Hello -
I have read with some interests the threads surrounding CarlosFM's use of snubbers in power supplies. I have done some listening tests and would like to report my (preliminary) results.
First of all, CarlosFM's uses the term "snubber" to refer to a series R-C network used as a bypass across the main electrolytic capacitors in a power supply. The goal is to reduce the impedance rise at high frequencies due to the parasitic inductance of the main electrolytic capacitors. Sometimes the term "snubber" is used to refer to a series R-C network across the rectifier diodes, but that is not the subject of this report.
In some of the threads the question was raised, "Why does CarlosFM recommend using an additional bypass C along with the series R-C snubber network?" This seemed a reasonable question to me, as simulations (and also measurements performed by JosephK) seemed to indicate that using the R-C snubber *without* the additional bypass C would give a smoother impedance curve. CarlosFM's reply was essentially that the combination C + R-C sounds better.
In my listening tests I compared:
1) Bypass with only capacitors.
2) Bypass with only R-C snubber networks.
3) Bypass with both R-C snubbers and capacitors.
(Many years ago I compared case #1 against the case with no bypass network at all, and found that using the bypass capacitor sounded better. I therefore did not repeat that experiment at this time.)
With case #1, the presentation was very natural and coherent. However compared to case #2, the size of the soundstage was reduced and the resolution was somewhat lower.
With case #2, in addition to the increased resolution, there was also a increase in dynamic contrast. (The music seemed louder at the same volume setting.) However, the snubbers alone made it so that less-than-perfect recorded CDs sounded quite annoying. To me this is a very important point, as my experience has shown me that equipment which exhibits this type of behavior has some sort of distortion that is exacerbating (or possibly even creating) a problem.
Finally, in case #3 I tried using both together. This seemed to give the best of both worlds, in that it combined the resolution and expanded soundstage of the snubbers with the naturalness and ease of the bypass capacitors.
I have studied the simulated impedance curves produced by these variations, and am currently unable to explain (even in a general, hand-waving sort of way) why these audible differences should exist. However, my conclusion is that CarlosFM is absolutely correct in his recommendations. I would urge fellow readers to try these tests, as they are quite simple to implement and the resulting sonic differences are far from subtle.
Please note that these tests were performed with a discrete, zero-feedback amp, and not a chip amp. Also note that I would consider these sonic differences to be quite significant, and of greater magnitude than the changes produced by many variations in the audio circuits themselves.
Hope this helps,
Charles Hansen
I have read with some interests the threads surrounding CarlosFM's use of snubbers in power supplies. I have done some listening tests and would like to report my (preliminary) results.
First of all, CarlosFM's uses the term "snubber" to refer to a series R-C network used as a bypass across the main electrolytic capacitors in a power supply. The goal is to reduce the impedance rise at high frequencies due to the parasitic inductance of the main electrolytic capacitors. Sometimes the term "snubber" is used to refer to a series R-C network across the rectifier diodes, but that is not the subject of this report.
In some of the threads the question was raised, "Why does CarlosFM recommend using an additional bypass C along with the series R-C snubber network?" This seemed a reasonable question to me, as simulations (and also measurements performed by JosephK) seemed to indicate that using the R-C snubber *without* the additional bypass C would give a smoother impedance curve. CarlosFM's reply was essentially that the combination C + R-C sounds better.
In my listening tests I compared:
1) Bypass with only capacitors.
2) Bypass with only R-C snubber networks.
3) Bypass with both R-C snubbers and capacitors.
(Many years ago I compared case #1 against the case with no bypass network at all, and found that using the bypass capacitor sounded better. I therefore did not repeat that experiment at this time.)
With case #1, the presentation was very natural and coherent. However compared to case #2, the size of the soundstage was reduced and the resolution was somewhat lower.
With case #2, in addition to the increased resolution, there was also a increase in dynamic contrast. (The music seemed louder at the same volume setting.) However, the snubbers alone made it so that less-than-perfect recorded CDs sounded quite annoying. To me this is a very important point, as my experience has shown me that equipment which exhibits this type of behavior has some sort of distortion that is exacerbating (or possibly even creating) a problem.
Finally, in case #3 I tried using both together. This seemed to give the best of both worlds, in that it combined the resolution and expanded soundstage of the snubbers with the naturalness and ease of the bypass capacitors.
I have studied the simulated impedance curves produced by these variations, and am currently unable to explain (even in a general, hand-waving sort of way) why these audible differences should exist. However, my conclusion is that CarlosFM is absolutely correct in his recommendations. I would urge fellow readers to try these tests, as they are quite simple to implement and the resulting sonic differences are far from subtle.
Please note that these tests were performed with a discrete, zero-feedback amp, and not a chip amp. Also note that I would consider these sonic differences to be quite significant, and of greater magnitude than the changes produced by many variations in the audio circuits themselves.
Hope this helps,
Charles Hansen
For a scientific perspective on what snubbers actually do to a circuit and what electrical phenomena they are intended to prevent, check that thread. It features waveforms captured from real circuits:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=66542&perpage=10&pagenumber=7
The most interesting stuff starts in page 6 or 7, altough it's worth read from the beginning to understand also the "vodoo" perspective.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=66542&perpage=10&pagenumber=7
The most interesting stuff starts in page 6 or 7, altough it's worth read from the beginning to understand also the "vodoo" perspective.
Charles,
Thanks for taking the time for such a report, as I am considering the use of a snubber circuit. My question is what component values did you use for the test? If you chose component values published from someone else or if you calculated them for yourself and if you did calculate them yourself then what formula did you use?
Many thanks.
Thanks for taking the time for such a report, as I am considering the use of a snubber circuit. My question is what component values did you use for the test? If you chose component values published from someone else or if you calculated them for yourself and if you did calculate them yourself then what formula did you use?
Many thanks.
Eva said:
Yes, some nice measurements. However, that thread is referring to the "snubbers" used on the rectifier diodes and/or power transformer. This thread is referring to the "snubbers" used on the electrolytic capacitors. I know the terminology is confusing, which is why I made the distinction in the original post.
duelbox said:
For the front-end power supply I used 3 x 1000 uF for each rail, each bypassed with a combination of 0.047 in parallel with a 1 ohm + 0.047 uF.
For the output stage power supply I used 4 x 6800 uF for each rail, with a single bypass network of 0.1 uF in parallel with a 1 ohm + 0.1 uF.
These values were selected simply because they were the easiest values I had to hand, and they seemed fairly close to what CarlosFM was recommending. I have no idea how he calculates his values. I have tried running simulations to see what would be optimal, but I haven't been able to make sense of it yet.
Oh, by the way, the thread about the diode snubbers explains how to select the exact R and C values in order to tame a certain resonance to the desired extent, but in order to do that, you have to take the oscilloscope and expend some time to find such a resonance first. The wrong component values may do nothing or may boost the resonance if the R value chosen is too small 
Interesting findings, Charles.
I was under exactly same impression when I tried snubbers first time in PS. I expressed it here: http://www.diyaudio.com/forums/showthread.php?s=&postid=550100&highlight=#post550100
Later however, I found some some defficiences that eventually put me off that aproach, which I'm not saying isn't good, it's just appeals to different tastes. Nevertherless, I may try it again:
http://www.diyaudio.com/forums/showthread.php?postid=574431#post574431
http://www.diyaudio.com/forums/showthread.php?postid=581230#post581230
I was under exactly same impression when I tried snubbers first time in PS. I expressed it here: http://www.diyaudio.com/forums/showthread.php?s=&postid=550100&highlight=#post550100
Later however, I found some some defficiences that eventually put me off that aproach, which I'm not saying isn't good, it's just appeals to different tastes. Nevertherless, I may try it again:
http://www.diyaudio.com/forums/showthread.php?postid=574431#post574431
http://www.diyaudio.com/forums/showthread.php?postid=581230#post581230
Upupa Epops said:
Nah, that's in the real world only...
Conceptually, all big electrolytics are very inductive, all foils compensate for that indutance yielding a flat impedance curve, all wiring has negligible inductance and all snubbers are just snubbers (component values are of little relevance, a matter of taste only). That's the key for success. (I like that method, I use it mostly to buy furniture).
Charles Hansen said:
Hello Charles,
Thanks for your time and work for trying this snubber thing.
It shows that you put audio performance above all, I'm not alone, then.
I'm sure that we are not alone, either.
Who's the car manufacturer that builds a new model without a test drive? Not only measurements, but test drives.
Building high-end audio gear is a labour of love, dedication, passion for music and always trying to make better.
You are right, the sonic improvement can be as drastic as removing a cheap electrolythic coupling cap from the signal path.
PS: don't be surprized that the conversation goes into diodes, I've been trying to xplain that I'm not talking about diodes, but...
as the proverbial partypooper i am feel i have to share my experience with the snubbers.
they did nothing, not a thing, nada, zip.
took me a while to understand the underlying pshycology, but in the interest of pursuing the truth, i did my own reserch in basic phsycoacustics and electronics, the voodoo was dispelled, and i finally saw the mechanisms that made me belive there was a difference, for what they were, and was able to come to a balanced conclution. learned a few things in the process.
It should be said offcource that i strongly belived in the snubber and a lot of other voodoo when i first enterd the DIY scene. and deffended it for all i was worth. funny how things change.
the objectivity has done me good.
-Marius
they did nothing, not a thing, nada, zip.
took me a while to understand the underlying pshycology, but in the interest of pursuing the truth, i did my own reserch in basic phsycoacustics and electronics, the voodoo was dispelled, and i finally saw the mechanisms that made me belive there was a difference, for what they were, and was able to come to a balanced conclution. learned a few things in the process.
It should be said offcource that i strongly belived in the snubber and a lot of other voodoo when i first enterd the DIY scene. and deffended it for all i was worth. funny how things change.
the objectivity has done me good.
-Marius
demogorgon:
I adhere to every word you have just wroten.
p.s.: However, altough it's off-topic, I have to admit that I can hear a difference with diode snubbers...
The turn-off pop when I unplug the amplifier is much softer (that's because the leakage inductance of the transformer is well damped now and can no longer radiate spikes).
I adhere to every word you have just wroten.
p.s.: However, altough it's off-topic, I have to admit that I can hear a difference with diode snubbers...
The turn-off pop when I unplug the amplifier is much softer (that's because the leakage inductance of the transformer is well damped now and can no longer radiate spikes).
If for argument's sake we assume there to be "something" introduceed into the DC power by the diodes, and if we further assume that it is potentially audible -- an additional variable is the PSRR of circuitry thus powered. The concerns of someone whose preference is for minimalist SE circuits can be pretty understandable. On the other hand those who like amplifies loaded up with current mirrors, cascodes, and the like that shield the precious signal from the cruel outside world might be faily cavalier with respect to what's buzzing around in their rails.
Then there are the obssesive-compulsive "belt & suspenders" type who just worry about everything as a matter of course.
Then there are the obssesive-compulsive "belt & suspenders" type who just worry about everything as a matter of course.
demogorgon said:
Am I allowed to post in the same language as this guy that calls psychos to those who KNOW that the PSU has a large impact on the performance of any amplifier?
Or am I going to Texas again, and the above post doesn't?
It is reasonable to say that you don't like, you were unable to get good results, etc.
But NOTHING?
Geez... suddently all amps sound the same.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.