Hey Dave, thanks for taking the time to make the video, I don't think I have every heard anyone talk that long on the subject of comparators, however, I think you well covered all the bases. The LM111/211/311 was the first monolithic low cost IC comparator and thus, had all the optional "tweak" pins, just like the uA709; the first Op Amp. I first began using the LM2903 in 1975 and have used it in numerous designs since then. The PNP input allows it to sense down to ground (as you mentioned). The broad voltage range and open collector output is nice in that the output can be taken above the Vcc, so long as you don't exceed the process voltage (about 36 volts). That component has always had the problem where the output capacitance couples into the positive input, causing it to be unstable if you are comparing millivolt changes. I consider the definition of a "Jelly Bean" as a part that has been around for at least 20 years and at least 5 manufacturers make it. I am partial to Jelly Beans as, most of the circuitry I designed over the years had a manufacturing lifetime of 20 years. You could probably say I am gun shy about using the latest super high speed/low voltage/low power part for that reason. The only thing I had a question about with your talk, was the "Jewel Comparator". Being from the US, that's a new one on me. Otherwise, keep up the good work Dave!
The LM339 datasheet is off by a few years. I was using LM339s in a project at Motorola in 1975-6, judging by the patent application that had my name on it, rather than 1979. The LM339 did easily put a substantial current spike on the 5V and ground rails which caused many problems till the, hand taped, board 5V and ground traces were cleaned up and bypass caps moved closer to the LM339s. Seems so obvious now.
THANK YOU THANK YOU THANK YOU!!! Please continue this series!!! Any moron on TH-cam can do a crappy teardown video, yes you are NOT a moron and you give good comments and critique when doing teardowns, but we need you more for full-on TEACHING us electronics!!!
I still have bins of ancient late-1980s to early-1990s jellybean parts purchased during the end of the dotcom bubble. The main spec differences compared to recent production are the ESD tolerance and power consumption. Meaning they also sometimes become "silicon fuses".
Enjoying the Jellybean series. At the beginning of each series, I try to guess the 5 parts to be mentioned. For this one I guessed LM339, LM311, LM319 and LM393 and a question mark. I've probably seen the LM339 used more than any other chip (except the 7805), it is used in a such wide variety of applications. One not mentioned that I have seen used fairly often is the LM319 dual 'fast' comparator. So many of those early National Semi parts became industry standards, everyone cloned them. I have 339 chips here from at least five different manufacturers. I was glad you finally mentioned 'input includes ground', that has become a more important spec for me in recent work. Kudos for another well done video.
Very useful information. Can you do a TOP 5 Jellybean A/D converters. Yes, lots of modern processors have built-in (slow) A/D converters, can you go over some good quality (fast) jellybean discreet A/D converters
26:53 "Common Road Mange" haha... This video comes in handy, as I have a circuit in mind, and was thinking I'd just use op-amps to do comparator work. Now it seems extra silly since I have some boards coming that call for an LM393 anyway, and I can just increase my order size and save a millipenny.
Enjoying these, cheers Dave. It's like picking up old basic electronic book and reading because it's easy to follow but always still relearn something forgotten or learn something new that didn't see before 👍
"Common-mode range includes ground" is a MUST-HAVE for all newbies. I also favor, whenever possible, a totem-pole output drive. It's symmetrical and faster (yes, you can't do logic hardware ORing obviously). It's sad they don't give you split switch high and switch low pins, that would allow you to connect the two to get totem pole, or omit one to get open collector, etc.
Some gate driver ICs give this option, and some of those have comparator inputs. They also tend to have fairly high current drive/capacitive load capability. They might have some other drawbacks that make them unsuitable for this role, but I don’t know what they are.
Biggest drawback of using one of a multipack opamp as comparator is that they all share the same bias supply, and thus you can have the comparator change the others when it saturates internally, either by causing a big offset drift depending on state, or actually running the other outputs to a rail when the bias saturates.
I v been in an opposite situation: the speed and response of comparator was gaining the input noise very well. Switching to slow opamp device resolved the majority of noise problems without applying any filter schematics.
Since you asked on the live stream, I thought this jelly bean thing would have been way more interesting if there was actual breadboarded experiments comparing them against OP Amps to help illustrate the pros and cons
Hi Dave, I always struggle when picking parts, I want to choose jelly beans whenever possible but being inexperienced it's hard to know which parts are actually jelly bean... And I've made some mistakes in the past... This series is very useful but I think it would be absolutely amazing if there was a reference card somewhere like: "jellybeans4idiots" which lists jellybeans by category and use case with the basic major specs you can expect from the jellybean. Does such a reference exist somewhere?
Great overview. While you mentioned that overload recovery on OpAmps is often dead slow, you could also have mentioned, that for a lot of OpAmps, the time of positive and negative overload recovery is vastly differen, whereas comparators try to have approximately the same propagation delay for positive and negative edges. There are a couple of application notes by different manufacturers, like TI's SLOA067 going into furher details. You could have explicitly pointed out that the "V" in a model number is very often an indicator for a "low-voltage version" of a chip. A good point in time for that remark would be 16:00. While there is no "V" in the ACMP391, you get that hint in the TLV1391 and the LMV7219 directly below it. If you are searching for a jelly bean part substitute, the "V" should be an alert to make extra sure to check the supply voltage rail. A consequence of "V" meaning low-voltage, the parts with "V" are likely more modern and have less current consumption even if operated at the same supply voltage and possibly less bias current. This means on the other hand, if you know you don't care about high voltage capabiliy, and power consumption is slightly important to you, a part with the "V" might be worth a second look, even if you didn't hear about yet. Be careful, though, that the TLV431 has a different reference voltage than the TL431, though (which indeed makes sense on a voltage reference meant for "low voltage" circuits). So, basically LM393 + LM358 + TL431 = TSM102? Putting the specs of the classic jelly beans next to the combo part component specs and comparing some key properties like offset voltage, current drive capability and propagation delay could also add some insightful content to the video. They don't specify a maximum sink current, because that would mean: "We guarantee that this part, under any condition, is never able to sink more than xx mA". Why would you want to know that? More importantly: Why would you want vendors to spend money on determining and controlling that value? They *do* specify the maximum current you can rely on: It's the *minimum* guaranteed sink current. They also specify the ballpark of what current the part will reach without external current limiting as "typcial" value. Good enough for most purposes, too.
Tom Legrady, whites goods as in appliances which traditionally are white. Also called hardlines as they are typically not soft on the outside. Marketing people are uhm different
These are bloody excellent, keep em coming, in these days of component drought, it is key to design in jelly bean stuff and go discrete whenever possible. May I suggest, BJT/Mosfets/Diodes for the next edition? Perhaps simple solutions where discrete is an easy way to remove those regular monolithic stuff?
One trap with comparators: their response time comes with an 'overdrive voltage' condition.. usually 100mV. They respond quickly to relatively large voltage steps, but aren't as fast when it comes to slow ramps. That's why Bob Pease would add a few picofarads of positive feedback from the output to the positive input. As soon as the output starts to change, the comparator's own output stage generates the overdrive necessary for a fast transition.
I had a friend that used thw NE555 as a comparator/scmitt for microcontroller inputs. I think it was mainly because he bought a load back in the 80s and still had most of them!
But they are really good for that, with adjustable threshold as well and built in reference resistors, plus it only needs a 100uF capacitor right by pins 1 and 8 to decouple it, which the rest of the board also uses.
@@SeanBZA Actually, it is not a good solution. The NE555 generates a large current peak during switching of the output which causes a lot of noise on power supply.
@@petrpolasek5931 I know well about that pulse, so that is why I suggest a 100uF capacitor right by the IC, to decouple it. But only one pulse per edge, and it will not oscillate at all no matter how slow the edge is. Plus you can use the exact same IC as both a totem pole output driver, and also as an open collector driver, with 200mA source and sink capability as a bonus for free. Yes need bypass but that can also act as bulk capacitance for the entire board as well, right by each 555 IC.
Back in the day I designed a lot of high speed test equipment, go to comparator was the AM687... dual part, differential ECL output and differential latch control (that was kind of cool, it would latch the outputs in their current state), 8 nS max propagation delay, and a hefty 40 mA or so power drain. Good times.
Seems to me, if one wanted the "strobe" functionality, and only used 1 comparator on a 2-comparator package, the other comparator could be used to create the strobe pin. Signal on the + pin, "strobe" on the - pin, and use the output to drive the other comparator. (Unless I have that backwards. 🤣) Love the jelly bean series! 👍️
Definitely not jellybean, but I've been swimming in MAX9012's lately. Dual 5ns precision comparators. Not rail to rail, but close enough for my application :)
ST does not specify a maximum sink current, because the condition is an output voltage of 1.5V, so it makes more sense to guarantee a minimum current that will not increase the output voltage above 1.5V.
9:10 EXACTLY what I need to do. Drive a relay coil whenever a 12V PWM signal is present. So far i have Low pass filter to drive a transistor. Thinking comparator. I just need to figure out what values for resistor, cap, and which comparator. Vehicle system. Need to trigger a relay when there is signal for the dash lights to be on. A scope would help, but I just need to make it work at the moment.
Questions from a n00b: 1.) LM311 06:59 It's not tied to a ground terminal... But how does the NPN-transistor in the schematic work then? There should be base current flowing through the emitter, shouldn't there? If the emitter is floating - how can there be a base current??? 2.) TS391 14:14: The output drive capablity is typical 16mA, min. 6ma. Dave said that he doesn't get, why there isn't a max value... Why would there be any? I thought that this means that in the worst case, you could only drive 6ma, in the typical case 16ma... Why should I need a max value? 14:30 "The other biggie is the speed of the OpAmp". Why OpAmp (I assume a mistake like later in the vid). 3.) LMV331 24:40 You put a circle around an important fact in the datasheet - but we can only see your face :-D Besides of this: Thanks a lot for the vid!
The 3peak version of the LMV331 has improved specs compared to the regular ones. Sometimes take must be taken in designing according to 3peak specs and then using any other version since their version is rail to rail, has internal hysteresis and many others. Since they are high end chinese the prices are set accordingly, some jellybean grade OP-amps are the same or more expensive than western brands.
For low speed stuff where timing isn't critical I usually just use an op amp if available in the design. But a comparator is technically a better choice in a fair few situations, and cost usually isn't any different. However, the bill of materials can be larger so that is a hidden cost.
You didn't discuss the most controversial topic of which comparator pin-out is most jellybean or bestest! Also are you brave enough to admit to us how many bags of unused comparators you have in a drawer somewhere because you ordered open-collector instead of push-pull and vice versa or the wrong pinout? I may have a 'small number' of such bags...
I have a point in my life that I was very disappointed in myself after assembling a project, only for it to not work at all despite my efforts, when I was in my teens. I only found years later that I was not aware I needed a pull-up resistor...
When do we get the Jellybean resistor video? We need to know the standard industry picks for resistors! lol. Jellybean BJTs, FETs, and maybe even JFETs may be in order. I tend to think of 2N3904, 2N3906, 2N2222, 2N2907, 2N4401, 2N7000, 2N3055, 2N2955, for BJTs. (and that sneaky MOSFET in the that list). For medium power N channel enhancement mode MOSFETs I know that the IRF540, IRF640, IRFZ44N are pretty common. For high power ones, I've taken a liking to the HY1920P. It USED to be under $1 on LCSC and rated 200V 90A, as an "improved" IRFP260N, which is a jellybean part.
"Applications: vacuum robot". I love when the list of applications are absurdly specific. 😂 I only buy components that mention my own name "Applications: LoraWAN test boards made by Stefans or Patricks in Europe or Asian countries"
They don't teach this stuff at (my) uni. The only part number we've seen is the '741. No mention that comparators even exist aside from op-amps. After three years mind you.
One note on the LCSC topic: There maybe are some no name Asian brands with high speed capability. The problem here is that with „made in Asia“ brands LCSC does not put in all their infos to the website. So when you put in that 50ns filter there are all the comparators missing which have a blank field on that. (and maybe one of them fulfills the criteria)
Like these jelly bean series a lot. Not only to recognise them, but also learn how to read datasheets. Like the word “binning”. (English is not my first language) Came in really handy when I read a datasheet for an IL300, which is really something different. Just saying. I like topics around designing circuits / pcb’s a lot.
Just designed a headphones DC protection circuit with using comparators, that is able to be triggered with +-millivolts of 1-3-5-9-you-name Hz. This job is for a comparator only ;)
How did the electronic engineers solve connecting a relatively slow comparator or even an OpAmp configured to use as input for a logical input that has a fast transition time when sufficiently fast comparators were not invented yet? Did they use a Schmitt-trigger, or perhaps another solution?
I couldn't help but think that a TL391 or an older LM393, with external compensation, could replace the opamp in a lowish-noise voltage regulator. Noise should be comparable to a 4558 opamp, but with higher order compensation you may well get a higher bandwidth out of it. Whether it's worth it if you can even get an MC33078 for less than the 393 is another matter though...
It’s very good idea with the jellybean series. Could you create another video about integrated current mirrors, current sinks and current sources? Thanks.
Hey Dave, I tell my students to save trees, and besides always printing double-sided, to NEVER print out datasheet's first and last pages. The last page usually has all the legalese. The first page has all the marketing bullcrap that is usually contradicted by plots and tables later on in the datasheet. I then explain to my students such wondorous first-page jems such as claims for MOSFETs of 1 milliohm ON-channel resistance (at 25C junction temperature, VDS of 1V, and IDS of 1mA) (this is for a D2PAK sized device)! Best MOSFET in the world!!! Typical applications: robotics, blah blah blah. What do you think? Do you know who and how writes the listing of "applications" that are always laughably present on first page of datasheets???
"I tell my students to save trees" I remember when I was studying Chem. Eng. in the 80s, my EE buddies were walking around carrying stacks of data books. Talk about wasting trees! The EE department would handout the data books for free. Mostly TI stuff.
I always presumed it was the appease the engineering managers in large companies who might question why a chip is used. I'd love to know who actually collates and writes them! Any decent design engineer isn't going to pay it any notice, but hey, psychology works in mysterious ways. I do know they put some in there to target markets they know they aren't hitting yet, or markets that actually requested the variant chip design.
@@EEVblog I really, really wish that this information would be real. Like they would say "this IC is now hopelessly obsolete. Please use our competitors' chips for modern designs. Much better ICs are available in this century". :-)))) nah, never gonna happen....
@@cidercreekranch I LOVE the application notes books!!! Back when they would give you these on paper, for free, I would get every one I could. I have most of these still in storage. Why do you think I pay $200/month for climate-controlled 8'x10'x7' worth of plastic totes with books and electronics storage?
Does anyone know if there is a database of different industry standard “jellybean” parts? These videos have been really helpful, and having some suggestions on all other parts not mentioned (yet) would be a life saver.
What you referred to as a "push-pull" output stage was not so. A true push-pull output has the N-channel FET on the high side, P-channel FET on the low side. The pictured amplifier was an inverting common source voltage amplifier, but a push-pull stage is a noninverting source follower current amplifier / buffer.
Hey Dave, have you ever handled the feeltech fy6900 signal generator? Seems like a decent device for the price. I am still in university and I am thinking of buying it because of the price.
That TSM102 feels like LM358 + LM393 + TL431 in one package. If some Asian company would make this chip, it would likely be a multi-die package with three dies in it.
Just a Heads Up Dave.... Compared to the latest BigClive video the audio of this video sounds quieter and more muffled (highs missing perhaps). Still watching it though.
I'm not convinced you've quite got the right idea of what "Output sink current" means on the datasheets. The maximum would be the current sunk when the output is shorted to the positive supply rail, so it's not very useful for use in driving a load. Look for it in the _Absolute Maximum Ratings_ section if it is going to be given: for example for the LM393, output current is given as 20mA max, and duration of short to ground is unlimited (although that's not helpful for an open-collector output). When you're looking in the _Electrical Characteristics_ section, you're seeing the amount of current that can be sunk before the output voltage rises above a certain value. As examples, the LM393 datasheet states you can sink at least (i.e. as a _minimum_ ) 6mA when the output has risen to 1.5V; while the TSM102 shows that a minimum of 6mA, but typically 16mA, can be sunk before the output rises above 1.5V. Nobody cares what the maximum is, because for practical purposes you need to know what the _minimum guaranteed_ current available is. So that's why they don't give maximum values of output sink current in the characteristics of the datasheets.
What sort of circuits are people using hardware comparators in? Whenever I find I might need one I'm in fact just doing the comparison in code on a microcontroller...
That's the fun part, you don't know. It is not rated for the temperature, anything could happen. It might work, it might not work, it might work correctly, it might do wrong stuff. If you use the part outside of the temperature range specified, nobody guarantees for anything. Lot of electronics you will need to heat for them to work in low temperatures.
By jellybean, i thought that they are widely popular, well known, tried and tested affordable parts, but these are specialized and difficult to substitute. Btw what does jellybean mean in terms of opamp or comparator?
Oh man that mouse movement...May I suggest you use some youtube video about mouse optimization? You'll invest 10 minutes and feel the difference for life!! I use the one from Panjno.
Really enjoying the “jellybean” series. Thanks Dave!
Yeah me too!
Yeah me three
Yes me too!
Would work with "none if that rubbish" too 😊
Hey Dave, thanks for taking the time to make the video, I don't think I have every heard anyone talk that long on the subject of comparators, however, I think you well covered all the bases. The LM111/211/311 was the first monolithic low cost IC comparator and thus, had all the optional "tweak" pins, just like the uA709; the first Op Amp.
I first began using the LM2903 in 1975 and have used it in numerous designs since then. The PNP input allows it to sense down to ground (as you mentioned). The broad voltage range and open collector output is nice in that the output can be taken above the Vcc, so long as you don't exceed the process voltage (about 36 volts). That component has always had the problem where the output capacitance couples into the positive input, causing it to be unstable if you are comparing millivolt changes.
I consider the definition of a "Jelly Bean" as a part that has been around for at least 20 years and at least 5 manufacturers make it. I am partial to Jelly Beans as, most of the circuitry I designed over the years had a manufacturing lifetime of 20 years. You could probably say I am gun shy about using the latest super high speed/low voltage/low power part for that reason.
The only thing I had a question about with your talk, was the "Jewel Comparator". Being from the US, that's a new one on me. Otherwise, keep up the good work Dave!
The LM339 datasheet is off by a few years. I was using LM339s in a project at Motorola in 1975-6, judging by the patent application that had my name on it, rather than 1979.
The LM339 did easily put a substantial current spike on the 5V and ground rails which caused many problems till the, hand taped, board 5V and ground traces were cleaned up and bypass caps moved closer to the LM339s. Seems so obvious now.
Datasheet production date, the part was in use before, but the datasheet was updated 1979 probably with extended data.
LM339 is neat. A very simple design and quite a bit faster than the contemporary jellybean op-amps.
Always good to add bypass caps. It doesn't hurt and often helps power rail noise.
THANK YOU THANK YOU THANK YOU!!! Please continue this series!!!
Any moron on TH-cam can do a crappy teardown video, yes you are NOT a moron and you give good comments and critique when doing teardowns, but we need you more for full-on TEACHING us electronics!!!
I still have bins of ancient late-1980s to early-1990s jellybean parts purchased during the end of the dotcom bubble. The main spec differences compared to recent production are the ESD tolerance and power consumption. Meaning they also sometimes become "silicon fuses".
Actually the older devices on larger process are more ESD resistant, though often it was not actually specced. All silicon can act as a fuse.
Enjoying the Jellybean series. At the beginning of each series, I try to guess the 5 parts to be mentioned. For this one I guessed LM339, LM311, LM319 and LM393 and a question mark. I've probably seen the LM339 used more than any other chip (except the 7805), it is used in a such wide variety of applications. One not mentioned that I have seen used fairly often is the LM319 dual 'fast' comparator. So many of those early National Semi parts became industry standards, everyone cloned them. I have 339 chips here from at least five different manufacturers. I was glad you finally mentioned 'input includes ground', that has become a more important spec for me in recent work. Kudos for another well done video.
Very useful information.
Can you do a TOP 5 Jellybean A/D converters. Yes, lots of modern processors have built-in (slow) A/D converters, can you go over some good quality (fast) jellybean discreet A/D converters
Great vid. Next in the jellybean series you should do transistors, both BJT and FETs.
26:53 "Common Road Mange" haha...
This video comes in handy, as I have a circuit in mind, and was thinking I'd just use op-amps to do comparator work. Now it seems extra silly since I have some boards coming that call for an LM393 anyway, and I can just increase my order size and save a millipenny.
Enjoying these, cheers Dave.
It's like picking up old basic electronic book and reading because it's easy to follow but always still relearn something forgotten or learn something new that didn't see before 👍
"Common-mode range includes ground" is a MUST-HAVE for all newbies.
I also favor, whenever possible, a totem-pole output drive. It's symmetrical and faster (yes, you can't do logic hardware ORing obviously).
It's sad they don't give you split switch high and switch low pins, that would allow you to connect the two to get totem pole, or omit one to get open collector, etc.
Some gate driver ICs give this option, and some of those have comparator inputs. They also tend to have fairly high current drive/capacitive load capability. They might have some other drawbacks that make them unsuitable for this role, but I don’t know what they are.
LM311 for President.....well earned.
Biggest drawback of using one of a multipack opamp as comparator is that they all share the same bias supply, and thus you can have the comparator change the others when it saturates internally, either by causing a big offset drift depending on state, or actually running the other outputs to a rail when the bias saturates.
I v been in an opposite situation: the speed and response of comparator was gaining the input noise very well. Switching to slow opamp device resolved the majority of noise problems without applying any filter schematics.
Opamp/comparator. You had an early morning Dave, lol.
You kept mixing them up. 😄
I realised that at some point :->
@@EEVblog btw, another funky IC that is actually made of of two jelly beans, is the LM392. Which is basically a LM193 and LM158 in one package.
Just fantastic Dave. Sometimes you just need something that does the job, nothing fancy. Your video spells that out perfectly
Since you asked on the live stream, I thought this jelly bean thing would have been way more interesting if there was actual breadboarded experiments comparing them against OP Amps to help illustrate the pros and cons
You're so awesome in providing tipps and tricks when it comes to electronics design. Greetings from Austria/Europe 😁👍
Hi Dave, I always struggle when picking parts, I want to choose jelly beans whenever possible but being inexperienced it's hard to know which parts are actually jelly bean... And I've made some mistakes in the past... This series is very useful but I think it would be absolutely amazing if there was a reference card somewhere like: "jellybeans4idiots" which lists jellybeans by category and use case with the basic major specs you can expect from the jellybean. Does such a reference exist somewhere?
The EEVblog drinking game, take a shot each time Dave says op-amp instead of comparator.
And a double for every micro ment to be milli. ;)
Blackbelt electronics knowledge just shines thru - keep on being you, SIr.
Great overview. While you mentioned that overload recovery on OpAmps is often dead slow, you could also have mentioned, that for a lot of OpAmps, the time of positive and negative overload recovery is vastly differen, whereas comparators try to have approximately the same propagation delay for positive and negative edges. There are a couple of application notes by different manufacturers, like TI's SLOA067 going into furher details.
You could have explicitly pointed out that the "V" in a model number is very often an indicator for a "low-voltage version" of a chip. A good point in time for that remark would be 16:00. While there is no "V" in the ACMP391, you get that hint in the TLV1391 and the LMV7219 directly below it. If you are searching for a jelly bean part substitute, the "V" should be an alert to make extra sure to check the supply voltage rail. A consequence of "V" meaning low-voltage, the parts with "V" are likely more modern and have less current consumption even if operated at the same supply voltage and possibly less bias current. This means on the other hand, if you know you don't care about high voltage capabiliy, and power consumption is slightly important to you, a part with the "V" might be worth a second look, even if you didn't hear about yet. Be careful, though, that the TLV431 has a different reference voltage than the TL431, though (which indeed makes sense on a voltage reference meant for "low voltage" circuits).
So, basically LM393 + LM358 + TL431 = TSM102? Putting the specs of the classic jelly beans next to the combo part component specs and comparing some key properties like offset voltage, current drive capability and propagation delay could also add some insightful content to the video.
They don't specify a maximum sink current, because that would mean: "We guarantee that this part, under any condition, is never able to sink more than xx mA". Why would you want to know that? More importantly: Why would you want vendors to spend money on determining and controlling that value? They *do* specify the maximum current you can rely on: It's the *minimum* guaranteed sink current. They also specify the ballpark of what current the part will reach without external current limiting as "typcial" value. Good enough for most purposes, too.
LM111 applications: "White goods" ... bed sheets? Pillow cases?
Tom Legrady, whites goods as in appliances which traditionally are white. Also called hardlines as they are typically not soft on the outside. Marketing people are uhm different
These are bloody excellent, keep em coming, in these days of component drought, it is key to design in jelly bean stuff and go discrete whenever possible. May I suggest, BJT/Mosfets/Diodes for the next edition? Perhaps simple solutions where discrete is an easy way to remove those regular monolithic stuff?
Really like the fact you explained the difference between op amps and comparators. Love the jelly bean videos.
One trap with comparators: their response time comes with an 'overdrive voltage' condition.. usually 100mV. They respond quickly to relatively large voltage steps, but aren't as fast when it comes to slow ramps.
That's why Bob Pease would add a few picofarads of positive feedback from the output to the positive input. As soon as the output starts to change, the comparator's own output stage generates the overdrive necessary for a fast transition.
I had a friend that used thw NE555 as a comparator/scmitt for microcontroller inputs. I think it was mainly because he bought a load back in the 80s and still had most of them!
But they are really good for that, with adjustable threshold as well and built in reference resistors, plus it only needs a 100uF capacitor right by pins 1 and 8 to decouple it, which the rest of the board also uses.
@@SeanBZA Actually, it is not a good solution. The NE555 generates a large current peak during switching of the output which causes a lot of noise on power supply.
@@petrpolasek5931 I know well about that pulse, so that is why I suggest a 100uF capacitor right by the IC, to decouple it. But only one pulse per edge, and it will not oscillate at all no matter how slow the edge is. Plus you can use the exact same IC as both a totem pole output driver, and also as an open collector driver, with 200mA source and sink capability as a bonus for free. Yes need bypass but that can also act as bulk capacitance for the entire board as well, right by each 555 IC.
Back in the day I designed a lot of high speed test equipment, go to comparator was the AM687... dual part, differential ECL output and differential latch control (that was kind of cool, it would latch the outputs in their current state), 8 nS max propagation delay, and a hefty 40 mA or so power drain. Good times.
I love this serial.
PS: Fun drinking game - drink any time Dave says op amp Instead of comparator.
Do you WANT to die of alcohol poisoning?!??
Seems to me, if one wanted the "strobe" functionality, and only used 1 comparator on a 2-comparator package, the other comparator could be used to create the strobe pin. Signal on the + pin, "strobe" on the - pin, and use the output to drive the other comparator. (Unless I have that backwards. 🤣) Love the jelly bean series! 👍️
the propagation delay through the cascaded parts may be the problem with that approach.
Dave always gives us a wealth of useful Australian idioms. "As slow as a wet week!". Classic! 🇦🇺👍👍👍😂
Nothing wrong with the 1970's the music was great:-)
The muff was too hairy though.
Appreciating a lot of it, though I'm more into underground '80s :)
@@KeritechElectronics It maybe should have stayed underground😁
Definitely not jellybean, but I've been swimming in MAX9012's lately. Dual 5ns precision comparators. Not rail to rail, but close enough for my application :)
Comparatively speaking another interesting video Dave.
ST does not specify a maximum sink current, because the condition is an output voltage of 1.5V, so it makes more sense to guarantee a minimum current that will not increase the output voltage above 1.5V.
9:10 EXACTLY what I need to do.
Drive a relay coil whenever a 12V PWM signal is present. So far i have Low pass filter to drive a transistor. Thinking comparator. I just need to figure out what values for resistor, cap, and which comparator.
Vehicle system. Need to trigger a relay when there is signal for the dash lights to be on.
A scope would help, but I just need to make it work at the moment.
Questions from a n00b:
1.) LM311
06:59 It's not tied to a ground terminal...
But how does the NPN-transistor in the schematic work then?
There should be base current flowing through the emitter, shouldn't there?
If the emitter is floating - how can there be a base current???
2.) TS391
14:14: The output drive capablity is typical 16mA, min. 6ma.
Dave said that he doesn't get, why there isn't a max value...
Why would there be any?
I thought that this means that in the worst case, you could only drive 6ma, in the typical case 16ma... Why should I need a max value?
14:30 "The other biggie is the speed of the OpAmp". Why OpAmp (I assume a mistake like later in the vid).
3.) LMV331
24:40 You put a circle around an important fact in the datasheet - but we can only see your face :-D
Besides of this:
Thanks a lot for the vid!
The 3peak version of the LMV331 has improved specs compared to the regular ones. Sometimes take must be taken in designing according to 3peak specs and then using any other version since their version is rail to rail, has internal hysteresis and many others.
Since they are high end chinese the prices are set accordingly, some jellybean grade OP-amps are the same or more expensive than western brands.
For low speed stuff where timing isn't critical I usually just use an op amp if available in the design. But a comparator is technically a better choice in a fair few situations, and cost usually isn't any different. However, the bill of materials can be larger so that is a hidden cost.
Why would I want to compare Jellybeans? Oh...
You didn't discuss the most controversial topic of which comparator pin-out is most jellybean or bestest!
Also are you brave enough to admit to us how many bags of unused comparators you have in a drawer somewhere because you ordered open-collector instead of push-pull and vice versa or the wrong pinout? I may have a 'small number' of such bags...
Oh come on, what's wrong with ground on pin 12? :-/
I have a point in my life that I was very disappointed in myself after assembling a project, only for it to not work at all despite my efforts, when I was in my teens. I only found years later that I was not aware I needed a pull-up resistor...
@@EEVblog nothing if it's a 24-pin package, LOL
Thanks for an explanation on the opamp vs comparator. At the end of the day, I hope I learned something. See ya on the livestream :)
I would be great if we get bjt/mosfet jellybean episode as well. Thanks for the content Dave.
When do we get the Jellybean resistor video? We need to know the standard industry picks for resistors! lol.
Jellybean BJTs, FETs, and maybe even JFETs may be in order. I tend to think of 2N3904, 2N3906, 2N2222, 2N2907, 2N4401, 2N7000, 2N3055, 2N2955, for BJTs. (and that sneaky MOSFET in the that list). For medium power N channel enhancement mode MOSFETs I know that the IRF540, IRF640, IRFZ44N are pretty common. For high power ones, I've taken a liking to the HY1920P. It USED to be under $1 on LCSC and rated 200V 90A, as an "improved" IRFP260N, which is a jellybean part.
"Applications: vacuum robot". I love when the list of applications are absurdly specific. 😂 I only buy components that mention my own name "Applications: LoraWAN test boards made by Stefans or Patricks in Europe or Asian countries"
Used loads of LM339 mostly on daisy wheel printers on the optical encoders. Also on feedback from the hammer detection on the faster machines.
Used for comparing Rolf Harris to Crocodile Dundee...... If Rolfs didgeridoo is bigger than Dundees knife, Output A = 1
Interesting as always Dave... makes we wish I had done some formal electronics training instead of just teaching myself.
Your level of "teaching yourself" is outstanding...
@@thomasw6169 thanks Thomas
They don't teach this stuff at (my) uni. The only part number we've seen is the '741. No mention that comparators even exist aside from op-amps. After three years mind you.
Too many electronic devices use op-amp, this Dave's video adds knowledge about some type of op-amp chips.
The only 'jellybean' I like, and now is insane cost.. is the LM3909 flasher.
jellybean BJTs / mosfets? love the series!
One note on the LCSC topic: There maybe are some no name Asian brands with high speed capability. The problem here is that with „made in Asia“ brands LCSC does not put in all their infos to the website. So when you put in that 50ns filter there are all the comparators missing which have a blank field on that. (and maybe one of them fulfills the criteria)
Slow as a wet week, great phrase! I'm stealing it lol
Like these jelly bean series a lot. Not only to recognise them, but also learn how to read datasheets. Like the word “binning”. (English is not my first language) Came in really handy when I read a datasheet for an IL300, which is really something different. Just saying. I like topics around designing circuits / pcb’s a lot.
2SC3356 is pretty popular as a 7ghz RF NPN. Many many OEMs making it.
Just designed a headphones DC protection circuit with using comparators, that is able to be triggered with +-millivolts of 1-3-5-9-you-name Hz. This job is for a comparator only ;)
How did the electronic engineers solve connecting a relatively slow comparator or even an OpAmp configured to use as input for a logical input that has a fast transition time when sufficiently fast comparators were not invented yet? Did they use a Schmitt-trigger, or perhaps another solution?
I couldn't help but think that a TL391 or an older LM393, with external compensation, could replace the opamp in a lowish-noise voltage regulator. Noise should be comparable to a 4558 opamp, but with higher order compensation you may well get a higher bandwidth out of it. Whether it's worth it if you can even get an MC33078 for less than the 393 is another matter though...
LM392 is jellybean opamp/comparator in one package from ancient history.
It’s very good idea with the jellybean series.
Could you create another video about integrated current mirrors, current sinks and current sources?
Thanks.
Hey Dave, I tell my students to save trees, and besides always printing double-sided, to NEVER print out datasheet's first and last pages. The last page usually has all the legalese. The first page has all the marketing bullcrap that is usually contradicted by plots and tables later on in the datasheet. I then explain to my students such wondorous first-page jems such as claims for MOSFETs of 1 milliohm ON-channel resistance (at 25C junction temperature, VDS of 1V, and IDS of 1mA) (this is for a D2PAK sized device)! Best MOSFET in the world!!! Typical applications: robotics, blah blah blah.
What do you think?
Do you know who and how writes the listing of "applications" that are always laughably present on first page of datasheets???
"I tell my students to save trees" I remember when I was studying Chem. Eng. in the 80s, my EE buddies were walking around carrying stacks of data books. Talk about wasting trees! The EE department would handout the data books for free. Mostly TI stuff.
I always presumed it was the appease the engineering managers in large companies who might question why a chip is used. I'd love to know who actually collates and writes them!
Any decent design engineer isn't going to pay it any notice, but hey, psychology works in mysterious ways.
I do know they put some in there to target markets they know they aren't hitting yet, or markets that actually requested the variant chip design.
@@EEVblog I really, really wish that this information would be real. Like they would say "this IC is now hopelessly obsolete. Please use our competitors' chips for modern designs. Much better ICs are available in this century". :-)))) nah, never gonna happen....
@@cidercreekranch I LOVE the application notes books!!! Back when they would give you these on paper, for free, I would get every one I could. I have most of these still in storage. Why do you think I pay $200/month for climate-controlled 8'x10'x7' worth of plastic totes with books and electronics storage?
...Dave keeps saying "op amp" instead of "comparator"...
I love how he opens this by galloping off on a tangent.
These tangents are not only interesting, but essential if you are designing (or aspiring ) as a professional...
I always used to use the LM3302 quad comparator
Thanks for another great video! Really enjoy your "jelly bean" and "fundamental" series...
OT: What's the PDF viewer you are using?
nice keep them coming!
Fascinating, thanks!
Dave, for the Op Amps, what would happen if you replaced the feedback resistor with a Zener Diode?
Does anyone know if there is a database of different industry standard “jellybean” parts? These videos have been really helpful, and having some suggestions on all other parts not mentioned (yet) would be a life saver.
Late reply, but The Art of Electronics 3rd Edition is an absolute must have. They have tables comparing the specs of jellybeans and more.
What you referred to as a "push-pull" output stage was not so. A true push-pull output has the N-channel FET on the high side, P-channel FET on the low side. The pictured amplifier was an inverting common source voltage amplifier, but a push-pull stage is a noninverting source follower current amplifier / buffer.
Hey Dave, have you ever handled the feeltech fy6900 signal generator? Seems like a decent device for the price. I am still in university and I am thinking of buying it because of the price.
It does the job for the price.
The PSU is pretty noisy but you can make your own or look up the SDG channel. He's made a really good one.
No. But as John said, I'm sure it does the job for the price.
That TSM102 feels like LM358 + LM393 + TL431 in one package. If some Asian company would make this chip, it would likely be a multi-die package with three dies in it.
Awesome! Thank you!
Do also a video of parts that you need to know (but that may be not meet the requirements for "jelly bean")....
How many do you want??? Have you ever parsed DigiKey???
Hah! I tried that. Use an op amp in place of a comparator. Looked liked it worked until it mattered.
you basically covered all the parts i use, i even still use the LT1016
the damn thing is $12 one of
Just a Heads Up Dave.... Compared to the latest BigClive video the audio of this video sounds quieter and more muffled (highs missing perhaps). Still watching it though.
Eagerly waiting for Chinese jelly beans
Let's compare actual jellybeans
I'm not convinced you've quite got the right idea of what "Output sink current" means on the datasheets. The maximum would be the current sunk when the output is shorted to the positive supply rail, so it's not very useful for use in driving a load. Look for it in the _Absolute Maximum Ratings_ section if it is going to be given: for example for the LM393, output current is given as 20mA max, and duration of short to ground is unlimited (although that's not helpful for an open-collector output).
When you're looking in the _Electrical Characteristics_ section, you're seeing the amount of current that can be sunk before the output voltage rises above a certain value. As examples, the LM393 datasheet states you can sink at least (i.e. as a _minimum_ ) 6mA when the output has risen to 1.5V; while the TSM102 shows that a minimum of 6mA, but typically 16mA, can be sunk before the output rises above 1.5V. Nobody cares what the maximum is, because for practical purposes you need to know what the _minimum guaranteed_ current available is. So that's why they don't give maximum values of output sink current in the characteristics of the datasheets.
What sort of circuits are people using hardware comparators in? Whenever I find I might need one I'm in fact just doing the comparison in code on a microcontroller...
Sorry, I'm new to this jellybean series. Is it a brand or something else? What do you mean by that?
TLV3201 from texas instrument is way better than the ts3021. ts3021 is out of stock from a few sites.
Question for the 393: What actually happens if you use it below the rated temperature? (0 deg C) 😅
That's the fun part, you don't know. It is not rated for the temperature, anything could happen. It might work, it might not work, it might work correctly, it might do wrong stuff.
If you use the part outside of the temperature range specified, nobody guarantees for anything.
Lot of electronics you will need to heat for them to work in low temperatures.
Unrelated to video. When is the BM786 multimeter coming back in stock? I want one for a graduation gift.
By jellybean, i thought that they are widely popular, well known, tried and tested affordable parts, but these are specialized and difficult to substitute. Btw what does jellybean mean in terms of opamp or comparator?
Keep this jellybean series going :)
Hello, would like to complain, these jellybeans are a bit crunchy.
That can happen if you leave them in the sun for too long or wave solder them. Just soften them up with a high voltage sauce...
Haha, "But I grew up with the LM339!! "
Dave just assumes that i am a professional engineer that pays attention to the specs, instead of the filthy hobbyist I really am
Cries in impostor syndrome
I love this series but somehow your audio levels are quite low on this one
So what do I need to connect to the pins of that thing to actually _compare jellybeans_ , eh?
Oh man that mouse movement...May I suggest you use some youtube video about mouse optimization? You'll invest 10 minutes and feel the difference for life!! I use the one from Panjno.
where is the module you constructed??!!😇
Please review the LT1394
No LM319?
I did not even know there is a difference between an opamp and a comparator
It's basic knowledge. Not discouraging you, but just saying there is always plenty left to learn for all of us...
Yo've done much of it, but how about JB diodes! 1N4148/1N914, 1N4007, etc.
JB SCR & TRIACs maybe, but few if anyone, other light dimmer worshippers :)
16:50 1979 = oldschool. Im old 😢
Jellybean Transistors next video
Please
Tell about jellybean transistors
Nano-power search pulled up zero availability.