Back emf when unplugging coil?

We don't know your circuit. I could make a wild finger in the air guess, but it probably won't be what's going on. Why not... Tr switches off softly enough not to cause excessive kickback spikes Coil is unplugged while tr on. Coil sparks to connector, but by this time tr is fully off. Result: >1kV on tr.

Hopefully you've got enough protection in place though, and it's just a random part failure.

NT

Does the transistor get very hot? It could dissipate about 30 watts.

The stray capacitance on the coil terminals, and variations in 'bounce' times, make for a lot of non-repeatability here. Wiring a tiny MOV on the coil-side connector would be an easy fix, if it WERE possible to store enough coil energy to make a transistor go into runaway.

More likely, an RF source and accidental resonance might put inconvenient AC into the transistor. Were there transmitters nearby, and can you grid-dip the assembly to find its resonances?

Right... opamp through 5k ohm feeds base, npn darlington, collector to V+ emitter to coil (and 10 ohm series C.. good size alum. electro, in parallel. Zobel compensation) then 0.5 ohm I_sense resistor to ground. There's zero compensation on the feedback network... I was much younger, electronically then. seems to work fine, good enough is often good enough.

George H.

That is my guess. I didn't tell you* that for over ~1 amp I've got a switch that lets you plug in an external DC supply in the back. We use to supply a 36V 3A Kenwood... went away, so now a 60V 3A Volteq.. I'll ask the tech and prof what PS they have in the lab.

Thermally I haven't tested it... Tomorrow, (in the voice of foghorn leghorn) "Fortunately, I still have product on the shelf that can be used for testing."

George H.

What's the heat sinking like on the transistor?

Post a pic!

A 60V supply and a 10 ohm load peaks at about 90 watts in the transistor.

Right! there's an Al heat sink onto a big brass back panel, (with holes in the top) and inside a fan pushing air out the bottom.

With the 36V supply I think I got the whole thing up to 60-70C. (worst case)

With a 60 V supply that's going to be ~3-4 times! that's looking bad. Maybe there will be burn marks where the brass back plate attaches to the wooden box. :^) It could be something else,

George H.

I can't now, looks to be a 641A wakefield HS

onto a brass plate, ~0.10" thick, 12" x 8"

The fan mostly cools stuff inside and I don't think, does much to the back panel and heat sink.

GH

You (or actually, your customer) are likely frying the transistor.

That heat skin is a toy. An infinite sheet of 1/8" thick aluminum will be about 2 K/W, so a thinner non-infinite sheet of brass will be worse, wild guess 5-ish. 90 watts * 5 K/W = 450C above ambient.

Yeah, all the extra voltage of the new PS is across my poor transistor. I'll fry some tomorrow. I'll stick a thermal couple on the top of the to-3, post numbers. If things didn't keep changing....

George H.

How the hell does a coil UNPLUGGED hurt anything else ?

No, the EMF is generated across the unplugged part, which appears across the source plus the arc just described. In fact, the source thinks there is less voltage across the load, because the impedance is rising.

Could be any number of things, vulnerable design (1-2 decades ago, who knows right?..), ESD, mains surge, fat fingers shorting out transistors, ...??

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website:

What is the Duty Cycle of the coil? If momentary then the transistor should be just fine with a wee bit of thought - 2N6059s were used for Gottlieb pinball Pop Bumper driver boards for over a decade and they lasted until something else failed and took the transistor along with it. These were triggering coils of perhaps 5 ohms across 24VDC to ground. The pulse would perhaps be 500ms. The coils always had back EMF diodes installed.

These pop bumper driver boards had NO heat sink installed. Very short duty cycle even during play...

I expect that much of the risk to the transistor could be removed by simply adding a diode across the Emitter and Collector of the transistor to protect it against momentary back EMF. If the coil was unplugged while powered then contact noise would be enough to send a few hundred volts of back EMF through the transistor and as a 2N6059 is only rated around 100PIV. There is a back diode in the 2N6059, but putting an external 1N4004 (or higher PIV) would probably protect it even better. Another diode across the coil would be worth looking into as well. Can't have too many when dealing with back-EMF!

John :-#)#

You're saying it like it's a bad thing :)

NT

Put your hand on one & try it. :) You will regret it.

The coils are run CW. I was thinking of a diode across the output... (but if the problem is over heating a diode doesn't help. :^)

George H.

Hi all, Welp there is more than just thermal stress that is a problem. I put a thermal couple on the transistor case. V_supply I_coil V_coil Temp case

I then turned up the voltage ~52 V and it broke.

Here's the schematic

Well duh! The poor LT1013 can only take a 44 V supply. But I powered down, replaced the transistor and it works again. So it's more complicated than just the opamp breaking. The opamp got flaky and that broke the transistor?

Anyway there's not much to do but put out a warning to users not to exceed

George H.

I don't want to be too presumptive here but it seems like that thing would not be all that efficient. Unless you are playing some cool tricks with tha t coil I see too much across it. Like ten ohms and a 330uF ? I think that's alot. I can only assume you had reasons for ding it.

d not be all that efficient. Unless you are playing some cool tricks with t hat coil I see too much across it. Like ten ohms and a 330uF ? I think that 's alot. I can only assume you had reasons for ding it.

Oh that's the 'Zobel' network, makes the inductor look like a resistor to the drive circuit. We can estimate the inductance from that, (equal time constants) RC=L/R; L = C*R^2 (.33mF*100 ohm^2) = 33 mH Someone asked about the inductance up stream... it's probably a bit more than that.

The RHS turns an led on when the supply falls out of compliance. (measures c-e voltage.)

George H.