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

[wgh52]

Hi Folks,

I‘m following this Thread for a while now and am overwhelmed by the Expertise and Mountains of Development efforts!

I hope my question is acceptable: For my application, an old LM317/LM337 based Phono-Preamp supply, which I want to keep, I‘d like to try one of the noise reducers developed here. But which one do I take?

Thanks for your advice!
Winfried

[Trileru]

Depends. Is the power supply on a separate board? If you have the +/- rails on a separate pcb you can replace it with a dienoiser type pcb, there's two dual version here, smd and tht.
If not then you could add an add-on pcb for each rail, but I'd try to make the complete supply if possible.

[Trileru]

I tried to make a discrete regulator using the dienoiser for ac feedback. Seems to simulate ok. PSRR at around 93-95dB at 100Hz, output impedance at 1-2mOhm and .noise sim shows around 1nV/√Hz. I made two versions, a 1A one and 100mA one. I'm not sure if I made the temperature sim correctly but shows around 300mV between 25 and 75 degrees C. Vout delta between few mA and full load is around 200mV for the 1A version and around 40mV for the 100mA version. The min voltage across the pass transistor for full performance seems around 3.8-4V. I could tweak the difference in output voltage between light and full load but I think there's a balance with the PSRR.
I used MJE15032 for the 1A version and BCP56 for the 100mA version. For the reference I used a LM329. I tried to pulse the load and seems stable.
Seems like it should work, and performance seems interesting for a discrete regulator. Did I miss anything or is there anything simple that could improve the performance? Would it work in practice?

[wgh52]

Originally Posted by Trileru

Depends. Is the power supply on a separate board? If you have the +/- rails on a separate pcb you can replace it with a dienoiser type pcb, there's two dual version here, smd and tht.
If not then you could add an add-on pcb for each rail, but I'd try to make the complete supply if possible.

Hello and thanks for the initial answer!

We are talking about retrofitting an existing, space limited, purchased power supply. Available space for an add-on is about 5x3x3cm. The 317/337 board (both on one PCB) shall remain, as it fits the case. I plan to piggy-back a known, well working noise reduction circuitry and do not plan to experiment or optimise beyond maybe tuning a potentiometer (as my expertise does not seem suffice beyond that... ).





So I would be very thankful if I would be pointed to the posting(s) with the appropriate ciruitry, schematic or PCB, so I can take it on from there.

Thanks and Regards,
Winfried

[Trileru]

You could try the board in this post, I've tested it and it's working:
D-Noizator: a magic active noise canceller to retrofit & upgrade any 317-based V.Reg.
It's 43mm x 20mm. If you think you can solder the sot-23 parts there's a smd version as well.

[wgh52]

Thanks!

I'll have a look at the KiCAD project (got the S/W but only rudimentary experience with KiCAD...) and see what it tells me. Would boards be available for purchase at all? I'd just need two of the THT boards, as I have two of the power supplies.

Regards,
Winfried

[Trileru]

I diyed mine. You can get them made pretty cheap at fab houses like jlcpcb or others similar. The gerber files for that should be included in the zip file.

[Trileru]

I think I can make the discrete regulator with even more accessible/cheaper parts. I could replace the voltage reference with Walt Jung's GLED431 setup using a resistor/transistor/led.
I might make a diy board to test it.

[carlmart]

Can you upload the asc file for the later?

[Trileru]

I have since improved it a bit. I used a cross between the nonoiser and dienoiser. I have also used some random led model I found, seems to perform worse than the ones in the library but seems to be having a closer Vdrop to the normal tht leds I have around.
Noise sim shows around 0.62nV/√Hz, PSRR around 120dB, and output impedance also dropped to around 0.5mOhm.
PSRR can be a bit higher with a LM329, or maybe a proper LED model. For the sim I used BD139 as pass transistor, I think it's a Bob Cordell model, around 100mA.
I tried to keep around 5mA going through the leds. To me the currents seem optimal, through the other components.
There's a delta of around 160mV between 25 and 75 degrees C for temperature sim.
Stepping the load between 10mA and 100mA there's a delta of around 120mV.
I guess figures could be improved a bit. I was in the middle of playing with the sim now. I pulsed the load and seems to recover nicely. I don't know if it needs compensation or ferrite bead like the NoNoiser.
I made a diy pcb of the previous version but I will do one for this version as well, seems like it should perform good. Considering the performance and price of components I think it's interesting.