parallel rectifiers

danielnaveen said:
Hello,

I have two GBU802 rectifiers can i parallel them? how do i do that?

Thank you for your answers

Daniel

You can parallel them, but it wouldn't be advantageous. Because the diodes will have small different forward thresholds, one will take the majority of the current. You can't parallel two 5A diodes and expect to be able to run 10A through the combo.

BTW Why do you ask?

Jan Didden
 

Bobken

Member
2002-12-23 11:22 pm
UK
Hi Daniel,

Many years ago I read that Kostas Metaxas (of Metaxas Amplifiers) advocated the use of multiple parallel low-current diodes in his PS bridges. Some sonic improvements were claimed, although it was so long ago that I don't now recall precisely what the theory behind this was.

Accordingly, as an experiment I made up a discrete diode bridge for a power amp with about 8 or 10 - 1N400* diodes in parallel, and used 4 of these parallel diode arrangements for each of the sides of this bridge.

This was an improvement sonically over the 'standard' encapsulated diode bridge which was replaced by this discrete arrangement, and I used it in a 100W amp for probably 2+ yrs without any problems. In fact I only replaced it later when I found that some fast/soft-recovery types (and Schottkys) provided a further sonic improvement over the suggestion made by Metaxas.

I still come across these discrete bridges from time to time when I go through my 'reclaimed parts' box as (so far) I have not broken them down again to use in other locations.
Possibly an Internet search will reveal Metaxas' reasoning behind his suggestion from long ago, but it must have made some sense to me at the time, or I wouldn't have wasted my time in trying it out for myself.

Regards,
 

Bobken

Member
2002-12-23 11:22 pm
UK
Hi Daniel,

From my experiences before, there appeared to be no problems with current-sharing, but this is not to say that all diodes will behave in the same manner. This was in a 100W RMS per channel Class AB amp, which I used to drive hard without any problems, but I didn't ever measure the current drawn during heavy use. However, it must have been many times greater than a single 1N400* diode is rated for.

I certainly cannot believe that there will be so much variation in their electrical parameters that adding a pair of your suggested diodes together will not provide you with something getting close to twice the current-rating of a single diode, though, and it should be able to cope with greater currents than a solitary diode of the same type.

As you suggest, the initial surge at switch-on with large smoothing-caps is the most vulnerable time for any such diodes. It is this parameter which you should ideally check out on the data-sheet for these devices.

Good luck if you try this out, anyway.

Regards,
 

Bobken

Member
2002-12-23 11:22 pm
UK
Hi Daniel,

It is not easy to precisely predict the switch-on surges in cases like this, and depending on the mains supply, they can vary, anyway.

However when not being absolutely confident during any new trial, it is sensible to take care, and not risk anything unnecessarily. The same goes even for a known situation, as not all components behave as they should, and things can still go wrong in spite of everything being according to specs etc.

Apart from Transzorbs (a totally different type of dode) in my own experiences, all other diode failures I have seen have failed with the diode going open-circuit, which is better/safer than shorting out.
They can burst apart, though, like a mini-explosion, so it would be sensible to wear some eye protection for the first few trials, just in case, and don't stand over anything during any such experiments. The internal parts of diodes are 'dry' so there is no acid or anything to worry about being thrown over yourself or your room, and my guess is that the worst you will experience if there is a failure is a small bang, a bit of smoke, and the failed device going open-circuit.

Regards,
 

Bobken

Member
2002-12-23 11:22 pm
UK
Hi Daniel,

I am sure that you don't mean this, but please don't take any silly chances, just in case.

I have just looked quickly at the data-sheet for these diodes, and the max. surge current rating is 200A, so these are certainly not feeble devices, even if you try using one of them on its own.

However, I have experienced diodes blowing for no apparent reason, sometimes after some years of similar usage, but there is very little risk of any harm if this is inside an enclosure. The materials they are made of are not normally combustible, and this is also suggested on this device's data-sheet.

In this as yet untried instance, try to cover the area (the enclosure, if you already have this) with something metallic which should stop anything unwanted flying around if there is a failure, and be certain to cover your eyes up, just in case.

Regards,
 
If it is a new project and your first time powering it up then you may also want to use 'the light bulb trick'. Get yourself a 50W-ish light bulb and an appropriate holder. Wire this in series with the mains live, before the transformer primary. So it should look something like:

Mains Live --------- Light Bulb --------- Transformer Primary (L)
Mains Neutral ---------------------------- Transformer Primary (N)


The light bulb should flash brightly for a second while the capacitors initially charge, then go dim / not lit. If it remains lit up brightly then something is wrong, and you should probably turn it off and post about it here for some advice on what to do next in order to troubleshoot your problem :)

Good luck!
 

Bobken

Member
2002-12-23 11:22 pm
UK
Hi Daniel,

I am pleased to see that you have taken this seriously, as you are on your own in this, and there is nothing to be gained by trying to be a hero!

I would be surprised to hear that you have any problems with what you are suggesting, but, until you try this for real, no-one can be absolutely certain of the outcome. Different diodes have different characteristics, and some don't take too well to surges at all even though they will handle highish currents during normal use, and they are rated more conservatively in this particular parameter. IIRC, Schottkys are usually rated lower in respect of max. surges than some other diodes when compared with their steady-state ratings, and I don't know much about these diodes of your choice.

However, with a normal max. rating of 8A, and a suggested max. surge rating of 200A, this doesn't seem unusual here, although other matters like how long the surge lasts for until the caps charge up etc., also enter the picture.

The use of heatsinks will not affect this particular issue as the duration of these surges is so short, of course.

The suggested light-bulb trick is a very good idea initially, especially if you don't have a variac (which I use), but sooner or later you will need to try without this in series, and the surge characteristics will be considerably worse then.

I hope that things go well for you.

Regards,
 

Bobken

Member
2002-12-23 11:22 pm
UK
Hi Jan,

My experience in this instance didn't indicate that serious current-hogging is a problem with using parallel diodes, and the commercial amps I referred to seem to suggest the same.

However, I have gone to great lengths to point out that not all diodes enjoy the same characteristics, and that care should be taken in any case such as this.
I have not suggested anything reckless, and with a chip amp which continually draws in excess of 8A, I would be very interested to know what size the heatsinks are on such an amplifier.

The main risk here is at switch-on, and again I have suggested all caution should be taken, but with a capability of 200A surge, I will be surprised if one of these bridges cannot cope with the load.
Therefore, it will probably not require more than one bridge, anyway, but if it does, then again I will be surprised if adding a second bridge in parallel will not cater for these requirements.

If you are aware of the charging characteristics of this poster's smoothing caps and can precisely calculate the instantaneous surge requirements here, then I am sure that Daniel will be very grateful.

Regards,
 
Bobken said:
Hi Jan,

My experience in this instance didn't indicate that serious current-hogging is a problem with using parallel diodes, and the commercial amps I referred to seem to suggest the same.

However, I have gone to great lengths to point out that not all diodes enjoy the same characteristics, and that care should be taken in any case such as this.
I have not suggested anything reckless, and with a chip amp which continually draws in excess of 8A, I would be very interested to know what size the heatsinks are on such an amplifier.

The main risk here is at switch-on, and again I have suggested all caution should be taken, but with a capability of 200A surge, I will be surprised if one of these bridges cannot cope with the load.
Therefore, it will probably not require more than one bridge, anyway, but if it does, then again I will be surprised if adding a second bridge in parallel will not cater for these requirements.

If you are aware of the charging characteristics of this poster's smoothing caps and can precisely calculate the instantaneous surge requirements here, then I am sure that Daniel will be very grateful.

Regards,


Well, a chipamp that draws 8A continuously is a physical impossibility. It would explode long before the diodes even get warm, if its protection didn't kick in.

When there is a need to parallel diodes for current sharing, the way it is usually done is with a small sub-ohm resistor in series with each diode to make sure they share equally the current when it is needed: at high current.

I'm sure you are aware that a difference of a few 10's of mV in conduction characteristices between diodes means a factor 1:10 in current difference.

Jan Didden
 

Bobken

Member
2002-12-23 11:22 pm
UK
Hi Jan,

Of course, such a current draw by a chip amp is impossible, as I suggested, as you would need massive and quite impractical heatsinking amongst other things, which is obviously not the case here with (any?) chip amps. We are probably dealing with a max. of 25% of this 8A capability in reality, which is why I don't forsee any real problems here, even with only one diode device being used.
The surge potential is not quite so clear cut, though, and this is why I continually suggested that caution should be adopted, at least initially. If the switch is flipped just when the mains is at a peak of the AC, and the smoothing caps are seen almost as a dead short, who knows precisely what this can do by way of instantaneous current draw?

I don't measure the characteristics of very many diodes as I haven't found it necessary for any matching purposes or whatever (unlike most other semi-conductors which I match as a matter of course), but recently for another purpose I did check a batch of fairly mundane diodes, and found that their forward voltage drops etc. were remarkably consistent.

It is academic now, but although I don't presently recall quite why Metaxas advocated this multi-parallel diode arrangement, my guess is that it was related to keeping the mains impedance as low as possible, as this does have a sonic effect with amps (like Class AB) which don't draw current consistently. For this same reason I would also avoid deliberately adding any resistance in this path, however low this might be, in order to 'equalise' parallel diodes. In my experience, it should not be needed, especially in this instance.

I have always been keen to encourage others to try things out for themselves, as I (and I suspect many others) have learned a great deal about these matters by so doing, and in my own case over a period of some 40 yrs. In this situation where the voltage is low, and with very little current draw to consider, I don't forsee any serious danger or disasters occurring, providing a little common-sense and care is taken by Daniel.

I hope that things go well for him, but I am sure the exercise will teach him something worthwhile, even if for some obscure reason it transpires that his chosen diodes are not quite up to the requirements of this job.

Regards,
 

BWRX

diyAudio Moderator Emeritus
2005-01-17 5:29 am
Pennsylvania
janneman said:
When there is a need to parallel diodes for current sharing, the way it is usually done is with a small sub-ohm resistor in series with each diode to make sure they share equally the current when it is needed: at high current.

The best solution is just to do what Jan says and put a small resistance in series with each diode. This also may have the advantage of slightly damping out some of the ringing caused by the diodes switching.

As far as estimating the surge current through the rectifying diodes at turn on, you should assume that the supply caps are completely discharged and that the maximum rail voltage will be put across the caps in about 1/4 the period of the mains waveform (about 0.004s for 60Hz and 0.005s for 50Hz). Then use i=C(dv/dt) to calculate the current.

Using 26,800uF, 35V supply rails, and 60Hz mains equates to about 235A.