[GreenKeys] constant current

[gil smith]

Hey John, good to hear from you:

At 07:02 PM 1/23/2007, WB6BLV wrote:
>As far as I know, no bipolar transistor is either full on or full off.  My
>experience has been that there is some forward resistance wihen the device
>is 'on' as well as some leakage when the device is 'off'.

When a transistor is saturated (fully on), as in the classic use of the 
MJE340 (which I use in my circuits), the voltage across the transistor 
(c-to-e) is on the order of about 0.3V.  So the power dissipated is 0.3 x 
0.06 = 0.018 W, which is nuthin'.  However, the loop resistor does need 
heatsinking -- with a loop supply of 135V, you need a drop resistor of 135 
/ 0.06 = 2250 ohms -- it will get hot and needs to be a big-frickin' power 
resistor that can handle 135 * 0.06 = 8.1 watts!

In this application, the transistor needs no heatsinking, although the 
voltage spike at turn-off (v=Ldi/dt due to inductance of the selector 
coils) can reach past the max voltage limit (collector-to-emitter) and 
damage the device.  This is why snubbers and things are added.  I tried RC 
snubbing, but the waveform distortion reduced range.  So I just use a diode 
clamp to the HV supply in my stuff (you can't do this on a keyer that does 
not have a built-in supply).

In a constant-current loop design, the transistor will be dissipating this 
8.1 watts, so it need excellent heatsinking.  I would use a different 
transistor in a TO-3 power package, rather that the little tab-heatsink 
design of the MJE340.

gil




Vaux Electronics, Inc.
480-354-5556
(fax: 480-354-5558)
www.vauxelectronics.com