IXTN200N10L2 Linear MOSFET: Achieving Stable Current Control for Programmable Load

Hi everyone,

I did some tests on this linear MOSFET (IXTN200N10L2)

I'm trying to use it as a simple load to replace the resistor bank that we're using to test some battery cells.
So in short, we're trying to regulate the current that the MOSFET will dissipate.

I plainly connected the gate to a variable PSU and put another 20V/80A PSU between drain & source.
When I make the gate voltage vary, it does indeed control the current that goes through.


Now, first problem is that at Vds>9V (much more linear around 4V for example), when the gate voltage gets around 6V, the beahviour is quite exponential ==> with tens of mV difference on the gate voltage, the current goes from 10 to short-circuit very quickly.
 
Also second problem is that when I lower the voltage, the current doesn't come back to 0 and neither does it lower quickly (unless I short-circuit the gate pin to GND), I assume it's because the PSU used is not regenerative... Would a simple pull-down resistor fix that?
 
Bonus question : I was thinking of using a very simple µC (teensy?) with an analog output (+ OP-amp for the appropriate power), a hall sensor and a simple PID to regulate the current wanted. Do you guys see any mistake in that?


 

Thanks for your answers!

 

Marc

 

There is a lot of info on the web re building these things, some with practical circuits. Essentially you need a constant current sink, you put your fet inside the feedback loop of an op amp, with a small current sense resistor, which linearizes the fet with respect to drive voltage. There is good theory on the web re current sinks using op amps with external drivers, either fets or transistors(as opposed to current sources, there is often a lot more around about them)
Ignoring all of the additional circuitry, Linear Techs article covers the essentials at the start of the piece in this link, see figure 3. If you are doing static tests, then the pulse input is irrelevant, you adjust the static current using the pot in the circuit.
http://cds.linear.com/docs/en/article/WMSstory_From%20EDN%2009-22%20NoAds.pdf
You give no indication of the voltages and currents you have to deal with, so hard to give any further advice. Needless to say, you have to pay close attention to power dissipation in the fet and sense resistor. To keep the voltage drop across the sense resistor small, it will have a low value, and could be dissipating quit a bit of power. If you want to measure things accurately, it needs to have a low temperature coefficient(self heating could change its value dramatically), you have to do the maths to establish what it needs to be, and so on,
cheers,
Richard

Hi Richard,

Didn't have the time to pursue this project lately but I'm on it again.
I had found the excellent EDN article about the 100A load but didn't go into the details.

So I did delve a bit deeper but my electronic understanding is limited, sorry!

I did the two following schematics and hope that you guys can help me.

The first schematic (Schematic A) is the one from the article you linked but with the right polarity. Or at least, what I understand of it.
Problem is I don't really understand how the shunt voltage can regulate the gate voltage with it... Also no idea about the resistor values I should put since I don't know what rules to use.

The second one (Schematic B, you guessed it) is my understanding of a feedback loop on the shunt voltage.

The current setpoint would be set by the POT from 0-1V and the shunt gives a voltage (after amplification) of 0 to 1V for 0 to 200A. (I know I'd rather the largest voltage range possible but it was easier to calculate at first)
So if I set 100A as the current set point (0,5V), the current is 0 at first and the op-amp goes to 60* (0,5-0) = 30V and saturates at 15V, giving full gate voltage.

Then current rises and goes beyond 100A. Say 150A and gives 0,75V as shunt voltage.
The op-amp now gives 60*(0,5-0,75) = 60*-0,25=-15V.
But if I only supply 0/+15V to the op-amp, that'd be 0V, right? But can the op-amp output sink the gate current at all? But also, is it even needed to make the MOSFET current lower?

And so on and so forth, regulating towards the current setpoint wanted.


Then again, not sure it works at all and even less regarding the resistor values etc...

Thank you all for your help.

Have a look at this article,
http://www.allaboutcircuits.com/technical-art...r-an-op-amp-output-for-higher-current-part-1/
You can substitute a fet for the transistor, it works exactly the same way.
The Linear tech article end result is extremely complex, since the current is very high, and the test dynamic.
You need to be able to use ohms law, and have a bit of op amp theory to design the circuit, the first circuit you have has positive feedback around the loop, the sense voltage across the sense resistor needs to go to the inverting input.
If you can get that circuit working in spice, then you have the fundamentals needed to design it for 20V at 100A.
It is not trivial, you need kelvin sensing on R4 for example, and temperature coefficients have to be low for the sense resistor, or the current could potentially be wildly out,
cheers,
Richard

Hi everyone,

So I managed to have a half-baked circuit working.

It goes like this :

Hall sensor current (2,5V midpoint +/- 0,625V) goes to µC ADC (Teensy) which translates it into 0-3,3V analog output (DAC)

This info then goes to an OpAmp mounted as a current buffer with 0-3,3V coming from a pot (current setpoint) on one input and 0-3,3V coming from the µC (current info) on the other.

It does work in a way but since the code also has to display some things over I2C, it probably runs pretty slow and the whole thing oscillates between full current and 0 if I'm anywhere under full blast (POT turned all the way).

I'm guessing this is due to the processing time of the current info... maybe also the "knee" (is that the word?) of the opamp is too sharp... If I had something softer I guess it'd go better.

Well, I guess I must go back to analog but how can I remove the 2,5V midpoint before amplifying the 0-0,625V to 0-3,3V?


Thanks for the help,

Marc