I've been trying the op amp power supply for AVCC and indeed I found that an OP177 performs remarkably well. https://www.analog.com/media/en/technical-documentation/data-sheets/OP177.pdf I was wondering recently about this and actually wondering why this was so ;so the linearity aspect could indeed be a factor.
It was a shocker as I was not expecting it at all. Thus there could be something to this. More trials to see if this ends up as the endpoint for my AVCC.
Last edited:
Linearity... 0.3V/uS slew rate. This is a precision opamp for DC / very low freq
applications.
TCD
indeed "Speed" might not be your friend here - ALL depends on the size of the output rail capacitors...OK, here is the story of this circuit. I had purchased an Ebay board that was dual 12V and 5V rail for over a year, I used that on a PCM1794 Dac with the 5V circuit for the AVCC equivalent on the 1794. I then also built a 5534 op amp equivalent circuit to power the AVCC circuit. The Ebay circuit was superior. It was a compact design with short and wide traces using an NE5534 as the error amp driving a FET. I think the output caps are dual 100uF caps, nothing special.
Then I was trying to finish my Jung Super Reg for the 9028 board's AVCC. In my second try at supidity, I was testing other power supplies that I had around for other things and the + volt rails were all on the top. When I got the Jung Super Reg, I noticed the +DC on the top and connected the 10V rail to it and powered it up. Well as it turns out the markings on the Didden board was +DC GND and +DC in. I did not read the complete markings and I had used these same boards before without a problem. Well I connected +10V where it should have been ground. No smoke but I lost the op amp pass transistor and worst of all my LTC6655. I continued to use my Sulzer Circuit on a board that was since the early 80s. Long traces etc. Not good.
Then one day I noticed the same regulator board I had used on the PCM1794 but now it was also available with a 3.3V. What the heck! The order went in.
I had been listening with the Sulzer for over a year, and had used LME49710 ( not stable) 5534 and the AD817. I had settled on the AD817. 50V/us slew 50Mhz bandwidth. Plenty Fast.
I received the regulator board early this year and did nothing with it since I had the flu.
So I recently opened it up and decided to use it just to see what it would sound like. Then I noticed it was not an NE5534 on the board but an OP177G. I checked the specs and I was disappointed. Chinese substitutions again. Nevertheless I proceeded to put it in.
That same evening my wife who has no real interest in audiophilia remarked how the speakers completely disappeared now. Was this supposed to be the case?
Upon listening I immediately noticed the sense of space was much more apparent. My sub get down to the low 20s for real measured with REW and a UMIK. It is eq'd in the last octave using DSP. PRAT..it has it in spades. Cymbal tails...it's there. Backup vocals are extremely distinct to the singer. The detail I hear is not that you can hear certain things but more importantly how each instrument in the background is being played. How is the singer singing, how is the piano being played/ The biggest change is that you can actually turn DOWN the volume to get the same listening satisfaction. If you know about listening, this is significant. This Ebay reg does that.
I'm now considering finally boxing up my 9028Pro project but I am left wondering if the Super Reg would have sounded better. As an option, I could try and remove the OP177 and roll in something like an LME49710. However my skills and toolset on SMD is quite limited.
The question I have for myself is it good enough to call it a day...and the real answer is yes.
Note this is simply my opinion of what I heard and was expecting poor results. Yes, for me in my system it was a shocker. Then Mark mentions something about linearity of the error amp and my try using a highly linear op amp, preceded his comment by days. Isn't that something?
HiFi Dual Power Output Ultra Low Noise Linear Regulator Power Class A | eBay
BTW, I will try removing the Super Regs for the IV and installing the dual 15V portion and see what comes out of that.
Then I was trying to finish my Jung Super Reg for the 9028 board's AVCC. In my second try at supidity, I was testing other power supplies that I had around for other things and the + volt rails were all on the top. When I got the Jung Super Reg, I noticed the +DC on the top and connected the 10V rail to it and powered it up. Well as it turns out the markings on the Didden board was +DC GND and +DC in. I did not read the complete markings and I had used these same boards before without a problem. Well I connected +10V where it should have been ground. No smoke but I lost the op amp pass transistor and worst of all my LTC6655. I continued to use my Sulzer Circuit on a board that was since the early 80s. Long traces etc. Not good.
Then one day I noticed the same regulator board I had used on the PCM1794 but now it was also available with a 3.3V. What the heck! The order went in.
I had been listening with the Sulzer for over a year, and had used LME49710 ( not stable) 5534 and the AD817. I had settled on the AD817. 50V/us slew 50Mhz bandwidth. Plenty Fast.
I received the regulator board early this year and did nothing with it since I had the flu.
So I recently opened it up and decided to use it just to see what it would sound like. Then I noticed it was not an NE5534 on the board but an OP177G. I checked the specs and I was disappointed. Chinese substitutions again. Nevertheless I proceeded to put it in.
That same evening my wife who has no real interest in audiophilia remarked how the speakers completely disappeared now. Was this supposed to be the case?
Upon listening I immediately noticed the sense of space was much more apparent. My sub get down to the low 20s for real measured with REW and a UMIK. It is eq'd in the last octave using DSP. PRAT..it has it in spades. Cymbal tails...it's there. Backup vocals are extremely distinct to the singer. The detail I hear is not that you can hear certain things but more importantly how each instrument in the background is being played. How is the singer singing, how is the piano being played/ The biggest change is that you can actually turn DOWN the volume to get the same listening satisfaction. If you know about listening, this is significant. This Ebay reg does that.
I'm now considering finally boxing up my 9028Pro project but I am left wondering if the Super Reg would have sounded better. As an option, I could try and remove the OP177 and roll in something like an LME49710. However my skills and toolset on SMD is quite limited.
The question I have for myself is it good enough to call it a day...and the real answer is yes.
Note this is simply my opinion of what I heard and was expecting poor results. Yes, for me in my system it was a shocker. Then Mark mentions something about linearity of the error amp and my try using a highly linear op amp, preceded his comment by days. Isn't that something?
HiFi Dual Power Output Ultra Low Noise Linear Regulator Power Class A | eBay
BTW, I will try removing the Super Regs for the IV and installing the dual 15V portion and see what comes out of that.
Last edited:
Interesting comments.
I thought I'd share the board but from Aliexpress and quite a but cheaper there too.
#Aliexpress ￡11.46 | HIFI DAC power supply Class A dual power supply multiple outputs Dual DC ± 12V ± 15V ± 18V Single 5V 3.3V H151
AliExpress
I have ordered the board...intact I may get 2 and use on on my Ian Ian Canada stack!
I thought I'd share the board but from Aliexpress and quite a but cheaper there too.
#Aliexpress ￡11.46 | HIFI DAC power supply Class A dual power supply multiple outputs Dual DC ± 12V ± 15V ± 18V Single 5V 3.3V H151
AliExpress
I have ordered the board...intact I may get 2 and use on on my Ian Ian Canada stack!
Last edited:
I thought that was very interesting reading, thanks John. I'm using a combo of Roon and the AK4137 board and have been mulling over the use of Roon upsampling vs the AK4137 (it reduces network traffic, Roon server processing etc) so would the ideal 'blend' be for Roon to upsample to 192 and then use the AK4137 and upsample to DSD256? Alternatively I could just leave it as 44.1 and only use DSD64.
Mike,
Thanks for the write up. I have one of those little regulator boards you used. Sketched out a schematic of it once. Its kinda similar to a Jung but simpler.
Interesting to note the AKM eval dac boards appear to have used Jungs for the Reference Voltage even before AK4499 came along. Most people seem to ignore that and assume AKM doesn't know what they are doing
However, Jung regulators have their particular characteristics too, which can have an effect on sound quality if the regulator is not suitably matched to the load.
Also, as JohnW and I discussed very briefly, almost everything seems to affect sound quality. However, that may not become noticeable until your system is approaching ultra low distortion and noise. There are a lot of things that need may attention to get to that point unfortunately.
From what you said about how the sound quality of your dac cleared up, I can estimate about where your system is at now. From that, starting to get 'not too bad' might be how Jam would put it. He says things like, 'some systems that are so bad the sound ends at the speakers, its should extend way out past them to the sides.' (Of course, his standard for audio is relative to 'as good as it gets.')
From that point, it can still get quite a bit better as what is coming out of the speakers is further improved.
Good to hear you are making progress and are happy. It can be a long journey, still on that road myself.
Thanks for the write up. I have one of those little regulator boards you used. Sketched out a schematic of it once. Its kinda similar to a Jung but simpler.
Interesting to note the AKM eval dac boards appear to have used Jungs for the Reference Voltage even before AK4499 came along. Most people seem to ignore that and assume AKM doesn't know what they are doing
However, Jung regulators have their particular characteristics too, which can have an effect on sound quality if the regulator is not suitably matched to the load.
Also, as JohnW and I discussed very briefly, almost everything seems to affect sound quality. However, that may not become noticeable until your system is approaching ultra low distortion and noise. There are a lot of things that need may attention to get to that point unfortunately.
From what you said about how the sound quality of your dac cleared up, I can estimate about where your system is at now. From that, starting to get 'not too bad' might be how Jam would put it. He says things like, 'some systems that are so bad the sound ends at the speakers, its should extend way out past them to the sides.' (Of course, his standard for audio is relative to 'as good as it gets.')
From that point, it can still get quite a bit better as what is coming out of the speakers is further improved.
Good to hear you are making progress and are happy. It can be a long journey, still on that road myself.
Last edited:
The road ends when you are forced to for one reason or another to stop. It's like a never ending road that leads everywhere but nowhere. Every stop has its attractions, then you grow bored with it and want to move on.....maybe ending back up one day with a Shure M91ED spinning vinyl with a Sherwood receiver and Dynaco A25s.
I am using the 3.3V from the mentioned reg board at this time. I will test out the 15V supplies for the op amps against the Jung Super Regs some time shortly.
One thing I must mention is that the board was SUPPOSED to come with NE5534s but it came with a different op amp, the OP177. Whether this is worse or better than the NE5534, I am unable to tell. So watch out for this.
On the 5V version, I was also using it to power the XMOS through a Y cable and it did work well there...equal to an LT1963 I am now using. But the requirements there are different from that for AVCC. I can tell you that an LT1963 does not sound as good for AVCC, not even close in fact. Yet in linear stages, the LT1963 is supposed to be close to a Jung Super Reg. I think there there's a lot of stuff happening in circuits that is not COMPLETELYy understood BUT what is understood works predictably well to a certain point. Beyond that, no one is sure.
Yeah...like cables.
One thing I must mention is that the board was SUPPOSED to come with NE5534s but it came with a different op amp, the OP177. Whether this is worse or better than the NE5534, I am unable to tell. So watch out for this.
On the 5V version, I was also using it to power the XMOS through a Y cable and it did work well there...equal to an LT1963 I am now using. But the requirements there are different from that for AVCC. I can tell you that an LT1963 does not sound as good for AVCC, not even close in fact. Yet in linear stages, the LT1963 is supposed to be close to a Jung Super Reg. I think there there's a lot of stuff happening in circuits that is not COMPLETELYy understood BUT what is understood works predictably well to a certain point. Beyond that, no one is sure.
Yeah...like cables.
Last edited:
So if I understand you correctly, one 3.3V line for both AVCCL and AVCCR on the ESS dac chip? I wonder if a single psu for each side of the AVCC would behave differently with the same components? I assume the stereo separation would vary a lot from one solution to another, and also varying separation at different frequencies.. Anyways, looking forward to reading more discoveries from you!
The first solution I'm going to try next is LiFePo4 on AVCC. The dac chip on my board were slightly defective so I had to restart with a new ES9038Q2M board so my build have delayed quite a while now.
The first solution I'm going to try next is LiFePo4 on AVCC. The dac chip on my board were slightly defective so I had to restart with a new ES9038Q2M board so my build have delayed quite a while now.
This thread is reallly lng sorry i was not here continuusly jast last year 3-4 month anr no again.
Than you were right
my listening test confirmed too.May I ask how did you come to this conclution anyway?
are C0G exempt or not from MLCC in that statement?
LikewiseI found ceramics have sounded pretty nasty for bypassing compared to film caps, going by facts I thought it might piezo effects of most MLCCs, which C0G is free from.
When 0.1u bypasses are needed C0G could then be the best possible cap to use, but never tested it.
LikewiseI found ceramics have sounded pretty nasty for bypassing compared to film caps, going by facts I thought it might piezo effects of most MLCCs, which C0G is free from.
When 0.1u bypasses are needed C0G could then be the best possible cap to use, but never tested it.
Recommendations about bypass caps are based on calculations and measurements of bypass impedance vs frequency. Due to parasitic inductance and due to the very low impedance of the ground planes used at HF/RF (including for dacs), plain old X7R is still the recommendation.
The problem with very low ESR caps is that they make nice undamped resonant circuits when parasitic inductance of leads and PCBs vias are taken into account. That makes impedance of the bypass vary wildly and not always low.
Even putting caps of different values in parallel is now deprecated for reasons of interactive resonances between them making net bypass impedance higher at some frequencies.
You can pick your own poison. If it sounds better it may or may not actually be better. Sometimes people get power rails ringing and think they made it sound better. Unfortunately, they are never going to get their systems truly good sounding by doing such things. Its the wrong fix for why the system doesn't sound right.