Ultra-Low noise linear power supply 5VDC/0.7A

Hi everyone!
I'm using Li-Ion battery supply for a headphone amplifier. I would like to replace the batteries with a very good linear regulator with minimal noise. I need max. 0.7A output. I have a 10VA toroidal transformer with a 9VAC output and was planning to use that if possible. But I have no clue as how to find a very good regulator suitable for this. Ideally something I could buy already assembled or just a circuit diagram so I can assemble one (I'm not an electronics designer but I can handle a solder iron) 🙂
 
LM723/uA723 may be getting near the lower supply range limit with a rectfied 9VAC transformer, considering ripple and eventual voltage sag of the 10VA transformer.
you could have a look at the various 317-de-noizator variants here, for even lower noise!
it's a huge thread, but includes 5V variants and PCBs proposals.
 
Last edited:
LDOVR TPS7A4700 modules are so affordable that I can't make them myself for a lower price. Very low noise and LDO so working perfectly with a 6V 10VA transformer. Small, low heat, green, the works. I suggest to use a larger heatsink though.

LM723 works OK and although it is a dinosaur it still is competitive with regards to low noise. I know as I have designed and built single board 5V PSU's with it fed by 8V transformers. This reminds me I have ready soldered so complete ones (with onboard transformers) in stock. Minimal size, CLC filtering, RF proof, current limiter/short circuit protection, very low noise but 5V limited to 0.25A by the mini transformer. Transformer can be changed of course.
 
Last edited:
  • Like
Reactions: stv
LT3045 - 20V 1.5A / 500mA
Lt3080 - 37V 1.5A

Depends on the package. Those do well targetting mains hum with their PSRR curves. You may want switching noise filters targetting incoming noise for differential/common mode noise.

Ian canada has a LiPo power supply that may be worth looking if you dont want to get into designing something.
 
LDOVR TPS7A4700 modules are so affordable that I can't make them myself for a lower price. Very low noise and LDO so working perfectly with a 6V 10VA transformer. Small, low heat, green, the works. I suggest to use a larger heatsink though.

LM723 works OK and although it is a dinosaur it still is competitive with regards to low noise. I know as I have designed and built single board 5V PSU's with it fed by 8V transformers. This reminds me I have ready soldered so complete ones (with onboard transformers) in stock. Minimal size, CLC filtering, RF proof, current limiter/short circuit protection, very low noise but 5V limited to 0.25A by the mini transformer. Transformer can be changed of course.
I'm in Sweden now. Could make a few changes and 250mA would be sufficient but need 115VAC input. Any measurements on your supply that you can share? Price?
 
Almost verbatim copy from the LT3042 data sheet , page +/-17 IIRC.
Works as promised. noise density 2nV/rtHz. A few boards may be left over for postage.
I tested it with 6.6V in and 5V out, 1.5A
 

Attachments

  • lt3042NPN_xtraSmall.png
    lt3042NPN_xtraSmall.png
    84.3 KB · Views: 594
  • Like
Reactions: YashN
Excuse the self-promotion, but various iterations of the denoiser/dienoiser/nonoiser, including the developments by Trileru amongst others are very high performance, cheap, and use only ordinary parts: it will even work with generic, Chinese-made parts.
If you need really stellar specs, you simply have to use a ZTXxxx transistor as correction transistor, it will not not cost you an arm and a leg.
Regarding real HF PSRR, nothing can beat a good combination of ordinary passives.
 
  • Like
Reactions: YashN and vacuphile
Excuse the self-promotion, but various iterations of the denoiser/dienoiser/nonoiser, including the developments by Trileru amongst others are very high performance, cheap, and use only ordinary parts: it will even work with generic, Chinese-made parts.
If you need really stellar specs, you simply have to use a ZTXxxx transistor as correction transistor, it will not not cost you an arm and a leg.
Regarding real HF PSRR, nothing can beat a good combination of ordinary passives.
Which one of the 3?! What is the final version? There are many trees in the forest.
 
Indeed! This one is an improved nonoiser and manages ~0.2nV total noise, using just ordinary BC8xx:
https://www.diyaudio.com/community/...grade-any-317-based-v-reg.331491/post-7094882
With a ZTX, it would go below 100pV.
The regular, plain-vanilla denoiser already offers excellent performance for an unbeatable price/performance ratio.
The dienoiser with a ZTX input is also stellar: >150dB PSRR, ~100pV noise.
Let's ask @Trileru which of his iterations offers the best combo in his opinion
 
It is very hard to follow with regards to countless experimental versions (superseded within a short time) and I thought it was all meant as an upgrade for existing LM317 designs?!

Isn't choosing a modern low noise LDO regulator and small single PCB 5V 1A design for a 5V 0.7A HPA a more obvious choice? OP also needs mains AC to 5V DC so a complete PSU. It is not for a lab instrument, it is meant for a probably relatively small sized HPA.

10VA transformer isn't enough for this.
The power requirement of source is 9VAC 2A, so must to use at least 40VA transformer, else it will be too hot.
A lower AC voltage would be optimal when a modern LDO would be used. 10VA is more than enough unless one listens to headphones at maximum volume 24/7. Tip: don't do that.
 
Last edited:
A lower AC voltage would be optimal when a modern LDO would be used. 10VA is more than enough ...
Are you sure?
No such option (even 5VAC transformer), when 10VA transformer enough (not burn out) at 0.7A continuos load.

7.5 .. 5V raw DC.jpg

The first capacitor in the PSU loading the transformer more, than static load.
If you decrease this capacitor, the noise (sawtooth "hum") will be growing.
 
It is very hard to follow with regards to countless experimental versions (superseded within a short time) and I thought it was all meant as an upgrade for existing LM317 designs?!
Initially, yes. The dienoiser still falls into the category, but the nonoiser and "children" should preferably be built from scratch
Isn't choosing a modern low noise LDO regulator and small single PCB 5V 1A design for a 5V 0.7A HPA a more obvious choice? OP also needs mains AC to 5V DC so a complete PSU. It is not for a lab instrument, it is meant for a probably relatively small sized HPA
Yes, a good modern regulator can give excellent performance without the headaches and hassle of building a custom circuit, even if it is cheaper and higher performance.
It makes sense if you are after the ultimate performance (very rarely needed) or lowest cost (which makes sense if you are building tens or hundreds of them).
There is another aspect to consider though: LM317/37 have been there forever, and will remain available for a very long time. Newer, specialty circuits tend to have a shorter commercial life, and are easily discontinued.
Commodity items are almost immune to that kind of issue
 
Are you sure?
No such option (even 5VAC transformer), when 10VA transformer enough (not burn out) at 0.7A continuos load.

View attachment 1093779

The first capacitor in the PSU loading the transformer more, than static load.
If you decrease this capacitor, the noise (sawtooth "hum") will be growing.
If you build stuff in real life you would know this is nonsense. It is also not 0.7A continuous load but if it would be then it would just work ok.

Initially, yes. The dienoiser still falls into the category, but the nonoiser and "children" should preferably be built from scratch
Please point the OP to the final design and PCB.
There is another aspect to consider though: LM317/37 have been there forever, and will remain available for a very long time. Newer, specialty circuits tend to have a shorter commercial life, and are easily discontinued.
Commodity items are almost immune to that kind of issue
Assumption. LM317/337 could be discontinued as we speak. That newer regulators have a shorter service life is again an assumption. I have PSUs running for many years 24/7.

What do we expect from PSUs?! That they outlive us, our children AND grandchildren?
 
Last edited: