Buckeye Amps: 2nd Generation Purifi EIGENTAKT 1ET9040BA Amplifier

[David_M]

Their FR measurement is without a filter (as noted in the datasheet). We did/do our measurements with a filter (AES 17 standard).

At some point we will send one to Amir, yes. But likely not until we get more Micro Audio SMPS' back in stock.

Ok, the filter to suppress the switching frequencies?

BTW, I know this is not a fair test, but is it possible to have a frequency plot all the way to 1MHz at rated power? OrchardAudio does this custom designed amps on this forum.
It would be interesting to see the distribution of the class D noise. After all, the amp operates without the filter in real life.

 

[Buckeye Amps]

Ok, the filter to suppress the switching frequencies?

BTW, I know this is not a fair test, but is it possible to have a frequency plot all the way to 1MHz at rated power? OrchardAudio does this custom designed amps on this forum.
It would be interesting to see the distribution of the class D noise. After all, the amp operates without the filter in real life.

But why though?

 

[yys310]

The 1ET9040BA module has switching frequency at 600-700kHz and ripple ~280mV.
Similar to 1ET400A

 

[David_M]

But why though?

Investigating the possible use of a special notch filter to remove switching noise and reduce ripple. but also for the fun of seeing it. No one is going to blame you for Purifi's noise performance.

 

[DonH56]

Investigating the possible use of a special notch filter to remove switching noise and reduce ripple. but also for the fun of seeing it. No one is going to blame you for Purifi's noise performance.

It's an interesting idea. I tried a very simple LC notch filter, and also a high-Q resonator filter, to suppress the switching noise for the first class D amp I designed many years ago, but ultimately found the speaker itself had enough inductance to reject the high-frequency noise so it wasn't a problem, and there was no real mixing I saw to introduce modulation distortion in the driver, so I quit trying. The usual output filter reduced the level enough that an additional notch filter didn't help anything. I'd be curious if you could duplicate that sort of research and measurements to see if that is still true today or not given the much higher performance of current class D amplifiers.

I do not know the details of Purify or Hypex modules, but some class D designs do not tightly control the oscillation frequency over process and environmental changes, so a high-Q filter might not be needed or desirable.

 

[pma]

I do not know the details of Purify or Hypex modules, but some class D designs do not tightly control the oscillation frequency over process and environmental changes, so a high-Q filter might not be needed or desirable.

That’s it. Purifi and Hypex do not use constant switching frequency. They are self-oscillating intentionally and the switching frequency is thus not constant at all. High Q notch filter would not work. And is also unnecessary.

 

[yys310]

It's an interesting idea. I tried a very simple LC notch filter, and also a high-Q resonator filter, to suppress the switching noise for the first class D amp I designed many years ago, but ultimately found the speaker itself had enough inductance to reject the high-frequency noise so it wasn't a problem, and there was no real mixing I saw to introduce modulation distortion in the driver, so I quit trying. The usual output filter reduced the level enough that an additional notch filter didn't help anything. I'd be curious if you could duplicate that sort of research and measurements to see if that is still true today or not given the much higher performance of current class D amplifiers.

I do not know the details of Purify or Hypex modules, but some class D designs do not tightly control the oscillation frequency over process and environmental changes, so a high-Q filter might not be needed or desirable.

I connected the oscilloscope to the 1ET9040BA output with and without tweeter before.
The switching noise is the same whether the tweeter is connected or not. Perhaps this is a sign that switching frequency is too high for the inductance of driver coil to response.

That’s it. Purifi and Hypex do not use constant switching frequency. They are self-oscillating intentionally and the switching frequency is thus not constant at all. High Q notch filter would not work. And is also unnecessary.

Is this also the case when no music is playing? No wander I see different switching frequency/noise on the 2 modules.
I should swapped the power supply/signal input to see if the behavior changes.

 

[pma]

Is this also the case when no music is playing?

No. Frequency change is proportional to input signal amplitude, by nature of the self oscillating modulator. Please see below situation with no input and when output is 1kHz/13Vrms.

The wideband spectrum also depends on input signal frequency, so please do not try to make any simplified deductions, the mater is not trivial. Better forget thinking about it, it is a non-issue.

 

[yys310]

The wideband spectrum also depends on input signal frequency, so please do not try to make any simplified deductions, the mater is not trivial. Better forget thinking about it, it is a non-issue.

The 2 modules of mine shows different switching frequency and one of it shows high frequency around 15-17kHz that's why I have concern about.
But the high frequency noise I measured is by MOTU M4 without LPF. Not sure if it's trust worthy.

 

[pma]

The 2 modules of mine shows different switching frequency and one of it shows high frequency around 15-17kHz that's why I have concern about.
But the high frequency noise I measured is by MOTU M4 without LPF. Not sure if it's trust worthy.

I am sorry, but in my opinion your plots show either wiring or measurement error. One needs to avoid any capacitive ground currents (use optical fibres and/or isolation power transformers to supply PC). And, MOTU is far from perfect device. Purifi modules are free from any spuriae in the audio band, if properly wired and measured. BTW, I do not use LPF filter for measurements.

1ET400A noise with Topping DX5 1ET400A noise with input shorted

 

[Sokel]

The 2 modules of mine shows different switching frequency and one of it shows high frequency around 15-17kHz that's why I have concern about.
But the high frequency noise I measured is by MOTU M4 without LPF. Not sure if it's trust worthy.
View attachment 388075View attachment 388076

This tone at about 18kHz smells a lot like a HVAC ot something like it.
I have seen it time and time again when not using my dedicated electrical phase.

 

[pma]

All these very low level measurements need extra care. We are talking about sub microvolt voltages.

 

[pogo]

it shows high frequency around 15-17kHz

Do you have a monitor nearby that could emit this frequency?

 

[yys310]

I am sorry, but in my opinion your plots show either wiring or measurement error. One needs to avoid any capacitive ground currents (use optical fibres and/or isolation power transformers to supply PC).

Already use SPDIF to the DAC. I guess that's also isolated good enough? The DAC (D90III) to MOTU shows nothing on the spectrum.

And, MOTU is far from perfect device. Purifi modules are free from any spuriae in the audio band, if properly wired and measured. BTW, I do not use LPF filter for measurements.

Yeah MOTU's ADC aren't that good and I can see it in one of the module so that's even more concerning.

This tone at about 18kHz smells a lot like a HVAC ot something like it.
I have seen it time and time again when not using my dedicated electrical phase.

But it's strange that both modules are in the same aluminum chassis and I tried to swap the modules power supply, 20pin front end cable, mounting position and it is still the same one that has this behavior.

All these very low level measurements need extra care. We are talking about sub microvolt voltages.

Yeah...perhaps I should just forget about it. I think it won't be audible...

Do you have a monitor nearby that could emit this frequency?

I have a monitor nearby but if that's the case I think I'll see that signal in both modules right?

I'll try to measure it using cosmos ADC in the future. Right now the modules seem to work fine. They sound fantastic regardless of the strange peak anyway.

 

[restorer-john]

But it's strange that both modules are in the same aluminum chassis and I tried to swap the modules power supply, 20pin front end cable, mounting position and it is still the same one that has this behavior.

It will be 16kHz and that's a byproduct of USB. Your interface (whatever it is) and grounding (or isolation) will be the issue.

 

[yys310]

It will be 16kHz and that's a byproduct of USB. Your interface (whatever it is) and grounding (or isolation) will be the issue.

Somehow it only shows at one of the module.
Swap both modules to the same SMPS and same signal input it's still the same module that shows the HF noise.

 

[pogo]

I have a monitor nearby but if that's the case I think I'll see that signal in both modules right?

No.

 

[yys310]

No.

So the interference source will identify certain module and only interfere on it.

 

[McFly]

16kHZ is USB noise. Try different, longer or shorter cables.

 

[Buckeye Amps]

The Micro Audio SMPS1k-SN power supply upgrade is now available to be added during checkout on the website. ETA for the Micro Audio SMPS is 4-6wka but you can preorder now.