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

A rectification on the previous post, the sim was made for 12Vout from LM317. PSRR and output impedance take a hit at 3.3Vout. So for digital technically LT3042 and ADM7150 should perform better as PSRR/output impedance goes vs LM317+dienoiser. Not much else can be done to improve the output impedance.
For PSRR improvement one could add a cap multiplier in front of the regulator. BOM for inductor + complete cap multiplier + lm317n sot223 + dienoiser with everything included (protection diodes/caps/bjt) minus resistors is still a tad cheaper than a single LT3042 without anything else. So for cheaper than LT3042 we get better PSRR at all frequencies. This is the configuration I used on the previous posted DAC pcb.

Output impedance is worse but not by far, I mean there's 30mOhms or something like that at higher frequencies vs ADM7150. Those can easily be lost in traces. That's the resistance of a 6cm long 40mil wide track to give a sense of it. Not that far off that it puts it in a whole other league. Even so I still think it wouldn't make perceptible difference for audio. Might do, but I personally don't think it translates into audible differences.
What strikes me is the LT3082 performance, I'm starting to think the model is not accurate tho it's native to LTSpice. I've added it in both PSRR and output impedance sims, with cap multiplier and dienoiser at 3.3Vout.
So my bad on the previous sim, I wasn't careful at output voltage and got hyped. Tho I still think LM317+dienoiser is a very good performer and the small differences (at 3.3Vout) could easily be lost in pcb design.
edit: PSRR for LM317+dienoiser is with L and cap multiplier, the same for LT3082.
 

Attachments

  • Screenshot from 2020-12-16 17-33-30.png
    Screenshot from 2020-12-16 17-33-30.png
    45.2 KB · Views: 661
  • Screenshot from 2020-12-16 17-49-33.png
    Screenshot from 2020-12-16 17-49-33.png
    49.4 KB · Views: 658
Last edited:
thank you for this
worth a try
very nice as this can be used for negative polarity also Instead of LM337
Is there a BOM

And the same thing for only the LM317. You can also use LT108x for this and the previous board, forgot to mention. Just needs very low ESR on the output cap (will test with 1206 ceramic) and some compensation, I think around 40-50nF/3.9R or so. When I get to testing I hope to get values for most of them.
Board size is around 37mm x 93mm.
Also warning, this board has not been tested. You make it at your own risk!
 
I made a bom for all 4 versions, single/dual tht/smd. The tht versions are diy-able and also have gerber files for fab. The smd versions have separate projects for diy and fab house.
I removed the tht sense connectors as I don't think you'll get good results with simple wires for remote sensing. I added u.fl connectors for all sense lines. If you absolutely want to, you can solder wires directly on that footprint, but I recommend using u.fl pigtails, can be found real cheap on ebay. Or maybe scavenall versions.ge some old wireless router. Theoretically you should have better results. I'll test this when I get my ordered parts. On the sense side of the coax be sure to keep the shield unsoldered, it should cover the middle signal cable as close as practical to the sense point but make sure it's not touching the sense wire or other parts of the circuit. Solder just the middle wire of the coax to the signal you're sensing (gnd or Vout). I added the cheapest u.fl connectors I found on Mouser, I used that footprint on all boards.

I made some minor tweaks to all boards. THT versions were missing the protection diode for cap multiplier. I added a fuse to the single smd version. There's extra smd footprints for output caps on all boards for where low esr is needed (lm337 or lt3082/lt1085 etc). 10uF 1206-1210 caps should work fine.
I didn't include any resistor in bom, you get those in your preferred type, they should be either 6.3/2.5mm for tht either 0805 for smd. I specified the required values in the BOM and quantity for each value. Make sure to also get 0R 0805 for different bypasses, like making a denoiser instead of a dienoiser or bypassing lm317 output cap esr correcting resistor. The caps I put in BOM should have the correct ESR range for both tht and smd versions for LM317.
Dienoiser should work fine with 560R resistor for most voltage ranges. I wouldn't stress with this value. You could adapt it to the output voltage but gains are minimal in the grand scheme of things.
For the smd diy versions you must link the voltage setting resistor gnd to the gnd trace using a wire link. This has to be done on both regulators on the dual smd diy version. There's a pad left for this near each resistor. Also for the smd versions the smd variable pot is TC33X type, you get the value you need depending on your output voltage and resistor value you use in parallel with it.
For some reason Mouser does not stock BC560 anymore so I put 2N4403 instead. Using LTSpice seems to give almost identical result to BC560 in the cap multiplier.
THT version has 35V input caps. For smd version I was limited to 25V for the 220uF from cap multiplier. If you want to go higher there's a 150uF/35V max capacitance cap available in 6.3mm diameter, smd. But you also need to up the voltage rating on the 100uF cap at the input of the regulator. It's 25V atm.

For THT dual version I didn't add any to220 rectifying diode. You can choose what type you like, Schottky or regular etc. That's needed only if you want to use the dual supply without a center tapped transformer. Mind the current rating.
For the compensation network for dienoisers I recommend getting 10nF/22nF/33nF and 47nF caps just so you have enough to cover most needs. Get 35V rating at least, al 0805 package. I'd put the 3.9R resistor as well, but for denoiser you could bypass it.
For LM317 output cap esr correcting resistor I'd get a few values between 0.1R and 0.25R to cover any situation if you want to use other caps than the ones I put in the bom.
For the smd fuses you can use any 1206-1210 smd package, just that you need to figure out the current rating needed depending on your output current. Play with LTSpice to see the current spikes and figure out what value is adequate. If you have inline fuses at transformer then bypass with 0R 1206-1210 package.
If you don't want remote sensing you can close the smd jumpers on all board versions, you'll get the low noise on the output of the board. Should be good enough for most needs, but if you want you can go with coax pigtails. I'll report on this when I get to test it.
I consider these rev 1.1, I made a diy version of the rev 1 dual smd board and I'll test it once my parts arrive. I should be able to test LT3082 as well, and also the smd and tht versions of LM3x7N. I'll post the comp values I end up with for each. Rev 1.1 versions should work the same, they are not that different in layout.
The tht single version is a bit shorter now because I removed the sense tht connector.
I also increased the thickness of the output traces for lower output impedance at the connector. There's also some alternative 1206/1210 cap footprints for reg outputs, closer to the reg. Might test to see if it makes any difference.
BOM has 4 sheets for each version.
Also warning! these designs have not been tested yet. You make them at your own risk!
 

Attachments

  • dual_smd_fab.zip
    230.6 KB · Views: 53
  • single_smd_fab.zip
    125 KB · Views: 52
  • dual_smd_diy.zip
    201.1 KB · Views: 52
  • single_smd_diy.zip
    94.4 KB · Views: 56
  • dual_tht.zip
    298.6 KB · Views: 72
  • single_tht.zip
    186.3 KB · Views: 77
  • bom.zip
    34 KB · Views: 70
Last edited:
And traces layout for all versions and how the u.fl connectors look. The single tht version is now 88.5mm long.
 

Attachments

  • Screenshot from 2020-12-21 21-38-59.png
    Screenshot from 2020-12-21 21-38-59.png
    103.7 KB · Views: 124
  • Screenshot from 2020-12-21 21-39-16.png
    Screenshot from 2020-12-21 21-39-16.png
    113.9 KB · Views: 413
  • Screenshot from 2020-12-21 21-39-35.png
    Screenshot from 2020-12-21 21-39-35.png
    77.8 KB · Views: 411
  • Screenshot from 2020-12-21 21-40-12.png
    Screenshot from 2020-12-21 21-40-12.png
    100.1 KB · Views: 412
  • Screenshot from 2020-12-21 20-46-49.png
    Screenshot from 2020-12-21 20-46-49.png
    519.9 KB · Views: 128
  • Screenshot from 2020-12-21 20-46-17.png
    Screenshot from 2020-12-21 20-46-17.png
    322 KB · Views: 126
  • Screenshot from 2020-12-21 20-45-07.png
    Screenshot from 2020-12-21 20-45-07.png
    417.8 KB · Views: 102
  • Screenshot from 2020-12-21 20-45-36.png
    Screenshot from 2020-12-21 20-45-36.png
    347.7 KB · Views: 92
  • Screenshot from 2020-12-21 21-46-02.png
    Screenshot from 2020-12-21 21-46-02.png
    491.3 KB · Views: 150
  • Screenshot from 2020-12-21 21-45-25.png
    Screenshot from 2020-12-21 21-45-25.png
    603.9 KB · Views: 135
Last edited:
maybe you can think to add Mark Johnson snubbers
also Is this C L C filter so much better than say C R C
L part take much place than resistor


If the snubbers are before the bridge you could add them on the transformer output.
The L cuts HF noise from the supply. It also keeps the output impedance down into the regulator. The resistor can also get hot depending on the drop.
You can bypass the inductor if you feel you don't need it. I put it to cut down HF stuff. You don't really need a resistor considering the PSRR of the whole supply.
 

Attachments

  • Screenshot from 2020-12-22 17-40-29.png
    Screenshot from 2020-12-22 17-40-29.png
    106.6 KB · Views: 136
If you want it even shorter without R or L then you can try this. It's 80mm x 37mm.
Edit: that inductor current rating was for the 1.5A max output of LM317. Without it you can try higher output currents with LT1085 etc. I also made the tracks wider at 90mils. But diodes might be an issue at higher currents.
 

Attachments

  • Screenshot from 2020-12-22 18-13-16.png
    Screenshot from 2020-12-22 18-13-16.png
    686.4 KB · Views: 342
  • single_tht.zip
    175.7 KB · Views: 56
  • Screenshot from 2020-12-22 18-29-07.png
    Screenshot from 2020-12-22 18-29-07.png
    138.3 KB · Views: 334
Last edited:
For 3.5A and up you'd need a TO220 package for the diodes. There's options with two Schottky diodes in one TO220 package, you'd need to get one for each of the two configurations, common anode and common cathode. With two of these the board can be at 84mm, and there's room for an extra 12.5mm cap on the input.
edit: with this configuration you'd have to add to BOM around 3$ for both diodes and another 1.5$ for the extra cap. Apart from the LT108x which is more expensive. But now you start looking at 3.5-5A output. And with a dual secondary transformer or two smaller transformers you can go dual rail. Max input would be around 30V on the caps.
 

Attachments

  • Screenshot from 2020-12-22 20-14-52.png
    Screenshot from 2020-12-22 20-14-52.png
    26 KB · Views: 322
  • Screenshot from 2020-12-22 20-05-53.png
    Screenshot from 2020-12-22 20-05-53.png
    200.7 KB · Views: 319
Last edited:
I played a bit with LTSpice and the above configuration should work to 3.5A output. You'd need a transformer with 24VAC rms on the secondary and you'd get 24VDC output at 3.5A. You'd need a low resistance secondary, I didn't model that. The whole thing is going to dissipate towards 30W of heat at full tilt, and output ripple noise should be around 22nVrms at 3.5A.
5A would be a bit more difficult.
 

Attachments

  • LT1085_3.5A.asc
    6.7 KB · Views: 42
  • Screenshot from 2020-12-22 21-45-35.png
    Screenshot from 2020-12-22 21-45-35.png
    115 KB · Views: 105
  • Screenshot from 2020-12-22 21-48-55.png
    Screenshot from 2020-12-22 21-48-55.png
    111 KB · Views: 312
Hi Trileru,
I thought Inductor pcb Is also 80 mm long
Can you please try to add resistor to two 2200 uF caps to get C R C
I think that you can get some space If you short rectifiers legs a bit and also move back the two rectifiers closer to Input connector

If you want it even shorter without R or L then you can try this. It's 80mm x 37mm.
Edit: that inductor current rating was for the 1.5A max output of LM317. Without it you can try higher output currents with LT1085 etc. I also made the tracks wider at 90mils. But diodes might be an issue at higher currents.
 
I made a test now with 1R, and at 12V/1A output with around 18V input, and apart from the around 1.5W dissipation on it, the output ripple noise without it is around 20something nVpp vs less than 1nVpp with it. It does something, just that I don't think it matters to the end result. It also increases your input voltage needs. You can't afford any R there with the above configuration 24VAC in 24VDC/3.5A out with LT1085. You can increase the input voltage but I don't see it as being practical in this case.
But if you really want I could replace the inductor with a resistor, just tell me what footprint you want.
 
you are right - there Is no need for resistor
would you please confirm that single tht LM 317 regulator with dienoiser and C L C pcb from post 1866 Is 80 mm long
do you think Is a good Idea to move back few mm D5 and two diodes closer to Input connector and restrict their footprint In order to free space
do we need heatsink for TO220 transistor at cap multiplier and LM 317 at 12V / 1 A output as there Is no place for heatsink on pcb
 
Last edited:
For sure you need a heatsink, just that you mount it on the side of the pcb. You can mount them to a case, you can also drop them on the bottom of the case if it's good for moving heat away from them.
I made rev 1.2 without the inductor, and optimized the output trace. It's now shorter, I moved the output connector. The board is now also shorter at 74mm x 37mm.
 

Attachments

  • Screenshot from 2020-12-23 14-16-54.png
    Screenshot from 2020-12-23 14-16-54.png
    339.8 KB · Views: 120
  • Screenshot from 2020-12-23 14-18-36.png
    Screenshot from 2020-12-23 14-18-36.png
    106 KB · Views: 111
  • single_tht_rev1.2.zip
    177.7 KB · Views: 56
And the same output layout for the higher output current LT1085, with three input caps. This is only for fab house as it's two sided.
Rectifying diodes should be a dual common anode and a dual common cathode. Something like MBRF10100CTR and MBRF10100CTL should work. The board should allow for 24VAC rms input and 24VDC/3.5A max output.
I think these designs should cover most needs. I don't plan on making others at the moment. I made add-on boards smd and tht single and double, and also full supply boards single and double tht and smd format, so should cover most needs for lower current to 3.5A output with this one.
I'll try and post the smd version comp values, the THT versions should work with the values I previously found in my tests in earlier posts.
As usual warning! This design (and one in previous post) has not been tested, you make it at your own risk!
 

Attachments

  • Screenshot from 2020-12-23 14-47-36.png
    Screenshot from 2020-12-23 14-47-36.png
    336 KB · Views: 113
  • Screenshot from 2020-12-23 14-47-13.png
    Screenshot from 2020-12-23 14-47-13.png
    62.1 KB · Views: 111
  • lt1085_3.5A.zip
    166.1 KB · Views: 57