Well, the Purifi evaluation board is an evaluation board, and the buffer is for evaluation and study purposes as well. You could say that it is an "evaluation buffer" and not a "production buffer". Read here March Audio post on the Purifi buffer on ASR: "lack of any feedback capacitors to limit the amp bandwidth. Its wide open. There is also no input filtering."
I do not have the skills to modify the board in order to add these features. Also, bypassing the buffer gives you a board with an input impedance of 2200Ohm, which is quite low for an amplifier.
The Neurochrome board comes from a designed that has a great pedigree and that knows the LME49724 extremely well, to the point that his Modulus-686 amp is sort of a legend.
With an input impedance of 48 kΩ it will be quite easy to drive, and with an output impedance < 0.1Ohm it will have no problems driving the purify modules. I am even thinking of keeping to at unity gain, so instead of a power amp I have a power follower, or to give it a gain of just 6 Db, so that the modules will swing to full power with an input of 4V.
Yes, it is 140 dollars and then I need the power supply, but TBH a pair of boutique opamps cost more than the full Neurochrome board, and it will come to me already assembled. ergo at HTguide is experimenting with various types of buffers and this is another option for people with more time (I am carefully managing my time because my work is quite demanding).
Roberto
Where can I find more information? (out of curiosity). Anyway, if it has a high input impedance, and low output impedance, then it can be used as a buffer.
ummm...
Marchaudio's criticism was of the eval buffer for the NC-500, not the eval buffer of the Purifi amps. The Purifi input buffer does include input filtering for RF, unlike the NC-500 eval buffer. Mr. Putzeys has also paid special attention to the sensitivity to RF of the Purifi modules, as he was aware that this could be a problem with Ncore in some situations (with some DACs in other words, and DSD especially).
Although I do agree that it is likely the Purifi eval input gain stage (as this is not a unity gain device, i hate calling it a buffer) could likely be improved upon, i just do not think the Neurochrome board will do so. The 1612 OPA is very, very good.
I would expect that a really talented engineer could do better though, with a discrete implementation, such as the discrete input stage used by Mr. Putzeys on the Mola Mola Kaluga amplifiers (a discrete gain stage using Bruno's "single-ended differential" approach). If one is willing to pay some dollars for different input gain stage, I am interested in something really special, not just another IC OPA approach. BTW, if one is looking for a DIY gain stage solution based on ICs, I would recommend the a24, it is very flexible and could be configured to suit the gain requirements of the Purifi modules: it also has differential input and will convert single ended inputs to balanced, and includes an RF filter on its input.
Thanks for sharing your findings, and your implementation with the Connex SMPS, very interesting. I am going to order the eval kit soon and start experimenting. Maybe I will do something similar, and build a nice linear supply for the +/- 18 VDC input section. I have been using my NC-400 stereo amp here for many years (also have an NC-1200 based amp running right now) and am itching to get something fresh going, and have been inspired by the reports of the sound quality of the Purifi modules.
Marchaudio's criticism was of the eval buffer for the NC-500, not the eval buffer of the Purifi amps. The Purifi input buffer does include input filtering for RF, unlike the NC-500 eval buffer. Mr. Putzeys has also paid special attention to the sensitivity to RF of the Purifi modules, as he was aware that this could be a problem with Ncore in some situations (with some DACs in other words, and DSD especially).
Although I do agree that it is likely the Purifi eval input gain stage (as this is not a unity gain device, i hate calling it a buffer) could likely be improved upon, i just do not think the Neurochrome board will do so. The 1612 OPA is very, very good.
I would expect that a really talented engineer could do better though, with a discrete implementation, such as the discrete input stage used by Mr. Putzeys on the Mola Mola Kaluga amplifiers (a discrete gain stage using Bruno's "single-ended differential" approach). If one is willing to pay some dollars for different input gain stage, I am interested in something really special, not just another IC OPA approach. BTW, if one is looking for a DIY gain stage solution based on ICs, I would recommend the a24, it is very flexible and could be configured to suit the gain requirements of the Purifi modules: it also has differential input and will convert single ended inputs to balanced, and includes an RF filter on its input.
Thanks for sharing your findings, and your implementation with the Connex SMPS, very interesting. I am going to order the eval kit soon and start experimenting. Maybe I will do something similar, and build a nice linear supply for the +/- 18 VDC input section. I have been using my NC-400 stereo amp here for many years (also have an NC-1200 based amp running right now) and am itching to get something fresh going, and have been inspired by the reports of the sound quality of the Purifi modules.
Last edited:
^
Nice input barrows. I posted about the UGS stage as a possible contender for a fully discrete and balanced (SuperSymmetry) if it was useful and could be implemented. I don't want to market the preamp project because that is a full preamp and not just the gain stage, but the gain stage could potentially be optimized as input buffer for the Purify. No Op-Amp, rather JFet's cascaded to some BJT's and MOSFET's.
I don't know enough here so can't say or judge. So for that part you guy's - does active in this thread - probably knows more than I do, which is great
Nice input barrows. I posted about the UGS stage as a possible contender for a fully discrete and balanced (SuperSymmetry) if it was useful and could be implemented. I don't want to market the preamp project because that is a full preamp and not just the gain stage, but the gain stage could potentially be optimized as input buffer for the Purify. No Op-Amp, rather JFet's cascaded to some BJT's and MOSFET's.
I don't know enough here so can't say or judge. So for that part you guy's - does active in this thread - probably knows more than I do, which is great
yes...
I am sure it will not be bad! It will be interesting to see if it betters the eval board by any degree.
I am sure it will not be bad! It will be interesting to see if it betters the eval board by any degree.
Honestly, I think it will. The eval board has a very minimalistic buffer, whereas the Neurochrome buffer has a completely different design (yes, the "block diagram" on Tom Christensen's page is so minimalistic that it tells nothing, and he is right to do so, otherwise a host of chinese would have a copy on AliExpress by tomorrow) but it has two OPA on the the path, the LM49720 (IIRC same as the LM4562) and the LM49724. I do not know whether they are cascaded (which can increase the bandwidth, and indeed the claimed bandwidth is very large, maty will appreciate that) or a different topology. But it inspires confidence.
(Also, I need confidence. I have spent about 250 USD worth of buffer and power supply)
xrk971 is designed one to the new class A, Alpha Nirvana 39w 8 Ohms.
Aleksandar has one with different implementations. PCB, PCB assembled and in case.
* Only soft start: PCB for Soft Start (inrush current limiter) for toroidal transformers | ATL Audio Ltd.
* With DC Blocker too: Combined module Soft Start (inrush current limiter) & DC Blocker (trap, filter) – assembled and tested PCB | ATL Audio Ltd.
Aleksandar has one with different implementations. PCB, PCB assembled and in case.
* Only soft start: PCB for Soft Start (inrush current limiter) for toroidal transformers | ATL Audio Ltd.
* With DC Blocker too: Combined module Soft Start (inrush current limiter) & DC Blocker (trap, filter) – assembled and tested PCB | ATL Audio Ltd.
Yup...
Well the only way to know is to try it! The Neurochrome has a lot of potential uses, so no worries if you find out it is not any better than the eval board's gain stage. I look forward to hearing what you think. The 1612 alone has lower noise and distortion by spec than the 49720, for what ever that might be worth, but I hear ya as far as what could be over-simplicity on the eval board.
I went ahead and ordered a +- 65 V supply from Connex, and will order the modules soon, but am super busy with work stuff here. First thing I'll try is just with the eval board-I really do not like that layout much (would like to be able to separate the modules some) so I hope some enterprising DIYer with a small commercial interest developed separate input boards dedicated to the {Purifi modules, seems like a good project for someone clever enough.
BTW, it looks like the Connex uses the very basic 7815 Vreg for the aux output for VDR. I have heard some amp designers suggest that having a really clean supply for VDR makes a nice difference on Ncore amps, I wonder if replacing that 7815 with a discrete alternative would be a good idea (like a Belleson). Just a quick thought... Connex mentions that further filtering/regulating might be desireable if the circuit it powers is sensitive to noise.
Well the only way to know is to try it! The Neurochrome has a lot of potential uses, so no worries if you find out it is not any better than the eval board's gain stage. I look forward to hearing what you think. The 1612 alone has lower noise and distortion by spec than the 49720, for what ever that might be worth, but I hear ya as far as what could be over-simplicity on the eval board.
I went ahead and ordered a +- 65 V supply from Connex, and will order the modules soon, but am super busy with work stuff here. First thing I'll try is just with the eval board-I really do not like that layout much (would like to be able to separate the modules some) so I hope some enterprising DIYer with a small commercial interest developed separate input boards dedicated to the {Purifi modules, seems like a good project for someone clever enough.
BTW, it looks like the Connex uses the very basic 7815 Vreg for the aux output for VDR. I have heard some amp designers suggest that having a really clean supply for VDR makes a nice difference on Ncore amps, I wonder if replacing that 7815 with a discrete alternative would be a good idea (like a Belleson). Just a quick thought... Connex mentions that further filtering/regulating might be desireable if the circuit it powers is sensitive to noise.
Where...
Thanks for this, where can we find out more about it? I can design simple stuff like low level linear power supplies, but I have no capabilities when it comes to analog line level circuits like this! I am just hopping that someone really clever will come up with an input gain stage dedicated to the Purifi modules, which could plug right in. Also allowing one to ditch the eval boards entirely, so as to allow for mono blocks, or just separating the modules a bit in a stereo build.
Edit: OK, found your Scion thread, nice! This is the Pass UGS gain stage, I am sure it could be made to work with Purifi, as long as it can be adjusted to the right level of gain. Cool...
Thanks for this, where can we find out more about it? I can design simple stuff like low level linear power supplies, but I have no capabilities when it comes to analog line level circuits like this! I am just hopping that someone really clever will come up with an input gain stage dedicated to the Purifi modules, which could plug right in. Also allowing one to ditch the eval boards entirely, so as to allow for mono blocks, or just separating the modules a bit in a stereo build.
Edit: OK, found your Scion thread, nice! This is the Pass UGS gain stage, I am sure it could be made to work with Purifi, as long as it can be adjusted to the right level of gain. Cool...
Last edited:
Yes...
I have thought about that as well, with both of my DACs I can configure the output stages for more gain, but I would be pushing the limits. I run everything at DSD 256 for both my Buff PRO and DSC-2 DACs. When you oversample to DSD 256 you have to reduce level before the oversampling by about 5-6 dB to avoid oversampling clipping, so I already raise the gain of my DACs to compensate for that. With -6dB, my Buff PRO now outputs a bit above 2 volts with a 0 dB signal (I like to have some additional gain for audiophile recordings like Channel Classics and Reference which are recorded at very low average levels and require quite a lot of volume in playback). So I would have to raise the gain again another 13 dB or so... Not sure what this would do to the performance of the output stages. I would rather not be running them very close to their max voltage swings, for example...
But, with the eval board I can try it easily in the bypass position, so I will. My DSC-2 has an AMB a24 gain stage on its output, so also driven by the OPA 1632 (like the Mercury), it could also be raised in gain, but I would have to build new ones (the PCB is a little fragile and I do not like removing the through hole resistors too many times).
Have you asked Russ/Brian if there are compromises in running the Mercury at really high gains and asking it to produce high voltage swings? IE, has it been tested for that and measured?
I have thought about that as well, with both of my DACs I can configure the output stages for more gain, but I would be pushing the limits. I run everything at DSD 256 for both my Buff PRO and DSC-2 DACs. When you oversample to DSD 256 you have to reduce level before the oversampling by about 5-6 dB to avoid oversampling clipping, so I already raise the gain of my DACs to compensate for that. With -6dB, my Buff PRO now outputs a bit above 2 volts with a 0 dB signal (I like to have some additional gain for audiophile recordings like Channel Classics and Reference which are recorded at very low average levels and require quite a lot of volume in playback). So I would have to raise the gain again another 13 dB or so... Not sure what this would do to the performance of the output stages. I would rather not be running them very close to their max voltage swings, for example...
But, with the eval board I can try it easily in the bypass position, so I will. My DSC-2 has an AMB a24 gain stage on its output, so also driven by the OPA 1632 (like the Mercury), it could also be raised in gain, but I would have to build new ones (the PCB is a little fragile and I do not like removing the through hole resistors too many times).
Have you asked Russ/Brian if there are compromises in running the Mercury at really high gains and asking it to produce high voltage swings? IE, has it been tested for that and measured?
You need a soft start
No it is not that. There seems no need, the SMPS starts gracefully, provides power after a couple of seconds, and the amplifiers per se have nearly no on/off bump. It is a "crackly" one when I turn it on and off. But I observed the behaviour better.
When I turn it on from cold, no noise.
When I turn it off, there can be a little of crackle, the switch probably is not that clean and lets some little sparks? dunno.
If I turn it on again without waiting a minute or so (the blue led on the SMPS is still on, even if barely), then there is this crackle.
Nothing that worries me, but I wonder what that exactly is.
Exactly!
Well, I would like to have a different gain well. Not yet sure whether the amps will be a "power follower" (which means unity gain buffer, not bypassed buffer) or a low sensitivity power amp (bringing the gain of the buffer to 6Db so that a +-4V swing will give full power). Definitely I do not want to drive it with the buffer bypassed (weaker sound, softer bass) or with all this gain (it just amplifies the noise upstream and I have to use my DAC at -36Db, which is not too bad, but still). So even if not "better" (and it is unlikely it will be worse) it will be more useful to me.
Yes I could change two SMD resistors on the Purifi buffer to change the gain. I do not have the skills to do that, and trying to find someone and having it done professionally may cost me some pretty penny. So I prefer to play on a different field.
The resistor to add to the Neurochrome UB is through hole. I can do that. There are nice and large pads on the bottom, which is also mostly free so it should be difficult to destroy something, even for me.
Looking forward to report
The 1612 alone has lower noise and distortion by spec than the 49720, for what ever that might be worth,I know, 1.1nV/√Hz vs 2.7nV/√Hz, 0.000015% vs 0.00003%. Which means that in all cases they will not be the dominant noise contributors in a non-trivial circuit
Tom claims the UB has 0.00001 % (-140Db) THD. Which is better than the 49720 on its own, giving to be a hint that he is using his "linearisation" tricks (see modulus-686).
Do not forget the power supply for +-VUNREG. It does not actually need to be unregulated, but it should be between 16 and 24V, I tried a 15 and the amps buzzed a lot, 18 V such as I use is the recommended value. VUNREG however reminds people using a hypex PS to move jumpers to provide unrelated aux, otherwise it would be 12V which is not sufficient.
I like the compact layout. If Purifi went the GaN route for their SMPS this could become a really miniature amplifier.
True, but this is not the hypex module. Perhaps they have improved power supply noise rejection (I think they claim that). Are you suggesting to desolder the 7815 and replace it with a Belleson? Eek.
"Do not forget the power supply for +-VUNREG. It does not actually need to be unregulated, but it should be between 16 and 24V, I tried a 15 and the amps buzzed a lot, 18 V such as I use is the recommended value. VUNREG however reminds people using a hypex PS to move jumpers to provide unrelated aux, otherwise it would be 12V which is not sufficient."
No worries, my plan is to build a nice linear for this, with +- 18 VDC output. I design power supplies for Sonore's Signature Series line, so I have plenty of experience with this. Note that the onboard regs (on the eval board) are discrete designs, pretty fancy for just "eval"! feel Bruno did not want the eval board to be too far from what is possible for sound quality.
"True, but this is not the hypex module. Perhaps they have improved power supply noise rejection (I think they claim that). Are you suggesting to desolder the 7815 and replace it with a Belleson? Eek."
Yes, that is exactly what I am considering, but until I see the supply in person I will not know how possible it is. I have done plenty of mods like this before, and do not anticipate problems, but it would depend on how tight the fit is there on the board. IDK if the PSRR of the Purifi module that they talk about includes the VDR section or not, all I know is that there have been folks who have supplied a separate power supply feed (from an ultra low noise low impedance linear supply) to the Ncore modules and reported nice sound quality increases. if it is easy for me to replace the reg I will do so, I have a good relationship with Belleson, and have a lot of their regs at hand, so it is easy for me to try as long as the board allows it-they are direct pin compatible with 7815.
BTW, I fully agree with your take on gain structures and matching to the needs of ones' system. Many folks have too much gain and this just adds noise. But i generally need the gain, as mentioned in my post above, I run my DACs anywhere from -20 dB to -2 dB right now, with my NC-400 amp which is 26 dB. I would be looking to lower gain as well if I had to run my DACs at -36 dB.
No worries, my plan is to build a nice linear for this, with +- 18 VDC output. I design power supplies for Sonore's Signature Series line, so I have plenty of experience with this. Note that the onboard regs (on the eval board) are discrete designs, pretty fancy for just "eval"! feel Bruno did not want the eval board to be too far from what is possible for sound quality.
"True, but this is not the hypex module. Perhaps they have improved power supply noise rejection (I think they claim that). Are you suggesting to desolder the 7815 and replace it with a Belleson? Eek.
"Yes, that is exactly what I am considering, but until I see the supply in person I will not know how possible it is. I have done plenty of mods like this before, and do not anticipate problems, but it would depend on how tight the fit is there on the board. IDK if the PSRR of the Purifi module that they talk about includes the VDR section or not, all I know is that there have been folks who have supplied a separate power supply feed (from an ultra low noise low impedance linear supply) to the Ncore modules and reported nice sound quality increases. if it is easy for me to replace the reg I will do so, I have a good relationship with Belleson, and have a lot of their regs at hand, so it is easy for me to try as long as the board allows it-they are direct pin compatible with 7815.
BTW, I fully agree with your take on gain structures and matching to the needs of ones' system. Many folks have too much gain and this just adds noise. But i generally need the gain, as mentioned in my post above, I run my DACs anywhere from -20 dB to -2 dB right now, with my NC-400 amp which is 26 dB. I would be looking to lower gain as well if I had to run my DACs at -36 dB.
Last edited: