Rectifiers in parallel

[Annakin4]

So quick question, can I take two DSI32As and put them in parallel to double the amps it can take in?

DSI32A- Up to 3.2A
120v x 3.2A = 420 watts
420 watts x 2 packages = 840 Watts

There is a reason, don't question XD

[Ice-Tea]

No.

[mariush]

No, the current won't be spread evenly.

I can't find a datasheet for DSI32A but i suspect : No, the rectifier itself is rated for 3.2A only it has enough cooling (there's a large enough copper surface on the pcb to dissipate the heat) or you actually have a heatsink on top of the rectifier. In real world, the chip would probably die due to overheating, especially if you're laying it near a second rectifier.
No matter what you do, you have the [current x voltage drop on internal diode x 2 ] watts that have to be dissipated in the air.

For lower input voltages (let's say up to 72v AC), you may be able to use a LT4320 plus 4 mosfets to rectify the AC to DC in a very small area and with very minimal heat dissipation (which depends on the Rds(on) of mosfets selected). Expensive but a good choice if you don't want heat and it has to take as little pcb space as possible.

[nwvlab]

Hi!

It's almost always a bad idea connecting in parallel two diodes (without any compensating resistor or other techniques), to increase their ampacity.

For sure, there will be some mismatches, and one diode will carry more current than the other. 
The diode carrying the larger current will eventually heat up and its "equivalent resistance" will decrease. This will end up in a thermal run away, with one diode carrying a current larger than its ratings, with a consequent catastrophic and spectacular failure..

This effect could be mitigated (but not 100% solved!) by:
1) Mounting the diodes on the same heatsink, if possible (still take into consideration the exponential dependence of the current to the temperature, the thermal resitance and thermal capacity, the latter being dangerous during transients).
2) Using a lot of "headroom" (just as an example if you need 2 amps, you might try using 2 diodes with 1.5A ampacity... but at this time a single 2A diode could be cheaper. Still beware of the exponential I-T dependence).
3) Using a resistor in series to each diode (but you'll waste a lot of power).

However, I would never recommend you doing this... I have seen many chinese consumer products failing because of such a poor design practice...

Cheers

[wraper]

You can connect SiC diodes in parallel because they have positive temperature coefficient of the forward voltage. Paralleling conventional Silicon diodes generally is safe only if you do this with a dual diode in a single package.

[dom0]

You can connect SiC diodes in parallel because they have positive temperature coefficient of the forward voltage. Paralleling conventional Silicon diodes generally is safe only if you do this with a dual diode in a single package.

+1

But even when paralleling dual-packaged diodes you can't have them conduct twice the current. 140..160 % might be okay.

[Seekonk]

I've used 8 diodes in parallel with good results without a common thermal package.  I've seen this done in small welders.  They don't share evenly but in some cases it can be a solution to a problem.  I would be comfortable with numbers quoted prior on the low side.

[langwadt]

I've used 8 diodes in parallel with good results without a common thermal package.  I've seen this done in small welders.  They don't share evenly but in some cases it can be a solution to a problem.  I would be comfortable with numbers quoted prior on the low side.

just using separate wires to each diode might add enough resistance to reasonably balance the currents

[Seekonk]

It is a race between the forward voltage dropping with increase in temperature AND forward voltage increasing with higher current.

[Ice-Tea]

Who the hell is voting "yes"?

Please look at the question:

So quick question, can I take two DSI32As and put them in parallel to double the amps it can take in

The answer to that is no. Yes, maybe more but never 2x the original current. 

[TimFox]

When you put two devices in parallel, the voltage across the devices is equal, and the total current equals the sums of the two device currents.
However, and this is a big but, the current through a PN diode is an exponential function of the voltage, and a small change in the co-efficient will make a large change in the current.  This is why equalizing resistors are used when it is absolutely necessary to parallel two diodes, but given the wide range of diodes on the market there is no good reason to parallel rectifier diodes.

[c4757p]

Why the hell would you use a poll for this? Everyone and his dog can vote. Ask for answers with reasoning and disregard the stupid ones.

[Seekonk]

given the wide range of diodes on the market there is no good reason to parallel rectifier diodes.

And now for something completely different.

I'm just cheap and I needed a 100V SCHOTTKY diode.  At the current I needed that is pricey. 5A SB5100 were dirt cheap and I was looking for the lowest forward voltage by running them at much lower current than rated.  This solution seemed to work for me.  Maybe in a couple months I'll run this test when I get to my camp.  It would seem you could calculate the maximum current thru any one diode by looking at the forward voltage of all of them in parallel.  While not what the OP asked, this seems like an interesting way to lower forward voltage if space is not critical.

[TimFox]

If you measure the forward voltage (very carefully) across the parallel set in your photo, you will need to know the voltage-current relationship of each individual diode to find the maximum current through any one diode.  The typical I-V curve in the data sheet is not accurate enough to find the current.  Also, the I-V curve depends on the junction temperature.  Had you measured the I-V curve for each individual diode (soldered by itself to the heat-sinking termination) before doing the parallel measurement, you might have a chance.

[Seekonk]

I did a little test with some SB5100 diodes.  House temperature was 16C (wife likes it cool)
so voltages are higher than the chart. First test is a single diode.  In all tests power was
removed to minimize heating between readings.  There were still heating effects.

0.1A    .5181
0.2A    .5706
0.3A    .6017
0.4A    .6216
0.5A    .6367
0.6A    .6467
0.7A    .6530
0.8A    .6575
0.9A    .6610
1.0A    .6645

Next test paralleled more diodes at 1A.

X2   .6264 compare at 0.50A
x3   .6005 compare at 0.33A
x4   .6787 compare at 0.25A

These numbers seem to indicate pretty good sharing.

Next test was with four diodes in parallel.

2A   .6215 compare at 0.50A
3A   .6354 compare at 0.75A

The test could have been constructed better, but they seem to be playing pretty well
together.  This would be a rare thing to do, but could be done if the designer was
mindful and had a compelling reason.