Home ›
Site Links
Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
Categories
In this Discussion
Please review the site Rules, Terms of Service, and Privacy Policy at your convenience. Rules, TOS, Privacy
Get familiar with the reaction system: Introducing the Reaction System
Remind me why Class-D Amps/SMPS are bad for Creating FR Sweeps/XO Design Data . . .
. . . JR had made a comment a while back [that I cannot find] as to why Class-D amps and SMPS powered amps were not suitable for acquiring accurate FR/Z data for use in XO design.
I know my ears prefer Class A/B amps [room heaters] but I am at a loss right now as to why? [He made a good argument against them].
Thanks for any good feedback.
Comments
Some, not all.
Class-D amps have a square wave PWM output, which then runs through a passive low pass filter to create audio. Some use pre-filter feedback, so the filter response is not compensated for by the amp output. This results in a response at high frequency that will vary depending on the load, usually result is a slightly rising response towards 20kHz. You will not find this problem in Hypex or Purifi amps. None of this is a problem if using 2 channel measurements. Impedance measurement is dual channel by design, so no problem there either way.
Many cheap class-D “chip amps” use BTL configuration, not to be used with dual channel measurement jig at all. Negative terminal of amp speaker connection gets connected to line input ground, so you can see that this will be problematic for BTL amp. This is true for any BTL amp, not strictly a class-D problem.
SMPS has nothing to do with anything. Modern high end SMPS offer lower noise and hum than linear supplies, so a good choice regardless of what you’re doing.
I'm going to have to absorb this for a while to distinguish the difference between "cheap chip amps" and the others but it sounds like dual channel test jigs bypass "ALL" of the Class-D chip amp issues concerning FR/Z testing? Am I understanding this correctly, Sir?
Additionally, do SMPS or chip amps introduce more or less measurable distortion into the audio stream than their linear counterparts?
Thanks for the reply.
This is an overly broad statement that is generally false. IMO the best class-D amps available these days (from Hypex and Purifi) are better than pretty much every class-AB amp ever produced. On the other end of the spectrum, some AliExpress garbage amp you get for $30 with free shipping is not what you want to use anywhere, anyhow. In general there is nothing bad about class-D or SMPS technology, but there is a wide range of quality out there in both products. For measuring frequency response such amps and PSs are perfectly fine. If there is some impedance dependency, that will be only up near 20kHz where the output filter of the amp can interact with the load impedance, but this depends on the design and implementation. If you will always keep the amp and speaker together, that doesn't matter either.
As decibel pointed out, if the amp is configured to always operate in bridged mode (like many lower power class-D amps are) then you can't make a practical impedance measurement with it because both + and - terminals are active, neither are at ground potential. So keep that in mind.
"How does a dual channel measurement work?" Glad you asked. It is a differential measurement comparing the measurement input to a feedback reference. This has the effect of normalizing, or self-calibrating one measurement to the other, for everything included within the feedback loop, as well as providing invaluable timing information. This can compensate for any non-flat amplitude response resulting from the amplifier, cabling, or any filter included within the feedback loop.
For filter design, if you are designing an active system that will be dedicated to a particular amp, it may be beneficial to include the non-linear amplifier response in the measured data so it can be compensated for. For speakers intended for use with any amp, response of only the speaker without any upstream change in response should be the goal.
Distortion varies amp to amp, irrespective of class. High end class-D from Hypex or Purifi are among the best performers available period.
dcible - that's some great information. I read some designs had issues at higher frequencies with respect to load, but never knew why. Thanks!
Hypex has some great information on Class-D and UCD post-filter feedback at their site:
https://www.hypex.nl/img/upload/doc/an_wp/WP_All_amps_are_analogue.pdf
https://www.hypex.nl/img/upload/doc/an_wp/WP_AES118BP_Simple_self-oscillating_class D_amplifier.pdf
https://www.hypex.nl/p/application-notes-white-papers/
I believe the Starkrimson amp is BTL and it's not really a cheap amp. I had the ultra stereo version at my house, and it's very good, but I doubt that you would consider using a $2500 amp for measurements.
Smoke 'em if you got 'em
. For $2500 it should at least have a basic electrical safety certification label on the back.
Thread over on PE/TT that seemed to confirm this effect:
https://techtalk.parts-express.com/forum/tech-talk-forum/1490039-what-s-wrong-with-my-c-notes
See the last page.. The rest is all the spitballing that lead up to it.
I think that bashing all amplifier of class D is unwarranted. There is a reference to an audiosciencereview thread that shows how the AIYIMA A07 response has some significant load dependency above ~5kHz but in the PETT thread the FR rise he shows happened much lower in frequency, too. So maybe he happened to get a very poorly example of this rather cheap amp. The AIYIMA amps are generally not very well designed and executed. The specific problem in this case is that the output filter uses cheap parts (like an inductor with a ferrite core instead of a powdered core), and this causes the load to interact with it within the audio passband. Better designs either move this up in frequency above 20kHz and reduce or eliminate it alltogether (e.g. Hypex) but those are more expensive.
Caveat Emptor.
I have very little confidence in Josh's measurement ability or the conclusion he arrived at. For example in post 49 he mentions measuring with Aiyima A07, Yamaha AS301, and SMSL SA36 and getting the same result. The Yamaha is a class-AB amp.
Keep in mind as well that FR measurements in amplifier reviews are usually completed using purely resistive load, so the FR is what you can expect from 10kHz+ with 8 ohm or 4 ohm impedance. Depending on the speaker, an 8 ohm speaker may have 10 ohm + impedance in this range, and it can vary quite a bit from speaker to speaker, all depending on the drivers used and crossover topology.
Of course, it's easy to measure for yourself with any amp and any speaker. With a 2-channel measurement jig, just measure the reference channel and see for yourself. For a dedicated jig for this purpose, all that's needed is a couple resistors and some scrap wire to step down the amplifier voltage to line level to measure it.
Also it wasn't clear to me that he was using the correct calibration file (or one at all) for the mic, and a rising top end with a bump near 20k is exactly what the response of an uncalibrated ECM-8000 or EMM6 looks like.