1. Home
  2. Electronics Design
  3. Overrated transistors

Overrated transistors

On the topic of overrated transistors:

I recently wrecked my dummy load which I use for testing power supplies. Because of its age and bad (too compact) mechanical design I decided to built a new one from some leftover parts instead of fixing the old one. In theory this should be the quickest way to get back to testing. One of the parts I had was a large heatsink with 4 transistors rated 1000V 100A 625W (QM100DY from Mitsubishi). Ideal for dissipating some power.... So after a weekend of cutting, milling & drilling I was ready to test it. Much to my surprise the transistors fail after a few seconds while dissipating only 100W (50V 2A) *. I'm wondering what kind of failure mode is behind this. The datasheet doesn't warn against using the transistor in its linear region.

  • I connected only one transistor to begin with.
--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
nico@nctdevpuntnl (punt=.)
--------------------------------------------------------------
Reply to
Nico Coesel
Loading thread data ...

Age? Seriously, did you put in heat sink grease? Sometimes, its necessary for proper heat transfer.

Reply to
linnix

See the middle of page 4 "FORWARD BIAS SAFE OPERATING AREA":

formatting link

DC SOA issue.

Looks like Id

Reply to
Winston

Well, it shows the SOA at DC to include 100A Ic at 50V Vce, so that shouldn't be it.

_How_ did you connect the 'transistor'? Forcing current to flow through the internal driver transistor rather than the output device might be a good way to fry it.

Reply to
Spehro Pefhany

(...)

5 KW D.C. dissipated by an 800 W device? :)

--Winston

Reply to
Winston

(...)

Avoid the 'Mocha'. It is wicked bad. AMHIKT

--Winston

Reply to
Winston

Oops, you're right, I misread the graph.

Reply to
Spehro Pefhany

Yup, mea culpa.

Reply to
Spehro Pefhany

I misread too. I meant "Ic < 1.3 A at Vc = 50 V"

The 'Maximum Ratings' Giveth, the 'D.C. Line' Taketh Away.

--Winston

Reply to
Winston

No way it's hitting any V or I limit. It's the heat, stupid (don't mean any of you).

Reply to
linnix

El 18-07-11 20:37, Nico Coesel escribió:

Hello Nico,

The SOA says 100V at 5A, so this shouldn't be the problem (assuming sufficient cooling).

It is a darlington, maybe you have a pulse-type oscillation (low average current, but high low duty cycle peaks). Do you have a large capacitor parallel to the load that can have sufficient energy to fry the device?

Other thing I can think of is that you drive it with too high dIbase/dt, so the main current must be carried by the internal smaller transistors during the on-switching event (as the main transistor needs time to take over the current). Maybe you find some application info for these heavy darlingtons.

Best regards,

Wim PA3DJS

formatting link
Please remove abc first in case of PM

Reply to
Wimpie

Under extremely ideal cooling conditions.

How big is your heatsink? Do you have fan on it? Do you have plumbing on it? To cool 2KW, you need to hook it up to your water heater.

Same for my 60A generator. Using 20A and thermo shutdown in 10 minutes. I need to hook it up to cool it with 40,000 gallons of water.

It's the economy, stupid. It's the heat, stupid.

Reply to
linnix

Although I don't understand a Non-Repetitive DC SOA.

John S

Reply to
John S

Reply to
John Larkin

If you are willing to part with the other half of the module, reduce your Vc to 10V and increase your current to 10A. If it doesn't fail after a few seconds, the problem was with the DC SOA. If it does fail, then heat is the likely problem.

All: If this seems to be bad advice, please jump in and let us know. I'm interested in this, too.

John S

Reply to
John S

Ofcourse.

The heatsink is the original heatsink. 140x290x59mm

Winston is dead on. I didn't bother to check the SOA. But the SOA graph looks weird. If I apply 10V the device can handle 70A (700W). If I apply 40V then the device can handle 3A (120W). This doesn't make much sense to me. Its almost as if the first two stages are not on the heatsink.

According to the datasheet the Rth of the junction to heatsink is

0.23C/W. If I multiply this by 100W I get a temp. rise of 23 degrees. Even if the Rth is half as good as they say it would mean a rise of less than 50 degrees. Nowhere near the maximum temperature. I could not feel something getting warm during my tests either.

Anyway, I opened one. The transistors are mounted on a ceramic substrate which in turn is mounted on the base plate which gets bolted onto the heatsink.

--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
nico@nctdevpuntnl (punt=.)
--------------------------------------------------------------
Reply to
Nico Coesel

Good first step, but should also test at the limit. For example: 20V

30A. Closer to the upper right corner.
Reply to
linnix

You are right. It is a P=V*I limit.

--Winston

Reply to
Winston

It's not going to handle 2KW, unless liquid cooled.

Can't argue with the data sheet, if they designed it that way. Just build your circuit close to the middle of the upper right corner.

Or not over 40V. Depends on the semi.

Reply to
linnix

I'm guessing it has to do with the pulsed part of the SOA curves. For example, at 1ms the curve shows 70V and 200A. They don't say how long you have to rest after that kind of pulse. Obviously, the max dissipation would be one indication. But it's just not stated in the SOA. So they take the easy route and just write "non-repetitive" to save their ass on it. With "non-repetitive" written there, they are taking the safest possible position. You use your brain and what else you can dredge up from the data sheet and experience for the rest.

Jon

Reply to
Jon Kirwan

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.