• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

DIY Purifi Amp builds

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
Not to give anything away in advance of my published review/measurements of the Purifi EVAL1, but I will say that if the Neurochrome boards actually increase the performance, it will take very specialized equipment to tease that out.

"Do you need 100,000 times lower distortion than you can hear, or 500,000 times lower?"
 
Last edited:

barrows

Active Member
Joined
Jan 24, 2020
Messages
150
Likes
219
I am one of those
I have a pure dual mono of nc400, and have heard in the past the difference between an amplifier with a single power supply and two power supply.
it was clearly my doubt with the Eval1
well no: no deviation in scene or depth with the Eval1 and its simple power supply versus the pure double mono of nc400!
this is the first time that I have an amplifier with a single power supply with such a scene, Purifi has made a real technical feat

I have experimented quite a bit with dual mono vs. single power supply set ups in various scenarios (DIY and other ways). I found, that as long as the power supply is very well designed, and has appropriate low output impedance, and plenty of power headroom, that single power supply amplifiers appear to perform equally to dual mono approaches. At lot of time this is not true in practice, usually because single supply set ups are often compromise din the total amount of power available. For example, a single SMPS 600 driving two NC-400 module sis clearly a compromised set up as the supply is inadequate in power output for two modules. With two Purifi modules, and a SMPS 1200 though, there is more than enough power available from he single supply to easily power both modules to full output, with many watts to spare, especially considering the high efficiency of the Purfi amp modules. It could be interesting to test this though by monitoring the SMPS voltage rails during high level playback, and to actually see if the power supply is being modulated significantly by the power draw.
 

ChrisPa

Member
Joined
Dec 16, 2019
Messages
35
Likes
27
My dual mono application is simple.
My amps are behind my speakers with about 40cm of speaker cable (which will be shortened one day :))

I can't understand long speaker cables

I'm very interested in applying, and the stability of, the remote feedback sensing :cool:
 

mocenigo

Major Contributor
Joined
Dec 8, 2018
Messages
1,274
Likes
1,034
I am one of those
I have a pure dual mono of nc400, and have heard in the past the difference between an amplifier with a single power supply and two power supply.
it was clearly my doubt with the Eval1
well no: no deviation in scene or depth with the Eval1 and its simple power supply versus the pure double mono of nc400!
this is the first time that I have an amplifier with a single power supply with such a scene, Purifi has made a real technical feat

Another application is to provide power independently to the two boards. If one is using a non-hyper power supply (the connex have their own advantages, including lower profile) then one must take care of supply pumping. One countermeasure is to invert inputs and outputs, which can be done also with the FE02 board included in the EVAL1 kit. Alternatively, or in addition, one can add 4700uF per rail on the main supply. Even better, split the 4700uF between the two 1ET400A modules, so for instance take four 2200uF caps and use one per rail per module, with mono adapter boards.

Why? Probably it is overengineering, but at least it is not audiophoolishness like using only silver cabling with pure silk sleeving and cryogenically treated by virgins under a full moon.
 

barrows

Active Member
Joined
Jan 24, 2020
Messages
150
Likes
219
Another application is to provide power independently to the two boards. If one is using a non-hyper power supply (the connex have their own advantages, including lower profile) then one must take care of supply pumping. One countermeasure is to invert inputs and outputs, which can be done also with the FE02 board included in the EVAL1 kit. Alternatively, or in addition, one can add 4700uF per rail on the main supply. Even better, split the 4700uF between the two 1ET400A modules, so for instance take four 2200uF caps and use one per rail per module, with mono adapter boards.

Why? Probably it is overengineering, but at least it is not audiophoolishness like using only silver cabling with pure silk sleeving and cryogenically treated by virgins under a full moon.

BTW, I added additional C (4700 µF per rail) to the Connex supply, but it did nothing to eliminate the pumping. I have moved on to using the SMPS1200, but still with my linear supply for the OPAs/modulator supplies...
 

mocenigo

Major Contributor
Joined
Dec 8, 2018
Messages
1,274
Likes
1,034
BTW, I added additional C (4700 µF per rail) to the Connex supply, but it did nothing to eliminate the pumping. I have moved on to using the SMPS1200, but still with my linear supply for the OPAs/modulator supplies...

Good to know. Did inverting a channel solve the problem (at least for music)?
 

barrows

Active Member
Joined
Jan 24, 2020
Messages
150
Likes
219
I am not willing to do this, as having the black (or white) speaker channel actively driven on one channel is just wrong to me. The easier solution is to use the Hypex PS. Plus unlike you, I need to be able to access the full power output of the Purifi modules without any issues (my speakers are not highly efficient).
 

tomchr

Addicted to Fun and Learning
Audio Company
Forum Donor
Joined
Nov 5, 2018
Messages
984
Likes
2,607
Location
Calgary, Canada
EVAL1 is about 125$, this is 2x149$.
$149 is on par with other Purifi / Hypex buffers.

And it ships from Canada and doesn't clear customs i.e. +24% VAT in my case.
That sounds like something you should take up with your government. I have no control over your taxes.

Tom
 

tomchr

Addicted to Fun and Learning
Audio Company
Forum Donor
Joined
Nov 5, 2018
Messages
984
Likes
2,607
Location
Calgary, Canada
Not to give anything away in advance of my published review/measurements of the Purifi EVAL1, but I will say that if the Neurochrome boards actually increase the performance, it will take very specialized equipment to tease that out.
I agree with that statement. I offer a lot of flexibility and opportunities for tweaking for those who desire that. A mono buffer board makes it much easier to build compact mono blocks or amps with an odd number of channels.

Tom
 

Zoomer

Senior Member
Joined
Oct 5, 2019
Messages
323
Likes
468
I can recommend one of these if you don't like the sensation of electric sound
winding.gif
 

tomchr

Addicted to Fun and Learning
Audio Company
Forum Donor
Joined
Nov 5, 2018
Messages
984
Likes
2,607
Location
Calgary, Canada
Someone on DIY Audio brought up that the gain structure of the Input Buffer (and Purifi EVAL1) isn't optimal, so I'm considering a design change.

I would be curious for you folks' input on this:

By default my Input Buffer has 13.2 dB gain (4.57x). The 1ET400A input specs are (Purifi data sheet Rev. 1.0):
  • Differential input voltage (pos to neg input): 12 V
  • Common-mode voltage: ±5 V
So for differential input, the max input voltage in the default configuration is: 12/4.57 = 2.62 V (or just shy of 2 V RMS).

For the single-ended input, the max input voltage is limited by the common-mode voltage spec. So: 5/4.57 = 1.09 V peak (770 mV RMS).

I can easily change the 'custom' gain option to allow for 4 V differential, 1.5 V single-ended. That's the "fast to market" option. Unfortunately, this option would make the 'custom' gain option a bit harder to use, as you would need to remove an 0805 resistor and replace it with another 0805 resistor to change the gain.

Alternatively, I can redesign the Input Buffer to include a sort of "differential level translation". That would increase cost (maybe by $15), but would mean that the Class D amp always "sees" a differential input - even if the source connected to the Buffer is single-ended. That would be a nice market advantage.
I might be able to shave off a few cents from the board cost by eliminating the two 3-pin connectors for the optional external regulator and just leaving the 6-pin connector/jumper block in place. You can always connect a ribbon cable there if you want an external regulator.

I won't make this a design-by-committee project, but I would rather the Buffer provide good value and satisfy unmet market needs than be a "me-too" product.

Tom
 

barrows

Active Member
Joined
Jan 24, 2020
Messages
150
Likes
219
The gain structure is optimal at 26 dB for the entire amp. Anyone who says it is not is crazy. The standard gain for almost all power amplifiers ranges from 24 dB on the low side, to 30 dB on the high side. 26 dB is perfect.
Tom, please have your standard gain as 26 dB, allow for bypass option, and then users with unusual system requirements for gain can adjust via tuning not he board via resistor changes as you already allow for. Please do not be dissuaded by a few folks who have unusual set ups, outside of the norm (super sensitive speakers, sources with very high output levels, etc), you already support these fringe cases through gain adjustment via resistor changes.
 

tomchr

Addicted to Fun and Learning
Audio Company
Forum Donor
Joined
Nov 5, 2018
Messages
984
Likes
2,607
Location
Calgary, Canada
I will always support 26 dB gain. That's the THX standard, optimal or not.

I do see value in the single-ended to differential conversion for those who use single-ended sources. The question is whether it adds enough value that people are willing to pay more for my board.

Tom
 

mocenigo

Major Contributor
Joined
Dec 8, 2018
Messages
1,274
Likes
1,034
I am not willing to do this, as having the black (or white) speaker channel actively driven on one channel is just wrong to me. The easier solution is to use the Hypex PS. Plus unlike you, I need to be able to access the full power output of the Purifi modules without any issues (my speakers are not highly efficient).

Why wrong? Are you chain connecting a sub or do you have Polk speakers (those that cross mixed the signals of the two channels out of phase)?
 

barrows

Active Member
Joined
Jan 24, 2020
Messages
150
Likes
219
Why wrong? Are you chain connecting a sub or do you have Polk speakers (those that cross mixed the signals of the two channels out of phase)?

Red speaker terminal means "hot" or actively driven, white/black means "cold" (or inverted in the case of balanced output amplifiers). Any other way of doing it is "wrong". For example, if one were to sell an amp built this way, it could result in problems for the buyer, and it is certainly confusing. Of course this does not mean that a DIYer cannot do it that way, it just rubs me the wrong way to have an amp built incorrectly. And yes, I do typically connect my sub to the speaker outputs, and yes that of course creates another problem. Additionally, given how little I needed to turn up the amp with the Connex supply before it started having problems, I am not confident that it is right for my set up (I wonder if it really meets its spec at this point). Add to that that I have had very good results over a bunch of years now with my Hypex SMPS600s, and I just figure the best option for me is to use Hypex.
It is just my preference to build my amplifier "correctly", with red binding posts being actively driven. I understand the approach, I just do not agree with it. For me, the better solution so to use the Hypex supply, and have no worries, but I will still use my linear supply for the +/- 18 VDC rails.
 

ChrisPa

Member
Joined
Dec 16, 2019
Messages
35
Likes
27
I will always support 26 dB gain. That's the THX standard, optimal or not.

I do see value in the single-ended to differential conversion for those who use single-ended sources. The question is whether it adds enough value that people are willing to pay more for my board.

Tom
Isn't your buffer already inherently a single-ended to differential convertor?

For those who already understand the benefits of differential, an additional $15 sub circuit isn't going to offer anything.

So I guess it's down to what proportion of your customers will use single-ended sources. And can't everyone implement a custom gain anyway with the design as it is, however many ends their sources have ;)

Edit: quoted wrong post
 

tomchr

Addicted to Fun and Learning
Audio Company
Forum Donor
Joined
Nov 5, 2018
Messages
984
Likes
2,607
Location
Calgary, Canada
Isn't your buffer already inherently a single-ended to differential convertor?
My Universal Buffer has a differential input and offers two outputs: differential and single-ended. Its input can be configured for single-ended operation by moving a jumper. So, yes, the Universal Buffer can work as a single-ended to differential converter.
My Purifi/Hypex Input Buffer is a differential in/differential out buffer. If you provide it with a single-ended input, it will provide a single-ended output - just as the Purifi EVAL1 or many of the other buffers on the market. What I am currently considering is to include a proper differential output, thus feature proper single-ended to differential conversion in the Purifi/Hypex Buffer.

For those who already understand the benefits of differential, an additional $15 sub circuit isn't going to offer anything.
True. Those who use differential sources will not see a benefit.

So I guess it's down to what proportion of your customers will use single-ended sources.
I agree. I suspect the vast majority use single-ended sources.

And can't everyone implement a custom gain anyway with the design as it is, however many ends their sources have ;)
Also correct.

I realized I had an error in the math earlier. The bottom line is this: If you want the full output power of the Purifi 1ET400A, you have to provide 9.6 V RMS (13.6 V peak) to the Purifi module. This is not possible to do with a single-ended source, as 13.6 V exceeds the 12 V max spec of the Purifi module. With a single-ended source the max output power will be about 340 W into 4 Ω (theoretically - in practice more like 300-310 W). To get the full 400 W, you'll need to use a differential source.

Or... If I add the "level shifter" I talked about earlier, I can make the output of the Input Buffer differential regardless of whether you provide a single-ended or a differential input signal. I think that idea has merit. Of course, 300 W and 400 W are both damn loud, so the difference is (mostly) academic, but some will definitely care and want all the power they paid for.

Tom
 
Last edited:

tomchr

Addicted to Fun and Learning
Audio Company
Forum Donor
Joined
Nov 5, 2018
Messages
984
Likes
2,607
Location
Calgary, Canada
...and just like that, Revision 2.0 of the Purifi/Hypex Input Buffer was born. I added a proper differential output. This will allow folks with single-ended sources to drive the amp to clipping, which is something the Purifi EVAL1 (and similar buffers) cannot do.

Added components -> added cost. I don't know exactly how much. Until I've done the math, I'll leave the prices alone and will honour the pricing on all orders placed.

I wasn't able to fit in the legend that showed the jumper settings, so you'll have to ... (gasp) ... read the instructions for that.: :)

I'm expecting to hit my mid/late April target for having boards in stock.

Tom
 

Attachments

  • Screen Shot 2020-03-20 at 22.18.35.png
    Screen Shot 2020-03-20 at 22.18.35.png
    58.2 KB · Views: 235
Top Bottom