There is a linear selling on ebay which is working fantastic for a application I have, but, its abnormally noisy and a bit unstable. I made a schematic for it and I am confuzed.
Does this look sane ? Its weirdly unstable. Certain loads cause a oscillation. No load has it too. It has a 4mV pk-pk very random looking noise.
Its VERY picky on which mfgr of TL431 and even grade.
I can't find any reference to this layout on-line.
Does this look sane ? Its weirdly unstable. Certain loads cause a oscillation. No load has it too. It has a 4mV pk-pk very random looking noise.
Its VERY picky on which mfgr of TL431 and even grade.
I can't find any reference to this layout on-line.
Attachments
Why would you feed the TL431 from the input via the 220R? Why not from the OUT pin? That 100nF to GND from its adjust pin is almost certainly going to make it oscillate wildly or maybe just amplify its self-noise crazily.
The 100n cap is not from the adj pin but from the Vref pin of the 431 to ground.
The regulator is unstable because 431 is no compensated, that 100n cap does nothing good where it is put.
But, as i have said, why would you use a 431 to adjust the adj pin of a 317?
It's stupid...
LM317 is it's self an adjustable and stable regulator, it does not need another EA...
The regulator is unstable because 431 is no compensated, that 100n cap does nothing good where it is put.
But, as i have said, why would you use a 431 to adjust the adj pin of a 317?
It's stupid...
LM317 is it's self an adjustable and stable regulator, it does not need another EA...
That's what I meant - the pin which is used to adjust the output voltage of the TL431. I didn't use the datasheet word.
There would be some value in feeding the LM317's adjust pin externally if the voltage reference used were to be lower noise than the bandgap inside the LM317. So for example using a zener diode there can give lower noise than the usual resistors. But a TL431 is probably just as noisy as the internal bandgap, so no additional value there.
This made almost no sense to me either... I agree with all these comments.. It is no surprize its unstable... I agree about the .1uF cap to ground.
YEA I agree its crazy.. In a number of ways.. Its why I am posting.. Good to know I am not crazy..
Here let me post some board pics to confirm the schematic.
I will also post a vid of the noise..
YEA I agree its crazy.. In a number of ways.. Its why I am posting.. Good to know I am not crazy..
Here let me post some board pics to confirm the schematic.
I will also post a vid of the noise..
Attachments
Looks like something a 14 year old with too much internet time on his hands has dreamed up.
I call it garbage.
I call it garbage.
I will reverify the caps and schematic.. I admit its, uhhh, annoying to reverse engineer a board.. haha...
This seems to be landing either in the TL431 instability zones OR its just wonky and unstable and oscillation.
The weird part is, for what i am doing its terrific.. I have been playing with a LOT of linears, including fully restored and updated Lambdas,, but this one has really stood out..
The transistor they used is interesting too. Interesting part selection..
I was thinking that maybe using the weird properties of a TL431 was maybe being used in some odd way here in a way I did not follow..
It does like its running high gain and unstable under some loads..
This seems to be landing either in the TL431 instability zones OR its just wonky and unstable and oscillation.
The weird part is, for what i am doing its terrific.. I have been playing with a LOT of linears, including fully restored and updated Lambdas,, but this one has really stood out..
The transistor they used is interesting too. Interesting part selection..
I was thinking that maybe using the weird properties of a TL431 was maybe being used in some odd way here in a way I did not follow..
It does like its running high gain and unstable under some loads..
I know huh ? Except.. The part selection like a R-Core and that output transistor seem pretty good.. The TL431 is weird enough with odd modes of use, I thought I would post..
This seller seems pretty popular on ebay..
https://www.ebay.com/str/pinshuche?_trksid=p2047675.m3561.l2563
I am very happy with the linear. I have done a LOT of AB with a LOT of linears. I wanted to add remote sensing and that was when i noticed the odd topology. WELL, I had previously tested it and seen the odd high noise level..
Varying that .1uf WIMA above or below the 0.1uf value lowers the noise. It seems tuned. 0.1uF seems to peak out the noise. Like high gain at a specific freq..
Moving to a different type/brand of TL431 causes it to jump into 100mV of ringing with all the same parts. Yes I know different TL431 can be very different internally and have different specs.
For what i am using it with its perfectly stable and the noise is meaningless. Its just very odd..
https://www.ebay.com/str/pinshuche?_trksid=p2047675.m3561.l2563
I am very happy with the linear. I have done a LOT of AB with a LOT of linears. I wanted to add remote sensing and that was when i noticed the odd topology. WELL, I had previously tested it and seen the odd high noise level..
Varying that .1uf WIMA above or below the 0.1uf value lowers the noise. It seems tuned. 0.1uF seems to peak out the noise. Like high gain at a specific freq..
Moving to a different type/brand of TL431 causes it to jump into 100mV of ringing with all the same parts. Yes I know different TL431 can be very different internally and have different specs.
For what i am using it with its perfectly stable and the noise is meaningless. Its just very odd..
Last edited:
Yea I ohmed it out and took that cap off the board to confirm it.. It is as depicted in the schematic..
I did find a error tho. Which really makes it more bizzare.. The pot is 50K.. So that makes the total 70k across that.. Very high impedance for that TL431 adj pin.. This causes my brain to hurt.. So that makes it high gain ? Or, WTF.. I will have to plug this into the data sheet math..
How weird...
The parts are all good quality, for its price range, and clearly mindfully picked. This looks very much like it was well crafted and maybe well engineered.. Or not..
I have not tested things like transit response and impedance spectra.
Almost everything I have tried to do to improve it has pretty much failed. I did replace the rect diodes with Schottky diodes and that was a slight improvement..
This R-Core does have some leakage coming off its corners BTW, just a FYI.. It should be shielded from other components.
I did find a error tho. Which really makes it more bizzare.. The pot is 50K.. So that makes the total 70k across that.. Very high impedance for that TL431 adj pin.. This causes my brain to hurt.. So that makes it high gain ? Or, WTF.. I will have to plug this into the data sheet math..
How weird...
The parts are all good quality, for its price range, and clearly mindfully picked. This looks very much like it was well crafted and maybe well engineered.. Or not..
I have not tested things like transit response and impedance spectra.
Almost everything I have tried to do to improve it has pretty much failed. I did replace the rect diodes with Schottky diodes and that was a slight improvement..
This R-Core does have some leakage coming off its corners BTW, just a FYI.. It should be shielded from other components.
I FULLY Agree.. I have a HDPlex, a Clones audio, a Uptone LS-2, a pile of other linears and even got into restoring Lambdas.. But this cheap thing from China is very interesting..
I think until someone can at least explain this circuit I am not going to rush to judgement yet.
The TL431 is a strange part.
Under all normal loads for this linear, its stable starting at 50mA of load. The noise appears suddenly and stays the same at any load. ALSO 3mV pk-pk is not the end of the world..
It could be this thing has a very good transit response and to get there has a tradeoff of having a kinda high level of noise.
3mv of noise is not show stopping. Its just really odd. ALso the noise is evenly distributed. It looks like zener noise which is can be used to make a reasonable pink noise gen..
Its a very odd circuit and its why I am posting.. I do not understand it.. Maybe its junk, but, maybe not.
I have updated the schematic.. I will replace the initial one I posted as well. In my use I remove and jump out the inductor on the output.
Attachments
3 mV would not pass my criteria although it is better than most SMPS. Many DIY attempts with relatively simple circuits perform better than that. In less time. Please post some pictures of the casing and brand name. I recognize it and preventing fellow audio people from disappointment would be nice. Apparently many are mainly attracted to low price and believe features promised in the best of chinglish.
TL431 made by exotic brands are strange parts indeed, some even pretend they're not TL431 🙂
Restoring the Lambda... same over there. Decades old Lambda, Astron, Power One etc. all are better than 99% of switchers. For low power application in audio that is.
TL431 made by exotic brands are strange parts indeed, some even pretend they're not TL431 🙂
Restoring the Lambda... same over there. Decades old Lambda, Astron, Power One etc. all are better than 99% of switchers. For low power application in audio that is.
Last edited:
It did not pass my criteria and looked terrible on my initial tests.. I was getting 30uV of noise from Lambdas and remote sensing. I routinely get calibration lab drift and accuracy from some nice VRef chips..
Yet this little supply is doing something fairly interesting...
It feeds a board with some horrible buck/boost regulators ubiquitous on a lot of gear these days. This 3mV of random wideband noise had a odd effect on the PMW spectra of the regulators on the board. It dithered the PWM.. Remember a switcher ( including buck/boost ) varies its PWM based on load and LINE conditions. This noise on the line side jittered the PWM and this smoothed out the measured emitted spectra and noise coming from these regulators. Its not a big effect of course, but it was a interesting result..
I don't think its right to slam these yet.. As people have mentioned anyone can google up a "better" circuit.. WHy would they do something so odd ? There is at least some possibility there might be a reason..
I have not done any transit testing or impedance testing over a wideband. It could be this supply is doing something right.. Maybe even innovative..
OR.. Maybe not.. But if you can google up a circuit, why do this at all ?
Maybe someone went nuts with SPICE circuit design and had good TL431 and LM317 models and figured out something non-standard..
OR NOT...
But I am happy with the design result. I dont want to talk about how I am using it, but, its working better then anything else I have tried. WHICH IS A FREAKY RESULT.. I get it, it looks like junk, it measures like junk... Yet, in what i am doing with it, it was stand out and it caused me to look for why and I discovered this really weird design that I do not understand yet and brought me here to post as you guys are SMART and have lots of depth in esoteric design of this kinda thing..
This seller. A lot of these boards look like they have a similar topology. Hard to tell without getting them. But they seem to do a great buis with great reviews.. https://www.ebay.com/str/pinshuche
OR NOT...
But I am happy with the design result. I dont want to talk about how I am using it, but, its working better then anything else I have tried. WHICH IS A FREAKY RESULT.. I get it, it looks like junk, it measures like junk... Yet, in what i am doing with it, it was stand out and it caused me to look for why and I discovered this really weird design that I do not understand yet and brought me here to post as you guys are SMART and have lots of depth in esoteric design of this kinda thing..
This seller. A lot of these boards look like they have a similar topology. Hard to tell without getting them. But they seem to do a great buis with great reviews.. https://www.ebay.com/str/pinshuche
Last edited:
Yea I love the 1970-80's Lambdas.. I have a whole pile of LM-B and LM-A series that do 12-15V fully restored and modded for better performance I now have no use for in this application sadly....
The DIYers curse to buy/build way more than just 1 good linear PSU and then end up with many very good PSUs without any use. Familiar. I made/restored at least 12 of them during Covid/lockdown.
Well I use a few of them on daily basis so that is comforting 🙂 A propos, even in lockdown I had no trouble getting the parts but today....
Well I use a few of them on daily basis so that is comforting 🙂 A propos, even in lockdown I had no trouble getting the parts but today....