DIY Purifi Amp builds

[mocenigo]

Am I having serious problems? What are they? Thermal pictures were not posted to illustrate problems.

I do not know. Reaching 75° with the case open is a bit high. It is not a Class A design. Just ask Purifi.

 

[TimoJ]

If you're using balanced inputs, then simply short + to -
Ground has no meaning/relevance to the balanced input
... But you should get the same result - silence

I did connect all together and the result was silence (exactly the same as with pre-gain disabled with jumpers).
So what does that tell? Too much gain?

btw. earlier I wrote that with pre-gain disabled I could still hear some hissing near the tweeter. That is false, I had DDRC-88A connected then (and muted, it emits much more noise unmuted) and it was coming from it. With just EVAL1 connected, I can't hear any noise with my ear inside the tweeter horn.

 

[mocenigo]

I did connect all together and the result was silence (exactly the same as with pre-gain disabled with jumpers).
So what does that tell? Too much gain?

btw. earlier I wrote that with pre-gain disabled I could still hear some hissing near the tweeter. That is false, I had DDRC-88A connected then (and muted, it emits much more noise unmuted) and it was coming from it. With just EVAL1 connected, I can't hear any noise with my ear inside the tweeter horn.

Sounds like a bit too much gain, and perhaps a bit too much noise upstream.

 

[Karu]

Or this 4 channel monster from Audiophonics:
https://www.audiophonics.fr/en/powe...oies-class-d-purifi-4x400w-4-ohm-p-14452.html

Thinking of Bi-amping an F208 with this - too much power?

 

[ChrisPa]

Sounds like a bit too much gain, and perhaps a bit too much noise upstream.

@TimoJ
Too much noise upstream. Yes

 

[wineds]

Regarding input stage gain, the new manual also gives us what we need, and is essentially what @pos described for the NC500 evaluation board. But in contrast to that board, the FE02A has a low pass input filter, as noted by @March Audio. It seems awfully high, though (16MHz without source impedance?). Any gain rolls off above 48KHz.

The following has been updated per Purifi, Rev. 1.11, Jan. 4, 2020.

View attachment 44613

R17 is a 0603 part, and Purifi uses and recommends the Panasonic ERA-3A series. So, for less buffer "pregain" (7 dB, say, instead of 14.6, to give a total gain of ~20dB), one would need to remove R17 and replace it with 5.36K (using available 0.1% tolerance).

A picture of the confirmed location:
View attachment 44526
Not trivial to replace.

Are they any bandwidth implications if the gain of the input buffer is changed as described above?

 

[TimoJ]

@TimoJ
Too much noise upstream. Yes

The noise is there without anything connected, so how is it upstream? I do know that the connected device, MiniDSP DDRC-88A is a little noisy, but surely the amp should be silent without it?

But I just read @JimB earlier message and tested with 10k between XLR input's + / - and the idle noise without source connected is gone. With 20k there is still some noise left, but it's about the same level as with my previous amplifier. So I used 15k resistors, soldered inside the XLR connectors, later I will install those below EVAL1 PCB.

Next I should reduce pre stage gain a little bit, maybe for 3-3.5V input.

 

[JimB]

... but surely the amp should be silent without it?

But I just read @JimB earlier message and tested with 10k between XLR input's + / - and the idle noise without source connected is gone. With 20k there is still some noise left, but it's about the same level as with my previous amplifier. So I used 15k resistors, soldered inside the XLR connectors, later I will install those below EVAL1 PCB.

Note. For the audio band, the impedance of your source should be low enough not to need to reduce the input impedance if the FE02B. I would not use this technique to reduce noise as it should not have a significant benefit with a source connected. In fact, it could actually increase distortion and reduce bandwidth in extreme cases. I would not be bothered by the noise of an open input.

DC offset is a different matter. With an AC coupled source, you might want to reduce input impedance as described before.

Stage gain optimization certainly can improve total system noise when the power amp has more gain than needed to achieve desired output from the intended source.

 

[TimoJ]

Note. For the audio band, the impedance of your source should be low enough not to need to reduce the input impedance if the FE02B. I would not use this technique to reduce noise as it should not have a significant benefit with a source connected. In fact, it could actually increase distortion and reduce bandwidth in extreme cases. I would not be bothered by the noise of an open input.

DC offset is a different matter. With an AC coupled source, you might want to reduce input impedance as described before.

Stage gain optimization certainly can improve total system noise when the power amp has more gain than needed to achieve desired output from the intended source.

MiniDSP specs says: Output impedance balanced: 1.12 k (560 ohm each signal pin)
So (in theory) is this the equivalent if I connect a 1.12k resistor between +/- pins? i.e. Purifi amps should be silent (since they are with 10k) and all the noise is from the DDRC88A? Looks like DDRC88A is AC coupled.

But isn't EVAL1's input impedance still quite high when compared to other power amps? Like 2-3x higher.

 

[ChrisPa]

The noise is there with anything connected<snip>

Do you mean without anything connected?

 

[TimoJ]

Do you mean without anything connected?

Yes. Nothing is connected, noise is there.

 

[ChrisPa]

Yes. Nothing is connected, noise is there.

So with infinite source impedance it produces noise - picked up through input traces possibly. This isn't how it's intended to be used/will be used in real life
With zero source impedance it's silent

If you want a final sanity check, connect 1.2k between + and -
It should still be silent. With finite source resistance it should be silent.

If not silent, then there's a problem with the amp (I'd be surprised)
If silent, then noise is being generated upstream

 

[TimoJ]

So with infinite source impedance it produces noise - picked up through input traces possibly. This isn't how it's intended to be used/will be used in real life
With zero source impedance it's silent

If you want a final sanity check, connect 1.2k between + and -
It should still be silent. With finite source resistance it should be silent.

If not silent, then there's a problem with the amp (I'd be surprised)
If silent, then noise is being generated upstream

Ok, thanks! It was already silent with 10k, so 1.2k probably will be just more silent, but I'll test it. Looks like I need to reduce input impedance and reduce gain to get the best results with DDRC-88A. Or I may just continue to use bypass mode, zero noise problems and I still get enough power for my usage.

 

[JimB]

MiniDSP specs says: Output impedance balanced: 1.12 k (560 ohm each signal pin)
So (in theory) is this the equivalent if I connect a 1.12k resistor between +/- pins? i.e. Purifi amps should be silent (since they are with 10k) and all the noise is from the DDRC88A? Looks like DDRC88A is AC coupled.

But isn't EVAL1's input impedance still quite high when compared to other power amps? Like 2-3x higher.

A high input impedance is generally better. There CAN BE an issue with DC offsets using AC coupled sources. It depends on the specific parts. I had channels with negligible offset and others that were considerably higher.

Don’t worry about open input noise- you’ll never listen to music that way. Since you know the amp set is quiet with a quiet source (example is your input resistors), any noise you hear with a source connected is from the source. In that case, using the minimum required amp gain is best (lowest noise).

 

[ChrisPa]

Ok, thanks! It was already silent with 10k, so 1.2k probably will be just more silent, but I'll test it. Looks like I need to reduce input impedance and reduce gain to get the best results with DDRC-88A. Or I may just continue to use bypass mode, zero noise problems and I still get enough power for my usage.

Leave the input impedance alone.

Look at the amp's gain as part of overall system gain/signal-to-noise

See if you can increase the signal to noise ratio of the preceding stage

 

[TimoJ]

Leave the input impedance alone.
Look at the amp's gain as part of overall system gain/signal-to-noise
See if you can increase the signal to noise ratio of the preceding stage

There probably isn't much I can do to the DDRC88A that is now the preceding stage. I may try it's 8V output level and check if it just adds more noise or if SNR gets better. Probably it's just more noise.
I do intend to upgrade my preamp to a model with build-in Dirac, so at that time everything changes. I'll leave R17 change to that time and just use pre-gain stage bypass.

I also have to do comparisions to my old amp to verify few differences I'm noticing. Not sure if I'm imaging things...

 

[mk1classic]

What to expect when sending a +12v signal to the Hypex SMPS1200A400 J5 pin1 SMPS Standby" Input ?
I have tried to give Pin 1 on the J-5 connector 12v signal but no reaction or turning off power in the SMPS or the EVAL1 card.

 

[wineds]

My EVAL1 one was delivered today. A question about the back panel XLR hole size. It looks like 22mm diameter is required but two adjacent holes of this size will surely run into each other? The gap between XLRs seems to only be a 1mm of so?

 

[ChrisPa]

My EVAL1 one was delivered today. A question about the back panel XLR hole size. It looks like 22mm diameter is required but two adjacent holes of this size will surely run into each other? The gap between XLRs seems to only be a 1mm of so?

Is that a problem?

 

[JimB]

Is that a problem?

https://www.audiosciencereview.com/...y-purifi-amp-builds.10478/page-16#post-308190

https://www.audiosciencereview.com/...y-purifi-amp-builds.10478/page-24#post-330480