Love when you and Amir throw the kitchen sink to those little chipamps. It's very fun. A challenge for you... do something similar with around $500 in test equipment, that would be AWESOME. (doesn't have to be THAT accurate)
10:51 That bump at about 40kHz to 45kHz is the output filters inductor and capacitor reacting with the 8 ohm load. The default filter with two 10uH inductors and the two 1uF capacitors in series on the output would work better with a 4 ohm load and should not give you that hump. A better filter for BTL and PBTL modes is to use one single capacitor across the output after the inductors instead of the two in series that are ground tied in the middle. It's fewer parts, half the capacitance as the two series caps version _and_ its the same sound... . Some values for inductors and capacitors in BTL and PBTL mode that give a fairly flat response with -3dB down at ~45kHz (and no hump): @~2ohms: L1,L2 = 5uH, C1 = ~1200nF @~4ohms: L1,L2 = 10uH, C1 = ~600nF @~8ohms: L1,L2 = 20uH, C1 = ~300nF @~16ohms: L1, L2 = 40uH, C1 = ~150nF With those impedances and those component values the plot should be the same all the way up to 45kHz ;)
This seems to be a massively overlooked issue with these TI chip amps. Manufacturers of these boards could in theory build in a switch or even automatic detection.. or just sell bespoke boards for your application.
Thanks so much for this great feedback!! As the speaker impedance is not flat - I wonder which LC combo would be a good fit to cover a swinging impedance? But also to remember is this happens outside the hearing range.
Good Bodes 👍. Would have been good to see a full sweep for distortion across the whole audio band and at a mumber of different amplitudes instead of just 1khz, and some intermodulation tests too. I have seen amps that perform the same at 1khz be very different above 10Khz.
@@KissAnalog Hi Eddie, yeah I saw the bode plot out to 1MHz, but I was referring to doing a full distortion plot, at least 20Hz to 20KHz (by that i nean sweeping the test signal 20hz to 20khz) both close to full power and at 1 or 2 watts, because in previous videos on the class D amp you had only measured distortion at 1khz on your Keighly meter (from memory it was a while ago). I suppose I should have made the comment on the other video really apologies for causing confusion. I was just wanting to emphasise that when making comparisons on subjects like class D vs A/B or powering an amplifier with linear vs switching power supplies, it would be better to have other measurements as well as bode plot, and distortion at 1KHz in order to arrive at a more detailed conclusion regarding the performance of each. When the new switching powesupply is done it would be cool to see some really in depth measurements done with the JAT amp and to compare it with how well it does with a linear powesupply. Maybe also include intermodulation distortion etc etc so we can see the whole story Cheers 👍
Hi! When i watch the other videos about noise levels for power supply. I don't know if it is working as a class D amplifier. There is a big discussion about that class D "looks"/perform nice if you looking at 20Hz-20khz range. But the switching noise the designer moving out of that range and put it much higher up in the frequency range! (For example 100kHz) The questions are: * When looking at noise should we not also look at a more generic broader frequency range than only the specific hearing range? (There might be amplifiers or (super-)tweeters that play beyond 20kHz that might be affected. So the conclusion COULD then be that linear power supplies has a lower noise in total than switching power supplies has when looking at a broader frequency range!🤔 Thanks for your videos! I like to hear your take or measurements on this?!🥰
Look at Burno Putezy work on noise in class d if you want answers. But to make a long story short no classd is silent to your ear the residual noise is negligible as it's 100mv switching voltage to force the loop to perform self oscillation.
Great questions! As the graphs show, this class D is flat across the hearing range and beyond in both magnitude and phase. it can be argued that switching power supply noise is outside the hearing range. This has actually been a work in effort since they came out, and MOSFETs have been a huge help so that they can switch faster - well beyond the hearing range. Linear power supplies are actually low frequency switching power supplies - unless they are regulated - linearly. I compare power supply noise between actual linear and switching power supplies. th-cam.com/video/kPHduHM7V1k/w-d-xo.html
if you really wanted to show what that Bode 100 does you should have probed the circuit where the class d integrator loop is located, you could then perform analysis of each loop in the each control loop and figure out what the bandwidth and loop gain looks like which is far more interesting and of educational value as well which does justice to the bode 100.
Thanks for the feedback! The TPA3255 is an integrated chip so this was video to show the performance of this amp. I'll show the Omicron in videos to come.
@@KissAnalog honey I know that, I'm saying the loop gain aspect and loop frequency response is of measurement interest. I do wonder what an audio precision have over a bode 100.
I stumbled across your videos early today while researching how to power a little project amp. Super informative videos, Can't wait to see how the amp project comes together. Quick question, the little amp I'm playing with requires positive, negative, and a ground. I have been told that it is for battery only but Id like to use it with a power supply. The hurdle is its operational voltage is 58-70 volts. If I use a traditional power supply like a 60volt like you have linked, and I do the resistor/capacitor bit to create the virtual ground I reduce the voltage by half and it is no longer the correct operational voltage. Any suggestions to help me navigate this rabbit hole? its an irs1092 board.
@@KissAnalog I appreciate the response! It's been quite a while, I did figure it out, but I don't exactly remember what the solution was and why, but It was something like it needed a special power supply that had 2 different types of DC power, almost like AC Hot and neutral kinda, there was a required 3rd leg that was weird, but they do make a power supply like that it was quite expensive and it turned out cheaper to go a different amp board.
Thanks for the great question! I will start a series that I will design my own Class D, but I've also shown a discreet Class D in the past. th-cam.com/video/97p6BkAMTik/w-d-xo.html
@@KissAnalog was wondering if i gen get a free one of those if i make them a couple of thousand videos starting with the philosophical meaning of logarithms
![POLYCAT - ข้อความรูปยิ้ม [Music Video]](http://i.ytimg.com/vi/lBhIgPGCgbI/mqdefault.jpg)
Love when you and Amir throw the kitchen sink to those little chipamps. It's very fun.
A challenge for you... do something similar with around $500 in test equipment, that would be AWESOME. (doesn't have to be THAT accurate)
Thank you - I'll see what I can do;)
That's a pretty fancy VNA
Yes it is - thank you;)
10:51 That bump at about 40kHz to 45kHz is the output filters inductor and capacitor reacting with the 8 ohm load. The default filter with two 10uH inductors and the two 1uF capacitors in series on the output would work better with a 4 ohm load and should not give you that hump. A better filter for BTL and PBTL modes is to use one single capacitor across the output after the inductors instead of the two in series that are ground tied in the middle. It's fewer parts, half the capacitance as the two series caps version _and_ its the same sound...
.
Some values for inductors and capacitors in BTL and PBTL mode that give a fairly flat response with -3dB down at ~45kHz (and no hump):
@~2ohms: L1,L2 = 5uH, C1 = ~1200nF
@~4ohms: L1,L2 = 10uH, C1 = ~600nF
@~8ohms: L1,L2 = 20uH, C1 = ~300nF
@~16ohms: L1, L2 = 40uH, C1 = ~150nF
With those impedances and those component values the plot should be the same all the way up to 45kHz ;)
This seems to be a massively overlooked issue with these TI chip amps.
Manufacturers of these boards could in theory build in a switch or even automatic detection.. or just sell bespoke boards for your application.
Totally agree that's the first thing I do when implementing a class D amplifier is modifying the the lowpass filter for the load
@@davidkclayton the impedances curve of speakers is also by no means flat. So using the headline figure could be a bit of a trap too 🤷♂️
@@FelixRisingOriginal most speakers impedance stabilized at the high-end (@ fc)
Thanks so much for this great feedback!! As the speaker impedance is not flat - I wonder which LC combo would be a good fit to cover a swinging impedance? But also to remember is this happens outside the hearing range.
Thanks for the video. I'm looking forward to your series on impedance measurement. :)
Thank you! That is coming very soon;) I’ll introduce a very important concept that might not be known to many;)
Everyone likes a very nice box
LOL I agree;)
You had me at box. 😉
Hi Eddie, I really appreciate you taking the time to share your knowledge and videos. You are amazing . Thank you.
Thanks - I appreciate your support!
Nice plot! also interest to know how we measure the output impedence of the amp at different frequency output
Thanks! I plan to show the output impedance measurement capability of this Omicron Lab unit very soon;)
Good Bodes 👍. Would have been good to see a full sweep for distortion across the whole audio band and at a mumber of different amplitudes instead of just 1khz, and some intermodulation tests too. I have seen amps that perform the same at 1khz be very different above 10Khz.
Thanks Ian for your feedback! Did you see where I scanned it out to 1MHz?
@@KissAnalog Hi Eddie, yeah I saw the bode plot out to 1MHz, but I was referring to doing a full distortion plot, at least 20Hz to 20KHz (by that i nean sweeping the test signal 20hz to 20khz) both close to full power and at 1 or 2 watts, because in previous videos on the class D amp you had only measured distortion at 1khz on your Keighly meter (from memory it was a while ago). I suppose I should have made the comment on the other video really apologies for causing confusion. I was just wanting to emphasise that when making comparisons on subjects like class D vs A/B or powering an amplifier with linear vs switching power supplies, it would be better to have other measurements as well as bode plot, and distortion at 1KHz in order to arrive at a more detailed conclusion regarding the performance of each. When the new switching powesupply is done it would be cool to see some really in depth measurements done with the JAT amp and to compare it with how well it does with a linear powesupply. Maybe also include intermodulation distortion etc etc so we can see the whole story
Cheers 👍
Only you Eduardo vould calm thr masses from the Omicron Lab Bode 100 strain. Hahaha jk. Great video as always on a really cool piece of gear.
Thanks Joey! LOL! It really is a cool piece of equipment. I will show much more soon;)
Hi! When i watch the other videos about noise levels for power supply.
I don't know if it is working as a class D amplifier.
There is a big discussion about that class D "looks"/perform nice if you looking at 20Hz-20khz range.
But the switching noise the designer moving out of that range and put it much higher up in the frequency range! (For example 100kHz)
The questions are:
* When looking at noise should we not also look at a more generic broader frequency range than only the specific hearing range? (There might be amplifiers or (super-)tweeters that play beyond 20kHz that might be affected.
So the conclusion COULD then be that linear power supplies has a lower noise in total than switching power supplies has when looking at a broader frequency range!🤔
Thanks for your videos! I like to hear your take or measurements on this?!🥰
Look at Burno Putezy work on noise in class d if you want answers. But to make a long story short no classd is silent to your ear the residual noise is negligible as it's 100mv switching voltage to force the loop to perform self oscillation.
Great questions! As the graphs show, this class D is flat across the hearing range and beyond in both magnitude and phase. it can be argued that switching power supply noise is outside the hearing range. This has actually been a work in effort since they came out, and MOSFETs have been a huge help so that they can switch faster - well beyond the hearing range. Linear power supplies are actually low frequency switching power supplies - unless they are regulated - linearly. I compare power supply noise between actual linear and switching power supplies. th-cam.com/video/kPHduHM7V1k/w-d-xo.html
Looks good to me !!
Thank you! Great to hear from you!
if you really wanted to show what that Bode 100 does you should have probed the circuit where the class d integrator loop is located, you could then perform analysis of each loop in the each control loop and figure out what the bandwidth and loop gain looks like which is far more interesting and of educational value as well which does justice to the bode 100.
Thanks for the feedback! The TPA3255 is an integrated chip so this was video to show the performance of this amp. I'll show the Omicron in videos to come.
@@KissAnalog honey I know that, I'm saying the loop gain aspect and loop frequency response is of measurement interest. I do wonder what an audio precision have over a bode 100.
Ok thanks! Let me see what I can do;)
I stumbled across your videos early today while researching how to power a little project amp. Super informative videos, Can't wait to see how the amp project comes together. Quick question, the little amp I'm playing with requires positive, negative, and a ground. I have been told that it is for battery only but Id like to use it with a power supply. The hurdle is its operational voltage is 58-70 volts. If I use a traditional power supply like a 60volt like you have linked, and I do the resistor/capacitor bit to create the virtual ground I reduce the voltage by half and it is no longer the correct operational voltage. Any suggestions to help me navigate this rabbit hole? its an irs1092 board.
Great question! Does it want a plus and minus 58-70 volts? If so, then get two power supplies and stack the outputs like batteries. Make sense?
@@KissAnalog I appreciate the response! It's been quite a while, I did figure it out, but I don't exactly remember what the solution was and why, but It was something like it needed a special power supply that had 2 different types of DC power, almost like AC Hot and neutral kinda, there was a required 3rd leg that was weird, but they do make a power supply like that it was quite expensive and it turned out cheaper to go a different amp board.
can you get any class d or smps, amplifiers that uses discreet components not ICs?
Thanks for the great question! I will start a series that I will design my own Class D, but I've also shown a discreet Class D in the past. th-cam.com/video/97p6BkAMTik/w-d-xo.html
Did you post the link to the power supply?
Thank you for reminding me! Yes I added the link in the video, and here it is: ebay.us/LqwFj6
how loud is your power supply?
Great question! Have you seen this video where I compare noise from different power supplies? th-cam.com/video/kPHduHM7V1k/w-d-xo.html
only 5 grand?! insulting... I gave all my change to the homeless and starving
LOL Thanks for the feedback!
@@KissAnalog was wondering if i gen get a free one of those if i make them a couple of thousand videos starting with the philosophical meaning of logarithms