Rectifiers for power supplies

William

Member
2001-07-24 11:44 am
I recently have been told that by replacing the rectifier bridge in your PSU with ultra fast rectifiers, one could improve the sound quality of one's amplifier - better resolution and more details. I did so and found audible improvements.

Can someone enlighten me with the reasons behind it?

Would one expect the same result for doing the same to a pre-amplifier?

Many thanks,

William
 
Nelson

I don't disagree with you, I merely offered the link as 'further reading' for William, since it contained a possible explanation for his perceived improvement (there are many other reasons why he may believe the sound is better).

My own view is that perhaps manufacturers would like persuade people to buy higher spec'ed devices, at higher prices, and with greater profit margins. If this is the case, then they are going to try to justify the additional expense by using all sorts of claims and explanations. Unfortunately, the audio world is full of such practices (we won't get into the subject of cables etc. here :).

Geoff
 
From the application note, I can't understand how replacing standard diodes with ultra-fast diodes in a bridge rectifier can change the sound.

On the other hand, there might be an issue with electrical or RF noise being spit out when the diodes switch. I saw this idea in Randy Slone's amplifier book. He suggests that each diode in the bridge have a 0.1uF cap in parallel with it. It might be interesting to compare the results of this method to the results of the 'fast diode' method.

[Edited by thoth on 07-24-2001 at 06:35 PM]
 
My assumption has always been that sufficient capacitance in the main bank would shunt the high frequencies (after all, it's a spike, hence Fourier analysis would indicate scads of high frequencies) to ground. Arguably, film caps (having a better high frequency response) would do a better job, either in conjuction with electrolytics, or by themselves.
A question I've always wanted to play with is whether any residual high frequency hash made it through a regulator--but then you get into a thousand questions about the 'sound' of whatever regulator circuit you might chose.
Personally, I have no opinion about the audibility of various sorts of diodes, as I've never tried them out, one against the other. I have, however, used both kinds (normal & soft recovery, but I've not used Schottky for anything) over the years, not out of any sense of conviction either way, just whatever mood I was in when I was buying parts for what I was building at the time: It's only a couple of bucks...why not? vs. Fooey, probably can't hear the blasted things anyway, put the money into the caps instead. Nelson's position, at least at his day job, would have to lean towards parts cost I'm sure, unless there was a fairly large sonic difference that the eventual customer would definitely notice. Since I'm only answerable to myself for the amount I spend on the rectifier, I have more freedom...but don't always vote the same way from one project to the next. One of these days, I'll take the time to fiddle the rectifier back and forth--at that point I may be able to resolve the question (at least to my own satisfaction). For the moment, that'll have to take a back seat whilst I'm rebuilding my system.
I don't take Audio Amateur, so I didn't see the article referenced in the paper. Did anyone read the original article? Is it, perhaps, available somewhere on the web? That might give us a firmer basis for discussion.
I generally bypass the diodes in rectifiers with smallish (~.1 uF) caps just as a matter of principle--not necessarily because of sonics (I haven't A-B'ed this either), but just as a matter of engineering. It serves as a good way to preserve diodes from high voltage spikes coming in from the AC line. It's another trick I picked up from the ham radio people. If you were to assume that the faster diodes did make a difference, the cap would presumably serve to reduce the spike still further. Yes, you're putting in four more parts, but .01 to .1 uF caps aren't that expensive, at least compared to the cost of these diodes.

Grey
 

Joe Berry

Member
2001-03-15 6:15 pm
USA
John Curl has commented on this on several occasions in the Tweaks section of the Audio Asylum. His position is that standard solid-state power rectifiers exhibit a glitch at turn-off which may find its way into the ground system of the host audio device.

He asserts that this glitch can be seen by using a differential scope probe or RF spectrum analyzer to monitor the current through the diode. He further notes that the cleanest behavior is shown by "soft recovery" diodes, as opposed to "high-speed" types which may form a secondary glitch or induce high-frequency ringing.

A number of these comments apparently are still available for review in the Audio Asylum archives. I found several by going to http://www.audioasylum.com and using the Search feature to do Boolean search in the Tweaker's Asylum archives for John+Curl+Schottky, John+Curl+Hexfred, and John+Curl+Diode.
 
Grey,

Unfortunately electrolytic capacitors have exceptionally poor performance at high frequencies, and will do little to attenuate switching noise. The normal method of reducing this noise is ceramic snubber capacitors immediately across the diodes i.e. as close as possible.

While I am increasingly impressed with the performance of circuit computer simulators, I believe they do tend to immerse us in a terribly artificial environment. The packages do assume, for example, “perfect” performance from components. It is possible that we are tending to assume this same type of performance in the real world. In addition, there are many “components” that simply never appear on a diagram simply due to the fact that we have to place the parts on something and hook the whole lot up together.

While I can’t really offer any valid explanation as to why switching noise should affect audio circuits, factors such as IMD and ground paths should possibly be considered. Ok, this is a fancy way of saying “dunno”, but I am dubious of an environment which mixes RF and any form of non-linearity.

Switching noise is a fact, I come from a HF background, and it is well established there. I’m with Grey, if there is any doubt as to whether less than ideal conditions will affect the circuit, I’ll steer on the safe side; it’s often only a few cents difference in low voltage applications.

I have used Schottky diodes in a CD power supply to good effect. In that case the PS supplied both AF and DAC circuits. DACs are particularly sensitive to line noise, hence I assumed the improvements were due to a reduction in this noise (indeed I still feel this way), however it’s not to say that it didn’t also improve the audio stages.

Cheers,

Pete
 

William

Member
2001-07-24 11:44 am
Many thanks, Geoff. I had read the technical notes and it sounds to be an explanation. I should also thank those who participated with this discussion.

However, I went one step further and changed the rectifiers for my valve pre-amp (the high tension part), again with ultra fast rectifiers. This time, instead of making a bridge, I used only two because I got a centre tap for the secondary coil. The result is appalling. It does not sound like valve at all. It sounds tight and the mid frequencies are less (or should I describe as gone). I then switched back to my IC pre-amp [which had only ordinary bridge rectifier] and it sounded more pleasing (to the changed valve pre-amp). Did I overdo it because I had already changed the ones for my power amplifier?

Technically, two rectifiers and centre tapping result in full-wave rectification as well as a bridge rectifier. So the above phenonemon should not be blamed for not using a bridge rectifier.

Any further comments please?


William
 
Pete,
We are in agreement about electrolytics--that's why I threw in the part about film caps, noting the superior high frequency performance. However, that's in the main bank. As far as caps directly across the diodes, I tend to use film there, too (polyester--nothing fancy or expensive, just garden variety critters), since I regard ceramic caps as suspect.
William,
You said you changed the diodes in your preamp...were there four diodes there to begin with, or two? If you went from a four diode rectifier to a two diode rectifier, your rail voltage changed (cut in half, actually), which would definitely give you sonic changes.
Yes, if a center tap is available, you can have full wave rectification with only two diodes. Both diodes should point the 'same way', i.e. with both cathodes (or anodes) towards the filter section, and the center tap serving as ground.
On a philosophical note--if it sounds better, it should sound better everywhere. I've always been wary of trying to balance colorations against one another to get a system to sound right. There are exceptions, viz. tube amps are rarely a good idea for subwoofers (regardless of strengths in midrange), although a good tube amp can hold its own down into the woofer range; poor ones give the classic bloated bass syndrome. If (and for me, the jury is still out) super-duper diodes are better in amps, they should also be better in preamps, CDs, etc. with the same strengths evident. I just haven't had time to play with it yet.

Grey
 
So far, we seem to have a lot of 'fence sitting', with very few people admitting to having tried the rectifier substitution. I will therefore give my experiences, for what they are worth.

Several years ago, I replaced the rectifiers in a pair of well-regarded, commercial Class-AB integrated amps that I was using in a bi-amped system. The replacement diodes were Schottky with a similar size and rating to the original standard diodes.

On completion of the modification (and no other changes were made at that time), the amplifiers appeared to be audibly better. The sound seemed 'smoother' and 'cleaner'. There are several possible reasons for this.

1. I had spent some hard earned money and wanted/expected an improvement.

2. In order to fit the diodes, all connections to the amps (mains, phono, speaker) had to be broken and remade and so any contact contamination/deterioration should have been removed (I had previously determined that regular cleaning of all connectors, including the mains plug, gave a perceived improvement)

3. The Schottky diodes actually gave an improvement.

About 12 months after this modification, I replaced the Schottkys with the original diodes, prior to selling the amps. On re-auditioning, in the same system, to check that the amps were working correctly, I could hear no deterioration in the sound quality due to reverting to the standard diode type. This tends to rule out options 2 and 3 above, and I am rather loath to accept that option 1 is the reason as I am sure I heard a difference!

This leaves me 'fence sitting', along with many other people.

A quick comment to Grey. If mains borne interference and noise can get through a power supply (simple or regulated) to affect the sound quality of an amp, then I'm sure diode switching noise can do so as well.

Geoff



[Edited by Geoff on 07-25-2001 at 08:22 PM]
 

William

Member
2001-07-24 11:44 am
Dear Grey,

Thanks for your comments. I did have 4 rectifiers in my pre-amp before (that was before the old transformer failed). Since the replacement transformer has a centre tap, so I used two rectifiers instead as it would still have full-wave rectification. I did connect up correctly to produce 350V d.c. Yet, the result ended up to be unexpectedly poor. A friend of mine said that I should use 4 to make a bridge and he believed it would be better to have 4 working instead of two. As I have no electronic background, I just could not figure out the reason.

My expectation is that I may find some improvement (if any) but not sounding worse. As I do not have testing instruments, I can only believe my ears.

Regards,

William
 
G’day Grey,

While I would agree with your conclusion of ceramic capacitors when used at AF, I am hesitant to project my prejudice into RF applications. In this instance we are looking for capacitors capable of performing their intended function well into the MHz range, and I believe the characteristics of ceramics make them the right choice for this purpose. Personally, I would be hesitant to use anything but a very small value Polyester capacitor in this instance given their relatively low resonant frequency compared with ceramics.

Cheers,

Pete
 
Geoff,
Your point about AC line noise vs. diode switching noise is well taken. Without trying to steer the thread in that direction, I threw together some quick-and-dirty AC conditioners a few years ago using a cap across the AC hot and neutral (bear in mind that I'm in the US with 120VAC, not 220V as you have), a surplus isolation toroid, followed by another cap and some other miscellaneous whatnots that I had on hand. What I got was a surprise--*much* quieter 'space between the notes.' Frankly, I hadn't expected it to do that much sonically, as my primary intention was to tame an unruly AC situation here at the house. Given the difference that such a simple (and cheap) circuit made, especially on the other side of the power transformer, I'd lean towards the idea that something closer to the circuit (both electrically and physically [possible air transmission of RF from the spikes?]) could be detrimental to the sound.
In the same vein...I would think that this would be (yet another) argument for inductors in the power supply (either L or PI), as the Z at such high frequencies is surely going to be a major impediment to any noise from diode spikes trying to travel downstream. What say ye?
William,
I hadn't realized that the transformer had been changed, too. Egads, I'm not ready to enter into a discussion on the 'sound' of power transformers, as I've done absolutely no fiddling in that direction at all. I don't recall ever hearing anyone say much along those lines, excepting a vague recollection that the old Peerless power transformers were reputed to sound better than their competitors. And--setting aside the issue of recovery time--what if the diodes themselves have a sound. It's true for transistors; why not diodes? If the power transformer failed, and presumably the diodes as well, did you lose the caps in the main bank? That might provide yet another variable. If you have the parts, the time, and the patience, by all means try a four diode rectifier if you can get the voltage to work out properly. If you do so, please report, as I'd be curious about the result.
Pete,
You're trying hard to complicate my life. (And succeeding.) I might need to sit down and try ceramic vs. polyester in situ to see if my mistrust of ceramics in that part of the circuit is justified.

Grey
 
Dunno!

Grey,

Sorry for the curve ball :)

Regarding transformers I have read reports (I can’t recall who from, possibly John Curl, probably not!) claming quite significant differences between different transformers. In this case not the type of transformer (although that brings to mind the different coupling characteristics at higher frequencies with regard noise injection … ergh, let’s not go there!), however the power rating of the transformer. There appeared to be a direct correlation between size and audible performance, with the larger transformers providing a significant improvement. Absolutely nothing to do with the thread, but I do like tangents.

Jam,

I really don’t have any experience with these capacitors in this application so I’ll have to hand this off to somebody more knowledgeable than myself. Personally I believe ceramics have pretty much ideal characteristics as snubbers so don’t see any reason to use anything else.

I certainly don’t want to hold myself out to be any sort of expert in this area, just base my judgement on experience and what I read. You may be interested in Walt Jung’s article if you haven’t already read it http://capacitors.com/pickcap/pickcap.htm however bear in mind that he is referring to audio frequencies. Another useful resource on this subject can be found at http://www.analog.com/publications/magazines/Dialogue/Anniversary/21.html indeed ADI do some very interesting articles on all sorts of toys.

Bit of a non-answer, sorry ‘bout that!

Cheers,

Pete
 
(Just for the record, I'd like to note that Jam positively *loves* to torture me with oddball ideas that I ain't got time to follow up on. Now he's got Pete helping out. You guys gotta promise to come visit me when they cart me off to the insane asylum...just bring red wine [is there any other kind?], and/or single malt Scotch and something that makes decent music, okay?)
Jam,
I had thought about silver-mica caps also, but they're generally pretty small, and you'd have to parallel several to get up to the .1 uF range.

Grey
 
Grey

Your digression from rectifier noise to power supply/mains noise is not much of a tangent so I think we can continue discussing it here.

Your point regarding choke input or inductive Pi filters is very relevant with regard to mains borne and power supply noise. The amplifier PSRR and the voltage regulator (if used) ripple rejection is usually very good at the normal ripple frequencies of 100 or 120Hz. However, the situation changes considerably at higher frequencies. For example, the venerable LM337 has a ripple rejection of over 80dB at 100 to 400Hz but less than 20db at 1MHz (as low as 10db if the adjustment pin isn't bypassed). The inductor(s) should help to attenuate the higher frequency noise and this may well be why a number of people prefer choke input filters to capacitor input types.

Your findings on the suppression of mains noise are similar to my own, though I haven't gone to the extreme of a full mains conditioner. I have had good results from the simple (and cheap, £0.80 here in the UK) modification of adding an MOV transient suppressor across the mains input. This can be wired in the amp itself, in the mains plug or even in a separate plug inserted into an adjacent wall socket if there is one spare. Like your conditioner, this mod gives 'cleaner spaces'. In fact, a friend of mine was having considerable problems in that his system became virtually 'unlistenable to' at certain times during the day (it sounded 'rough' to say the least) and nothing he tried would rectify the problem. I fitted a £0.80 transient suppressor to his mains distribution block and he hasn't had a complaint since. The problem, I believe, was caused by mains borne noise from an adjacent semi-industrial area.

Geoff

[Edited by Geoff on 07-26-2001 at 11:35 AM]
 

ALW

Member
2001-06-12 10:15 am
UK
Re : Mains Noise

I've had good results from transformers (toroidal) that incorporate an electrostatic shield - they can provide some significant attenuation of noise, but you don't often find them on off-the-self parts.

I also discovered recently that I get sonically better results from a full-wave bridge rectifier using a centre tapped transformer (2 diodes) than a more conventional (and efficient) bridge circuit.

Whether this relates to the lower diode drop, or the fact that the earth reference is moved back to the transformer on the centre tapped design I've yet to investigate.

I haven't yet had the opportunity to look for measured differences, but the audible ones are definitely there and they are not subtle.

Andy.

P.S. Both designs used the same standard bridge rectifier module.

P.P.S Both supplies were regulated


[Edited by ALW on 07-26-2001 at 11:07 AM]