort of... maybe.
The point that Mike and I were trying to make was NOT that it is the be-all and end-all device for audio, but rather that it could be used for that purpose **IF** the proper circuit was designed around it.
However, quite honestly, NO ONE would be able to actually hear any potential improvements in slew rate and distortion as compared to a device such as the LME49720.
So in the end it is just not worth the added time and effort.
Stick with NE5532s, LM4562a, LME49720s, or similar family devices as you may like.
Second-
Unfortunately even if you did solder it to a DIP adapter and piggy-back the necessary bypass capacitors literally ON the device package (to keep it from breaking into oscillation) , it still cannot be a drop-in replacement for a 5532, 4562, 49720 or other similar spec devices.
For example, its input bias current is about 1000 times (yes, 1000X..!) higher that of an LME49720.
That alone means that the circuitry around it would have to be re-designed for it to even function properly.
Finally-
The real point of this video is to show just how slow 1970's vintage "audio" chips such as the LM358 were, and how very much the state of the art has improved, even as recently as with the LM4562/LME49720.
In fact, if you have elderly hearing (such as ours), and assuming that the remainder of your hi-fi system is at all decent, you CAN hear the difference in audio played through an LM358, an NE5532 and an LME49720 using the very same amplifier circuit.
You absolutely do NOT have to have magic ears or be a rich audiophile.
More detail and clarity mainly.
And those differences are primarily due to the faster slew rates.
It is not that the detail was not there already, it was always there in the recordings.
It is that slower devices and circuits mask it.
The faster devices are not adding things, they are simply taking away less.
So- please DON'T go buy OPA2891, OPA891 or OPA892 chips expecting to drop them in and have magical things happen (other than your money disappearing into thin air) because they just won't. 🤔
@@ronschauer839 maybe it's a good opamp to use on a power supply as the super regulator and others
@@Mrhifitunes Nope. Just like in audio applications; it has no virtues that would be useful for a PS regulator and has too many issues that would make it almost impossible to work properly. Jut use it as intended; for video signals. A perceived "better" is not always better!
@ronschauer839 Absolutely right! Whatever the point of this YT is it's silly at best and harmful to the less experienced at worst. This "rolling" of opamps is often a bunch of BS with people claiming to hear differences between (hopefully) similarly speced devices. Between a LM358 and a LM4562? Sure. Between an NE5532 and a 4562, I don't think so. The one constant is the omission of double blind or ABX tests. If you're not evaluating the swaps that way you're just kidding yourself thinking you hear a difference.
@@Obsfucation If you think the audible differences between ICs are small (which I agree 100%) If you bring in differences in these so-called discrete devices a supper high price. Then the snake oil really starts to flow! Most do sound great but no better than the LM4562 but costs 100 times more money.
@@MichaelBeeny
I'm not talking about using it as audio opamp but as power supply.
It has the advantage of fast switching, low noise and low supply voltage. It can be used for 5v linear power supply, which cannot be with the lme.
The Texas Instrument datasheet says:
7.1.1 Driving a Capacitive Load
The OPAx891 are internally compensated to maximize bandwidth and slew-rate performance. To maintain
stability, take additional precautions when driving capacitive loads with a high-performance amplifier. As a
result of the internal compensation, significant capacitive loading directly on the output node decreases the
device phase margin, and potentially lead to high-frequency ringing or oscillations. Therefore, for capacitive
loads greater than 10pF, place an isolation resistor in series with the output of the amplifier. Figure 7-1 shows
this configuration. For most applications, a minimum resistance of 20Ω is recommended. In 75Ω transmission
systems, setting the series resistor value to 75Ω is a beneficial choice because this value isolates any
capacitance loading and provides source impedance matching.
So if a 20 ohm resistor can solve the problem, we're good!
Spec sheet suggests that you can simply put a 20Ω resistor in series with the output if you need to drive a larger load. If you drive 100 pF, the RC time constant ends up being 2ns, which is still a small time constant compared to the intrinsic 100 V per µs, or 1 volt per 10 ns.
So, yeah, you probably can use this part with an added output resistor and it will work just fine.
I made the preamp using Analog Devices AD815 ,which is a video amp with two channels, and I liked it immediately with its very broad image though some describe it as grainy. But to experience that level of image I liked it regardless. It was back in 2010 Carlos FM introduced the video ic AD815 on many forums. Thank you Michael and thank you Carlos!
Howdy michael, yet another great video. It will be very interesting to see where this eventually ends, and what solution/selection will become the «final draft» for the preamp. Best regards to you and your family. Take care.
Dear Mr Beeny thank you for your always excellent and very valuable analysis
There is an old chip that was considered one of the best for a line stage preamp The ad797 from Analogue Devices
Its noise and distortion are very low when it is used correctly
Hey Everybody! I just had a fascinating discussion with chat GPT to help understand something. Why on earth does it make such a big difference to put the sparkos ss3602 in these TPA3255 amps (fosi V3, V3 mono etc.). I think I found the answer and it’s more than “it’s a better opamp.”
See the TPA3255 chip is highly capable but has a severe limitation. It requires a great deal of signal power and drive current to feed it. It’s not an amplifier that can be driven with .5V input to any satisfactory volume level without a huge boost from an opamp. This differs from most amplifier designs that simply do not need this boost and only use opamp as a buffer stage for volume or tone controls. In the Fosi V3, you will find two dual opamps. One is the preamp the other is the amplifier driver which boosts the signal after the preamp. The driver opamp is being pushed to deliver quite a lot of current similar to driving headphones at higher volume levels. This is the reason why the amplifier benefits so dramatically when you stick an SS3602 in that position because it has a 50ma output current capability which is really impressive and it can handle that gracefully.
For the preamp socket, you could probably do just fine with something like the LME49720 which is about $4 and is about equally capable in situations where drive current isn’t needed. HF oscillation doesn’t seem to be an issue in these circuits because they are actually pretty well optimized. At least with the V3 I’m testing.
I think it’s important to consider this because these amps are very cheap and you may not want to spend twice as much on opamps as you are spending on the amplifier and you probably don’t need to.
Now what’s interesting is that the ne5532 has much higher drive capabilities as the LME49720 but the LME49720 has better sound quality. So this means if you just buy 2 LME49720s and populate both spots, it might be a downgrade at higher volume levels because it will run out of juice as a driver but will be great in the preamp circuit.
Interesting right? I thought so. Also it’s freaky how much chatGPT understands about this stuff.
Revision:
Turns out chatGPT might have been wrong. The TPA3255 should only need 1-3ma of drive current so it seems like the LME should be just fine with its 26ma of drive current although ChatGPT still debates me on this saying that higher is better in this application and that it could be an audible difference.
Funny i did the exact mod lme for pre and ss3602 for latter. It's fantastic and i don't push loud levels thus perfect for me.
@@wric01 That’s so cool! Just curious, did you try it the other way? And what did you notice about the change vs the stock ne5532s?
I had enough with ChatGPT after I asked a simple question about a common base stage with bjt. Instead of answering my question "he" tried to convince me that common emitter is better.
@@jjcale2288 ai can listen that way. I am comfortable with discreet components. I have human ears. Amplifiers are not the problem. Microphones and speakers are.
My question is: if almost all recordings and mixes are produced using devices with "normal" slew rates (like a 5532), do playback devices using much higher slew rates "reclaim" detail and perceived speed/dynamics or do they just effect the signal in such a way that it adds it's own fast/clear character?
once it's gone, it's gone forever...
@@torstenjohann9204 Any use of a slow opamp in the reproduction chain will subtract even more detail from the music
The object in a music **reproduction** system should be to not take anything more away, but rather to reproduce whatever actually made it to the recording as faithfully as possible.
You should not be adding anything to the signal (unless that is your goal of course, such as bass boost).
Regardless, you cannot re-create things that were lost in the studio because they are not there to be reclaimed, if that makes sense.
Not a chance. And there is no way that you are ever even going to get the potential bandwidth of even a middling good op-amp by the time the content gets to you. The bandwidth (and therefore the slew rate) of the playback from any source you care to name will be, by a wide, wide margin, the limiting factor.
Accuracy is illusion.
Once many years ago i was for accuracy but now i realize that it's not important for enjoying music , accuracy is important for building bridges not for music ,it is impossible to know what is accurate, starting from chossing microphone for vocal or instruments in the studio and continues along the chain sources amplification and loudspeakers, every piece of equipment is different and have it's own flavor, accuracy is subjective.
Once i had toshiba tv that was for me quite natural colors then i bought sony tv that the colors were much more vibrant and possibly more saturated. At first i tried too make sony looks like toshiba as used to but it wasn't possible. After some time i realized that picture is more enjoyable that way although not perfectly true to life as my eyes sees.
No need to seek accuracy for music enjoyment.i have both valves and solid state amplifiers of any classes , although ss amplifier tend to be more accurate the valves brings more enjoyment and engagement, there is no true to life music and even the man who did the mastering in the studio don't remember after a year how it sounded then.
Many people think that audio hi-end is about an extreme precision, but in fact (aside of top notch connector, cables, switches, and a wow-design) it’s about a unique sound. And people with million$ are able to find their own one. For us an accuracy remains. We can chase zeroes in THD
There is a small firm in the US that have an interesting product on the market where they produce these miniature surface mount circuits which are the size and pinout of the classic 741 so they are a direct plug-in replacement for a single op amp. The idea is for about 50 quid you can take any commercial amp which uses a chip with that pinout and swap it in for an instant improvement in quality. They claim it is impossible to create a chip as good as their circuit because the integration of the transistors in the chip is inferior to a circuit where you use separate components. Two possible reasons are the noise problem of chips and the ability to accurately match transistors for the longtail pairs and current mirrors.
Thanks a lot - very interesting! 👍 Best regards from Germany! 🍀
Very nice find! I'll be watching with interest your follow up videos with circuit diagrams etc. One very good chip is the Harris HA 5002/3 which has higher slew rate low noise and lower output impedance. I have been using these for years for both a preamp and many paralleled together to use as an audio amp direct.
Michael, this opamp is not perfect for audio (thats why its marketed for video😉) I glanced through the datasheet of this opamp and it has a few problems. First.. It has very low voltage noise but it has EXTREMELY HIGH current noise, it is highly dependent on low values for the feedback resistors and circuit topology. Secondly the frequency response is also highly dependent on the feedback and circuit topology. Michael, promise you slew rate and speed isn't the end all be all. Think of your analogy of using dual transformers for a preamp. As long as you have 'enough' slew rate with some margin is good enough. Monolithic IC designers spend thousands of hours and hundreds of thousands of dollars to come up with a solution for individual problem. Like everything in physics there may be an optimum solution but not one to fit every problem. If you're really looking for a solution for an audio preamp I would look at a composite amplifier do some research and feel free to message me or respond to this comment and I'll get you in the right direction
Exciting video!
Would be good if you would state the bandwidth of the chip at 10 pF load, 100 pF and 1 nF.
If it is so good at no load, then it must also be good at small loads and you get a long way with 1 nF.
For power amplifiers, the ADA4700-1 Opamp is really to be preferred. It can be used in the voltage amplification stage up to +/- 50 V, and also can set accurate bias.
There is a limit on the maximum slew rate you will see in a real world audio signal. The slew rate will be proportional to the frequency and maximum signal. Slew rate is in essence the first time derivative of the audio signal.
There is a section on the datasheet regarding capacitive loads, 7.1.1 Driving a Capacitive Load. I would add that, in principle, OPA2891 would be a very good solution for the first stage of amplification as the issue of noise is more pressing on a low level signal like one from a phono MC input, once the level is in the hundred milivolts, then potentially almost any modern op-amp can further amplify the signal for driving a power amp. But there is a potential issue with OPA2891, the voltage noise level below 100Hz becomes increasingly higher and the current noise is much higher than OPA1612 for example, which could be important depending on application. Important thing is to experiment and check what works best, on paper it could be perfect but in practical application it could sound worse of the same as a cheaper op-amp, it happened to me several times.
only slew rate dosn"t say annything about sound quality, the slewrate you need for audio is not that very high, even if you go up to 40.000 Hz , look at your scoop , see the flank, thats all you need, more is not needed.( a faster slewrate does not exist in 20 to 40.000 hz, a flank of a square will never be faster in the audio range)
Thank you Michael! As always, you've suggested quite a bit of gear. What do you think about the TL072? I was planning to use it in a guitar pedal it has high accuracy and sounds nice; sadly only available as SMD these days. However, I've never heard of the OPA2891, I'll give it a try
Try ne5532 it's good
I have an video opamp in my analog oscilloscope from the 90ties by the way.
49720 seems ok for the audio I think.
I agree that a LM358 is objectively terrible. Beyond demonstrating that point, I think the focus on slew rate here is misplaced. The things that make this op amp interesting for audio applications are its commendably low voltage noise spec, 0.95nV / sqrt(HZ), and its low distortion. The exceptional bandwidth only matters because it preserves a decent amount open loop gain clear out to 20 kHz, where it's still 80 dB. That's 10 dB better than an LM4562 or LME49720 at the same frequency.
You do of course realize, there is NO musical instrument or recording microphone that has a slew rate better than 1V/us? Perfectly fine for any preamp driving a power amp with 2V input sensitivity at 40kHz (>90% of all consumer products).
Fantasy vs real world... it's not about speed; rather, the noise floor spectrum and out of band interference/intermodulation products and interface impedances! (that is where the DETAIL is hidden) And let's not forget those really big dollar bills always sound better, eh?
Graph paper is wonderfully good at confusing audiophools. { short grain Egyptian papyrus only handled by virgins }
...and while I'm busy here screaming about nonsense - take a walk in the studio of your favorite recordings - YUP, daisey chained 5532's
Just chucking my two cents here: I use a 9U 104hp eurorack synthesizer which has a section of video modules and modules that use preamps that are specifically compatible with video rate signals as well as audio. And for either my audio or my control signals, running them through one of those modules enhances detail, and the breadth of nuances that change with a knob turn connected to one of those preamps has much more detail in a smaller section of a knob turn than what comes from audio preamps.
I should stick to hem on the scope and sniff around, but just going by ear and a surface trace of things using them, I 100% agree that they introduce incredible detail because they’re over spec
I like the lm6172. It sounded really good and I think it was made for video and high speed application.
It is best to make air wire connection or make pcb with Rogers which is usually for microwave pcb.
Feedback wire should be air wire connected.
I hade a gain peaking with feedback on pcb layout.
FR4 can easily do 100Mhz square wave with fast rise time and no overshoot.
For me THD and noise in the audio pass band are more important than slew rate beyond a relatively low point, especially for a preamp, and then if you want to drive cable you have a 600R resistor in the output that will isolate the cable capacitance from the output and feedback stage. Then you're down to cable capacitance limiting your bandwidth only which should make it out to 20kHz give or take 15kHz is good enough for pop music which is somewhat band limited anyway.
But a faster opamp in a power amplifier, driving the power transistors, with benefit from a high slew rate to correct the cross-over distortion of the transistors, resulting in a much lower THD of the entire amplifier.
Hi, grateful for this interesting presentation. What would be your choice (and why) of opamp for performing I/V conversion after a DAC chip ? Thank you
Nothing wrong with the 5532 so far. As long 'audiophiles' (with a hearing range up to 12-15 kHz) listen to crappy bookshelf speakers randomly placed in untreated rooms, swapping OpAmps is just a fetish :)
However, I like your videos. Is your new digital amp working fine? What about getting better coil filters with higher ratings for it? Guess this makes way more difference when power rises than any other parts on these boards.
It's hard to believe the 5532 is still as respected as it is. I should really say it's hard to believe something hasn't come along that would make its performance as outdated as the LM358 or the (gak!) 741. I remember upgrading our studio mixing consoles to 5532 opamps in the mid 1980's, so 40 years ago. It was a noticeable improvement if you knew what to listen for. Then about 10 or 15 years ago, OPA2134 opamps were the chip of choice. I couldn't hear much difference (I've been listening professionally for decades) but they had a positive placebo impact so that's what I've stuck with.
@MichaelBeeny I thought this chip was going to have an amazing slew rate, but at 105V/uS it's way behind the OPA861 that is already in some audio designs. Try 900V/uS
Dear Michael, certainly is a perfect opamp, but there are a lot of other excellent.
My point of view :
The circuit is very importance
Although it cannot be used as a drop in replacement, PCB's from China are dirt cheap. The 10pF output is simple to overcome when designing a small dedicated board. Simply isolate the output of the op-amp with a 20Ω resistor. This isolation method is commonly used with all new op-amps.
If only life was that simple. The biasing requirements are VERY different. It really is not a drop-in replacement, even with a conversion PCB.
That 20 ohm resistor only compensates for overshoot. It impedance matches the pcb trace to the opamp and will require experiment to match perfectly ie could be between 10-30 ohms.
Interesting, not that I have but a very basic understanding of electronics, but how is the OPA2891 actually used in a video circuit that the very low capacitance isn't a problem and why that isn't suitable to use in an audio circuit?
I am not a designer of video circuits, but I can guess that they use a discrete line driver buffer stage to decouple the OPA's output from downstream loads such as coax cables.
There is even such a buffer technique shown in the LME49720 specification sheets (to boost its capabilities) from TI, and there are app notes for that sort of thing.
OPA 891the best ( fast, smooth, natural ) sounding audio OPamp for me. I started to use it in super regulators, preamps.
Rise time of 25 minutes!! Oh that killed me 🤣🤣🤣🤣🤣
Accurate and fast. As fast as 105V/μS. We shall need plasma speakers then. Have no mass inertia at all.
Michael and Ron are such teases. First they promise me a headphones amp, then they withhold Gerber files for a board accommodating a crazy fast chip. 👻
Well not exactly true. We did show the Accuphase clone preamp used as a great headphone amp, that works really well. We have no circuit using this chip. Not very practical, more a theoretical idea. We are working on a new preamp design, a bit far off yet. Measures well but still not done any listening tests. Might sound dreadful lol
@@MichaelBeeny eheh, I know.
@@el_arte There are two preamp circuits in the works.
One is a Baxandall-style active volume control with gain (in other words, a preamp).
The other is a simplified preamp based upon either the NE5532 or the LME49720 (your choice) but having a hybrid output buffer to decouple the effects of the load capacitance of successive stages from the op amp.
So far so good in testing but we are not ready to divulge the circuit and results just yet. Sorry...
Oh, and the "Accuphase C-3850" clone Mike reviewed works fine as a headphone amp with headphones as low as 32 ohms impedance.
The final output transistors get a little warm, but nothing unmanageable.😀
@@ronschauer839 Thanks for the update, Ron.
Would love to see good/bad the chip preforms with higher capacitance. Is it still beter than the lme with 100pf?
Looks exciting but I agree probably no real time gains in audio, as for that piece of junk LME49720 well I'm not saying a word Lol! Now for serious audio with that dash of Burr Brown goodness I think that the OPA1656 is the leader of the pack ! ask your friends I would like to here yours and there's response to my statement and yes I love your videos but there isn't a button marked 'Love' so you will have to put up with a 'like'
Yes Andy, love might be a bit over the top lol I'm happy with LIKE.
@@MichaelBeeny :)
also a lover here of the OPA1656, i still have to hear something better in terms of op amp chips
I use these as well along with the OPA1642 etc from the OPA16XX 'Sound Plus' range in my pre-amps and they sound great for the price they charge for them. Even on DIP adaptors I prefer them now compared to LME497XX.
Go grab an OPA209 - lowest noise below 100Hz (however, many people prefer the 'warmth' and fuzziness of lo freq. noise like vinyl, cheap LF347 keeps them content)
Where is the input waveform and the 2891 waveform? The overshoot/undershoot looks like an impedance mismatch to the pcb trace, solved using series termination. Fast outputs have low drive capabilities and a suitable buffer could make this a brilliant (perfect) gain stage.
If speed is so inportant why not used LT1226. More expencive ($11.34 single vs $6.67). LY1226 have 1GHz bandwith. Another thing: there is leaded verion, so you could used socket.
What about potentiometers which can quickly have a negative effect on the sound. (PS Audio) user is a kind of digital potentiometer. not quite sure what the principle is
nice video. years ago, me and my friend experimented with the samples we got from Ti. we used the lme49990 and added the lme49600 as buffer just for fun.
If we need a speed (and if that means an accuracy) we should use a discrete pre-amp with local negative feedback only. In a current amplification mode. Some 300MHz bjt, and it will amplify PAL signal with a good linearity (0.01% this is what nobody hears). I remember when tape recorders switched to opamp driven pre-amplifiers, a sound became dead.
If I still need an opamp for a sound these days, that would be muses02, burson v6, or sparkos 360x
I really don't understand why some people think IC don't sound good. Afterall it's full of transistors. The devices are very precision indeed. Much more accurate in component tolerances than any discreet circuit can be. And at a fraction of the cost. But here comes the snake oil. One is often in a shiny metal can, the other in a black plastic mold and looks very boring and just like a hundred other chips. So, they cannot be as good, can they?
@@MichaelBeeny For my opinion opamp is "worse" for two reasons. First, because it's a differential amplifier. DA affects a dynamic (however perhaps not all of them, as an internal structure of all those current mirrors can affect). Secondly, because it's a " devices are very precision indeed". That means that a stability is more important than a dynamics. If we measure a sine wave, we see nothing (because by ear we hear a difference between 5532 and 797, so it's about a processing of a complex wave). Third, like a bonus, most of opamps were designed for a measurement, not for a sound, when an input bias means more than a transient characteristics.
I'm pretty confident that a linearity is not the most important for a sound (well, for a sound people like).
Interesting.
What about midrange or Mid bass or bass? not only treble that is important.
5k 3k 1k 800hz 500hz 300hz 100hz 80hz 50hz 20hz
CMRR 90 dB min, Open Loop Gain 93 dB min, Voltage Noise 1.9 nV @ 40Hz, I would only use it with low gain (10 max), I haven't found where the 10 pF problem is hidden in the data sheet. At 10nF and 20KHz, the permissible output current is only used to about 10%. All my opinion😉
That's how it goes. We old farts we have the knowledge and wisdom, but the hearing accuracy is long gone. The younger may have golden ears (I doubt though) but not so much experience and hands-on for the practical part of hi-fi. They grew up with plastic PC loudspeakers and 5.1 home theater. They don't need high slew rate for Bluetooth speakers.
Only measurements will show which is "best"........but whether the improvement in measurements,slew rate etc over the rival would be audible is a different matter.
So only listening will tell which is best
@@Pete.across.the.street No, it's the exact opposite. Humans are poor sensors and defining "best" with only the hearing experience as a metric only means not quality matters but everything that is subjective by nature, therefore opinion based. This is why measurements are the only way to make everyone objectively agree on facts. Listening will maybe tell you the one you *like* the most. But definitely not which one is "best".
@@nashaut7635 the one i like to listen to the most would be the best. The one I don't want to listen to would not be the best.
@@nashaut7635 With any audio project, I always start with, in some cases many weeks of listening tests. Only then does it go in the bench. Often the measurements confirm what I've heard but as circuits and ICs get better, the differences are also harder, if not imposable hear OR measure.
Could be wrong but I don’t believe that slew rate is a factor at audio frequencies.
It's not. Slew rate only matters when high frequencies are amplified to high voltages. A few volts per microsecond is more than fast enough, even in a thousand-watt (90VRMS at 8 ohms) power amp playing at 20kHz. For a preamp operating at only a couple of volts RMS, even less slew rate is necessary.
The slew rate of a sinewave (in volts per second) is (2π * Vp * f) where Vp is the peak voltage (not peak to peak or RMS) and f is the frequency.
For 90VRMS (1000 watts into an 8-ohm load), Vp is 127.3 (rounded to a tenth). At 20kHz, the slew rate is approximately 16 million volts per second, or 16V/uS. For 2VRMS (the output level of a CD player at 0dBFS), the slew rate is only 0.35V/uS. To avoid issues, the slew rate of a circuit should be at least a couple of times the slew rate of the signal being reproduced. The slew rate of a 4558 opamp (very common, very cheap audio opamp) is 2.2V/uS, which is six times the slew rate of a 20kHz signal at 2VRMS. That's plenty.
Yes, you're wrong.
@@MrDmjaysee above exhaustive explanation of why slew rate does not matter at audio frequencies.
I agree that it's generally not an issue (anymore). For example, the LM358: if I did the math right, it can't produce a 10 V peak to peak signal over 16 kHz. So that is a realistic flaw but there's no need to use that IC.
Rise time definitely has an effect at audio frequencies but he only refers to slew rate.
Hope this isn’t a bother but do you recommend any prebuilt amps on Ali? Or in general do you have any recommendations under 1,500
I'm not 100% sure I understand the question! Are you talking preamps, integrated amplifiers or just power amps? Is the 1,500 dollars US of pounds UK? you gotta help me lol
@@MichaelBeeny USD$ mainly looking for a power amp to hook straight into a streamer.
@@jakegibson6096 there are many options, a lot depends on your preference and the speakers you are pairing it with
I Like the AD797 myself.
Wasn't that the bees knees about twenty years ago? Never got to try it as single channel though.
I've seen dip8 adapters you could solder it to
They will add more capacitance, that's (one) of the issues. Think I will stick to the LME49720
What is the best OP-Chip for Audio Pre-amplifier? Thx.
The LME49721 is our favorite right now. The results are clearly shown in this video. It does need a capacitor mod, again shown in past videos, to avoid stability issues.
@@MichaelBeeny I ask for Audio not for Video.thx.
@@FDKAUSW hes talking about audio. hes saying he did a video on the best ones for audio.
@@Pete.across.the.street Hi Pete, you will be shocked, like I was to hear, not everyone watches all my videos!! and one or two don't watch them all the way though. I know that's hard to believe. I was devastated.
Video typically uses 75 ohm coax, so they must have some way of driving the capacitance...
I doubt this IC would be the final driver in any video circuit. My knowledge and experiences in video are almost zero.
When used as an output driver, the chip won't be connected directly to the cable. It will have a 75 ohm resistor in series with it to match the cables characteristic impedance. Of course, one must use careful design and board layout to keep a chip like this happy. Do that, and it's a monster capable of pushing 200 mA.
Once you're past 40 your ears wont know the difference.
Why being so obsessed about slew rate?
What is the bandwidth of your ears? You're not the youngest any more, so if you’re lucky, you might still hear 15 or 16 kHz.
A large signal rise time of 1 µs results in a large signal bandwidth 350 kHz!
Your oscilloscope already shows a slightly better rise time for the NE5532.
So the large signal bandwidth of the NE5532, being >350 kHz, is already miles beyond the bandwidth of your ears. So why bother for more slew rate and bandwidth?
Even if the rise time is only 5 µs, the bandwidth is still 70 kHz, way better than your ears.
What about accuracy?
What would be the accuracy of all the equipment between a musical instrument and your ears?
That includes: microphones, equipment for recording, editing, mixing, ADCs for CD/DVD, your audio chain with DACs and speakers and the acoustical properties of your room.
Any accuracy, if ever present, will be long gone in this long chain.
Adjust the controls of your audio chain as you like it and enjoy the music!
Op-amps have a very high gain, that is reduced by feedback from end to start in the application. Many designers of amplifiers therefore don't like op-amps and insist on discrete transistors, where each stage can have its own feedback loop (emitter resistor) which is reacting immediately on this stage, not delayed over many stages. These designers avoid an over-all feedback loop from output back to input, or at least use only a small part of the total feedback over the long way.
I never designed audio amplifiers, but I can understand the motivation to avoid or at least reduce over-all feedback. However - if the op-amp is extremely fast compared to the frequency band needed - that critic can be forgotten. The delay from output back to input is negligible.
your ears can hear a phase difference of microseconds. At 5kHz, 5 degrees of phase shift is 2.8us. So yeah you can't hear 20kHz, but you can hear the phase shift of an amp that rolls off at 20kHz. Or a speaker for that matter.
Whats with this obsession with slew rate?
Listening to ultrasonics? 10-20v/us is plenty enough for anything audio.
Capacitance doesn't matter as long as you push enough current.
Parallel is the answer
Kind off, experiments are under way. Not as simple as JUST putting 2 in parallel. Tests are underway. We shall see. Coming soon.
Why can't you just solder it onto a IC socket the socket will plug into another socket it's just too small?
I think the LM358 has never been used in any commercial piece of audio equipment. Regarding slew rate. A 20 V 20 kHZ sine wave would require a slew rate of only 2.5 V/µs so I fail to see why you would need over 9 V/µs for audio applications. For a 20 V sine wave, the NE5532 would work fine up to 70 kHZ until slew rate would start to be a problem. Audio processing in the analog domain is still done with TL072, NE5532, RC4560, RC4580... nothing exotic at all. The signal from a microphone in an old analog mixing console would pass trough maybe 50 op amps (the ones everyone seems to hate because they seem to sound bad) before it reaches the output of a mixing console. Even your most beloved audiophile recordings mixed on some world famous analog mixing consoles are all equipped with ordinary op amps and many of them. Schematics of those famous mixing consoles are out there. The trick in those consoles is not in the op amps but the reference structure of the mixing console. This structure you will probably never find in home audio equipment not even the expensive units. (at least not to my knowledge)
I've seen the LM358 used in the audio stages of some Tait Electronics base station transmitter designs from the 1980s. Perhaps not the sort of audio that some people would be interested in, but it certainly was audio. I seem to remember these were quite sensitive to strong RF fields which sometimes meant the transmitter could leak back into the audio stages with some undesirable side effects. One quick fix for this was to replace them with something like a TL062. The LM358 was cheap, but not my favourite Op Amp.
This is audio, right? The enemy is not lack of high frequency, it is noise. We don't need vhf capability.
Kâğıt üzerinde opa 1612 cok iyi degerlere sahip,ama kör dinleme testinde muses02 ile yarışamaz, ayrica opa2134 vee lm4562 opa1612 ye göre kör testte kulağa daha iyi gelen sesi verir! Kör test nihai objektif dinleme testidir yalan söylemez ve muses02 yi ancak akliam lc5 geçer ve a sınıf çalışan bir op amp tir ve cok ısınır! Opa 1656 denemedim ama lm4562 seviyesinde olacagini tahmin ediyorum
Hi Michael. While I like some of your reviews of the various cheap audio kits available, I believe your assertions that high slew rates are required for good audio amplifiers is non sensical. The required response range for "hi-fi" is generally regarded as 20 Hz to 20 kHz. Human voice, musical instruments, microphones and loudspeakers don't respond much outside this range and certainly human hearing covers a smaller range which further decreases quickly with age. Maximum slew rate is a simple function of the peak voltage and the frequency being reproduced. Maximum slew rate i.e. rate of change of a sine wave occurs at the zero crossing and can be found from a straight forward differential equation: ∂/∂t [Vpeak * sin(w*t) where w = 2*π*f. Do the maths and you will see what slew rates are actually required which can be easily verified with a signal generator and an oscilloscope. For example, the output stage of a power amp specified to provide 100 W into 8 Ω (which corresponds to about 40 V peak) needs a slew rate of: 2*π*20000*40 = 5.02 V/µs to reproduce 20 kHz. ICs that might be used in preamplifier stages only ever need to produce output voltages that peak in the 100s of millivolts otherwise the power amp will be overdriven so have a maximum slew rate requirement of less than 1V/µs to be able to full reproduce signals in excess of 20 kHz. One of the first lessons I learnt as an electrical engineering undergraduate was do not design anything faster than it needs to be. An amplifier that is faster than necessary requires extra design effort to guarantee stability and minimise noise. If you study the work of some of the reputable practitioners of good audio design such as John Linsley-Hood, Douglas Self, Peter Walker etc you will see a realistic approach taken to design requirements.
You stick to pure theory, but when you hear the difference in sound clearly you have missed something. Many amplifiers measure the same but sound way different, sine-wave measurements tell nothing as it is the easiest form of signal to reproduce up to the levels of 0.001 THD and below, you forget about phase delay of low frequences relative to high frequences, settling time also in some circuits it can be included in a feedback loop and have an effect on damping the drivers, an increased delay in this case will be detrimental but will not be apparent on pure sine-wave as it will stay a pure sine-wave. Probably this chip has also other parameters giving him an edge over other audio op-amps.
My current discrete build preamp has a slewrate of 200Vu/s. Bandwidth is way beyond human hearing, yet it sounds so good and natural. Cutting the bandwidth would essentially reduce the slewrate. Since the preamp runs happily without problems, I'll keep my +1MHz range and enjoying the 20-20000hz music it produces with the enormous headroom that the slewrate provide.
@@kenielsen I guess like driving a Ferrari in a school zone.
@@glasslinger why? You can actually hear this even at low level music. Many DIY'ers (and often expensive consumer amplifiers) aim for a high slewrate, it's just a matter of design and parts selection. Often no extra cost if done right.
@@kenielsen The comment left out the :) I was making a funny, having all that power and not needing it. I'm 79 so my listening days are muted. I can use a 358!
Gimmi an old 12ax7
Well, if that's the sound you like, go for it!
He doesn’t like it
👍❤️🫂
You really don't seem to understand preamp design using opamps. It is NEVER good practice to use directly the output of an opamp to an output. NEVER EVER. Typical is the use of 100R from output of the opamp to the output. If you don't understand why, I can be bothered to explain it, it has something to do with the term "infinity" which is impossible..
You clearly have a fundamental misunderstanding of how opamps work if you think that dropping them into an otherwise identical circuit is going to give a fair comparison of how well each opamp can work. I suggest that anyone who's serious about getting the most out of actual audio opamps consults the works of a Mr Doug Self.
You clearly have not watched my video with care. In fact, I have stated you CANNOT do this. In this video alone I've said if you swap the 5532 with a LME49720 you need extra caps across the supply pins 4 & 8 to keep it stable. This whole video is about that very fact that in MANY chases you cannot just swap them over. Many people do just that. I'm fully aware of Doug's bible, I have a copy of a number of his books right here. You will note, I only swapped 3 explaining the need for extra decoupling. The OPA2891 was the main subject of the video, explaining you CANNOT JUST DROP IT IN. It looks like you watched 3 mins of my video and started typing!
Maalesef türkçe altyazi yok michael en ilgilendiğim sevdigim konular saf ses üzerine bu op amp ses için en iyi değerlere sahip ama pratikte iyi ses vermiyormu anlamadim?!
I'm so sorry, but subtitles are 100% out of my control. TH-cam just do it (sometimes) I have written to them about this. I hope I receive a reply soon, or better still, they fix it.
Regarding the IC. It's not really practical to build for audio. Really just a talking point. Should it be possible to build a working circuit, I don't doubt it WILL sound good.
Just chucking my two cents here: I use a 9U 104hp eurorack synthesizer which has a section of video modules and modules that use preamps that are specifically compatible with video rate signals as well as audio. And for either my audio or my control signals, running them through one of those modules enhances detail, and the breadth of nuances that change with a knob turn connected to one of those preamps has much more detail in a smaller section of a knob turn than what comes from audio preamps.
I should stick to hem on the scope and sniff around, but just going by ear and a surface trace of things using them, I 100% agree that they introduce incredible detail because they’re over spec