D-Noizator: a magic active noise canceller to retrofit & upgrade any 317-based V.Reg.

Member
Joined 2019
Paid Member
Well, this low ESR capacitor load will be connected with some length of wire and there will be some pcb traces as well, meaning some inductance in between. That could provide stable conditions, …. or not. Hard to say without real conditions test. I don’t have such low ESR capacitor at hand to make test but will order one.
 
Thanks to all for the recent updates. The impedance for the final output electrolytic has been very useful. Seems like Nichicon PW or Panasonic FC in 63V 100uF guise would work. The series resistor is a smart move which I've put into the next revision.

My board arrived, going to populate it for 24V over the new few weeks. Not in a hurry! Have a BOM ready to fire away.

I've sent one to Nisbeth, and have two spare if anyone is interested to test them out (non-profit fabcrication cost only). They are 120x100mm on board transformer, single rail, 230v. More info a few posts back
 

Attachments

  • PCB-LM317DN-1.jpg
    PCB-LM317DN-1.jpg
    551.1 KB · Views: 728
  • PCB-LM317DN-2.jpg
    PCB-LM317DN-2.jpg
    594.7 KB · Views: 726
Last edited:
Thanks to all for the recent updates. The impedance for the final output electrolytic has been very useful. Seems like Nichicon PW or Panasonic FC in 63V 100uF guise would work. The series resistor is a smart move which I've put into the next revision.

My board arrived, going to populate it for 24V over the new few weeks. Not in a hurry! Have a BOM ready to fire away.

I've sent one to Nisbeth, and have two spare if anyone is interested to test them out (non-profit production-only cost). They are 120x100mm on board single rail, more info a few posts back

Here is the BOM
 

Attachments

  • LM317DN - BOM.pdf
    49.2 KB · Views: 137
@Grunf
I guess you are thinking to use denoiser for some DAC supply. We're you thinking about using just denoiser or add some regulator near DAC? If you think adding local regulator then i think denoiser is not needed cause modern regulators like LT3045(or similar) don't need really noiseless input cause they are good as they are.
I'm not sure if i said that to you on local forum but i consider denoiser cheap,easy build PSU with really good specs but for some specific use where you need regulator near circuit then i would go like you and put some modern LDO so PCB tracks are really short and it takes less space then denoiser.

And if you wanna see how this tests with os-con build one yourself, it's cheap and easy to make. ;)
I can send you PCB if you wanna. :)
 
Member
Joined 2019
Paid Member
Dienoiser stability under very low ESR load

There was question whether regulator will be stable with attached load containing a capacitor with very low esr (organic polymer for instance). Short answer is: yes, it is rock stable.
After receiving today 1000 uF/8 mΩ organic polymer capacitor, I prepared test contraption worthy of best Grunf’s achievements. BTW, Grunf is comic book character famous for his inventions made of junk that spectacularly fail during secret agent’s missions.

Output load was 250 mA constant load + 2 kHz 50 mA square wave dynamic load. Even with 6 cm long supply leads to the load with 1000 uF/8mΩ capacitor, it was impossible to spot any oscillation using oscilloscope and small AM radio as sniffer. There was some small AM noise, as square waves are involved, but noise was originating from the proto board with load and not from the regulator.

Organic polymer capacitor was then moved directly to the regulator’s output connector and after that directly in parallel with 100 uF output capacitor. Again, no oscillations at all.

Output voltage is stable and visible noise is in fact environment noise picked up by crappy oscilloscope probe that came along with the device. I will order decent one.

All measurements were done with 1 MHz low pass filtering.
With 50 mA change, output voltage was varying around 1 mV.

As we were discussing instability/oscillation problems recently, those findings paint another picture.
 

Attachments

  • Test contraption.jpg
    Test contraption.jpg
    643.7 KB · Views: 621
  • Load current.jpg
    Load current.jpg
    317.6 KB · Views: 612
  • Ouput constant load.jpg
    Ouput constant load.jpg
    326.7 KB · Views: 489
  • Voltage variation dynamic load.jpg
    Voltage variation dynamic load.jpg
    337.7 KB · Views: 230
Last edited:
Do you have some LNA connected to denoiser output or did you connected it directly into scope? That is question just for ripple measurement, for oscillations you can spot them from miles distance as i did when i placed untested value of bias resistors for BC337. Oscillations we're few Vpp and you couldn't miss them.
 
@Grunf
I guess you are thinking to use denoiser for some DAC supply. We're you thinking about using just denoiser or add some regulator near DAC? If you think adding local regulator then i think denoiser is not needed cause modern regulators like LT3045(or similar) don't need really noiseless input cause they are good as they are.
I'm not sure if i said that to you on local forum but i consider denoiser cheap,easy build PSU with really good specs but for some specific use where you need regulator near circuit then i would go like you and put some modern LDO so PCB tracks are really short and it takes less space then denoiser.

And if you wanna see how this tests with os-con build one yourself, it's cheap and easy to make. ;)
I can send you PCB if you wanna. :)

i have no intention of using this regulator in my DAC, i am just asking for others as there is a discussion going on about the characteristics of the output capacitor and the oscillation problem so i wanted to point out a possible problem.
 
There was question whether regulator will be stable with attached load containing a capacitor with very low esr (organic polymer for instance). Short answer is: yes, it is rock stable.
That confirms my own experience (only for the 317): a good layout combined with a single, good cap does not seem to be problematic.
In fact, with its compensation, the denoiser might act as an active compensation, and contribute to the stability.
There might however be a penalty: a noise peak could appear in the tens of kHz range.
I didn't check it, because it would be complicated with my current instruments configuration, but if someone has the means to do it simply and easily, it might be worth checking

All measurements were done with 1 MHz low pass filtering.
With 50 mA change, output voltage was varying around 1 mV.
That's not really bad, but it leaves to be desired: it represents ~0.02ohm, which must be about the same as the naked 317.
The denoiser should push that below 1milliohm.
There might be some wiring/connection issue degrading the output impedance
 
This all seems to show that have the lowest impedance, on both Denoiser and Dienoise, it's mandatory using Kelvin interface.

I didn't imagine that could be so. I first heard of such a wiring on the Jung Super-regulator, but it seems an important requirement.

Why other low-impedance chips, also low-impedance, like LT3042/3045, do not require such wiring and still get low impedance output figures?
 
Member
Joined 2019
Paid Member
Just not to leave the cat both dead and alive, I made another observation on voltage variation under load.
Using 500 mA square wave load at 2 kHz, oscilloscope probe was placed directly at the regulator’s output connector. No discernable voltage variation was possible to spot, even using average of 32 to 128 measurements and low pass filter. So, voltage variation is below acquired environment noise level which was 240 µV PP with filtering and averaging. Therefore I assume that voltage variation is, at least, lower than 0,2 mV.

That gives the worst case output impedance of 0,4 mΩ @ 2 kHz, what is in line with expected.
 
Hi Elvee,
Does the implement of TL783 need a high voltage transistor?
No, not really: this has already been answered, but to summarize the transistor is isolated by the coupling caps, and statically the bias forces a lowish Vce.
Under surge conditions, the voltage could become higher, but if you use protection diodes on the denoiser and regulators, it should be immune
Just not to leave the cat both dead and alive, I made another observation on voltage variation under load.
Using 500 mA square wave load at 2 kHz, oscilloscope probe was placed directly at the regulator’s output connector. No discernable voltage variation was possible to spot, even using average of 32 to 128 measurements and low pass filter. So, voltage variation is below acquired environment noise level which was 240 µV PP with filtering and averaging. Therefore I assume that voltage variation is, at least, lower than 0,2 mV.

That gives the worst case output impedance of 0,4 mΩ @ 2 kHz, what is in line with expected.
Thanks for taking the time to sort that out
 
Here is the answer to a question asked via PM: how to implement remote sensing with the denoiser.
This has already been answered earlier in various forms, but here it is again:

attachment.php


That said, it is probably preferable to locate the denoiser remotely instead of using long wires to connect it: that is the way it has initially been designed for, and using long connecting wires could result in VHF instability (a few cm is OK, tens of cm could be problematic).

BTW, for general interest questions, it is preferable to post them directly on the thread, as it can be useful to other members, and PM's do not allow attachments
 

Attachments

  • Roby.png
    Roby.png
    304 KB · Views: 913
There was question whether regulator will be stable with attached load containing a capacitor with very low esr (organic polymer for instance). Short answer is: yes, it is rock stable.
After receiving today 1000 uF/8 mΩ organic polymer capacitor, I prepared test contraption worthy of best Grunf’s achievements. BTW, Grunf is comic book character famous for his inventions made of junk that spectacularly fail during secret agent’s missions.

Output load was 250 mA constant load + 2 kHz 50 mA square wave dynamic load. Even with 6 cm long supply leads to the load with 1000 uF/8mΩ capacitor, it was impossible to spot any oscillation using oscilloscope and small AM radio as sniffer. There was some small AM noise, as square waves are involved, but noise was originating from the proto board with load and not from the regulator.

Organic polymer capacitor was then moved directly to the regulator’s output connector and after that directly in parallel with 100 uF output capacitor. Again, no oscillations at all.

Output voltage is stable and visible noise is in fact environment noise picked up by crappy oscilloscope probe that came along with the device. I will order decent one.

All measurements were done with 1 MHz low pass filtering.
With 50 mA change, output voltage was varying around 1 mV.

As we were discussing instability/oscillation problems recently, those findings paint another picture.

Thank you for the measurements.
What about sound quality have you tested the supply against a super regulator / sulzer / lt3042/45??

Have you more pcb's left?

Thx
 
Member
Joined 2019
Paid Member
The same question bothers me as well. There seems to be considerable number of regulators built, but we have no reports, measurements or opinions on how they affect sound.
I am not yet in the position to make comparisons on sound quality as I am rebuilding everything from the scratch, after long time absence from DIY audio.
Regarding measurements, keep I mind that results are bound to the equipment precision. I am sure that output impedance is, at least, ten times lower than I was able to measure.
I have to order corrected pcb’s for another project and already took decision to order another batch of LM317/LM337 pcbs. Those will be again offered free of any charge to the forum members. Call it seed planting and fruits will come.