How can El Cheapo bridge rectifiers be so small yet rated at 50 Amps?

[Chris Wilson]

I had one leg of a 35A 600V bridge rectifier module go short the other day. Only 100V on it, and less than 12 Amps, bolted to a large sheet metal case with thermal paste, but never felt the region get even mildly warm during the months it worked fine. It got me thinking how a similar size, alloy potted case device can be rated 100V at 50 Amps?

I have some 10A10 diodes and I doubt 4 of them would fit in the case. is it down to heat dissipation when potted and thermal transfer from the bolted case to sheet metal? Thanks.


The bridge feeds two 10,000uF smoothing caps in parallel, but the module failed not at switch on, but during constant state usage.

[TheUnnamedNewbie]

Let's say that there is a .5V voltage drop over the diodes inside (which I think is quite a realistic number for a schottky bridge rectifier) and you try pulling 50 amps through it, you are going to be dissipating 25W... I don't know just how big the block is, but that sounds like a lot. You should be able to find the Rtheta for the rectifier and do the math... Gonna need a lot of metal to dissipate that much heat tho.

[exe]

> is it down to heat dissipation

I think so. And heat dissipation depends on voltage drop. But different diodes have different forward voltage curves. It is also temperature-dependent (e.g., Schottky diodes conduct better when warm, normal diodes not). My impression that higher current diodes have more silicon in them. Therefore, they have lower voltage drop and can sustain higher currents (apart from bigger packages that better transfer heat.

My impression is that for high current diodes are no good, better use "ideal diodes" with mosfets or syncronus rectification (again, with mosfets). But if you have to use a diode bridge check voltage drop for your current. The problem may be not with a diode, but with insufficient cooling. It's a common gotcha with power products: they often need huge heatsinks to realize full potential.

If you are asking about cheap Chinese semiconductors: they may not meet the specs and can be ripoff and dodgy.

PS I'm no specialist, just my £0.02.

[f5r5e5d]

"the spec" can be unrealistic too

many power semis have a P_max stated for "25 C case temp" - requires a water cooled cold plate to match

then you need derating curves, realistic thermal design - often achieving even 1/2 the optimistic P_max requires a quite substantial air cooled heat sink

[Synthetase]

Consider also that the thermal resistance from junction to case and case to air of your average diode will be much higher than those block-shaped rectifiers. The rectifiers have large dies bonded to metal cases, bonded in turn to a heatsink. Discreet round plastic-cased diodes obviously don't have any of these things so they will only be able to dissipate a small amount of power for an equivalent temperature rise.

[T3sl4co1l]

It's even more impressive when you consider the construction.

Normally you see a metal plate or box on the end of one of those parts.

If you ever take the time to smash one up and peel it apart, you'll most likely see:

The heatsink plate is just that, a tab of metal glued into the package.  The lead frame is the first heatsink, having the diodes soldered onto it.  Tabs are bent up, so that pins/terminals exit the top of the package.  The lead frame is really just floating in the package, a small height above that backing plate.

So despite appearances, it's really just a TO-220FP or whatnot, in a different shape, using diodes.  Full-packs are notoriously poor at dissipation (e.g., 30W max for a TO-220 held at room temperature!).

It just goes to show how robust junction diodes are, even at high operating temperature.  And they'll use every bit of their 175C ratings, too!


Related story:
Back when I was building a 5kW induction heater from scratch, I had a 35A 800V FWB on the supply (rectifying 240V mains).  I had it up around 20-30A DC output, on a not-very-good heatsink.  Eventually it failed, drawing fault current through one unlucky diode, once its companion died shorted.  Amazingly, despite that ~2000A surge, on top of already being at ~175C maximum temperature (or more), 3 out of 4 diodes still measured well (correct Vf, low enough reverse leakage).

Junction diodes, and similar parts like thyristors (SCRs and TRIACs), are impressively robust.  They are the only semiconductors that are robust enough to actually be able to protect with fusing: given that you use the correct type (a so-called semiconductor fuse has such an aggressive fast-blow characteristic, that they can clear faults within milliseconds -- on that time scale, it doesn't matter if it's AC or DC!).

(Semiconductor fuses can be used on IGBTs and the like, but only as a safety feature -- the transistor is past the point of no return, after maybe 0.02 ms, but in the ensuing 0.98 ms, plasma expands, fracturing encapsulation and launching sharp bits of shrapnel.  The blast and arc flash is reduced when such fuses are used.)

Tim

[Seekonk]

I bought some 50A 100V bridges on ebay.  Even the seller noted in the listing that he has complaints about them shorting out.  They were super cheap and I needed something quick.   I needed them for a 20A 48V battery charger.

I did two in parallel with 8 inch leads of #20 so they would current share and even added an additional heat sink on each one.   With all that at a 14A charge they ran hotter than hell.   One of them filed 6 months later.  Forward voltage was about 1.5V only at 2A.  They weighed about half of what my 25A bridges weigh. 

[SeanB]

I have some old bridge rectifiers that are rated at 6A 400VAC, and they are in a massive package, around 3 times the volume of a "35A 400V" bridge rectifier package. They run cool, and I never had trouble with them aside from them being too large to fit on some heatsinks.

[GNU_Ninja]

Welcome to the wonderful world of Wun Hung Lo component specifications 

[Jeroen3]

Were the diodes protected with mov's? Since voltage kills diodes way faster than current does.
Since they were not hot at all I assume a voltage spike, which only lasts a few uS, killed it.

[Chris Wilson]

Just to revive this old thread, and to say a thank you as I thought I had already posted that "in period", sorry. I had another fail yesterday, a supposed 50 Amp one, again one leg shorted. I will probably go for stud rectifiers on a custom heat sink affair, rather than try and improve the thermal transfer from a tint encapsulated package.