Purifi 1ET400A with linear PSU tested to comply with FTC rule

[JustJones]

I don't look at the regulation from an engineering point of view but a legal one. Viewing from that perspective then the regulation means what it says all frequencies from 20hz-20khz at 5 minutes. Not 1, 3, 10 or 32 which is why under the petition to reopen the FTC is being asked for clarification or guidance on these tests as well as whether the regulation applies retroactively.

 

[Sokel]

pma tests here are good for my rule except my tests freqs would be 20Hz,100Hz,1KHz,6.6KHz
The 20Hz one may seem low for music but if it can do that it can also do 30's which is my concern and as far as i can see I'll stick to full-bridge amps as they seem to tame (my) lows a little better no matter the power (apart the actual power of course) .

 

[Lars Risbo]

thank you, good work @pma

 

[pma]

pma tests here are good for my rule except my tests freqs would be 20Hz,100Hz,1KHz,6.6KHz
The 20Hz one may seem low for music but if it can do that it can also do 30's which is my concern and as far as i can see I'll stick to full-brigde amps as they seem to tame (my) lows a little better no matter the power (apart the actual power of course) .

Test frequencies may be change according to DUT behaviour. 20Hz, 1kHz and 20kHz I use always, anything in between may be added if there is a good reason.

 

[Sokel]

Test frequencies may be change according to DUT behaviour. 20Hz, 1kHz and 20kHz I use always, anything in between may be added if there is a good reason.

6.6KHz is an interesting one for all amps as apples to apples as it's H2 and H3 are measured even with the limited 20Hz-20KHz or filtered BW which we use for class D.

Edit:for example,icepower promised us great improvement there with their new Conductor,and it seems they delivered:


link

(someone here must cross-check this! )

 

[pma]

6.6KHz is an interesting one for all amps as apples to apples as it's H2 and H3 are measured even with the limited 20Hz-20KHz or filtered BW which we use for class D.

I do not use the 20 kHz LPF filter when testing class D amplifiers. Fortunately, E1DA Cosmos ADC does not suffer from input slew rate problem and measures THD < -120 dB even without that filter. I find measuring THD and also THD+N with 20kHz limited bandwidth as pointless for signal frekvencies >= 5kHz. HF distortion and noise is interesting from engineering point. I do not say it is audible, but if we stick with audibility of nonlinear distortion, then we may close the whole distortion measurement chapter.

 

[Lars Risbo]

I was expecting a comment from @restorer-john ?

 

[pma]

Happy New Year!

 

[amirm]

I do not use the 20 kHz LPF filter when testing class D amplifiers. Fortunately, E1DA Cosmos ADC does not suffer from input slew rate problem and measures THD < -120 dB even without that filter. I find measuring THD and also THD+N with 20kHz limited bandwidth as pointless for signal frekvencies >= 5kHz.

Seems like there is some confusion with respect to AES 20 kHz filter. It does NOT have 20 kHz of bandwidth. Spec mandates that it be flat in that bandwidth. Given that, the actual cut off frequency is far higher. Here are the AP measurements for the AUX-0040 that I use:

As seen, it is nearly flat to 100 kHz which is plenty of bandwidth to use for harmonic distortion. Their 20 kHz one is not as extended but even there, the bandwidth goes much higher than 20 kHz:

As we see, it is only down a couple of dBs at 50 kHz. So no problem in capturing harmonics with a 45 kHz bandwidth.

As to E1DA not needing it, when it has an autoscalar front-end with ability to measure somem 155 volts input, then come and brag about it. Until then, AP also does quite well without the filter. The ultrasonic noise mostly confuses that auto-scalar than other artifacts. Here is the Purifi reference design FFT without the filter:

As you see, spikes are easily visible down to -120 dB. If you used a software THD analyzer as you are doing with E1DA you could also run the AP without the filter. AP's dedicated THD analyzer is not as sophisticated so may get confused with ultrasonic noise.

 

[KSTR]

Maybe Pavel meant the 20kHz brickwall analog filter (S-AES17) as present in the previous generation of APs?
That one would compromise THD(+N) quite a bit at higher frequencies.

And/or the original 25kHz version of the AUX-00xx filters?
While not as steep as the above filter some effect would be seen on harmonics falling above 20kHz.

 

[pma]

Maybe Pavel meant the 20kHz brickwall analog filter (S-AES17) as present in the previous generation of APs?
That one would compromise THD(+N) quite a bit at higher frequencies.

Yes. Strict 20 kHz BW limiting is used to create nicely looking THD+N plots (however pointless and irrelevant) of class D amplifiers, including the Purifi. Below is the proof, including AES17 mention.

To @amirm
Would you mind posting the graphs like the one above (switching frequency amplitude) with calibrated Y axis? Like dBV, V, anything. The plot in dBrA does not allow to read the switching frequency amplitude, it is important to know.

 

[HaveMeterWillTravel]

As someone who has professionally tested electronics to a multitude of standards over the years, the FTC's standard has less information than the marketing blurb from IEC, EN, MIL-STD, ANSI, AMSE put out about their standards.

 

[HaveMeterWillTravel]

I don't look at the regulation from an engineering point of view but a legal one. Viewing from that perspective then the regulation means what it says all frequencies from 20hz-20khz at 5 minutes. Not 1, 3, 10 or 32 which is why under the petition to reopen the FTC is being asked for clarification or guidance on these tests as well as whether the regulation applies retroactively.

How do you propose to measure THD using white noise as the test stimuli in a practical manner?

 

[Devnull]

I don't look at the regulation from an engineering point of view but a legal one. Viewing from that perspective then the regulation means what it says all frequencies from 20hz-20khz at 5 minutes. Not 1, 3, 10 or 32 which is why under the petition to reopen the FTC is being asked for clarification or guidance on these tests as well as whether the regulation applies retroactively.

Nothing in the regulation says every frequency needs to be tested. It's up to the manufacturer to decide what frequencies from an engineering view allows them to make a factual statement. In an enforcement action it's up to the FTC to prove the amplifier is incapable of meeting the stated spec.

 

[MaxwellsEq]

Nothing in the regulation says every frequency needs to be tested. It's up to the manufacturer to decide what frequencies from an engineering view allows them to make a factual statement

The regulation requires all frequencies and measured with distortion.

 

[pma]

The regulation requires all frequencies and measured with distortion.

Interpretation is to be wise and making sense. There is an infinite number of frequencies between 20Hz and 20kHz, depending on smallest frequency resolution used. Again, only a nit picking, posts for posts, without practical sense.

 

[Geert]

Nothing in the regulation says every frequency needs to be tested. It's up to the manufacturer to decide what frequencies from an engineering view allows them to make a factual statement. In an enforcement action it's up to the FTC to prove the amplifier is incapable of meeting the stated spec.

https://www.ecfr.gov/current/title-16/chapter-I/subchapter-D/part-432/section-432.3

The rated power output is measured in compliance with the standard test conditions in § 432.3; Any power level from 250 mW to the rated power shall be obtainable at all frequencies within the rated power band of 20 Hz to 20 kHz without exceeding 1.0% of total harmonic distortion plus noise (THD+N) at an impedance of 8 ohms after input signals at said frequencies have been continuously applied at full rated power for not less than five (5) minutes

Don't know what a signal that contains all frequencies at full power continuously for 5 minutes looks like, and how you measure THD for such a signal. White noise doesn't comply (doesn't guaranty it contains all frequencies at full power continously).

 

[KSTR]

Repeated medium fast log sweep as used for IR measurement would apply, minimum crest factor. If you need it completely gapless, Newman sweep would be a solution, which is a form of periodic white noise with phases set up to result in a sort of endless sweep with a very low crest factor.
And then we have FSAF: https://www.audiosciencereview.com/...nd-adaptive-filtering-fsaf-measurement.58697/

 

[pma]

Don't know what a signal that contains all frequencies at full power continuously for 5 minutes looks like, and how you measure THD for such a signal. White noise doesn't comply (doesn't guaranty it contains all frequencies at full power continously).

Yes. It is only a misunderstanding and misinterpretation based on lack of knowledge in the field.

 

[pma]

Medium fast log sweep as used for IR measurement would apply, minimum crest factor

Yes but they will argue with 5 minutes at the fixed frequency. The debate is pointless, as most debates.