Class D Amp

The place to discuss general topics about Soundeasy.
Post Reply
User avatar
Chad Gray
Site Admin
Posts: 46
Joined: Wed Dec 02, 2020 7:26 pm
Contact:

Class D Amp

Post by Chad Gray »

Has anyone gotten a Class D amp to play nicely with Soundeasy?

Last time I tried I got really strange results and switched back to my good old 90's Kenwood basic receiver.

Chad
Jacq.
Posts: 3
Joined: Thu Dec 03, 2020 3:15 am

Re: Class D Amp

Post by Jacq. »

Do not so far really know, to make a small story short I'm trying to re acquaint with SE after 10 years. My last use was with SE was in with v11/14 and win xp.
I have now upgraded with a Dell desk top, win10/i7, gen9 and SE24 and the amp in use right now for measurments is a 50w IcePower . The sound card is a Behringer UCA202??? .
So far not a pc user thought I'm having negative results trying to make decent measurements. Scarlett 2i2 interface is on it's way but from your post I'm now suspicious the IcePower could also be the culprit?
User avatar
Chad Gray
Site Admin
Posts: 46
Joined: Wed Dec 02, 2020 7:26 pm
Contact:

Re: Class D Amp

Post by Chad Gray »

Here is an impedance measurment with my Class D amp
https://www.amazon.com/gp/product/B071XQGYRJ/


ClassDAmp.png
ClassDAmp.png (757.19 KiB) Viewed 738 times


This is a measurment with my old Kenwood receiver.


Kenwood.png
Kenwood.png (754.24 KiB) Viewed 738 times

The phase and impedance are way off compared to my Kenwood amp.
User avatar
Chad Gray
Site Admin
Posts: 46
Joined: Wed Dec 02, 2020 7:26 pm
Contact:

Re: Class D Amp

Post by Chad Gray »

Found this thread on the PE board. Looks like class D amps are a no go.


http://techtalk.parts-express.com/forum ... asurements

Re: Need a smallish amp for measurements

Originally posted by tyger23 View Post
For measurement related activities, I would really encourage you to stay away from any Class-D amplifier (inclusive of the Topping, Lepai, or Class-T based stuff). The reason for this is that all class-D (class-T) amps will include a low-pass filter on the output path that has to make some kind of assumption as to what the speaker load is. This low-pass filter is required to remove the switching frequency from the audio output (for FCC compliance and improved THD performance).

As an example - many of the Class T amps (like the Lepai) assume that the end user will be connecting 8-ohm loads. In this case, the treble will roll-off at around 24KHz. This is OK for most measurements. However, if you connect up a 4-ohm tweeter and try and take measurements with that amp, the output filter will start causing the treble to roll-off at around 15Khz, which will adversely affect the accuracy of your measurements.

Class-AB amplifiers (like the AudioSource AMP100, Pyle PAMP1000, and others) will often include similar type filters to help avoid RF interference. However, the filters that are included in the Class-AB amplifiers are NOT dependant on the speaker load and will not change just because you've changed what speaker you're trying to measure. My modifications that I've posted to the AMP100 and PAMP1000 will remove these RF filters and will extend the frequency response (and thus the accuracy) of your measurements.

So - IMHO, I would really recommend a class-AB for measurement purposes. The Class D and class T amps are fine for normal, everyday use, but could introduce some variances in your measurements.


I completely agree Ty, but I would like to add something else... The corner frequency of the filter will change dependent on the load applied, but the filters are usually very high Q, which means you may also get a very high boost before the roll-off. On top of all that, even if the filter is optimized for a 4 ohm speaker, any inductive rise in the higher frequency (where the corner frequency of the amplifier is) will cause the speaker to look like a higher impedance. In summary: Using an amplifier for a measurement setup whose response is dependent on the load impedance is a very bad practice.

I have one of the sure 50W x 2 amplifiers, and the corner frequency of the output filter was originally picked to be 17 kHz (horrible idea to put the corner frequency in the pass-band). If I disconnected the load on either channel and fed anything in the 17 kHz range, the amplifier went in to shut down immediately because the output hit the filter resonance and there was virtually zero damping. With a high impedance load in the 17 kHz range, the output would spike 6 dB or more. I changed the filter to push the corner frequency out above 20 kHz and got the 20-20 kHz response flat within a dB or two.

Another thing to keep in mind is that the output of most of the low power switching amplifiers with low voltage supplies are BTL outputs. If you use a measurement package that has the ability of looping the amplifier output back to a soundcard input (i.e. ARTA), extreme care must be taken in doing so. Feeding a BTL amplifier back to a single-ended soundcard input will short one side of the amplifier output through your soundcard, potentially damaging both. Amplifiers with BTL outputs should only be looped back into sound cards with balanced inputs.
dcibel
Posts: 39
Joined: Sun Dec 13, 2020 6:39 am

Re: Class D Amp

Post by dcibel »

Class-D amps are okay for a true 2-channel measurement system like SoundEasy, any non-linearity in the amp output is compensated for by the reference channel measurement. And espeically if you are designing a speaker for use with a specific class-D amp like a plate amp, you may want to include the amp non-linearity in your design as it is a constant part of the system.

What is very important is that the amp has a single ended output, many low power and low cost class-D solutions are internally bridged, so the negative speaker terminal is not a 0V reference, which will be a problem for the measurement probe system used here.
User avatar
Chad Gray
Site Admin
Posts: 46
Joined: Wed Dec 02, 2020 7:26 pm
Contact:

Re: Class D Amp

Post by Chad Gray »

OK, thanks for the reply! Do you think this is why i get such a strange reading in the snapshot above? The speaker terminal is not a 0V reference?

I was just about to start testing a powered subwoofer with Soundeasy. I might start another topic on this subject as i look over my plan to set it up.

Thanks again!
Chad
dcibel
Posts: 39
Joined: Sun Dec 13, 2020 6:39 am

Re: Class D Amp

Post by dcibel »

Easy enough to check with a multi-meter from each speaker terminal to ground. I would not be surprised if the small amp you linked to has bridged speaker output. Speaker negative voltage reference must be the same as your measurement probe 0V reference which won't be the case with a bridged amplifier. It's the last paragraph in the above quoted text
"Another thing to keep in mind is that the output of most of the low power switching amplifiers with low voltage supplies are BTL outputs. If you use a measurement package that has the ability of looping the amplifier output back to a soundcard input (i.e. ARTA), extreme care must be taken in doing so. Feeding a BTL amplifier back to a single-ended soundcard input will short one side of the amplifier output through your soundcard, potentially damaging both. Amplifiers with BTL outputs should only be looped back into sound cards with balanced inputs."
The rest of that text really only applies to single channel measurement systems like the USB mics that many use. Anyway, class-D is not the issue here, I've used my Hypex UcD180 amp perfectly well, but for low cost solutions there's a plethora of LM3886 and similar class-AB chip amps that will do the trick, or used receivers.
dcibel
Posts: 39
Joined: Sun Dec 13, 2020 6:39 am

Re: Class D Amp

Post by dcibel »

Pulled up the datasheet for the TDA3116D2 chip that is in your Amazon amp, very common IC for low cost class-D amps.

https://www.ti.com/lit/ds/symlink/tpa3116d2.pdf

It is a BTL amp, do not use for measurements.
User avatar
Chad Gray
Site Admin
Posts: 46
Joined: Wed Dec 02, 2020 7:26 pm
Contact:

Re: Class D Amp

Post by Chad Gray »

Great information! I always learn something and I appreciate your response and explanation. I want to update the Wiki with some of this information.

Chad
Post Reply