Paralleling transistors...

[Tim Williams]

I'm working on an induction heater circuit, since my function generator died
I have to make it self resonant and since I don't feel like op-amps right
now that means self-excited class C BJT's, which I've done successfully
before on a smaller scale (namely, two 2SC3519's in parallel with 0.15 ohm
RE's[1]). Having already blown a few MOSFETs before the function gen. died,
I'm guessing I can't very well make those work in the same class just yet...

First of all:
http://webpages.charter.net/dawill/tmoranwms/Elec_IndHeat1.html
This was upgraded to a 200W (40V 5A as measured) model:
http://webpages.charter.net/dawill/Images/IndProgress2.jpg
(Heating a strip of transformer iron; you can see the heatsink and emitter
resistors below my hand.)
The power source is a rewound MOT. I rewound it again for 100V 10A output,
which gets +130VDC at the capacitors. I found six 2SC2625's laying around
so paralleled them all in a nice bank and tried the exact same circuit, but
alas, one blew (shorted, fun) after another. It was always the one on the
end, so I'm wondering if it had parasitic oscillations?
http://webpages.charter.net/dawill/Images/IndHeater.gif

Yes, long post, I'm being thurough... Comments?

Tim

--
"I've got more trophies than Wayne Gretsky and the Pope combined!"
- Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

 

[John Popelish]

Tim Williams wrote:
(snip)

The power source is a rewound MOT. I rewound it again for 100V 10A output,
which gets +130VDC at the capacitors. I found six 2SC2625's laying around
so paralleled them all in a nice bank and tried the exact same circuit, but
alas, one blew (shorted, fun) after another. It was always the one on the
end, so I'm wondering if it had parasitic oscillations?
http://webpages.charter.net/dawill/Images/IndHeater.gif

In order to get reasonable sharing and good behavior you need 3 heavy

busses for the transistors with the signals connected to the middle of
the busses. Better yet would be something like ring busses connected
to their signals with three parallel wires like spokes to the center.
The devices can be connected all around like flower petals. Keep in
mind that your induction coil sprays out a lot of stray field, and the
transistors should be several diameters away from the coil to reduce
stray coupling that varies device to device. Connect the coil to the
transistor and supply busses through a transmission line.


--
John Popelish

 

[Tim Williams]

"John Popelish" <jpopelish@rica.net> wrote in message
news:41AE46B2.6F90BB2D@rica.net...

In order to get reasonable sharing and good behavior you need...

Just emitter resistors, right? Aside from the wiring.
How about not blowing the MOSFETs?

3 heavy
busses for the transistors with the signals connected to the middle of
the busses.

I don't think "heavy buss" is quite applicable here, I intend on having up
to 200A reactive flowing, maybe more, but not necessarily that much in the
transistors. At least what I recieved today from Digi-Key, 10 x MJE18008,
rated for 16A peak (so 160A total) and 1000Vcbo. Meh. In the final version
I intend on using a strip of 1/4" tubing for it. This is only supposed to
be 1kW or so, nothing too big.

Better yet would be something like ring busses connected
to their signals with three parallel wires like spokes to the center.
The devices can be connected all around like flower petals.

Last setup I had was two devices per (small, only for intermittent testing)
heatsink, with the leads going out to buss wires (18ga.), so three units
strung together.

Keep in mind that your induction coil sprays out a lot of stray field

Heh... first model I made, I could actually feel the magnetic pull.

and the
transistors should be several diameters away from the coil to reduce
stray coupling that varies device to device.

Ok. Shouldn't be a problem...

Connect the coil to the
transistor and supply busses through a transmission line.

TL? We're at 50kHz here.

Tim

--
"I've got more trophies than Wayne Gretsky and the Pope combined!"
- Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

 

[John Popelish]

Tim Williams wrote:

"John Popelish" <jpopelish@rica.net> wrote in message
news:41AE46B2.6F90BB2D@rica.net...
In order to get reasonable sharing and good behavior you need...

Just emitter resistors, right? Aside from the wiring.
How about not blowing the MOSFETs?

I think wire wound source resistors would be a good idea, also. The
inductance helps as much as the resistance.

3 heavy
busses for the transistors with the signals connected to the middle of
the busses.

I don't think "heavy buss" is quite applicable here, I intend on having up
to 200A reactive flowing, maybe more, but not necessarily that much in the
transistors.

Not heavy (low resistance) to carry hugh current without overheating,
but heavy (low inductance) to keep differences between devices small.

At least what I recieved today from Digi-Key, 10 x MJE18008,
rated for 16A peak (so 160A total) and 1000Vcbo. Meh. In the final version
I intend on using a strip of 1/4" tubing for it. This is only supposed to
be 1kW or so, nothing too big.

Better yet would be something like ring busses connected
to their signals with three parallel wires like spokes to the center.
The devices can be connected all around like flower petals.

Last setup I had was two devices per (small, only for intermittent testing)
heatsink, with the leads going out to buss wires (18ga.), so three units
strung together.


Connect the coil to the
transistor and supply busses through a transmission line.

TL? We're at 50kHz here.

I am not concerned with impedance matching, just field containment. A
big wire inside a copper pipe would work. Two big wires twisted
would, also. This will make up part of the tank capacitance so teflon
insulated wire would keep the losses from melting the insulation.

--
John Popelish