So true. I've seen first hand evidence of this in RF signal handling equipment (isn't that what digital audio gear is really?).
I'm not quite sure why audio designers haven't adopted 3 terminal decoupling caps to combat power scheme resonances more widely, like RF engineers do.
https://www.murata.com/en-eu/products/emiconfun/emc
But then I'm not a proper engineer...
an update to the power supply i added a few threads previous. Remember I had said I had 12 V versions of it at my disposal. So what I did was I used it as a PreRegulator and also used a LM317 reg as a prereg as well. The LM317 was hardly an improvement if any. However the Jung type in series was another revelation. The Bass articulation and dynamic structure of the sound was very distinct now.
By far the biggest sonic signature of the DAC has been the implementation of the AVCC. Aspects considered as overkill are not really. Three terminal devices so commonly used in so many DACs are in my book not worthy at all anymore.
There will be new DAC chips but one needs to ponder whether these are as sensitive to the AVCC as the current ones or whether they are equally so or more sensitive.
The AVCC supplies I tried are the LT1963 ( stock), Current boosted LT3042 with stability cap, Jung Sulzer with LM317 Preregs, using NE5534, AD817. Jung type OP177 error amp driving Fet and now series connected Jung Type with the first using an NE5534 error amp driving FET outputting 12V powering Final stage of Jung Type OP177 error amp driving FET outputting 3.3V. Maybe this is the second to last attempt before I box it all up in Al cases.
My final attempt is to rebuild a Super Reg but this time using 2 of the Jung Type Regs I have outputting 12 V as Preregs. One will power the error amp LME49710 in the Super Reg and voltage reference using an LT3042-3.3V. The second 12V will power the Pass transistor circuit driven by the LME49710. So the LME49710 error amp and 3.3V voltage reference will hardly be modulated by the AVCC current needs. My last final attempt..... hopefully no more messups.
By far the biggest sonic signature of the DAC has been the implementation of the AVCC. Aspects considered as overkill are not really. Three terminal devices so commonly used in so many DACs are in my book not worthy at all anymore.
There will be new DAC chips but one needs to ponder whether these are as sensitive to the AVCC as the current ones or whether they are equally so or more sensitive.
The AVCC supplies I tried are the LT1963 ( stock), Current boosted LT3042 with stability cap, Jung Sulzer with LM317 Preregs, using NE5534, AD817. Jung type OP177 error amp driving Fet and now series connected Jung Type with the first using an NE5534 error amp driving FET outputting 12V powering Final stage of Jung Type OP177 error amp driving FET outputting 3.3V. Maybe this is the second to last attempt before I box it all up in Al cases.
My final attempt is to rebuild a Super Reg but this time using 2 of the Jung Type Regs I have outputting 12 V as Preregs. One will power the error amp LME49710 in the Super Reg and voltage reference using an LT3042-3.3V. The second 12V will power the Pass transistor circuit driven by the LME49710. So the LME49710 error amp and 3.3V voltage reference will hardly be modulated by the AVCC current needs. My last final attempt..... hopefully no more messups.
Depends what you mean by AVCC. AKM calls their equivalent voltage the Reference Voltage (which is 5v), and they use one Jung regulator for that supply for each of the four dac channels on AK4499 evaluation board. Also use 4 of NJM7805 regulators (one for each channel) for another supply pin that doesn't sound good with LT1963 either.
Last edited:
Mark, the Jung Regulator is loosely used. I'd like some clarification on the circuit topology that AKM has used. It is interesting as I look at it because the Jung Super Regulator is somewhat unique in that it the error amp and reference voltage of the error amp is powered by what the error amp outputs itself. So intrinsically it uses the output of what it is to control to power itself and thereby why some op amps do not start up properly. The circuitry in front of the pass transistor is to help it power up into an operable state.
This is an elegant solution to regulating the error amp but it also presents some challenges. I don't know if this was done for performance or for potential simplicity and reduced parts count.
This was the source of my difficulty because I needed to find a decent op amp that could operate on a single rail at 3.3V.
The circuitry that we loosely called the Jung Type regulator is indeed simpler in that the error amp is powered by the unregulated input to to the pass transistor prior to the output of the regulator. Is this the topology that AKM has also used?
This is an elegant solution to regulating the error amp but it also presents some challenges. I don't know if this was done for performance or for potential simplicity and reduced parts count.
This was the source of my difficulty because I needed to find a decent op amp that could operate on a single rail at 3.3V.
The circuitry that we loosely called the Jung Type regulator is indeed simpler in that the error amp is powered by the unregulated input to to the pass transistor prior to the output of the regulator. Is this the topology that AKM has also used?
Last edited:
Mark,
How would you rate that power supply
here Is being offered at good price
HIFI DAC power supply Class A dual power supply multiple outputs Dual DC +- 12V +- 15V +- 18V Single 5V 3.3V H151 on AliExpress
HiFi Walt Jung Super Regulator DC Power Supply Board Linear PSU Audio DAC Preamp | eBay
The above is a true Jung Super Reg. It uses LEDs as the precision reference voltage. I have one in my hands but I have not directly compared it to a Diyaudio store one using LM329s as the reference. I did power it up and noticed a slight amount of drift as the LEDs warmed up. Once it stabilized it seemed OK. It did ship with an LME49710 which is at least as good as the AD797 but more stable in startup. The other positive aspect is that uses a tight SMD layout which is potentially better. However I have not listened to it as yet. It was supposed to power up an MC Head amp I have in the oven.
For the asking price.....if it sounds good, there's no reason not not use these in linear circuits.
Then there are these...DirtyPCBs.com
A Fellow Diyaudio member in Toronto pointed these out to me. He has used multiples on his projects and they work. Again at the asking price.....for 10 boards. Less than a couple bucks per board shipped. The parts required for the Super Regs are not exotic at all and all quite inexpensive...depending on the op amp you choose.
The above is a true Jung Super Reg. It uses LEDs as the precision reference voltage. I have one in my hands but I have not directly compared it to a Diyaudio store one using LM329s as the reference. I did power it up and noticed a slight amount of drift as the LEDs warmed up. Once it stabilized it seemed OK. It did ship with an LME49710 which is at least as good as the AD797 but more stable in startup. The other positive aspect is that uses a tight SMD layout which is potentially better. However I have not listened to it as yet. It was supposed to power up an MC Head amp I have in the oven.
For the asking price.....if it sounds good, there's no reason not not use these in linear circuits.
Then there are these...DirtyPCBs.com
A Fellow Diyaudio member in Toronto pointed these out to me. He has used multiples on his projects and they work. Again at the asking price.....for 10 boards. Less than a couple bucks per board shipped. The parts required for the Super Regs are not exotic at all and all quite inexpensive...depending on the op amp you choose.
Their opamp is powered by the regulated output voltage, but that voltage is 5v making opamp selection easier than if 3.3v.
Its not a real Jung. It is much simplified and IIRC the opamp is powered from the input voltage side. That topology came from before Jung's design.
Not as good, I would say.
A thing to watch with Jung regulators is how they behave for something like AVCC as the output current is raised. You may notice sound quality changing if you put a dummy load resistor from a regulator output to ground. For some regulators, that may make them more linear and better sounding. IMHO Jung can tend to get overly bright.
Last edited:
What voltage are the zener diodes in this AKM "Jung" circuit? I do not see these values in the Ak4497 evaluation board manual, and the AK4499 evaluation board manual is not available...
Do you mean VDDL pins?
That's interesting. When I was wondering about the OP177, this is exactly what was going through my mind, thinking about it in an analog non EE fashion which is me. The thought ran through my head that if the error amp was not linear or did not respond such that the needed current would be deficient, then the result would be a thinner and possibly sound tending on the bright side IF the speed was fast but current was wavering in the lower frequencies.
But would the feedback not correct for it?. So I was then confused where do Precision Op Amps really belong? What really are the benefits and what aspects might they be of benefit here? Still trying to understand this. Then there is the possibility that it could be the pass transistor that might be the weak point as well.
At what conditions? What shoud we take care off? With a bad opamp, bad bypassing, too big output capacitor?
Thank you for explanation.
Hey guys. A while ago I found a real nice graph with output impedance against frequency for a couple of voltage regulators including LT3042 and AD797 (direct output). Might have been in this forum, might have been elsewhere. I tried all sorts of google searches but can't find it. Anybody knows what I'm talking about, please tell me. Thanks in advance!
Have a look at this.
ADM7151 vs LT3042 (Output Impedance) by Analog Dicovery: アナログ回路のおもちゃ箱
First he started to measure an lt3042 output impedance vs frequency, with his beautifully tuned (painfully tuned and calibrated) Analog Discovery setup.
Then he had noticed that the values are maybe 'falsified' by the few mm of copper - on the testboards themselves!
Then he goes on and searches further. Look at the final values.. Much below 1mOhm up to 100kHz.
The initial test had given this result:
ADM7151 vs LT3042 (Output Impedance) by Analog Dicovery: アナログ回路のおもちゃ箱
First he started to measure an lt3042 output impedance vs frequency, with his beautifully tuned (painfully tuned and calibrated) Analog Discovery setup.
Then he had noticed that the values are maybe 'falsified' by the few mm of copper - on the testboards themselves!
Then he goes on and searches further. Look at the final values.. Much below 1mOhm up to 100kHz.
The initial test had given this result:
Attachments
Last edited:
I would say: one can nicely just forget any other comparisons, regulators. Will never arrive at this performance. Because a Jung/ Schulzer, usually big like a soccer field, will just never be possible to place in that vicinity to the load, to be in such an intimate close coupling with the load, than any of the modern LDO regulators.
Attachments
There might be more to just output impedance alone, it might for example be the symmetry of response to load changes (over- and undershoot) and linearity of the relation between load changes and output voltage changes. And we're also dealing with current demands way above 100kHz in case of the modulation schemes of modern DACs. Trying to wrap my head around this whole topic before laying out a prototype PCB where we can test all sorts of regulators and filtering/bypassing schemes. This seems to be either uncharted territory or simply undisclosed because of "trade secrets" by people who have mastered the topic. Either because of commercial interest or whatever (ego?). If a relatively "slow" Jung reg is really better according to listening tests, the question remains as to WHY this is the case, I mean the theory behind all that.
An lt3042 is active up to 1MHz. One will never have symmetrical response with series type regulator. Always the load is pulling down.. An opamp do push/pull, - - and dies at 10KHz.. And oscillates with load capacity.
And anyway, no regulator can take care of the MHz range of the fast changing load currents into a dac reference pin. One must apply proper local bypassing. The bypass capacitors are taking over the low impedance job. The problem is to pass over from the regulator (with it's virtual output inductance) to the output cap, without resonance peaking.
But, it is a quite different task in case of a slow regulator (opamp) having uH equivalent of virtual output inductance, or an LDO with nH range..
Ciao, George
And anyway, no regulator can take care of the MHz range of the fast changing load currents into a dac reference pin. One must apply proper local bypassing. The bypass capacitors are taking over the low impedance job. The problem is to pass over from the regulator (with it's virtual output inductance) to the output cap, without resonance peaking.
But, it is a quite different task in case of a slow regulator (opamp) having uH equivalent of virtual output inductance, or an LDO with nH range..
Ciao, George
