ThermalTrak transistor application.

[Jeb-D.]

Well, the samples arrived yesterday. There is defiantly a voltage drop difference between the diode and the BE junction of the transistor per given current. I love how the datasheet really explains nothing about the relationship between the diode and BE junction of the transistor.

Steady state measurements (I do not have to tools to do test for temperature tracking). It's looking like if I use degeneration resistor values that allow for a 1V drop, it will give 3% imbalance in current (which is 1.8mA in this particular circuit). Degeneration resistors that allow for a 2V drop give a 1.5% imbalance (which is .9mA in this particular circuit).

I'd consider the above usable for this app, but wonder if a standard 2 device mirror would be much more accurate. It would for sure, if both devices were the same temp, but since there will be thermal resistance between the two. I am just wondering if the temperature mis-match in a 2 device configuration would cause a similar current imbalance to the thermaltrak configuration.

Perhaps shrinking the bias resistor and going with some rather large degeneration resistors (that drop like 1/2 the bias voltage) would provide good match. Though in this case, voltage drop across the silicone wouldn't matter much at all. Resistor tolerance would be the primary cause of imbalance. A 1n4007 paired with some random power NPN could probably do the job accurately.

[tomchr]

How about a transistor array like these? 160 V, 200 mA, 300 mW, matched within 2 %.

Tom

[revintage]

I can see any problems using this solution. Add 15ohm under the diode and a 20ohm pot under the emitter to trim fo equal currents through both tubes. This only needs to be done once and for all. As I showed earlier in our SEPTOR it works like a charm IRL. Used two BD139s in that one.

[Jeb-D.]

I can see any problems using this solution. Add 15ohm under the diode and a 20ohm pot under the emitter to trim fo equal currents through both tubes. This only needs to be done once and for all. As I showed earlier in our SEPTOR it works like a charm IRL.

I had thought about that. But, was thinking. You set the currents equal by using unequal resistances. Now what happens if say, the operating temperature changes (or ambient), which causes both BE junction V drops to change by and equal amount. It will cause an un-equal current change between the two. Though I'm sure in most practical cases, the imbalance caused would be very small.

[revintage]

Do not worry Jeb! Tracking will stay within 1mA from 25 to 90 degrees C when resistors are 15 to 12ohms. A realistic temp diff would be 30 degrees at most.

Even if you use a mainstoroid for OPT you will have an open window of at least 5mA.

Even if the Thermaltrack solution looks nice, it seems like a waste of money as two dirtcheap BD139s at the same heatsink probably will do the job better.

[Wavebourn]

Originally posted by tomchr
How about a transistor array like these? 160 V, 200 mA, 300 mW, matched within 2 %.

Tom

It looks interesting Tom,
http://www.datasheetcatalog.org/data...es/ds30436.pdf

but I bought already few hundreds of 2SC1815 from Toshiba.

[Jeb-D.]

Even if the Thermaltrack solution looks nice, it seems like a waste of money as two dirtcheap BD139s at the same heatsink probably will do the job better.

That's pretty much what the testing I did yesterday indicated. The voltage drop difference between the diode and B-E of the BJT was larger than the difference between the B-E's of different transistors.

It is close enough to work decent, but the diode and BJT were obviously optimized for different quiescent currents. This kind of throws out any advantages the Thermaltrak may have had for this application.

Unfortunately, the Thermaltrak idea is one that doesn't seem worth pursuing at this point. If the diode were identical to the B-E portion of the BJT, it would be. But, being that they aren't, I'm probably going to use a pair of cheap NPN's as you suggest.

[Sheldon]

With apologies to JB for adulterating his work: http://tubecad.com/2009/04/blog0163.htm

Version for direct coupling?