When I connected 2 diodes (30A) in parallel, the current was flowing only through only one of them. Why didn't it flow through both?

My input source is 2 solar pannels (550W each) connected in parallel (each panel is protected with the same kind of diode (30A) at the roof).
I feed this input (that would be 43V , 25A current for both panels) into 50A step down module, which outputs 29.2 V . The output wires of the step down module I route it through two 30A diodes (which I borrowed from solar panel installation) before they go into the BMS. The BMS will make the current on the bus to flow both directions, so I have to put a diode to prevent backflow (otherwise the buck converter will burn), but since i don't have 50A diode (or bigger) I decided to wire the diodes in parallel. 30A + 30A = 60A, I thought it was enough. (See attached picture for reference)

Now, when measuring the current with clamp ampere-meter I notices that current (25A at the time of measurement) was flowing only through one of them, the other was getting somewhere like 400 milliamps of current or so. Then the diode that was doing all the work broke (I heard the "click" sound) and it smelled like smoke was going, I checked with clamp meter and the diode had 0 current, but at the same time, the other diode entered to work, and I was seeing the same 25A current flowing . So, first diode burned, the second diode entered operation immediately.

According from what I have read about Kirchhoff's Law, if I have a node with 1 input and 2 outputs, the current must be flowing both sides, a portion of current will flow on one wire, the other portion will flow on another. Why didn't this happen in my case??? Yes, I can understand that maybe wire will get 10amps, another will get 15 because they have different resistance. But 0 amps on one wire and the other one burning??? This is very weird to me.

How do I make it so current flows both wires ?

 

The two diodes do not have exactly the same I-V characteristics.

Draw the I-V curve of two diodes. Now pick a diode forward voltage on the X-axis.
Follow up until you intersect with the I-V characteristics to find each diode operating current. You will find two very different currents for the two diodes.

 

You will find two very different currents for the two diodes.

I can understand that, but the diodes are most likely from the same manufacturer. How can they differ so much in specification? I believe it is 30A10 diode.

 

I can understand that, but the diodes are most likely from the same manufacturer. How can they differ so much in specification? I believe it is 30A10 diode.

Semiconductor characteristics are not normally well controlled. Think of each parameter as a random variable with a mean and a variance. In any sample of two parts from a lot the chances of finding two identical parts are rather small.

This is why for example; designers do not design circuits that depend on the value of a specific parameter. Transistor beta is the classic example.

 

I would suggest using current sharing resistors in a pinch, but you are better off just getting a proper diode.

 

The two diodes do not have exactly the same I-V characteristics.

I think I am starting to understand what you mean. See, on the buck converter I have the output voltage set to 29.2, however it doesn't show that on the display, it shows 27.1 volts. This is because the BMS (which is a 300A capable BMS) sucks all the current and the voltage drops. So on one side of the diode I have the BMS with 27.1V and on the other side of the diode (buck side) I have 29.2 , which gives me 2.1 volts of potential difference. Now the chart starts to make sense. Because when I saw the chart, I said, "well, I don't have voltage in 1-5V region, I have 29V", but it is not the absolute voltage value, it is the potential difference what is important, right? So when first diode opens, all the voltage drops and the other diode doesn't have enough time to open because there is not enough current anymore (because the BMS sucks all the current immediately, it is 300A but the buck is only giving 25A, only 10%, nothing for a battery in a depleted state), am I getting it correctly?

 

Do you have a model or part number for those diodes or, better yet, a link to the data sheet for them? I'm not coming up with anything that I'm confident is what you are using based on the 30A10 number you gave previously.

Semiconductor parameters can vary quite a bit from one part to the next, even from the same lot or even the same die. The exponential nature of the V-I characteristic of diodes means that it doesn't take much difference in the characteristics to result in one diode hogging all of the current. This is aggravate by thermal runaway in these parts, since the voltage across a silicon diode goes down for the same current flowing through it. So, even if the two diodes initially conduct almost the same current, one will heat up a bit more than the other, which results in it's voltage dropping, which results in more current going through it and less through the other one, which results in it heating up even faster than the other one -- and so on, and so on. Soon, you have one diode with nearly all the current flowing through it, followed by it failing, usually followed by the other diode failing shortly thereafter because now all of the current goes through it.

 

I can understand that, but the diodes are most likely from the same manufacturer. How can they differ so much in specification?

Process variation. Even diodes on the same die will have different characteristics; though diodes that are adjacent to each other on the same wafer will tend to be more closely matched.

When you buy diodes from a retailer, you aren't even guaranteed they're from the same lot. At the company I worked at, a lot was usually 50 wafers (6", 8", or 12") but we were manufacturing microprocessors that tended to be about the size of your thumbnail.

Manufacturers can bin (sort) diodes by certain characteristics. The company that used to be Motorola did this when they put zener diodes in parallel to create TVS diodes.

 

Do you have a model or part number for those diodes or, better yet, a link to the data sheet for them? I'm not coming up with anything that I'm confident is what you are using based on the 30A10 number you gave previously.

I bought everything on AliExpress, they sell cheap stuff. That diode was like 1 or 2 dollars (including plastic holder). I don't have datasheet , they sell it as standard diode for solar panels. it says 30A10 on the diode.

So, even if the two diodes initially conduct almost the same current, one will heat up a bit more than the other, which results in it's voltage dropping, which results in more current going through it and less through the other one, which results in it heating up even faster than the other one -- and so on, and so on. Soon, you have one diode with nearly all the current flowing through it, followed by it failing, usually followed by the other diode failing shortly thereafter because now all of the current goes through it.

that's exactly what happened! Does this behavior is the same for MOSFETs? Because I could use a power MOSFET instead of a diode. But if MOSFETS suffer of the same problem, then how do they make it work with parallel MOSFETS? I have opened my BMS enclosure, it has about 20 or 30 MOSFET transistors all over the place, and it doesn't even have a FAN!! And it doesn't heat much. Buck converters/ inverters, they all have transistors in parallel, how does it work for them? Is it all because of heat only? Then I could fill diodes in a case with silicon adhesive (mixed with some dust improving heat conduction) and they would be submerged in a substance that would level the temperature for all diodes? Will this fix it?

 

this is the datasheet of the diode

 

I bought everything on AliExpress, they sell cheap stuff. That diode was like 1 or 2 dollars (including plastic holder). I don't have datasheet , they sell it as standard diode for solar panels. it says 30A10 on the diode.


that's exactly what happened! Does this behavior is the same for MOSFETs? Because I could use a power MOSFET instead of a diode. But if MOSFETS suffer of the same problem, then how do they make it work with parallel MOSFETS? I have opened my BMS enclosure, it has about 20 or 30 MOSFET transistors all over the place, and it doesn't even have a FAN!! And it doesn't heat much. Buck converters/ inverters, they all have transistors in parallel, how does it work for them? Is it all because of heat only? Then I could fill diodes in a case with silicon adhesive (mixed with some dust improving heat conduction) and they would be submerged in a substance that would level the temperature for all diodes? Will this fix it?

They are probably using different MOSFETs for different things -- it's unlikely all of them are in parallel. But some of them likely are. There are a number of techniques to make this work, and the manufacturer likely used some combination of them.

First, the MOSFETs could have been "binned", meaning that they were individually measured and sorted so that they are in groups of devices with very similar characteristics. For volume production, you can pay a small premium and have the device manufacturer supply you with binned parts (since they have to measure every device anyway).

Next, the MOSFETs that are in parallel are likely coupled pretty tightly thermally so that they operate very close to the same temperature. Your idea of coupling them is the right idea, but I doubt your dust-filled silicon adhesive idea is going to do the trick.

Third, the circuit topology can incorporate devices to provide negative feedback and force equitable current sharing. There are a number of techniques to do this. One of the simplest is to incorporate a small ballast resistor in series with each device. Another is to mount a thermally-sensitive component on each device and use it to rob the device of drive signal as it heats.

 

this is the datasheet of the diode

If you are looking at currents of 50 A, you should go ahead and get a diode that is rated for at least 100 A. DigiKey has them for <$4 in single quantities. Think of the time and headache you can save yourself for just a few bucks.

 

I made this video that touches on the question you asked.

 

Does this behavior is the same for MOSFETs?

Not necessarily.
The difference between a diode and a MOSFET is that the forward drop of a diode reduces with a temperature increase, while the on-resistance of a MOSFET goes up.

So with two diodes in parallel, one will tend to take slightly more current due to normal variations, and the temperature increase from the higher current the causes it to carry even more current and further heat up, leading to a runaway current hog condition.

Just the opposite happens with two MOSFETs in parallel, since any on-resistance increases with temperature will tend to cause more current to flow in the other MOSFET, so they tend to self-balance the load between them.

 

That claims to be a 30A diode - diodes in that package are generally rated at 6A.
The forward voltage at full current is 1.15V, which means it will be dissipating 34.5W.
Its thermal resistance is 6°C/W in free air (but it's in a plastic case)
That would be a temperature rise of 207°C in free air (but it's in a plastic case). It's never going to work!
The forward voltage drop falls with temperature, so the warmer one takes more current, so as they warm up, the warmer one ends up taking all the current (as you discovered)

 

I made this video that touches on the question you asked.

Very nice video. Well worth watching even if you already know the material. For the neophyte it will be very informative about the content; for the potential mentor it offers a great example of a cogent, lucid presentation of the subject matter.

Really well done.

 

Indeed. Beautifully done...

 

Thank you all, guys for the great input you have provided! I am now a lot more knowledgeable than before.