Capacitors in Parallel with diodes in a Bridge rectifier

['LIVEWIRE?']

TBH, as I type this I've forgotten why small value caps. are often fitted in parallel with each diode in a rectifier, but what seem strange to me is that in neither of two HiFi amplifiers (a HH Electronics VX-150, and a Cambridge Audio A3i) are such caps. fitted, and I wonder if any improvement would result in adding these components. If so, what value & type of caps. would other member suggest? I have to replace the rectifier in the HH amp. anyway. This is a BR82 bridge, bolted to the chassis, so the caps. could simply be soldered to each pin of the device with the leads kept as short as possible. The A3i uses four individual silicon diodes type PW04 on a PCB, so the caps. would probably have to be added under the board. Alex Nitikin, who designed the A3i, has replaced the PW04s with Shottky diodes - he doesn't specify what types, which he says results in an improvement in performance. Would this be worth doing, too? Alex's thread can be found over on www.diyaudio.com

[Herald1360]

The small caps are there to shut up RF hash from the reverse recovery snap off of slow rectifier diodes like 1N400x series. If fast, soft recovery types are used, there's less of a problem. The noise from slow diodes is more likely to be an actual problem with radios than AF amplifiers, but nowadays there's the requirement to keep RF emissions under control as well.

Schottky diodes, where otherwise suitable (voltage ratings etc etc) don't have the problem to start with, though any "improvement" in performance may just be down to the slight increase in DC output voltage.

Edit: Having browsed the diyaudio thread the improvements seem to make the already very low THD even lower. I can only guess at the lack of general hash and noise from the new rectifiers being at the root of this improvement.

[G6Tanuki]

The capacitors are either to 'equalise' any voltage-spikes when the several dipdes are wired in series (same as how resistors are fitted in parallel with the diodes in a series-chain to equalise the reverse-voltages across the diodes) or they're for RFI-suppression.

In the latter case the issue is 'charge storage' - the diode is not perfect - the charge stored in the junction as the supply-voltage reverses can cause interference, which the capacitor essentially short-circuits rather than allowing the resultant noise to propagate into the rest of the circuit.

It was a big problem with older semiconductor-diodes (the 1N4xxx-series for example!) but modern semiconductors (Schottky diodes) have no depletion-layer so there are no stored charges in the junction.

Schottky diodes are generally no more-expensive these days than the older ones so there's no reason not to use them.

As to whether you'd notice any change in performance, well I doubt a change of diodes would increase the output-power of the amp to a noticeable extent.

[Position-statement: I don't have 'golden ears' so minute differences in amplifier-performance pass me by. I do !Loud! though.]

[alayn91]

Hello,
https://www.diyaudio.com/forums/powe...-test-jig.html

[stevehertz]

Many decades ago I made a mains to DC converter for 30s battery sets. It introduced noise. I showed my cct diagram to an experienced electronics engineer at Thorn where I worked and after a few minutes perusing he suggested I fit low value caps across the bridge diodes. It did the trick.

['LIVEWIRE?']

Thanks, all for your comments and advice. I haven't studied the thread on DiyAudio.com in detail, alayn91, but may peruse it further later. Chris (Herald), I agree that the very low THD of the A3i is very good as it is, so I may leave the circuit as it is.

[Terry_VK5TM]

Back when I worked for Rank-Arena for a while, there was one particular model of TV that used caps across the diodes in the power supply.

When one or more of the caps failed you used to get all sorts of strange things happen to the picture and replacement (according to the service sheet) had to be a matched set of 4, ie replace all 4 caps.

[broadgage]

BTW, it is not just silicon rectifier diodes that can produce RF hash when switching on/off.

Discharge lamps, including fluorescent tubes can do this. It can be worth shunting the lamp with a small value high voltage capacitor.
Sometimes this component is incorporated built into the starter.

I believe that the cause is that the lamp goes out when the supply voltage drops below a critical voltage. As the voltage rises again on the next half cycle, the lamp can suddenly start conducting and RFI can result.

Lamps near the end of life are more prone to this.

[Radio Wrangler]

The main beneficiaries of filtering rectifier noise are any radio sets in the vicinity. Not having any sources of interference which you can avoid is good housekeeping. Good manufacturers have used such capacitors for a long time, but they really got going once products needed EMC tests for CE marking.

Don't bother adding capacitors where non were before, but use faster diode. It sounds bizarre but old types like the 1N4xxx series are too slow to work well as 50Hz rectifiers!

David

[Diabolical Artificer]

If you look at a PSU rail of amps on a scope with a very low mv/div you can see the diode noise/spikes sometimes. If you listen really closely you can sometimes hear it.

I put 4 100v,100n caps across all my PSU's BR's I make as standard whether there's an issue or not, I read it was good practice ages ago, though at the time didn't know why and now do it as it eliminates one small component of N and as David says, it's good house keeping, would that I was as concientious regarding actual housework : )

It costs nowt in comparison to all the other components of an amp build; any home built amp can afford a few bob extra as we haven't got bean counters on our case, the bain of commercial amplifiers.

Andy

[Craig Sawyers]

The reason that the diode turn off transients cause problems is that they excite a resonance between the interwinding capacitance and leakage inductance of the mains transformer. That will generally be in the range 100kHz to 1MHz, and can be quite high Q. So every kick of the diode recovery transient has a long high frequency ring associated with it. Which is why it interferes with radio reception, particularly in the MW.

Putting capacitors across the diodes does help, but the PCB trace and capacitor inductance mean that this is not as effective as it might be.

The way to deal with it (other than using Schottky or soft recovery diodes) is to snub the transformer secondary. There is theory here, and simple measurements to enable calculation of values http://www.hagtech.com/pdf/snubber.pdf and also an easily built measurement jig to experimentally measure your transformer and find the correct circuit values here https://www.diyaudio.com/forums/powe...-test-jig.html

Craig

[GMB]

When silicon diodes first became readily available I am sure I read somewhere that the idea of putting capacitors in parallel was to protect the diode from voltage spikes on the mains. This being when the PIV ratings available were only just good enough.

[lesmw0sec]

Originally Posted by GMB

When silicon diodes first became readily available I am sure I read somewhere that the idea of putting capacitors in parallel was to protect the diode from voltage spikes on the mains. This being when the PIV ratings available were only just good enough.

Agreed - that is the only time I have encountered the practice. It was common to replace the selenium HT rect. in TV sets with a couple of diodes and I recall the recommendation to fit the caps.

[Craig Sawyers]

That would figure.

Nowadays putting a capacitor across the diodes is intended to handle reverse recovery transients, and assumes the diodes just act like switches (which they don't). All that does is make a high amplitude circulating current through the capacitor as the diode switches off. So you still get a transient condition which still stimulates the transformer ring.

I've been through that route a couple of decades ago, and was surprised to still get the ringing, hardly altered. I only twigged what was going on when I started reading around (as above).

Craig

[kalee20]

They're there for possibly 3 reasons:

1 to bypass mains spikes which may cause diode breakdown (as GMB has said)

2 to reduce RFI caused by diode reverse recovery transients (as many have said).

3 to reduce modulation hum in an AM radio, by making the diodes look like a low-impedance at RF in both directions, so that any signal picked up by mains wiring doesn't get modulated by the diode changing its impedance at mains frequency.

[Ian - G4JQT]

Is there any advantage in putting appropriately rated capacitors across receiver valve rectifiers, either in AC/DC sets or AC only full-wave valve power supplies? Particularly regarding mains-bourne interference and modulation hum?

[ms660]

Yes, that was done quite often.

Lawrence.

[Al (astral highway)]

Originally Posted by Craig Sawyers

The reason that the diode turn off transients cause problems is that they excite a resonance between the interwinding capacitance and leakage inductance of the mains transformer. That will generally be in the range 100kHz to 1MHz, and can be quite high Q. So every kick of the diode recovery transient has a long high frequency ring associated with it. Which is why it interferes with radio reception, particularly in the MW.

Hi Craig, that is a very interesting observation and thanks for sharing it.

[Nuvistor]

Further to Lawrences’ post, a snip from the circuit of an Ekco radio.

[Craig Sawyers]

That is not to do with reverse recovery, because a thermionic diode doesn't have any of that.

Power seems to come from a resistor chain, and not have a mains transformer, as was common back in the day. I agree that it is probably to do with modulation hum, where the capacitor bypasses the diode and stops any potential demodulation of the mains lead acting as an aerial.

Craig