Vera Audio Class-D Amp Build Quality

[restorer-john]

I'm curious which loudspeakers you have (or anyone else for that matter) that require anything near full rated power above 10khz, and does not melt the tweeter in the process?

It doesn't matter. The FTC rules are clear. I shouldn't need to remind you of them. To sell legally into the US market, they must be compliant. No excuses.

Class D simply fails to perform at high power and high frequencies. We all know that. I know it, you know it, and most other sensible people know it.

Be honest about the rated THD across the audible bandwidth (20-20k), especially at high frequencies and high power.
Be honest about the complete inability to deliver continuous rated full power at HF.
Specify compliant minimum THD from 250mW to full rated power THD, not cherry picked bottom of the plot numbers.

 

[sigbergaudio]

It doesn't matter. The FTC rules are clear. I shouldn't need to remind you of them. To sell legally into the US market, they must be compliant. No excuses.

Class D simply fails to perform at high power and high frequencies. We all know that. I know it, you know it, and most other sensible people know it.

Be honest about the rated THD across the audible bandwidth (20-20k), especially at high frequencies and high power.
Be honest about the complete inability to deliver continuous rated full power at HF.
Specify compliant minimum THD from 250mW to full rated power THD, not cherry picked bottom of the plot numbers.

And what about this measurement? Is this a failure to perform?

 

[restorer-john]

And what about this measurement? Is this a failure to perform?

Do you actually believe 273.5W @4R at 0.034% THD is SOTA in 2021? LOL.

We had amplifiers in the 1980s at least an order of magnitude better. In the 90s, they were almost two orders of magnitude better.

I've got power amplifiers here from 1983 that will do 0.003% at 20KHz at ~300W/4R (200kHz BW)

But, with respect, his bandwidth is actually too great- he's picking up non harmonics and making his numbers worse, as he's heading into 400kHz...

 

[sigbergaudio]

Do you actually believe 273.5W @4R at 0.034% THD is SOTA in 2021? LOL.

We had amplifiers in the 1980s at least an order of magnitude better. In the 90s, they were almost two orders of magnitude better.

I've got power amplifiers here from 1983 that will do 0.003% at 20KHz at ~300W/4R (200kHz BW)

But, with respect, his bandwidth is actually too great- he's picking up non harmonics and making his numbers worse, as he's heading into 400kHz...

There's a pretty wide gap between "failing to perform" and SOTA. Also I'm not sure why we are interested in a) Inaudible distortion or b) theoretical situations that will never occur in real life (like drawing 300W of power at 20khz).

I understand that it's the FTC standard, but then perhaps it's time to change it.

 

[Armand]

Where in the FTC standard does it specify that the THD+N measurements must be done with a bandwidth of at least 250kHz?

I can only find this:
(e) Rated power shall be obtainable at all frequencies within the rated power band without exceeding the rated maximum percentage of total harmonic distortion after input signals at said frequencies have been continuously applied at full rated power for not less than five (5) minutes at the amplifier's auxiliary input, or if not provided, at the phono input.

https://www.law.cornell.edu/cfr/text/16/432.3

 

[RichB]

Wide-band distortion numbers are not meaningless.
Many amplifiers increase distortion with frequency.

Folks into MQA (yuck) and Hi-Res audio should be interested in measured performance up to 48 kHz.

Using the now somewhat famous "Keys" tests, notable members have passed a blind. Not because they can hear above 20 kHz, so it must be due to modulation of some kind into the audible range. This could be modulation from an amplifier, preamplifier, DAC, basically something in the chain. If Hi-Res audio is your thing, then the amplifier candidates should include high bandwidth measurements.

Amplifiers should certainly post standard FTC 20 to 20 kHz unweighted and weighted measurements as well as wide-band that wish to claim superior performance for customers to make valid comparisons. Weighted measurements can mask issues delivering power to low frequencies as well.

I sent an ATI AT522NC (NCore) amp to ASR for testing and at first it appeared that power was limited to 22 watts at 20 kHz, due to to ASR stringent distortion and wide band testing.

(1) ATI AT522NC Stereo Amplifier Review | Audio Science Review (ASR) Forum

In order to get close to the specification, the bandwidth was reduced to 45 kHz, down from 90 kHz, and the THD+N target loosened to THD+N 0.03 (SINAD 70 dB).

(1) ATI AT522NC Stereo Amplifier Review | Page 5 | Audio Science Review (ASR) Forum

Even then, the 250 WPC spec was measured below 10 kHz. at 15 kHz power dropped was almost spec at about 220 WPC.

FWIW, I was able to distinguish the AT525NC and AHB2 level matched to .1 dB (using a Fluke voltmeter). The MiniDSP has 4 DACs that was used to provide dual outputs level matched using its DSP. A speaker switch was used for instantaneous switching. I was able to pass a SBT test as well. Listening levels did not exceed 10 watts driving Revel 126Be bookshelf speakers.
The AHB2 was better a resolving cymbal and other percussion instruments.
Feist's "Metals" contains many tracks well suited for this comparisson.
This is not scientifically significant but suits my needs.

- Rich

 

[Armand]

I agree that wideband measurements are a useful addition in properly characterizing an amplifier. We can see if there are potential harmful high levels of RF present etc.

The point of the FTC is to avoid misleading of customers and is most welcome.
John's constant complaints about THD+N measurements with 250kHz bandwidth is flawed and in fact misleading!

Let me illustrate with some examples:

Let's say a less tech savvy customer wants an amplifier that has a very low noise level for his high sensitivity tweeters operating in the 2-20kHz band. He takes a look at the noise number measured with 250kHz bandwidth.

Amplifier 1 has a noise level of 250uV and Amplifier 2 has a noise level of 1000uV.
Amplifier 1 must be better than amplifier 2 right?.... Wrong!

Here is a spectrum of the very very expensive class A amplifier in the 50.000 USD price range I had in for repair a year ago (name removed). A noise level of 250uV is very poor. It is measured only up to 20kHz, but is does not matter since the noise is declining above that. Perhaps the noise would be around 300uV if measured up to 250kHz.

The noise is in the audible range and while spikes of -100dBV @ 2kHz is not very much it will be audible on a sensitive tweeter.

Here is the spectum of Vera P150/600 in bridged mode measured with 250kHz bandwith:

All the noise up to 250kHz is included in the measurement and if only the noise number is presented to the customer he will assume the expensive class A amplifier will be more silent than the Vera P150/600 class D amplifier.


When limiting the bandwith to a very conservative 40kHz which is much higher than we actually can hear and what the tweeter actually can reproduce the table is turned and the noise level on the bridged Vera is about 50uV. The spectrum now looks like this.

Which one of these amplifiers will give the customer the actual audible lowest noise level? Which one of the presented information gives the customer the best advice?
250kHz bandwidth or 40kHz bandwidth? John?

Reducing the bandwidth to a more realistically 20kHz gives us 30uV:

About THD
(All noise measurements above is in bridged mode by the way. )

Because the feedback is reduced at high frequencies, there is no argue that high frequencies are the weak point for class D amplifiers.
But how severe is it?
Here is the output when delivering 8kHz with 600W power into a 8 Ohm load. One with 40kHz bandwith and one with 250kHz.

Which one of these two measurements gives the customer the most informative information about THD+N?
The one with 0.0007% up to 40kHz, or the one with 0.0183% up to 250kHz?

Unless you are a bat, the 40kHz bandwith is far more relevant than the other one, and is not that exactly the whole point?

John, you can argue all you want about this, but please do not accuse us for cherry picking measurements and misleading customers. I hope the measurements presented here show that class D is not so bad. Is it?

 

[Feyire]

@Armand, for my own curiosity, could you share amirm style THD+N measurements (4 Ohm @ 5W) of your P150/600 with varying gain, say at 12, 21 and 30?

 

[Armand]

Sorry for the late reply. It's summer, and we have to make the most out of the very few days we have per year with 25 degrees celcius and sun.

I did you requested measurements from serial number 18 today.
Note that the second order harmonic is from my analyzer. Not everybody is blessed with a AP555x like Amir
Here is 12dB, 21dB and 30dB gain with 4,47V into 4 Ohm load (5W).

12dB gain. 1120mV input

21dB gain. 400mV input

30dB gain. 142mV input.

Basically the same result as Amir got on the reference amplifier from Purifi (0.0006%). But his analyzer does not have the second order problem my analyzer have so we end up on 0,0008%.

We can see remains from the european 50Hz mains frequency at 150Hz, 300Hz, 350Hz at the highest gain setting. The maximum level of these are -132dB.
The signal is 4.47V (13dBV) so the 250Hz noise at -132dB is effectively at -119dBV. This is 1.1uV and will not be audible at any speaker.

 

[Matias]

Could you use an analog notch filter to attenuate the main tone when capturing, so that the ADC harmonics are reduced as well? A classic trick.

 

[KSTR]

Could you use an analog notch filter to attenuate the main tone when capturing, so that the ADC harmonics are reduced as well? A classic trick.

It is notched out, as can be seen from the dip in the noise floor.
The R&S draws thick lines for the (notched) fundamental, and the full fundamental is overlaid with a thin line. Or maybe there are just two thin lines seperated by one pixel.

 

[tmtomh]

It doesn't matter. The FTC rules are clear. I shouldn't need to remind you of them. To sell legally into the US market, they must be compliant. No excuses.

Class D simply fails to perform at high power and high frequencies. We all know that. I know it, you know it, and most other sensible people know it.

Be honest about the rated THD across the audible bandwidth (20-20k), especially at high frequencies and high power.
Be honest about the complete inability to deliver continuous rated full power at HF.
Specify compliant minimum THD from 250mW to full rated power THD, not cherry picked bottom of the plot numbers.

This is nonsense, and you know it - and you don't care because you keep banging this drum over and over... and over and over and over.

If bandwidth to 400kHz is too great, then so is bandwidth to 100kHz. Just because the FTC picked a particular ultrasonic frequency 50-60 years ago does not mean it's relevant to an amp typology that wasn't used at the time (and that did not mature until recently - yes I know Sony introduced a version of Class D in the late 1970s...). If it were illegal to sell Class D amps in the US, then the big brands like NAD and others wouldn't be selling them.

No amplifier - or DAC, or any other piece of equipment - needs to be able to reproduce the 9th harmonic of a 20kHz signal. Or the 6th harmonic, or the 3rd, or the 2nd. Beyond 20kHz, it doesn't matter.

I and many others have said before that you are 100% correct about the top-line wattage rating of Hypex and Purifi modules: 400 watts is indeed misleading. Based on Amir's and others' testing, the full audible bandwidth rating of these higher-end Class D modules should be something like 250w into 4 ohms and 150w into 8 ohms, not 400 into 4 ohms as they say.

But other than that, the rest of what you keep banging on about is irrelevant, misleading, and accompanied by inaccurate, borderline libelous value judgments that you smear these individual amp makers with based on the issue you have with Purifi and Hypex.

 

[David_M]

What is the gain of a Class D amplifier with DC power rails at +/- Vs and maximum input (to get rated power) of Vrms? I tried 20*log(Vs/Vrms) and couldn't get the gain numbers given by Purifi for their 1ET400A module. Does anyone know?

 

[restorer-john]

What is the gain of a Class D amplifier with DC power rails at +/- Vs and maximum input (to get rated power) of Vrms? I tried 20*log(Vs/Vrms) and couldn't get the gain numbers given by Purifi for their 1ET400A module. Does anyone know?

Power supply rails are not constant (unless the supply is regulated). The supply rails will sag significantly at full rated power. Not only that, the losses across the output FETs will increase as the current through them increases.

Also, bear in mind when doing calculations, the maximum voltage swing over the load is the peak of the waveform, not the RMS value used for power calculations.

 

[David_M]

Power supply rails are not constant (unless the supply is regulated). The supply rails will sag significantly at full rated power. Not only that, the losses across the output FETs will increase as the current through them increases.

Also, bear in mind when doing calculations, the maximum voltage swing over the load is the peak of the waveform, not the RMS value used for power calculations.

My mistake, I realize. So what's the gain if peak input is Vp and rails are Vmax? The gain is fixed, regardless of the load current draw or power-supply sag, which for a well-regulated supply, should not be an issue.

 

[Doodski]

Class D simply fails to perform at high power and high frequencies. We all know that. I know it, you know it, and most other sensible people know it.

For me I dislike the non-repairability of the class D system. Either replace the entire module or toss it in the garbage. As a Techy type that's poison for me. I guess it is progress although we will see over the long term how these class D amps hold up.

 

[David_M]

For me I dislike the non-repairability of the class D system. Either replace the entire module or toss it in the garbage. As a Techy type that's poison for me. I guess it is progress although we will see over the long term how these class D amps hold up.

But how often do you repair your standard class A/AB power amplifiers nowadays? And what exactly do you repair, I'm curious, apart from capacitors (maybe) after decades of faithful operation? Modern amplifier designs have large design margins, such that these devices are able to work for many many years without failure.

 

[Doodski]

But how often do you repair your standard class A/AB power amplifiers nowadays?

I don't repair professionally anymore. I've moved on into industrial stuff. When I did repairs I repaired many thousands of class AB amps both home and car audio with PCM power supplies that required repair about as much as the amplifier sections did. I replaced many output transistors, little little annoying fusible resisters, gain stages, heat related issues with the PCBs, power supply and regulation issues, ground issues, broken PCBs, displays, display drivers, key command issues, capacitors and of course lotsa cleaning and lubricate controls is most of the stuff. Sometimes I handled weird issues like turning on in the middle of the night and the radio stations switching with the volume going to full output (Looking at you Sony.) Class AB amps and receivers where very serviceable and where a money maker for those that had the patience to repair them such that they didn't self destruct on power ON. That was back in the days when a tech could order a service manual and get ALL the parts from the manufacturers. Keep in mind that there where many thousands of techs repairing full time and so the amount of damaged and blown gear was huge. Some models and manufacturers have iconic models that stand up to time, last a long time and others are prone to issues and have many service bulletins advising techs and explaining MODS and stuff that was required. I estimate over the 15 years that I repaired about 55,000 units of repair (from several categories of gear not just amps and receivers) and another 8000 units of modification to both mechatronics and electronics. It sounds like a lot, it is a lot and it's not superhuman, the numbers simply add up that way. So... with that kind of quantity you can see that class AB stuff was prone to requiring service in massive quantities. I never saw a class D amp although I heard about a Sony model that was apparently rare as hens' teeth. I have lotsa experience servicing switching power supplies and became very proficient at servicing them. From that experience I have doubts about class D stuff and the power supplies in them but then again now the power supplies in PCs seem to be pretty reliable these days.

 

[David_M]

I bet that if something is to go wrong in a modern class D amp, it'll be in the switching power supply since that's the part that gets the hottest in the amp, notably the output coils, which in turn get too hot and damage the output filter capacitors, which may not be properly rated for the temperatures they'll experience in actual product use. Heat is the enemy of all electronics.

 

[Doodski]

I bet that if something is to go wrong in a modern class D amp, it'll be in the switching power supply since that's the part that gets the hottest in the amp, notably the output coils, which in turn get too hot and damage the output filter capacitors, which may not be properly rated for the temperatures they'll experience in actual product use. Heat is the enemy of all electronics.

When the human factor becomes involved anything is possible. Peeps run the gear at 2-3 Ohms, provide little to no ventilation, put plants on top and water them, spill rum and coke on them with OJ too and generally run the bag off the gear. If guessing I would say that they seem to have the protection circuitry in place now where as some time ago the protection circuitry was pretty basic and power supplies and amplifier circuitry would fry before the amp went into protection mode. I think both the power supply and the amp sections can both go up in smoke under the right circumstances or with a faulty part.