failing transistors

[bfhamr]

Newbie here. I repair machines for a living, I'm currently working on an EDM. Normally I would contact the manufacturer but I'm told they are no longer in business.
The machine has a set of 6 transistors (Hitachi 2SC1343). These can be switched into and out of the circuit depending on how much power is wanted while machining. The problem is these transistors are failing (shorted) and I'm not sure why. I had replaced the board which contained these and all 6 have failed again. Another board which had been in the machine had 4 of the 6 shorted.
I'm not looking for specifics on this machine but in general what would cause a transistor to fail shorted.

Thanks
Dan

 

[Comcokid]

The transistor - 2SC1343 - Silicon NPN, 150V, 10A, 100W TO-3 case.

OK, you are dealing with a power device in some kind of circuit. Transistors fail for the following reasons - too much voltage, too much current, too much heat. Most power transistors normally fail shorted.

Not much information provided. You need to indicate more how the transistor is used in the circuit.

I once permanetly repaired a very old CNC mill that regurarly blew the drive transistors for it's steppers. The design used a mag-amp arrangement, and everytime the incomming power had a short interruption while the machine was operating it would blow. By carefully examinaning the mag-amp transformer arrangement and measuring the voltages, I selected and placed large MOVs to clip any transients from the transformers before they reached the drive transistors. Also replaced the house-numbered drive transistors with some easily available high voltage TV sweep ones. Sometimes a design engineer is called upon to do the unusual.

 

[bfhamr]

Sorry for the lack of info, I guess I was looking for general info. I'm not in front of the machine, actually I'm in the office which is an hour away.
Here is what I know about how the machine works:
--power is supplied to the transistors in question.
--the transistors are switched on and off rapidly to get the proper burning at the electrode.
--when more power is needed more transistors are switched into the circuit.
--the transistors are current limited by large wire wound resistors. Each transistor has a seperate resistor.
--I measured the voltage from the emitter to the collector at 70-100 volts DC
--the transistors are all switched on and off, whether they are needed or not. When needed, a relay is energised closing a set of contacts in series with the wire wound resistor and collector. This wil then increase the power available.

Sorry for the poor description,

thanks
dan

 

[bfhamr]

For those of you that don't know what EDM is, it is a spark erosion prosess. An electrode is brought in contact with the part to be machined. An arc is established and the metal is burned away. The machine controls the arc and moves the electrode up and down to maintain the arc. More power results in a faster burn but a poorer finish.

 

[itsmoked]

And, it is usually done under a liquid.

Well, wire wound resistors can have inductance that then cause a 'kick' which can be very much higher than 100V.

Also you realize that the transistors are depending heavily on good cooling. If a fan is bad, or not spinning at rated speed, or a filter is dirty, or a door is left open, or the heatsink grease is wrong, or different, or the ambient temperature is high, or the electrode is bigger, or the liquid is conductive, or, or, or... You will blow a transistor which will heap the load onto its neighbors which will all blow in sequence faster and faster.

I suspect the design is poor since each T's current is not being monitored to prevent recurring sequential burnouts. If something obvious in my list isn't present you may need to do as Comcokid mentions and actually hunt down the problem and make changes to the design.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[bfhamr]

I don't believe the burnout will be sequential, the T's are seperate with their own seris resistor. They are then in parallel with the electrode. I'm wondering if maybe the "kick" from the wire wound resistors may be whats causing the problem. There are diodes in the circuit to bypass reverse voltage spikes, these I believe all terminate at the same spot. I guess I need to check the integrity of that part of the circuit.

By the way, the machine is 10-15 years old. I'm told it's been working properly until recently.

Thanks for the replies.

Dan

 

[MacGyverS2000]

I would definitely make sure a diode is across those wire-wounds, and check the relay coils to make sure...

Transistors aren't big fans of operating in parallel... if one burns out, the load is shifted to the remaining operational transistors. Unless they are perfectly matched, one is going to heat up just a bit more than the others, causing a larger portion of the current to flow through that one unit. Thermal overload quickly ensues on that one unit, blowing it, and a dominoe effect is seen on the remaining units.

When it comes to current sharing, FETs are a more reasonable choice. IGBTs are another possible choice, though I have nearly zero experience using/designing with them and could not provide much useful info on them.

I'm curious as to why the original designers kept the transistors turned on all of the time and switched them into the circuit with relays. Seems like adding in an unnecessary step, and the relays are probably bulky/pricey.


Dan - Owner

http://www.Hi-TecDesigns.com

 

[OperaHouse]

I don't think IGBT's would be fast enough. Definitely you should replace these with higher voltage transistors. Do these transistors get warm? Another possibility could be there is a problem with the driver. If these transistors do not go into saturation they will overheat and fail. The driver could be the common element. Check for a resistance in the driver that may have changed. Look for a resistance problem from the emitter common to the driver board. I had a cheap China machine that had a problem in the looping common of the power supply caps. Either cheap or stupid, they didn't have any redundat paths in case one failed. I used to import and repair EDM machines and have made supplies up to 1MHz 400V @ 6A for a specialized application.

 

[bfhamr]

I've gotten a little more info.

-The transistors are not getting hot, only lukewarm when touched. All cooling fans are working.
-The transistors and the associated switching circuitry are all located on the same board. This board has been replaced and has failed since being replaced.
-There is a single wire with the on/off pulse stream coming into this board, all the transistors are switched on/off at the same time.
-Each transistor has a large wire wound resistor in series with the collector terminal, these resistors are 13 ohm. The voltage between the collector and emitter is 100v DC.
-each resistor has a reverse voltage diode across it, these all are on the board with the transistors, so they have been replaced.
-The only other voltages coming into this circuit board are two lines with +8 volts dc and -8 volts dc. These supply power to a few transistors used to turn on/off the power transistors.

thanks
Dan

 

[Comcokid]

OK, sounds like too much heat is not the issue. This just leaves overvoltage or current spikes.

Make sure the reverse diodes across the transistors are fast enough to conduct quickly. 1N400x power diodes are fine for 60 Hz rectification, but don't always do the job for fast inductive spikes.

The general specs I found on the transistor indicated 150V. If you're measuring 100 Volts, this doesn't leave much margin. I went to Digikey, searched "transistors" and filtered for the highest voltage and current TO-3 case NPN transistors and saw there were heafty choices available at around $5 each. If you choose this route, do a little more research on your original transistor before selecting a better replacement.

 

[itsmoked]

What happens when the wire/electrode touches the work for an extended period like 2 seconds, 13Ohm resistor is supposed to take the hit? 100V/13Ohm = 7.6A Doesn't seem too abusive.

It's sounding like a voltage hit or bad drive. 100V is pretty easy to come up with voltage wise. You can get 250V 16A NPNs.

But the drive problem could easily explain the multiple failures as it would effect all the transistors. It might only be an intermittent drive problem that would toast the transistors in a second or two but otherwise they run cool.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[bfhamr]

itsmoked,

I would agree that a drive problem could explain the failures except that one of the boards (with the 6 transistors) has 2 of the 6 transistors still working perfectly. Wouldn't all of the transistors fail? Maybe it's justs coincidence that 2 haven't.

dan

 

[itsmoked]

Dan,

[highlight #FF99FF]"The only other voltages coming into this circuit board are two lines with +8 volts dc and -8 volts dc. These supply power to a few transistors used to turn on/off the power transistors." [/highlight]

It depends on how the drive is realized. Could be that once a few Ts are cooked the weak drive may then be enough to properly drive the remainders.

Can you say more about the drive? Is there a drive transistor per main T or are they all driven by one?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[bfhamr]

I've tried tracing out the circuit on the board which has all 6 transistors. The 6 are controlled by 1 transistor, but that one is also controlled by several others. I've not been able to completely trace things out. Not being an EE, what may be a familiar circuit to you, takes me awhile to understand.

thanks again
dan

 

[Comcokid]

One other thing I would check out. Most electrolytic capacitors dry-out over time when they are in constant use reducing the capacitor value significantly. Check out the capacitors on the 100 volt supply. Maybe they're not holding the 100 volts steady enough as the EDM does it's thing. Replace any suspect caps with good quality ones.

You might also check out the +/- 8 Volt supply caps as well and any electrolytics in the drive circuits, but if the issue was here you would probably be complaining that the transistor were not turning on hard enough rather than blowing.

 

[MacGyverS2000]

You might also check out the +/- 8 Volt supply caps as well and any electrolytics in the drive circuits, but if the issue was here you would probably be complaining that the transistor were not turning on hard enough rather than blowing.

Well, if the transistors weren't being driven hard enough, then they would overheat quite quickly and blow. Sounds to me like a very valid possible explanation.


Dan - Owner

http://www.Hi-TecDesigns.com

 

[Comcokid]

However, he has already established that the transistors are not overheating.

 

[bfhamr]

After a bit of searching I was able to find a schematic of the circuit board. I'd post it but I don't have any way of scanning it. I could take a picture of it if anyone is interested.

The last time I was able to work on the machine I noticed that the negative side of the +/-8 volts was low, about 5.5 to 6.0 volts dc. The transistors (the two left that hadn't failed) still ran cool.

Dan

 

[MacGyverS2000]

bf, when were the temp measurements made, before or after some of the transistors blew? If before, it's a non-starter, but if you were only able to take the measurements after, well...

If drive power is flakey, all 6 running may make for a set of hot transistors. Once a few blow, it all cools down.


Dan - Owner

http://www.Hi-TecDesigns.com

 

[itsmoked]

A picture would be great!

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com