V. linear drivers and fast - 200v Vceo, 120mA

[Kevinbd]

Hello,

Have been trying to source some complimentary driver transistors much like video transistors.

Looked at 2SC4474 video transistor for example. Looks very good (linear) upto 70mA Ic but was looking for 120mA.

Power T-03P types like 2SC5200 look bad at these currents.

Rails are +/- 55v.

Any Ideas ?

[EWorkshop1708]

MJE15034/35 On-Semi transistors are 350V, and are more linear around 100mA or so.

[wrenchone]

You might want to look at the 2SA1535/A and 2SC3944/A from Panasonic. They are fast and have very linear HFE vs current in the range you mention. The A version of the device are rated to 180V. They are in isolated TO-220 packages, so heat sinking should be a snap. The only down side is that I cannot find junction to case thermal resistance for the devices, so sizing a heat sink may be problematic. This could be an issue, as you are talking about dissipating almost 6W per device.

[AndrewT]

Hi,
re 2SA1535/3944
calc Rjc from Tj-Ta = 150- 25 = 125
Rjc = 125 / 15W = 10 Cdeg /W
At 55v these can take about 12w.
You will be running a high Tj putting 6w through them. This will apply to most fast drivers since the max power will be about 20w to 25w. Slower drivers can go to about 40w to 60w but either of these driver solutions will require large heatsinks (almost output stage type sinks).
If you run hi Tj then you have little spare capacity for low frequency current demands particularly if the peak current repeats over a short period.
Two solutions come to mind
1. reduce Iq to reduce Tj and increase peak current capacity.
2. use separate driver for each output transistor thus reducing Iq demand and total current demand. (assuming a push pull output stage)
regards Andrew T.

[PH104]

How about MOSFETs? I'm looking at a very similar problem right now. I tried a pair of IRF 610 and 9610s this morning, +/- 60VDC rails, about 100 mA current. They look like they will work fine.

Phil

[wrenchone]

I agree with Andrew T's method of calculating junction to case thermal resistance, though the actual answer is 8.3C/W. This is still pretty poor, and is likely the result of the fully encapsulated tab and a relatively small die. To be conservative and limit failure due to thermal cycling, you would want to limit the max junction temperature at worst case ambient to around 100C. The 6W of dissipation gives you a 50C junction to case temperature rise. If you assume a 40C worst case local ambient temperature, this means that the heat sink must have a thermal resistance of (100-50-40)/6, or 1.7 degrees/watt. This is a big, big heat sink. As was said, either you need some forced air, other drivers, or lower quiescent current. You can cheat a little bit and allow a higher temperature rise or a lower max ambient (though a 15C rise inside the box is being optimistic), but the numbers still aren't too nice.