Thanks for your efforts, I concerned you may simply be measuring the reactance between the power supply and the switching amplifier. A low pass is helpful between the pwr supply and amp with a ultra low esr cap bank on the amplifier. You are probably measuring a portion of your bench setup. Excellent presentation. We use an ultra low esr, low impedance linear supply when characterizing the circuit in the lab with very short leads.
The high frequency noise that appears just before clipping is due to the effective switching frequency being lowered as you approach 100% duty cycle in the class-D. The off time cannot go below a certain fixed time, so the feedback in the amp forces the output to stretch the ON time past a normal 450KHz cycle. Another thing that happens is mentioned in the datasheet: TI injects narrow pulses at clipping to keep the bootstrap supply replenished. All of that stuff is outside the audio band, even though it looks like noise on the scope. One thing that you might want to do is try some sharp cutoff low pass filtering between 40KHz and 100KHz or so between the class-D speaker output and the Keithley distortion meter. You might get a better reading of in-band noise rather than measuring out of band ripple.
Thanks for the great feedback Frammis! I can filter the measurments, but I guess the part that I struggle with is wondering how much the filter capacitors are stressed with the switching noise. So, I think it might be worth the extra filtering to be sure that the parts are not over stressed. One good thing is that the high freq of 450kHz usually means low energy in the noise - so then maybe the noise or ripple isn't as bad as I make it out to be;) I do like the idea of the sharp cutoff of the amp at 40kHz. The the pass band is flat and phase is zero - and any concern about the tweeter being stressed is relieved. You obviously know a lot about the Class D, so I'd be happy to hear more. I want to design a Class D and so I ned to start by selecting the best control chip, and then deciding what rail voltages to use.
just fantastic ... so glad I stumbled upon this channel. quite a bit over my head, but makes no difference,... this is very educational, interesting and even relaxing for me.
It's always fun to do some "bench tinkering" to see what improvements you can make on something without doing a full blown redesign. I am also a big fan of using some kind of thermal imaging to do a checkup on any items I am working on, and surprising findings can arise, which, I plan to make some content on in the near future.
Really glad I found you here... A friend of mine suggested 2 caps. a 63v 10000uf Nichicon with a 1uf 63V Wurth Aluminum Electrolytic. These in a box basically on the DC line close to the Aiyima A07 which is a 3255 amplifier board very similar to what you have. Apparently the ripple causes the amp to heat up faster. I have the caps, but I have yet to implement them in my system. I'm looking for a cheap mains filter and I have to rebuild my SMPS enclosure in order to add an isolated DC module on one of the terminals so I can power a 5V 2.1A DAC Would love to have some help ironing all these details out.
Thanks for the feedback! I'd swap the 1uF with a poly 1uF for higher frequency filtering since that would be the benefit of the 1uF in parallel with a larger aluminum cap that isn't as great at the high frequency filtering;) If you have an isolated voltage, you can use an inexpensive buck to generate the 5V. ua-cam.com/video/FnEF_qzX0N4/v-deo.html
@@KissAnalog I'll admit that I'm too much if a novice to know better, so this wasn't really feedback per se.. Moreover, I'm like a kid raising my hand in class. I just picked caps... Not knowing what to get. Would I need AC/Pulse caps for this part of the project? What do I know.... Was buying the Nichicon a total newb move? Could you make some recommendations on which caps to pair for this part of the project? As far as the isolated converter my PSU needs - I bought a Meanwell isolated DC to DC converter with some filtering - PSD-15B-05. Here is the another I bought - xp power DTJ1548S05 When I use this XWST DC DC 8-85V to 5V USB converter from Amazon in the juicebox in the pic - It gives a terrible hum like tone on my DAC. So I searched and searched for isolated, etc. This is what I ended up with. Any way to get a little 1 on 1 help with this project?
Thanks for your feedback! Nichicon is a great brand, and they make a variety of series - each for a specific or general application in mind;) I want to start a weekly call or meeting with my Patreons and/or UA-cam members of the channel (I have to turn this feature on;)
A nice little amp to connect to a mobile device in a garage or a lab for some background music if you want a half decent sound on the cheap, I don't think I would do any more with it than that. Seems the PSU doesn't make much difference overall, it just pushes the limit a bit higher. It might be interesting to connect it to a linear supply and see if the "clipping behaviour" changes.
I set gains on my audio amplifiers w an o scope a (cheaper one)and in fact i was wondering what the fuzzy peeks are????on ab amps i have a clean wave form thank you great video ❤
That's a great way to do it. Thanks for sharing that! Are you asking about Class AB amps with fuzzy peaks? They should be clean, but Class D amps have a fuzzy or noise on the waveform. It looks bad but it is actually far outside what we can hear. It is the Class D switching.
Cheers for vid dude! Any way you can do a multi out (mixer) switch mode power supp - 5 /12/48v WITHOUT opto feedback?? So many Mixers are binned because you can't get Power Supps!!
Thanks for the great question Graeme Wilson! This sounds like a great video! What would be the input voltage? The output would be 5V, 12V, and 48V? What current requirement for each voltage?
@@KissAnalog For me in the UK. IN 230 Vac 50 Hz - OUT +- 12Vdc - 5volt digital supply (USB interface) and 48v dc phantom supply. Say 20watt for 8 channel mixer. (+-12Vdc could be +-15) Thanks for reply!
I think painting the component you want to measure the temperature of with matt black acrylic paint, or even a black sharpie, would give more accurate readings than a piece of black tape. Nice video, waiting to see it all come together.
Thanks Ian! I've tried different methods but the quick way is just to tape it and it does seem to come out about the same. The emisstivity can be adjusted on the camera, but even that doesn't make a huge difference. I find the camera to be a great way to get an overall idea, but a temp probe is more accurate when you need it.
Great question! I will do more testing with this when I get it in a box so that the wiring is cleaner and things are mounted so that I can run power longer. Technically I should be able to get closer to 150 Watts per side into 8 Ohms at 1% THD+N and higher with more distortion.
You still have your plastic sheet on your Micsig; cmon time to remove it... Class D, modulated switch mode power supplies, notorious RF noise soiurces. You can see why the audio enthusiasts aren't fussed by the technology. Keep up the good work, cheers!
LOL Peter ;) I do need to peel the plastic! I'm really curious as to the black box testing to see if one can really hear the difference between good designs.
I'm wondering what a Tantalum 63V 10uF ESR 100mOhm or Tantalum 63V 6.8uF ESR=75mOhms (Mfr. Part # T521D685M063ATE075) ($5.31 at Mouser) would do to it?
Gonna try this out, the way I know to remove noise almost completely from an audio signal is to run the signal through 2 small transformers back to back, thats what I would call "lifting the ground", I have an analog synthesizer that I bought recently that is powered via USB micro or via 4 x AAA batteries, now this thing just eats batteries so, its way less painful to power it via usb, I have tried many usb chargers (240v 60hz to 5.3v Dc) all of them introduced horrible noise, music and talking from radio sources, It is just as bad when powered from a usb port on my studio PC, it picks up every fan and drive spinning up or down ha ha! from looking here on youtube it is a well known complaint about the instrument and its easily fixed with a cheap n chearful ebay ground filter. I have ordered one, it will be interesting to open it up and see how the filtering is achived in a cheap Chinese unit.
@@KissAnalog That noise you're measuring is at the outputs, right? I was actually thinking more along the lines of some inductance in series with those power resistors to match the reactance of the speaker coils. As I'm sure you know, many Class D amplifiers rely on the voice coil as part of a low pass filter.
Uuum yeah, having the power amp suddenly cut out on the actual speaker is usually unsettling*. Maybe it does it every time while it is stone cold? ...then it is probably some pin with a poor solder joint in there. If the amp always behave good with higher input voltage then it could be an issue with internal over current protection/detection. Maybe the issue scales inversely proportional to input voltage? 🤔 Then surely it must be some over current detection issue. * Had a power outtage while playing at groovy levels, nearly shat my pants because the pop! from two subwoofers each having 2,5kW continuos power capability was pretty hair raising... I firmly believed it would be a bit more gracious, but no, horror levels of adenaline right there.
Thanks! It's so good to know how to improve the sound quality.
Thanks Johnson - I appreciate you!
Thanks for your efforts, I concerned you may simply be measuring the reactance between the power supply and the switching amplifier. A low pass is helpful between the pwr supply and amp with a ultra low esr cap bank on the amplifier. You are probably measuring a portion of your bench setup. Excellent presentation. We use an ultra low esr, low impedance linear supply when characterizing the circuit in the lab with very short leads.
Thanks for your feedback! I'll do more of these SMPS noise measuring and filtering videos very soon;)
The high frequency noise that appears just before clipping is due to the effective switching frequency being lowered as you approach 100% duty cycle in the class-D. The off time cannot go below a certain fixed time, so the feedback in the amp forces the output to stretch the ON time past a normal 450KHz cycle. Another thing that happens is mentioned in the datasheet: TI injects narrow pulses at clipping to keep the bootstrap supply replenished. All of that stuff is outside the audio band, even though it looks like noise on the scope. One thing that you might want to do is try some sharp cutoff low pass filtering between 40KHz and 100KHz or so between the class-D speaker output and the Keithley distortion meter. You might get a better reading of in-band noise rather than measuring out of band ripple.
Thanks for the great feedback Frammis! I can filter the measurments, but I guess the part that I struggle with is wondering how much the filter capacitors are stressed with the switching noise. So, I think it might be worth the extra filtering to be sure that the parts are not over stressed. One good thing is that the high freq of 450kHz usually means low energy in the noise - so then maybe the noise or ripple isn't as bad as I make it out to be;) I do like the idea of the sharp cutoff of the amp at 40kHz. The the pass band is flat and phase is zero - and any concern about the tweeter being stressed is relieved. You obviously know a lot about the Class D, so I'd be happy to hear more. I want to design a Class D and so I ned to start by selecting the best control chip, and then deciding what rail voltages to use.
just fantastic ... so glad I stumbled upon this channel. quite a bit over my head, but makes no difference,... this is very educational, interesting and even relaxing for me.
Thanks so much! Welcome! Let me know if there is something that I can do to help with educational vids.
It's always fun to do some "bench tinkering" to see what improvements you can make on something without doing a full blown redesign. I am also a big fan of using some kind of thermal imaging to do a checkup on any items I am working on, and surprising findings can arise, which, I plan to make some content on in the near future.
Thanks for the great feedback! I appreciate you!
Really glad I found you here...
A friend of mine suggested 2 caps. a 63v 10000uf Nichicon with a 1uf 63V Wurth Aluminum Electrolytic. These in a box basically on the DC line close to the Aiyima A07 which is a 3255 amplifier board very similar to what you have. Apparently the ripple causes the amp to heat up faster. I have the caps, but I have yet to implement them in my system. I'm looking for a cheap mains filter and I have to rebuild my SMPS enclosure in order to add an isolated DC module on one of the terminals so I can power a 5V 2.1A DAC Would love to have some help ironing all these details out.
Thanks for the feedback! I'd swap the 1uF with a poly 1uF for higher frequency filtering since that would be the benefit of the 1uF in parallel with a larger aluminum cap that isn't as great at the high frequency filtering;) If you have an isolated voltage, you can use an inexpensive buck to generate the 5V. ua-cam.com/video/FnEF_qzX0N4/v-deo.html
@@KissAnalog I'll admit that I'm too much if a novice to know better, so this wasn't really feedback per se.. Moreover, I'm like a kid raising my hand in class. I just picked caps... Not knowing what to get. Would I need AC/Pulse caps for this part of the project? What do I know.... Was buying the Nichicon a total newb move? Could you make some recommendations on which caps to pair for this part of the project?
As far as the isolated converter my PSU needs - I bought a Meanwell isolated DC to DC converter with some filtering - PSD-15B-05. Here is the another I bought - xp power DTJ1548S05
When I use this XWST DC DC 8-85V to 5V USB converter from Amazon in the juicebox in the pic - It gives a terrible hum like tone on my DAC. So I searched and searched for isolated, etc. This is what I ended up with.
Any way to get a little 1 on 1 help with this project?
Thanks for your feedback! Nichicon is a great brand, and they make a variety of series - each for a specific or general application in mind;) I want to start a weekly call or meeting with my Patreons and/or UA-cam members of the channel (I have to turn this feature on;)
Amazing video, as always...keep up the awesome work!
Thanks! Will do!
Hi Eddie, as usual, a very interesting video and great demo. Thank you Eddie
Thanks so much for your support!
A nice little amp to connect to a mobile device in a garage or a lab for some background music if you want a half decent sound on the cheap, I don't think I would do any more with it than that. Seems the PSU doesn't make much difference overall, it just pushes the limit a bit higher. It might be interesting to connect it to a linear supply and see if the "clipping behaviour" changes.
wowwww thanks for demo sir
You bet! I appreciate you!
It was a useful video.
Thank you!
I set gains on my audio amplifiers w an o scope a (cheaper one)and in fact i was wondering what the fuzzy peeks are????on ab amps i have a clean wave form thank you great video ❤
That's a great way to do it. Thanks for sharing that! Are you asking about Class AB amps with fuzzy peaks? They should be clean, but Class D amps have a fuzzy or noise on the waveform. It looks bad but it is actually far outside what we can hear. It is the Class D switching.
@@KissAnalog exactly on ab they're very clean thank you very much love the videos ❤️
Badass Hioki clamp meter!
It sure is Boyd! I appreciate you!
Good Video Eddie !
Thanks bud! I appreciate you!
@@KissAnalog Hell yeah man !!
Cheers for vid dude! Any way you can do a multi out (mixer) switch mode power supp - 5 /12/48v WITHOUT opto feedback?? So many Mixers are binned because you can't get Power Supps!!
Thanks for the great question Graeme Wilson! This sounds like a great video! What would be the input voltage? The output would be 5V, 12V, and 48V? What current requirement for each voltage?
@@KissAnalog For me in the UK. IN 230 Vac 50 Hz - OUT +- 12Vdc - 5volt digital supply (USB interface) and 48v dc phantom supply. Say 20watt for 8 channel mixer. (+-12Vdc could be +-15) Thanks for reply!
What is the summary of this test?
I think painting the component you want to measure the temperature of with matt black acrylic paint, or even a black sharpie, would give more accurate readings than a piece of black tape. Nice video, waiting to see it all come together.
Thanks Ian! I've tried different methods but the quick way is just to tape it and it does seem to come out about the same. The emisstivity can be adjusted on the camera, but even that doesn't make a huge difference. I find the camera to be a great way to get an overall idea, but a temp probe is more accurate when you need it.
Is 84 watts the most you at get at 8 ohms out of the power supply?
Great question! I will do more testing with this when I get it in a box so that the wiring is cleaner and things are mounted so that I can run power longer. Technically I should be able to get closer to 150 Watts per side into 8 Ohms at 1% THD+N and higher with more distortion.
@@KissAnalog It is something I wondered also. I picked up a TPA3255 in PBTL for subwoofer and 600w PSU from Aliexpress to find out myself.
Would you like to use our Brand Seesii Digital Multimeter in your video next time?
Thank you! I posted a 'short' video and will be doing a full review very soon!
You still have your plastic sheet on your Micsig; cmon time to remove it...
Class D, modulated switch mode power supplies, notorious RF noise soiurces. You can see why the audio enthusiasts aren't fussed by the technology. Keep up the good work, cheers!
LOL Peter ;) I do need to peel the plastic! I'm really curious as to the black box testing to see if one can really hear the difference between good designs.
I'm wondering what a Tantalum 63V 10uF ESR 100mOhm or Tantalum 63V 6.8uF ESR=75mOhms (Mfr. Part #
T521D685M063ATE075) ($5.31 at Mouser) would do to it?
Great question David! I like tantalum’s - they are the most dense capacitor that you can get, so small and East to fit more in parallel!
Gonna try this out, the way I know to remove noise almost completely from an audio signal is to run the signal through 2 small transformers back to back, thats what I would call "lifting the ground", I have an analog synthesizer that I bought recently that is powered via USB micro or via 4 x AAA batteries, now this thing just eats batteries so, its way less painful to power it via usb, I have tried many usb chargers (240v 60hz to 5.3v Dc) all of them introduced horrible noise, music and talking from radio sources, It is just as bad when powered from a usb port on my studio PC, it picks up every fan and drive spinning up or down ha ha! from looking here on youtube it is a well known complaint about the instrument and its easily fixed with a cheap n chearful ebay ground filter. I have ordered one, it will be interesting to open it up and see how the filtering is achived in a cheap Chinese unit.
This may be tl;dw, but you seem to just keep adding microfarts. What about adding a few millihenries?
Thanks Peter! That could work especially if it is multiple caps in parallel to lower the esr.
@@KissAnalog That noise you're measuring is at the outputs, right? I was actually thinking more along the lines of some inductance in series with those power resistors to match the reactance of the speaker coils. As I'm sure you know, many Class D amplifiers rely on the voice coil as part of a low pass filter.
Uuum yeah, having the power amp suddenly cut out on the actual speaker is usually unsettling*. Maybe it does it every time while it is stone cold? ...then it is probably some pin with a poor solder joint in there. If the amp always behave good with higher input voltage then it could be an issue with internal over current protection/detection. Maybe the issue scales inversely proportional to input voltage? 🤔 Then surely it must be some over current detection issue.
* Had a power outtage while playing at groovy levels, nearly shat my pants because the pop! from two subwoofers each having 2,5kW continuos power capability was pretty hair raising...
I firmly believed it would be a bit more gracious, but no, horror levels of adenaline right there.