The snubber is C1, C2 and R1.
You seem to be talking about the CRC filter preceding the regulator, aren't you?
The problem with the rectifier snubber is that is specific of the transformer you are using. That's why you need the Quadimodo + oscilloscope to select the snubber values.
Nope, talking about the coupled RC snubber pre-bridge.
CRC can be spoken of as a pi-filter but also snubber topology. It is not synonymous to one. I.e. you can have a CRC snubber or just RC snubber.
Edit: Having the snubber as an option is better than none at all. "Needing" Cheapmodo/Quasimodo/maths isn't correct. Optimising it is the most ideal, but even if you just throw in default Cx, Cs and 100ohm Rs and you're still better off than jumpering.
Last edited:
I’m not claiming that double copper lanes are essential or responsible for low output impedance, but I would not put there 1 mm copper lane with explanation that regulator will take care of everything (although it almost can).
So I officially change PCB description to “PCB uses both copper sides and has proper Kelvin connection.”
Now, that is taken care of this issue we can move forward.
Denoiser/dienoiser provides output impedance in range of mico to mili ohm. On standard LM317 layout, resistance of pcb copper lane between regulator output pin and output connector is directly added to the total output impedance. However, having Kelvin connection, this copper lane is inside regulation loop and is compensated. So my approach is that is better to have less to compensate for and there is abundance of unused copper on the pcb.
As for snubber, it’s there for anyone having means to use it. Others can skip that. I was little selfish during design accounting only for my needs.
Single rail LM317 denoiser/dienoiser gerber files
Here are gerber files in format accepted by JLCPCB, for anyone willing to risk a little (4$ + P&P).
Again I point out that this design is still untested and I’m awaiting my set of pcbs. I’m convinced that design is without any errors, but then people are convinced about many wrong things.
Here are gerber files in format accepted by JLCPCB, for anyone willing to risk a little (4$ + P&P).
Again I point out that this design is still untested and I’m awaiting my set of pcbs. I’m convinced that design is without any errors, but then people are convinced about many wrong things.
Thanks Tombo. Going to do a quick run of PCBs soon so will add this to the list and let you know.
Sadly, also won't be able to post results for a while, but will try my best.
Will also test out Sadface's version 2.
That is just great. I hope other members will benefit from that.
I have in the works dual rail versions with LM & LT regulators. They were completely finished but I decided to rearrange components a little and do some optimizations.
If one’s goal is to have very good voltage regulator for the audio circuitry purposes, using standard components, then I think that we should concentrate on LT versions (Nonoiser/Dienoiser).
That sounds like a reasonable plan. How far along are you? I'm happy to wait so can get it done in one batch with testing etc.
Looking at Sadface's V2 layout, it's going to be tricky to add snubbing given the input and bridge is quite crowded without increasing area?
Last edited:
It won’t be very soon as I have other priorities and no immediate need for those versions. Better go ahead.
I was considering adding the transformer snubbers as it’s something I have started using in my other power supplies. However I decided it was too hard without taking up more space.
It is also tricky to maintain the flexibility between a centre tap transformer and one with dual secondaries. My understanding is that with a centre tap the snubber goes rail to rail. However with dual secondaries we need dual snubbers.
It could be done be done but at a probably large space penalty.
Class x2 caps are a bit larger in footprint and my understanding is that one really should use x2 for transformer snubbers.
Am I incorrect?
That would be great if you intend to design a single rail LT version along with dual rail version. I mean, like LM version you shared.
And maybe a mixed version too? LT at positive rail, LM at negative rail.
Just a thought for V2:
LV mentioned a trimpot to set +V wouldn't be ideal due to component quality vs fixed resistor. What are your thoughts on sticking with single R4? Obviously can jumper the trimpot connections. Would welcome your thoughts.
Single rail LT version won’t be a problem after I finish dual rail version. But don’t expect that to be very soon. Dual rail LT version uses LT1963 / LT3015 and I don’t see what would be purpose of mixed LT/LM version?
I think that this excellent Elvee’s design deserves more attention and presence.
That is the reason I have put trimpot and resistor in series, so trimpot can be of small value, needed only for precise voltage adjustment, and of possibly smaller negative effect.
For next version, I will put pads for the fixed resistor along with pads for the trimpot.
I thought you were referring to LT1083/4/5 which can accompany with LM337 instead of not so common LT1033. I just saw that LT1083 is pin compatible with LM317, thus dual LM pcb would be enough for both 3pin LT and LM's.
I was just checking my simulation files, and also doing a search (not page by page) of the D-Noizator topic, and I couldn't find a good implementation, better than with 317, for the LT1963. Is there one?
