DIY Purifi Amp builds

[tomchr]

The best would be if @tomchr could make an adapter board for the pin output <> universal buffer connection.

I'm looking into making such an adapter board.

Does it take any part in the feedback loop? I've read that the soldered sockets are part of the feedback loop, so I guess it has something to do with it.

It should. The Purifi modules have remote sensing, so bringing that out to the speaker connections would make a lot of sense. That's what I've done in my Hypex NC500/Purifi 1ET400A adapters that I've made for OEMs.

Tom

 

[JimB]

I'm looking into making such an adapter board.


It should. The Purifi modules have remote sensing, so bringing that out to the speaker connections would make a lot of sense. That's what I've done in my Hypex NC500/Purifi 1ET400A adapters that I've made for OEMs.

Tom

Purifi 1ET400A adapters for OEMs? Do tell...

 

[tomchr]

Purifi 1ET400A adapters for OEMs? Do tell...

If you'd like me to design one for you, just toss me an email. Take my username here and add @neurochrome.com.

The adapter I'm thinking of putting together will be better suited for the DIY crowd.

Tom

 

[mocenigo]

The best would be if @tomchr could make an adapter board for the pin output <> universal buffer connection. But what kind of features does the EVAL1 board do actually, with bypassed gain? Does it take any part in the feedback loop? I've read that the soldered sockets are part of the feedback loop, so I guess it has something to do with it.

Well, then any connection boards must connect the speaker out signals to the sensing inputs, that's all. Not a very difficult thing to do

 

[JimM]

What does every one expect to gain by a new buffer board, other than a single board per amp for mono use, and possible better placement which I would go for? Please do not think I'm denigrating the Neurochrome design, I have used the 45/49 series opamps as replacement of other types and they very good sounding Opamps. The main thing is probably the power supply implementation used in the amps. But in my case I have bypassed the front end buffer and running with no gain into the ET400 via the EVAL unit. Does the ET400 use opamps and if so what are they? In fact I am driving the Eval board with a Benchmark HGC Dac that uses the 45 series opamps. I would not expect to see much improvement with a different buffer board using same amps as the benchmark. A different buffer using discrete fets would be a good test to see if any real difference can be heard. I personally think very good opamps with proper implementation are as good or better than discrete and have a more neutral and transparent sound. Which in my opinion is a reason a lot of people prefer the discrete preamps. Please understand I am not trying to start a problem. but I do have question as to why people are wanting a different buffer, do they not like the sound they are getting from the EVAL and ET400? I also understand this is what DIY is all about. Thanks again for all the input on this forum it is a great discussion and help, so far my impression of the EVAL unit and the ET400 is very close to the Benchmark AHB2 I owned but with more power. This is not what I expected from a class D amp having always owned or built AB or class A. Very pleased so far for a DIY for around $1200 that competes or exceeds $3000 amps.

 

[JimB]

If you'd like me to design one for you, just toss me an email. Take my username here and add @neurochrome.com.

The adapter I'm thinking of putting together will be better suited for the DIY crowd.

Tom

Thanks, @tomchr. It was a tease. I am DIY. But, I do have opinions about what would be best for such a board... And interest in purchasing a few, if suitable and available at the right time and price.

 

[JimB]

... Does the ET400 use opamps and if so what are they? In fact I am driving the Eval board with a Benchmark HGC Dac that uses the 45 series opamps. ...

Sure. You have a couple, right? Take a look at the little carrier board - there's an OPA1612A there.

 

[hyperknot]

What does every one expect to gain by a new buffer board, other than a single board per amp for mono use, and possible better placement which I would go for?

Actually, I'd be thinking in a stereo build, here are some of my objective points:
1. The EVAL board isn't exactly cheap. It costs 123 USD of you subtract the prices inc. VAT on the website. The Universal Buffer is sold for $149. (OK the sockets are included in the EVAL, so not exactly straight comparison).
2. It's exactly the gain stage which looses 4 db in the SINAD value of the amp, dropping it behind Hypex NC400's stock buffer.

3. The universal buffer might be a state of the art buffer for not much more money, bringing the level of 1ET400 close to the no gain measurement.

+1 @tomchr is one of the very few entrepreneurs who create no-BS DIY audio modules with solid engineering and objective performance, I'd like to support him by purchasing his product.

 

[JimB]

What does every one expect to gain by a new buffer board, other than a single board per amp for mono use, and possible better placement which I would go for? ...

Those are the high points for me. For me, "mono" use includes configurations like 3 channels in a box (or more). Also, relatively easier gain change (for example, by jumpers or by paralleling with a good through hole resistor).

 

[JimB]

...
2. It's exactly the gain stage which looses 4 db in the SINAD value of the amp, dropping it behind Hypex NC400's stock buffer.
...

Implying that with the Neurochrome configured for 13 dB of gain added as a front end to the 1ET400A, it would then measure 108dB in Amir's test? Maybe. Maybe not. I'll predict not. It's still an added gain stage (adds some noise, even if unity gain). What is the best SINAD Amir has measured for any amp in this power class with ~27dB total gain?

Thinking it through another way, for your use, how low is low enough noise for you? Are you using extremely efficient speakers? If so, using no extra gain, and accepting lower power output from 'limited' sources, would be best.

Added note: The EVAL1 board nets about $100 in the US: $300 amp module, vs $700 for a pair + EVAL board ($700 - 2*$300 = $100). I am looking forward to some good alternatives, but, if we are going to be using gain, I don't expect a significant improvement in SINAD over the EVAL1 - certainly not as much as 3 dB. As always, the minimum gain for the job is the winner, and I'm now leaning (again) towards reducing the gain of mine for my application (power stages for AVRs). More justification later...

 

[JimM]

I also agree with different gain changes per buffer, I tried reducing the DAC gain by -10db leaving EVAL in buffer bypass but had to go back to 0 db on DAC to get sufficient gain. So right now the DAC volume runs from 9 o'clock to 11 o"clock ideally I would prefer 1 to 4 o'clock. But this works ok just have to be careful to watch the level. The Benchmark DAC has a mode that puts gain into line out rather than variable so this could be a problem in this mode upon power with active source as it would get really loud. I still would like for peoples impression of how the amp sounds in their systems. After all it's the music we are after.

 

[JimB]

I also agree with different gain changes per buffer, I tried reducing the DAC gain by -10db leaving EVAL in buffer bypass but had to go back to 0 db on DAC to get sufficient gain. ...

I'm surprised that 10 dB is too big a change to work for you and forces you to run from 9-11 on the dial. My thinking on a preferred gain is still what I described, earlier, and what several "wished for": 20 dB, total.
https://www.audiosciencereview.com/...iy-purifi-amp-builds.10478/page-8#post-296784

Unfortunately, as noted, reducing gain means increasing resistance, so one cannot just 'tack' an additional R in parallel to what is there - you must replace one. The parts are cheap enough, and readily available, but a PITA to replace "by hand". I've done it on resistors one size up, but in a tighter location (the Hypex NC400) - a task made worse by a vision issue from a previous head injury. Still, I'll have to get some parts in and have a go.
https://www.digikey.com/product-detail/en/ERA-3AEB5361V/P5.36KDBCT-ND/3075997/?itemSeq=315489480

 

[mocenigo]

Implying that with the Neurochrome configured for 13 dB of gain added as a front end to the 1ET400A, it would then measure 108dB in Amir's test? Maybe. Maybe not. I'll predict not. It's still an added gain stage (adds some noise, even if unity gain). What is the best SINAD Amir has measured for any amp in this power class with ~27dB total gain?

Thinking it through another way, for your use, how low is low enough noise for you? Are you using extremely efficient speakers? If so, using no extra gain, and accepting lower power output from 'limited' sources, would be best.

Your arguments resonate with me. My speakers are very efficient (about 95Db/w/m) even though not extremely efficient - there are 110 Db/W/m monsters out there. But I am quite sure that the EVAL1 with those speakers would not produce significant noise. If anything, such a configuration would be cruel to anything upstream – almost unjustifiably so – whose slightest noise would be amplified and audible.

There are two solutions: use good inline attenuators – the SHURE A15AS are apparently quite good, and I found also the way less expensive JTS MA-123 to be completely transparent to my (fallible) ears – or reduce the gain of the system. For the second scenario either bypass the buffers if your source can drive 2k2 SE or 4k4 BAL well, or reduce the gain by changing SMD resistors as described in the EVAL1, or use a different buffer where the gain can be more easily set.

With such a system one would be using the amplifier at very low output power anyway, and therefore the efficiency would drop to 50-70%. Not too far from less powerful and well designed AB/AH amplifiers. So, again, an amp like the Benchmark AHB2 would perhaps become a more attractive proposition, or perhaps even a simpler DIY class A design. W.r.t. the Benchmark, the Purifi solution wins on price, and w.r.t. class A the Purifi will win on measureable distortion and on efficiency, but the savings may not be that great considering the total ownership costs. YMMV.

 

[mocenigo]

What does every one expect to gain by a new buffer board, other than a single board per amp for mono use, and possible better placement which I would go for?

DIY is also about having fun. Testing a different buffer is part of that. Also, I trust Tom Christensen's design skills. But, for me, most importantly, to give the buffer a gain of 6-7 Db, so that the total gain will be 19-20Db. This way, 4Vrms will drive the amps to full power, which is fine with any modern pre, and also my DAC (not that I need that, but psychologically at least I *can* do that, even though if I did I would probably break my windows, with speakers of 95Db/w/m of sensitivity in a small room). I believe that the old DIN standard of "set the gain so that 1V drives the power amp to full power" belongs to a different era (it is even european standard IEC 60268-2) when sources were weak.

 

[tomchr]

My original idea for an adapter board was to provide a breakout board. This would allow you to experiment with different buffers, regulators, whatnot. It would also allow greater flexibility in the placement of the Purifi modules, including the option of building mono amps.

So here's a crazy thought: What if this breakout board also included a bypassable buffer with, say, 7.2 dB gain with an option to increase the gain by adding a resistor? This would allow those who want to play with buffers, regulators, whatnot to do so and those who want a plug-n-play solution to have that. The only drawback is that adding the extra buffer parts probably add ~$20-25 to the price of the board.

I found myself mulling over how one would use a breakout adapter board with my Universal Buffer. I found it silly that those who are building mono blocks should have to buy two stereo buffers and only use one channel of each. I mean... I don't mind selling more Buffers, but it seems a bit wasteful.

Tom

 

[mocenigo]

My original idea for an adapter board was to provide a breakout board. This would allow you to experiment with different buffers, regulators, whatnot. It would also allow greater flexibility in the placement of the Purifi modules, including the option of building mono amps.

So here's a crazy thought: What if this breakout board also included a bypassable buffer with, say, 7.2 dB gain with an option to increase the gain by adding a resistor? This would allow those who want to play with buffers, regulators, whatnot to do so and those who want a plug-n-play solution to have that. The only drawback is that adding the extra buffer parts probably add ~$20-25 to the price of the board.

I found myself mulling over how one would use a breakout adapter board with my Universal Buffer. I found it silly that those who are building mono blocks should have to buy two stereo buffers and only use one channel of each. I mean... I don't mind selling more Buffers, but it seems a bit wasteful.

On one hand this idea is great. On the other hand what would people that already have the Universal Buffer do? Of course the perfect solution would be to have two versions, one purely passive and one with a little buffer, in a perfect world where everything is easy including managing stock and orders.

Maybe providing a mono version of the universal buffer is a better option? After all, to connect them one needs just three very short wires (per channel) and a little screwdriver. Of course, since only the balanced out is used, then we would not be using the single ended out. I suppose that the buffer in this hypotetical adapter board would use the same circuit as the Universal Board except the circuitry for the single ended part would be removed, making it potentially cheaper (and maybe even cleaner?!?).

I fear I have just added more doubt...

 

[tomchr]

Nah. The buffer on the adapter board would be a dual opamp and a handful of passives. That's much simpler than half of a Universal Buffer.

Folks with the Universal Buffer could still use it. Just bypass the buffer on the adapter board by moving two jumpers, and off they go.

Tom

 

[hyperknot]

@tomchr I think the main use case would be the following:
- Connect a 4V balanced input to the 1ET400
- Connect a 2V unbalanced input to the 1ET400

I don't understand the discussion about disabling gains and high sensitivity speakers and using specialized DACs like Benchmark DAC2 HGC, those are tiny tiny miniscule use cases. (I know disabling the gain doesn't add cost, I just want to prove my use case as well).

Most users don't have 95+ dB speakers (for example my Harbeth is 83 dB), nor specialized DACs, and want a high quality gain stage allowing them to use the 1ET400 modules possibly at the maximum performance, with all kind of DACs, with all kind of output impedance.

As long as your adapter board could bring better than EVAL1 performance, (which the Universal Buffer probably can), then I think it'd be an even better product for Purifi users, and probably very popular. One note: myself, as many other folks doing Class D "DIY" are not really doing DIY, we are just putting modules in a case. I think adapter board would need to be a populated board similar to the Universal Buffer.

Also, it'd probably be a big advantage if it could use the same SMPS as the ones being used for the 1ET400 modules.

 

[tomchr]

I don't understand the discussion about disabling gains and high sensitivity speakers and using specialized DACs like Benchmark DAC2 HGC, those are tiny tiny miniscule use cases. (I know disabling the gain doesn't add cost, I just want to prove my use case as well).

Most users don't have 95+ dB speakers (for example my Harbeth is 83 dB), nor specialized DACs, and want a high quality gain stage allowing them to use the 1ET400 modules possibly at the maximum performance, with all kind of DACs, with all kind of output impedance.

True that. But at the same time, I would also think that the average DIYer would be interested in having an amp that doesn't blow the speakers if you breathe heavily on the volume control.

Why anybody with 95+ dB efficient speakers is using Class D in a residential setting is beyond me. For PA use, yes. But in a living room? That blows my mind (and ears).

As long as your adapter board could bring better than EVAL1 performance, (which the Universal Buffer probably can), then I think it'd be an even better product for Purifi users, and probably very popular. One note: myself, as many other folks doing Class D "DIY" are not really doing DIY, we are just putting modules in a case. I think adapter board would need to be a populated board similar to the Universal Buffer.

My original plan was to give DIYers a modding platform. DIYers like to put their own flair on things. I then though that it would be pretty easy to accommodate the folks who put modules in boxes as well. As you point out, that's possibly a sizeable chunk of the market.

Also, it'd probably be a big advantage if it could use the same SMPS as the ones being used for the 1ET400 modules.

Oh, it would.

Tom

 

[mocenigo]

Nah. The buffer on the adapter board would be a dual opamp and a handful of passives. That's much simpler than half of a Universal Buffer.

Folks with the Universal Buffer could still use it. Just bypass the buffer on the adapter board by moving two jumpers, and off they go.

IMO the best option would be to have a board with the simple on board buffer you just described, configured to provide, say, ~7Db of gain and space for a thru hole resistor to increase the gain.

Of course you can also provide a version WITHOUT the SMD components of the buffer and maybe two bridges in place of the bypass jumpers, for those that already know they will not be using the buffer anyway, if you so choose, and that could be at a reduced price, but the PCB would be otherwise identical. This would allow you to cover all use cases without having to make two completely different SKUs (you still have two SKUs though).