You could also use these small smd boards for negative lm337 regulators, just that you need to flip the diode polarity and transistors need to be flipped upside down and rotated. It's doable but a bit more finicky. I suggest you are very careful for lm337, triple check everything and test it on an expandable circuit, don't add it directly into the final application if that happens to be a nice piece of gear.
edit: in this photo I didn't rotate the diode, just the bjt for illustration purposes. You need to rotate the diode and bjts ONLY for negative LM337!
edit: in this photo I didn't rotate the diode, just the bjt for illustration purposes. You need to rotate the diode and bjts ONLY for negative LM337!
Last edited:
I remember Elvee mentioning that it shouldn't matter. But maybe he can chime in on this as well.
Also get some 10R/22R/33R/47R for R6/R7 so you can adjust for best performance. I'm not sure what the best values are for these. Denoiser doesn't seem to use R7, and I don't think it's even recommended for Dienoiser on negative lm337 regs, only for Dienoiser on positive lm317 regs.
The PNP transistors need to be flipped and rotated? I will check that very closely and breadboard a LM337 to test the board before installing it. Another reason I just wanted to use the DeNoiser. BTW, my C3's I have on hand are some nice Nichon's 13mm in diameter. I have space on the Hafler board to lay them on their side, probably put a dab of silicone under each of them. I will put the DeNoisers right at the spot where the cap connections go into the board to shorten lead length.
Glad you mentioned flipping the diode, duh, I already knew the transisitor and caps had to be oriented correctly but hadn't thought about the diode.
Glad you mentioned flipping the diode, duh, I already knew the transisitor and caps had to be oriented correctly but hadn't thought about the diode.
The resistor should not be used with the denoiser. It is probably required for the dienoiser
The final recommended value is 10nF for the denoiser
The BC817 (SMD version of the BC337) has a slightly lower noise than the BC550 in this circuit.
For the denoiser, it makes no difference, but for the dienoiser or nonoiser, it matters
The gain does not really matter: a lower Hfe is correlated to a higher Early voltage, but in this circuit this will lower the Ic, thus the transconductance, meaning the gain available for the correction will remain ~the same
Denoiser uses a NPN transistor only. Dienoiser uses a NPN + PNP. For negative lm337 applications on any of the Denoiser/Dienoiser versions all BJT transistors and diode need to have polarity flipped. For the diode it's easy as you just mount it the other way around. For the NPN and PNP transistors it's a bit more fincky as they have to be flipped and then rotated as in the photo. It's doable tho.
You can lift the ground pin of C3 from the board and route it to the ADJ connection on the add-on pcb. But then if you don't have any C3 on the add-on board, you need to short the C3 pins on the add-on board, so the ADJ signal gets into the denoiser circuit. 13mm won't fit on the tiny add-on board. Or you could but it wouldn't be pretty. The quality of the cap shouldn't matter. Only the value, 220uF minimum, 470uF if you can afford the space. But low esr/boutique is irrelevant here. I say just use any that fit properly. I would just remove the onboard adj cap and use a 6.3-8mm one on the add-on board, and use the adj connection leftover from the original adj cap to tap into the adj connection on the original circuit.
Risking opening a can of worms, I recommend you just whack out any electrolytic capacitor on the output of LM317/LM337 and use a single (for each lm317/lm337 output) Panasonic FC 82uF or 100uF/63V rating. They seem to have the ESR in the ballpark for the best results with the De/Dienoiser circuits:
http://industrial.panasonic.com/cdbs/www-data/pdf/RDF0000/ABA0000C1209.pdf
Also remove any 100nF ceramic or film caps that are bypassing the output caps of the regulators. A single 82uF or 100uF is adequate. And more specifically the Panasonic FC I recommended is good quality and 105C rating.
I have 1000uF || 100nF on the regulator outputs of my Magni 3 and those will get removed and replaced with the above cap when I'll implement the denoisers on it.
Getting all these things right I think will make for a successful implementation of the circuit. Most people that had problems with this have not used the recommended values/layout.
You can lift the ground pin of C3 from the board and route it to the ADJ connection on the add-on pcb. But then if you don't have any C3 on the add-on board, you need to short the C3 pins on the add-on board, so the ADJ signal gets into the denoiser circuit. 13mm won't fit on the tiny add-on board. Or you could but it wouldn't be pretty. The quality of the cap shouldn't matter. Only the value, 220uF minimum, 470uF if you can afford the space. But low esr/boutique is irrelevant here. I say just use any that fit properly. I would just remove the onboard adj cap and use a 6.3-8mm one on the add-on board, and use the adj connection leftover from the original adj cap to tap into the adj connection on the original circuit.
Risking opening a can of worms, I recommend you just whack out any electrolytic capacitor on the output of LM317/LM337 and use a single (for each lm317/lm337 output) Panasonic FC 82uF or 100uF/63V rating. They seem to have the ESR in the ballpark for the best results with the De/Dienoiser circuits:
http://industrial.panasonic.com/cdbs/www-data/pdf/RDF0000/ABA0000C1209.pdf
Also remove any 100nF ceramic or film caps that are bypassing the output caps of the regulators. A single 82uF or 100uF is adequate. And more specifically the Panasonic FC I recommended is good quality and 105C rating.
I have 1000uF || 100nF on the regulator outputs of my Magni 3 and those will get removed and replaced with the above cap when I'll implement the denoisers on it.
Getting all these things right I think will make for a successful implementation of the circuit. Most people that had problems with this have not used the recommended values/layout.
Last edited:
View attachment Scan20201018113227.pdf
The Hafler circuit is a little different. See the attachment. C35&C37 are 100nF. The larger caps are way down the rail, on the phono section, C17&C18 are the 470uF filter caps. There is another set for the line section. I've replaced all the electrolytic caps in the preamp, none of them fancy, mostly Nicheon general purpose.
The Hafler circuit is a little different. See the attachment. C35&C37 are 100nF. The larger caps are way down the rail, on the phono section, C17&C18 are the 470uF filter caps. There is another set for the line section. I've replaced all the electrolytic caps in the preamp, none of them fancy, mostly Nicheon general purpose.
The only costly part was the 10nF caps, almost $2 a piece.
https://cpc.farnell.com/c/electronic-electrical-components/capacitors/10nf-capacitors
https://cpc.farnell.com/c/electronic-electrical-components/capacitors/10nf-capacitors
I'd remove D7 and D10 and replace them with a resistor for the needed output voltage. I think someone made sims earlier in this thread and with the denoiser the resistors had the best result.
I would remove the 100nF C35 and C37 and replace them with 47uF/100V Panasonic FC. Also check the esr of or replace the 470uF C17 and C18. How far are they physically in the circuit? If not far then the 82-100uF Panasonic FC should be ok to replace them with.
The idea is to have the total ESR of the output caps in the range of 200-230mOhm. Paralleling caps lowers the value. Small 100nF caps have very low esr. They should be removed.
If you are absolutely sure that resistor is 270R then a 4.7K resistor would give you bang on 23V.
If C17 and C18 have some tens of cm between the output of the regs and the phono stage then you could leave them and only replace C35 and C37 with a low value Panasonic FC, something like 68uF/100V.
Can you measure the voltage output on both regs? How close is it to the 23.2V in the schematic? I don't think .2V matters that much in this context.
Also do note that the Denoiser circuit has an initial higher swing. You could expect 24.5V at startup for a very short while. The Dienoiser circuit has a way lower swing, much closer to the nominal output, you can expect a few hundred mV at most. I'm not sure it matters, the denoiser shouldn't go past the 25V that some caps might be rated for, on those rails.
Do make a breadboard test and note voltages. Get some 270R and 4.7K 1% or better, so you can emulate the conditions on a breadboard.
If C17 and C18 have some tens of cm between the output of the regs and the phono stage then you could leave them and only replace C35 and C37 with a low value Panasonic FC, something like 68uF/100V.
Can you measure the voltage output on both regs? How close is it to the 23.2V in the schematic? I don't think .2V matters that much in this context.
Also do note that the Denoiser circuit has an initial higher swing. You could expect 24.5V at startup for a very short while. The Dienoiser circuit has a way lower swing, much closer to the nominal output, you can expect a few hundred mV at most. I'm not sure it matters, the denoiser shouldn't go past the 25V that some caps might be rated for, on those rails.
Do make a breadboard test and note voltages. Get some 270R and 4.7K 1% or better, so you can emulate the conditions on a breadboard.
Check post no 496 and 498:
D-Noizator: a magic active noise canceller to retrofit & upgrade any 317-based V.Reg.
Regarding diodes as R2. If you want to improve the performance of the regulators then you need to remove the diodes, or else you are not going to benefit from the denoisers.
Also another way to apply the denoiser/dienoiser in your case would be to tap the gnd/vcc directly on the caps from the phono, and run the adj wire from the add-on board to the adj cap that you remove from the regulators adj pin. I remember Elvee mentioning that you could move the whole denoiser circuit to the point of load. I'm not sure how much you can run the ADJ wire for, maybe Elvee can chime in with this as well, but it would be interesting to try. In this case you need to make sure that C17 and C18 are in the right ballpark for ESR, around 0.2R. Or just replace them with the Panasonic FC 100uF/63V cap if doing it like this. You'd need some measuring gear to check the result tho. Either way you need to replace D7 and D10 with resistors (4.7K) if using any of the denoiser versions.
D-Noizator: a magic active noise canceller to retrofit & upgrade any 317-based V.Reg.
Regarding diodes as R2. If you want to improve the performance of the regulators then you need to remove the diodes, or else you are not going to benefit from the denoisers.
Also another way to apply the denoiser/dienoiser in your case would be to tap the gnd/vcc directly on the caps from the phono, and run the adj wire from the add-on board to the adj cap that you remove from the regulators adj pin. I remember Elvee mentioning that you could move the whole denoiser circuit to the point of load. I'm not sure how much you can run the ADJ wire for, maybe Elvee can chime in with this as well, but it would be interesting to try. In this case you need to make sure that C17 and C18 are in the right ballpark for ESR, around 0.2R. Or just replace them with the Panasonic FC 100uF/63V cap if doing it like this. You'd need some measuring gear to check the result tho. Either way you need to replace D7 and D10 with resistors (4.7K) if using any of the denoiser versions.
Last edited: