Suggestions for high current CCS?

[dsavitsk]

I am in need of a constant current source that can put out about 250 to 300mA and that can operate while dropping minimum volts across it -- on the order of 4 to 5V or less would be great. I am currently just using a LM317 with a RESISTOR which works, but I am sure something else will work better.

I'll use these in several places, but in particular I need to supply a 5V 250mA TUBE filament and I have 9.5VDC to work with. I'll also need it for a shunt regulator that feeds a circuit that needs 15V at ~150mA from a ~20V supply.

Any ideas?

[Salas]

Shunt.

[Salas]

Isource. The R5 and DC meter are to simulate the heater. You connect the heater instead of R5.

[Mr Evil]

Depletion mode MOSFET. They are very simple to use - just short the gate to the source, like a JFET current source, optionally with a resistor to set the current. You will probably need to cascode it to keep the power dissipation down unless you can get a big enough one.

[dsavitsk]

Originally posted by Salas
Shunt.

Thanks. R1 sets the current?

Besides the filament one, I'll use the others in an update to this project: Differential TDA1543 based NOS DAC

Originally posted by Mr Evil
Depletion mode MOSFET.

I've used the 10m45's and the dn2540's, but I was looking for something a little better. Can you really get that much current from a cascode of these? I had assumed not, but maybe I'll try it.

[Salas]

Originally posted by dsavitsk
Thanks. R1 sets the current?

You are welcome. Yes, I=Roughly0.66/R1. Dissipation of CCS Mosfet = about Vin-Vout * I CCS. Dissipation of shunt Mosfet = (I CCS - Iload)*Vout. Arrange your sinking.

[Mr Evil]

Originally posted by dsavitsk
...I've used the 10m45's and the dn2540's, but I was looking for something a little better. Can you really get that much current from a cascode of these? I had assumed not, but maybe I'll try it.

You can get ones that will give you plenty of current at low Vds, suitable for using a pair in cascode (IXYS do some), but the trouble is they don't have enough power handling. I don't know of any larger ones, but there may be some out there. You don't need to use one for the cascode though, you could use an ordinary low-threshold enhancement MOSFET with a zener or diode string to provide a bias voltage for it.

This would end up more expensive than Salas's solution, but possibly with better performance.

[woody]

I thaught the Lm317t was fine for a current source? Why would
you need better for a filiment supply? But here is an idea use the
Lm317t as a ccs but use a R C network between the ccs and the
tube something like .2ohm and 100uf and .1uf film bypass.

[AndrewT]

have you read Walt Jung's views/results/measurements of CCS performance.

I'll bet most of your questions and many others are answered in there.