Diode switching noise on mains power supplies

[trh01uk]

Just been reading the thread on the rating of caps across psu rectifier diodes which are intended to suppress diode switching noise (which is here if anyone wants to check that out).

That thread discussed the voltage rating of a capacitor fitted across a diode rectifier. What I am interested in is whether the capacitor actually works?

Last time I did this, it was for the 1200V EHT power supply for a T1154 transmitter. So I had 3 x 1N4007 per leg of a full bridge rectifier. I don't have that psu any more, but I have a photo of the bridge rectifier which is attached. There was a 1n0 1kV AC rated cap in parallel with a 390K resistor in parallel with each 1N4007 diode. 3 of these per leg.

The parallel R+C was fitted when I noticed a sort of white noise on an AM radio tuned to the MW, and about 5' away from the PSU. What was particularly galling was that the noise wasn't reduced that much. The effect of the added R+C was marginal in fact, but in that instance I ran out of time to investigate further.

What are other people's experiences of this problem?


Richard

[barrymagrec]

I have heard that using a faster rectifier such as the UF4007 gives better results.

[Terry_VK5TM]

When I worked for Rank-NEC, there was one particular model tv that was prone to various problems due to rectifier noise from its mains supply.

The cure was to put caps across each diode in the bridge rectifier (it was individual diodes, not an encapsulated bridge rectifier) which fixed the problems (mainly bands of lines across the picture).

They did have a habit of failing and all of them had to be replaced as a matched set. (there were no series resistors with them).

[trh01uk]

Terry,

yes, the other thread noted the tendency for the caps to fail. Incidentally I didn't have series resistors, I had parallel resistors - an idea I picked up from somewhere and presumably intended to ensure the caps lost any charge quickly when the psu was switched off (In a transmitter psu it will be on and off a lot!)

Any recollection what value cap you had to fit to cure the problem?


Richard

[trobbins]

For mains AC rectification, where the dV/dt is relatively slow, and when using fast recovery diodes like UF4007, then there is no valid reason to add balancing caps or resistors across say 3xUF4007 in series to give a higher PIV diode. The caveat is to use diodes from the same batch, and to ensure the diodes are at the same temperature, and to provide some derating due to the series connection. The addition of any balancing C or R would have to ensure the C and R can support 1kV, which wouldn't be the case with those resistors in the photo.

The other caveat is that the load current is within the rating of the diode used. Although a 1N4007 and UF4007 are 1A average rated, that current rating is for a very different waveform than what is found in a power supply, and is at the edge of thermal stress on the diode. Those diodes are more rated for 30-40% of the datasheet average level.

Any balancing caps, or caps across diodes intended for emi reduction, is imho a bad idea nowadays with fast diodes like UF4007, and if extra attention to emi was needed then a CRC snubber across the secondary winding being rectified is the better solution as it is aiming to manage the secondary winding leakage inductance that is the root cause of any consequential emi.

[Terry_VK5TM]

A value of 470pF or 4700pF comes to mind for some reason, but it was so long ago, I can't remember exactly.

Parallel resistors (which is what I meant to say - long day), are a hangover from older times when the manufacturing tolerances of solid state diodes was much broader than with modern day devices.

They were to equalise the voltages across each diode to prevent one from possibly running outside it's PIV rating because of the broad tolerances.

These days, with much tighter specs, they are not needed.

[Radio Wrangler]

There are two issues here. Adding capacitors to prevent diodes blowing up, and adding capacitors to prevent snappy turnoff diodes generating lots of RFI. The same capacitors can do some of both jobs.

Modern high voltage rectifiers are much faster and turn off more cleanly. There is still some RFI but the level is greatly reduced. You probably still want some filtering in the mains path, but RIFAs aside, it's easier.

Modern rectifier diodes may also have controlled breakdown actions, to act to protect them from large spikes which can come down the mains. These sorts of diodes play well together when strung in series, and so long as you have enough of them, they will share voltage amicably.

So with the right diodes, the need for balancing resistors and capacitors disappear, the diodes do the job themselves.

David

[daviddeakin]

Originally Posted by trh01uk

Just been reading the thread on the rating of caps across psu rectifier diodes which are intended to suppress diode switching noise
What I am interested in is whether the capacitor actually works?

A capacitor alone reduces the frequency of the ringing, but not the total energy. But a lower frequency is good enough for many applications. For better suppression you need RC snubbing, although it doesn't need to be across the diodes, it can be across the secondary -same result with less parts. Using fast diodes helps even more.
https://books.google.co.uk/books?id=...page&q&f=false

[Craig Sawyers]

Indeed. The RF ringing is due to the recovery transient of the diodes inducing ringing in the mains transformer's winding capacitance/leakage inductance resonant circuit. It is usually a fairly high Q ringdown in the mid hundreds of kHz. Much worse in EI transformers, particularly those with split bobbins with the mains winding on one bobbin window and the secondary on the other.

http://www.hagtech.com/pdf/snubber.pdf
https://www.diyaudio.com/forums/powe...-test-jig.html

The basic idea is that you snub the transformer secondary, not the individual rectifier diodes.

Craig

[trh01uk]

Thank you everyone! First class answers revealing what is actually going on here, and a much better solution than the one I tried. Pity I don't still have that psu or I would try out the snubber across the transformer secondary. And I will remember the UF4007s for future reference.

Richard

[Diabolical Artificer]

On a lot of the amps I've built I've noticed spikes on waveforms, I'm talking about low voltage here often in the 20-40mV P-P range,but the spikes which are hard to see in entirety look to go higher in amplitude than the actual ripple. Whether these are diode induced I've no idea. I've never heard them though.

Andy.

[trh01uk]

Originally Posted by Diabolical Artificer

On a lot of the amps I've built I've noticed spikes on waveforms, I'm talking about low voltage here often in the 20-40mV P-P range,but the spikes which are hard to see in entirety look to go higher in amplitude than the actual ripple. Whether these are diode induced I've no idea. I've never heard them though.

Andy.

Andy,

to be "diode induced" in the context of this thread, they would need to be repetitive at 50Hz or 100Hz (at least in UK, other frequencies elsewhere), as they are related to the mains frequency of course.

Even then, such mains related spikes doesn't prove the local diodes caused them. They could well be spikes coming down the mains wiring, or picked up on other wiring (e.g. loudspeaker cables running near other wiring), which have originated in some other bit of equipment.

The serious spike generators - typically SMPS - should be filtered to avoid noise and spikes going back up the mains supply if they are EMC Directive certified. I've never heard of a standard diode rectifier power supply such as we are discussing here failing the conducted emissions test levels.


Richard

[lesmw0sec]

I may just have been lucky, but having built a large number of PSU's for audio equipment (from big amps. to sensitive mixers) in the past, I have never detected any problem from diodes or felt the necessity to surround them with caps or resistors. In truth, I was never aware of any such potential problem.

[kellymarie]

I was given a 20 amp 13.8 volt PSU that generated lots of diode noise especially on the long and lower mediumwave. As an experiment I plugged it into my mains filter which reduced some of the noise but as a full cure I wired a 0.01 uF cap across the diodes and it completely cured it. Was I just lucky?

[Radio Wrangler]

Originally Posted by lesmw0sec

I may just have been lucky, but having built a large number of PSU's for audio equipment (from big amps. to sensitive mixers) in the past, I have never detected any problem from diodes or felt the necessity to surround them with caps or resistors. In truth, I was never aware of any such potential problem.

You've probably not had to look in the right places. If you get involved in getting products to pass EMC specs like EN5022 etc, you'll soon run into rectifier noise. Some SMPS make more emissions from their rectifiers than they do from the switching sections.

The first time you run into it, it's an eye opener. Having to fit very fast rectifiers for 50Hz mains is so counter-intuitive, it's ridiculous.

If you trawl up the MF and HF bands you'll often come across broad swathes of angrily buzzing interference. It's mains synchronous Some of it is SMPS in LED bulbs some of it is rectifiers in virtually everything.

David

[KeithsTV]

I was once working on an amplifier with a MW radio on in the background. When ever I powered the amplifier up there was buzzing on the radio. I was working on the PSU so disconnected the amplifier then there was no noise. Suspecting it was the PSU I fitted a capacitor across the transformer secondary and when the amp was reconnected there was no noise on the radio.

Keith

[lesmw0sec]

Originally Posted by Radio Wrangler

You've probably not had to look in the right places. If you get involved in getting products to pass EMC specs like EN5022 etc, you'll soon run into rectifier noise. Some SMPS make more emissions from their rectifiers than they do from the switching sections.


David

Thankfully such enligtened specs. were not around in my day... Almost all electronics these days seem to be an excellent source of emissions - even my wife's electric blanket, so whether or not the regs. are actually achieving anything is a moot point!

[Radio Wrangler]

In one of mankind's greatest ever goofs the levels of permitted radiated and conducted interference were set assuming there would only ever be one interferer at a time within range of a victim receiver. They completely failed to see the explosive growth of doodads.

Then pile on top making self-certification allowed and not having effective enforcement.

Oh, and did I mention the spread-spectrum loophole where SMPS and CPU clocks are FM'd deliberately so most of their power falls outside of the bandwidth of the measuring receiver called up in the standard?

What a mess!

David

[G6Tanuki]

Diode 'commutation noise', and the effects of charge-storage in first-generation semiconductor power-diodes, can be a hell of an issue in some circumstances - such as 'direct conversion' receivers powered from a rectified-AC supply.

[In these, if you allow the signal-frequency RF or the local VFO to pass any of their signal through the power-supply's diodes, you _will_ get hum-modulation!]

In the 1970s I built a little mains-to-13V power supply so I could take one of my trusty 'Larkspur' R209 receivers away with me to University. I used a cheap-and-cheerful bridge-rectifier from Tandy; this was a mistake - despite wiring capacitors across the diodes and fitting filter-chokes in the DC output I never got the thing to be properly RF-quiet - the diode-hash was always louder than the sparking of the vibrator-contacts!

Until, that is, I learned on my course about the work done by Mr. Schottky on transit-time, charge-storage etc: I borrowed some 'modern-for-the-1970s' diodes from the lab and fitted them in my supply and the hash problem vanished!

Don't use slow diodes. Anywhere.

[turretslug]

Wayback when, I thought I'd be clever and make a little variable regulated supply for assorted transistor radios using between 4.5-12V with a reference Zener, 741 comparator, series pass and a Si transistor/series resistor current limit (i.e. a 723 before I'd heard of the 723)- and a W01 bridge.... Creased brow and puzzlement, every set had a background mush on MW/LW no matter how strong the signal. That introduced me to noisy diode switching, no doubt aggravated by the low output impedance of the 741/series pass setup.

There are so many different fast recovery diodes available for pennies that the 1N400x etc. veterans should be a fading footnote in technological history. My default HT diode has been the BA159 for so long, I expect it's also long been bettered!