AVR Freaks

Light bulb?

Tom perhaps inadvertently raises an important question. How "hard" is this limit? And how precise does this limit have to be?

For example, suppose that it starts limiting at 25mA but the limit increases to 30mA at maximum voltage.

Jim

Here are some details of a diode current limiting bridge that Jim alluded to :

https://www.radiomuseum.org/forum/ballast_tubes_solid_state_replacements.html?language_id=5

Item 4 ins interesting to eliminate the effect of diode drops.

Note, you can limit current using a plain old N-chan powerfet...for a given Vgs, that sets a current limit between drain & source (you can use a 200k pot & 9v battery to create a very long-lasting Vgs source).  However, this can be drifty/sensitive over a long time /temperature, but quick & easy to build. Two of them in opposite-series, of course current limits AC (you could even have different limits in each direction). The internal moseft parasitic diode from S to D could be paralled with a schottky for lower drop, or a more linear response when not limiting.

Finally, from the Godfather of analog, here is a Bob Pease solution...note figure 4 pins abc get pasted into fig 3 (cut out fig 3 abc circuit).

http://www.electronicdesign.com/analog/whats-all-current-limiter-stuff-anyhow  

avrcandies reference is exactly what I was thinking of. While 25mA is not much for most transistors, the dissipation of 2W+ at maximum transformer output voltage can be a killer. The problem is that almost anything you do will exhibit the same dissipation characteristic. So, using a FET or some diode, or something else, won't "save" you. On the other hand, 2W+ is not THAT bad and use of appropriately voltage rated power transistors combined with a modest heat sink should be able to do it. You would need to be VERY careful about line voltages and isolation between the power transistor and the heat sink.

Jim

A variation on this old gem might do

This arrangement might be better suited to low supply voltage (like <15V), since many fet gates are rated for Vgs 20v max.  With small/no load (no active current limiting), the gate is pulled up to Vdd...a Vgs issue if Vdd is 200V.  Of course a zener clamp on the gate might alleviate that  issue. 

I take it then a good DUT presents a constant load and a bad DUT presents a variable load?  Can you simply cut the power when the abnormal situation is detected?  Or does the test require the power remain applied?

I'd be inclined to look at a high-speed circuit breaker. Not of the electromechanical type but something solid state. I have a design in production which uses such a circuit and it detects an overload in around 100us with instant off. Getting it to trip at 25mA would be trivial.

Vdd is 200V

Somehow I missed that! 

Does this help: http://www.powersolve.co.uk/catalog/inrush-current-limiters ?

I looked at their datasheets. They are all based on a very simple Resistor, Relay & timer setup.

It can not be very difficult to design an opamp with a discrete amplifier stage just as in an audio amplifier.

Your voltage is a bit high, but if you only need 25mA dissipation is still quite low even at those voltages.

Have you checked the voltage rating of the components you've chosen?

Ah, the KREMS de la KREMS.

250V drive into 35k, intersects 5mA at 75V load.

250V drive into  12.5k, intersects 10mA at 125V

250V drive into  3.3K, intersects 15mA at 200V 

Not 100% sure I'd go that route...that would feed the load with a high impedance source, which it might not like, or cause it to operate differently than when provided a normal voltage source ("zero" impedance).     While the resistors do set an ultimate current limit, the supplied voltage will rapidly decline or oscillate as the load draws varying currents.   It could be "ok" depending on what exactly the load is.  A good current limiter looks like a zero impedance driver (a voltage source) until the limit is hit, then jumps into action.

but it's simple enough to try

Perhaps, but a bit dangerous.  If the load DUT was a transistor radio its current draw would go up & down with the sound....with a few K in series at the poer connection, the input voltage supply would be all over the map.  That was the point I was considering.

If the load is constant, of course, less of an issue.