http://www.zero-distortion.com/techno/powersupply/powersi_07.htm
Some may find the above rather interesting. 🙂
Some may find the above rather interesting. 🙂
Interesting link!
Thanks, I know I did.
Looks like something I would have written myself, in fact, wished I had.
Lots of other info there as well as some interesting links, worth a look.
Roger
classd4sure said:
Thanks, I know I did.
Looks like something I would have written myself, in fact, wished I had.
Lots of other info there as well as some interesting links, worth a look.
Roger
Re: Interesting link!
It seems like it's an old trick doesn't it? I like his note about it being a compensating network and so you should know what you're compensating for. However it appears from the forum here that we can get away with some sort of "rule of thumb" easily, with so many having tried the same values reporting good results.
sx881663 said:
It seems like it's an old trick doesn't it? I like his note about it being a compensating network and so you should know what you're compensating for. However it appears from the forum here that we can get away with some sort of "rule of thumb" easily, with so many having tried the same values reporting good results.
Absolute values?
I agree, absolute values don't seem to make that much differance as long as they are in range. Far less differance than having them or not!
Roger
I agree, absolute values don't seem to make that much differance as long as they are in range. Far less differance than having them or not!
Roger
Hi,
One note: In my opinion, snubberizing a UCD amp is not needed, because - it has been already done!
Bruno about his way of snubbing
Ciao, George
One note: In my opinion, snubberizing a UCD amp is not needed, because - it has been already done!
Bruno about his way of snubbing
Ciao, George
Hi George,
Edit: (redundant edit here)
Bruno was asked if it would be worth doing this in the hotrod thread a few weeks ago and I believe the response was similar to "try it, it might help", and then this thread popped up later on.
It's not at all to say what they've used is insufficient, people are just after the pot of gold at the end of the rainbow I think. What can you do?? I think I agree with you, and what you'd concluded in your linked thread at one point too. The on board electrolytic will impose it's own characteristics over any external snubber.
Cheers,
Chris
PS: I appreciate the work you've done in that thread of yours on snubbing. Nice effort, thanks.
Edit: (redundant edit here)
Bruno was asked if it would be worth doing this in the hotrod thread a few weeks ago and I believe the response was similar to "try it, it might help", and then this thread popped up later on.
It's not at all to say what they've used is insufficient, people are just after the pot of gold at the end of the rainbow I think. What can you do?? I think I agree with you, and what you'd concluded in your linked thread at one point too. The on board electrolytic will impose it's own characteristics over any external snubber.
Cheers,
Chris
PS: I appreciate the work you've done in that thread of yours on snubbing. Nice effort, thanks.
classd4sure said:
Snubberizing the main filter cap is also snubberizing the resonant circuit formed by the wiring from the PSU and the local capacitor decoupling at the IC.
You can also snubberize the local cap.
Try it. Doesn't cost much.
I had french reports that ça marche in class D amps. 😎
Chris, thanks. I also think that it's something which usually is not done, [for ex. Tripath] and as such is just another plus for the UCD
I also have some suspect that Bruno maybe was talking too much here..
Hi Carlos! nice to see You - and disagree..🙂
Ciao, George
I also have some suspect that Bruno maybe was talking too much here..
Hi Carlos! nice to see You - and disagree..🙂
Ciao, George
carlosfm said:
So in essence you're saying ... ideally you'd terminate both ends of the transmission line?
Bruno had recommended to place it as close to the board as possible, but it's already done on the module and we should NOT mess with that.
Joseph K:
Blasphemy!! lol..... shhhhhhhhh.
Maybe for Philips liking I agree, but I suspect he was somewhat stiffled there, either way he won't let slip anything he doesn't want us to know.
We sure appreciate the quality of info he freely gives though dont' we? I still say he should write a book, but then I'd like to see the 700 out too.
George,
I hope his silence is only temporaly.. and really is only because of the 700.. oops, which 700 am I talking about?
Ciao from ciaogeorge
I hope his silence is only temporaly.. and really is only because of the 700.. oops, which 700 am I talking about?
Ciao from ciaogeorge
Hi.
first some info for Carlos and others. From what I can see UcD400 is decoupled this way (from switching transistors to the power supply connector, talking about one rail only):
-smd ceramic cap paralleled by 100u high ESR electrolytic
-5mOhm noninductive smd resistor
-470uFmain filtering cap
-3parallel smd ferrite beads
-smd ceramic cap almost directly on power supply pins
So decoupling scheme is allready quite complex as is to be expected in class D.
I have made a fast crude simulation of the circuit. My first impression is that snubber on reservoir caps is sufficently decoupled from module by 500nH wiring inductance. I couldn't spot any change in voltage frequency spectrum on supply connector of the module. However, frequency spectrum of current drawn from the reservoir caps changes with snubbing, and I woud say for the better (less very high harmonics). Time permitting, I will fire up spectrum analyzer and measure spectrum of the voltage and current on the module with snubbers and without them.
Best regards,
Jaka Racman
first some info for Carlos and others. From what I can see UcD400 is decoupled this way (from switching transistors to the power supply connector, talking about one rail only):
-smd ceramic cap paralleled by 100u high ESR electrolytic
-5mOhm noninductive smd resistor
-470uFmain filtering cap
-3parallel smd ferrite beads
-smd ceramic cap almost directly on power supply pins
So decoupling scheme is allready quite complex as is to be expected in class D.
I have made a fast crude simulation of the circuit. My first impression is that snubber on reservoir caps is sufficently decoupled from module by 500nH wiring inductance. I couldn't spot any change in voltage frequency spectrum on supply connector of the module. However, frequency spectrum of current drawn from the reservoir caps changes with snubbing, and I woud say for the better (less very high harmonics). Time permitting, I will fire up spectrum analyzer and measure spectrum of the voltage and current on the module with snubbers and without them.
Best regards,
Jaka Racman
Any further news on snubbing the power supply for the UCD?
Re-reading this thread it sounds like in general a snubber on the power supply is a good thing. Only one person has reported using it and it has made some improvements.
Jaka, did you do your measurements with/without? Anyone else done any further before/after listening tests?
Regards,
Dean
Re-reading this thread it sounds like in general a snubber on the power supply is a good thing. Only one person has reported using it and it has made some improvements.
Jaka, did you do your measurements with/without? Anyone else done any further before/after listening tests?
Regards,
Dean
Snubbing,
Seems like with a few caps you can tune the response across the entire audio band.
I threw a guestimated snubber on my homebrew amp right next to the mosfet leads, 10nF+2.5R// 10nF// 470uF.
Originally I just had a few spares I threw in, like a 10uF, 22uF, and 100uF on each rail, even made the mistake of a 100nF ceramic.
Removed all that and installed the above + another 470uF in parallel to the other.
The second 470uF cap brought out a big sense of power in the base but certain tones had lost presence, in the mid/high range.
Taking the second 470uF cap out fixed that, but the bass is less subtantial, all very controlled, I wonder the tripath guys like the 1000uF squeel stoppers, it seems too big.
I haven't played with the value of the snubber I put in at all and am very happy with it as is, no measurements taken only ears. It's playing songs perfectly that I couldnt' listen to before, now in total control.
That was to decouple the rail at the mosfet though, not the PSU at the cap, I suspect a range of improvements there depending what kind of cap you have, might be a big help for the industrial sort.
Regards,
Chris
Seems like with a few caps you can tune the response across the entire audio band.
I threw a guestimated snubber on my homebrew amp right next to the mosfet leads, 10nF+2.5R// 10nF// 470uF.
Originally I just had a few spares I threw in, like a 10uF, 22uF, and 100uF on each rail, even made the mistake of a 100nF ceramic.
Removed all that and installed the above + another 470uF in parallel to the other.
The second 470uF cap brought out a big sense of power in the base but certain tones had lost presence, in the mid/high range.
Taking the second 470uF cap out fixed that, but the bass is less subtantial, all very controlled, I wonder the tripath guys like the 1000uF squeel stoppers, it seems too big.
I haven't played with the value of the snubber I put in at all and am very happy with it as is, no measurements taken only ears. It's playing songs perfectly that I couldnt' listen to before, now in total control.
That was to decouple the rail at the mosfet though, not the PSU at the cap, I suspect a range of improvements there depending what kind of cap you have, might be a big help for the industrial sort.
Regards,
Chris
Chris, try a lower value resistor, lower than 1R.
Then try again with bigger caps, like 1,000uf.
Then try again with bigger caps, like 1,000uf.
Hi Carlos
You have suggested snubberizing the PSU caps, by trying (between the PSU 10,000uf caps and the Local IC caps) 100nf (0.1uf) caps in parallel to the 10,000uf caps (bypass) then 0.1R + 3.3nf (MKT) in parallel to the 100nf cap. Is that correct?
What snubberizing would you suggest, between the local caps and the IC.
You have suggested snubberizing the PSU caps, by trying (between the PSU 10,000uf caps and the Local IC caps) 100nf (0.1uf) caps in parallel to the 10,000uf caps (bypass) then 0.1R + 3.3nf (MKT) in parallel to the 100nf cap. Is that correct?
What snubberizing would you suggest, between the local caps and the IC.
Hi,
Guys the reason I used two 5 ohms in parallel is because that's all I could scavenge. Maybe I'll have another dig later on see.
All the same Carlos you made a point and I got it. It was premature to judge the use of bigger decoupling caps based on the little experimenting that I did, and you got me there🙂
So I'd certainly consider trying it again and am curious enough that I Think I'll start hunting for more resistors just a few minutes.
All I set out to do with this originally was prove to myself that was worth the effort for one, and for another to experience the kind of improvement that research on it has led me to think it would make.
Even with just throwing the first few values I had at it it was highly worth it and yes also worth further experimentation.
Thanks,
Chris
Guys the reason I used two 5 ohms in parallel is because that's all I could scavenge. Maybe I'll have another dig later on see.
All the same Carlos you made a point and I got it. It was premature to judge the use of bigger decoupling caps based on the little experimenting that I did, and you got me there🙂
So I'd certainly consider trying it again and am curious enough that I Think I'll start hunting for more resistors just a few minutes.
All I set out to do with this originally was prove to myself that was worth the effort for one, and for another to experience the kind of improvement that research on it has led me to think it would make.
Even with just throwing the first few values I had at it it was highly worth it and yes also worth further experimentation.
Thanks,
Chris
- Status
- Not open for further replies.