Mosfet Ohmic Region Control

[Johnex]

Hi!
My first topic

So i am trying to design a battery discharger, very high amps. Checking online i saw a guy used a mosfet, water cooled, and with the right bank of mosfets he was able to dischange 100A (much more too im sure since with one mosfet he could do 50A no problems) at around 20V into the water, basically making a glorified water boiler.

For that specific mosfet the ohmic region was around 0.2V to 0.7V'ish.

So the question is, what would be the best way to get an arduino to generate between 0V and 1V with something like 12bit accuracy?
A MCP4725 12bit DAC runs on 3.3V even 2.8V down, but that means i still loose a third of the potential accuracy.
Is there any DAC that can easily be for example supplied 1V and also drive it with say 5V, so that i can get the full 12bit between 0V and 1V?

Maybe this is not the right way to generate those voltages with such high accuracy?
From what i understand the DAC would be the best, since an LC filter would have some ripple and so on...

What are your thoughts?

 

[ahbushnell]

So i am trying to design a battery discharger, very high amps. Checking online i saw a guy used a mosfet, water cooled, and with the right bank of mosfets he was able to dischange 100A (much more too im sure since with one mosfet he could do 50A no problems) at around 20V into the water, basically making a glorified water boiler.

What do you want to regulate?  The current to 12 bits?  What is your maximum current?  Why dump the power into the MOSFET why not use a resistor with the MOSFET?  That would be with a fixed current. 

If you want to change the current you could use a switching power supply.  That would require and inductor and a resistor.  You would need a pulse generator and gate drive for the MOSFET. 

Also the current the MOSFET can handle is when it is turned on all the way (ohmic).  So the current that is flowing is dependent on the Rd of the MOSFET and the resistance in the battery and the wires.  If you try to regulate the current by controlling the gate voltage you will greatly increase the losses and maybe destroy the MOSFET.

[capt bullshot]

Controlling the MOSFETs drain current by the gate voltage isn't practical. You'd insert a small sense resistor into the source and build a simple control loop to regulate the drain current. The setpoint of the control loop could come from any suitable source, like a pot, D/A-converter, whatever.
Look for "constant current" circuits, that's what is commonly used here.

[Johnex]

Yeah i want to be able to control the current via programming, so i can't have a fixed resistor bank or something like that.

For that i would just use some wire and make my own high current water boiler and just clamp it at different lengths to get the desired current, but its inaccurate and not repeatable so i need it programmable.

I also what to make it as simple as possible with basically no extra components other than a mosfet an IC and the arduino. It's ok if the mosfets don't last their rated MTBF.
I found a mosfet that can dissipate around 2000W, running that at a much lower current than max with the ohmic way would be ok no?

The maximum current could vary, lets say 200A just to have it set and done
This is why i need very high accuracy, so i can have even steps of like 50mA from the 0 to 200A.

[Mechatrommer]

meh, use power bjt for this sort of thing. i dont get why using mosfet as constant current is so famous around here. when i suggested a jellybean mosfet in another thread, people started talking about hotspot, balancing current imbalance and all sort of weird stuffs, and then they suggested using special (read expensive) linear region mosfet. in the end i used jellybean power darlington bjt... ymmv..

[Johnex]

If you can find me a BJT that can dissipate around 2000W i can use that How would i control the current to it? I have no idea 

This is the mosfet i planned to use:
http://www.mouser.com/ds/2/149/FDL100N50F-1007460.pdf

[alm]

Good luck keeping that case temperature at 25°C (or 50°C) while dissipating 2 kW.

[max_torque]

er, ok i'm lost!

What's the difference between a Mosfet dissipating 2kW and a BJT dissipating 2kW??  (other than both being highly unlikely without some sort of cryogenic forced cooling (ie liquid nitrogen etc).


Either device has to be operated in it's linear region in order to actually dissipate any power (ie, have a voltage drop) the mosfet is going to be easier to drive (being a voltage controlled device) but comes with significant disadvantages (such as a smaller SOA for typical devices), whereas a large darlington BJT will take a bit more driving (and you need to make sure the gate current doesn't interfere with your EUT discharge current measurement) but is likely to be significantly more robust (although require some small balance resistors when used with parallel devices, although you probably ought to put those in place with any parallelled mosfets as well.

[max_torque]

If you can find me a BJT that can dissipate around 2000W i can use that How would i control the current to it? I have no idea 

This is the mosfet i planned to use:
http://www.mouser.com/ds/2/149/FDL100N50F-1007460.pdf

I take it you've looked at Figure 9,10 and 11 on page 5 of that datasheet, right???     

[Mechatrommer]

If you can find me a BJT that can dissipate around 2000W i can use that How would i control the current to it? I have no idea 
This is the mosfet i planned to use:
http://www.mouser.com/ds/2/149/FDL100N50F-1007460.pdf

if you follow several constant current discussion. 2000W on a tiny single chip is not a wise thing to do, you'll introduce a whole lot of other issues, one of it is super special cooling system close to liquid nitrogen system. how about spreading the heat to say 20x TIP142, even those are on the verge of their power limit, you screw their SOA and derating curve they all burn, more so to the $16 FDL100N50F...

I take it you've looked at Figure 9,10 and 11 on page 5 of that datasheet, right???     

page 4, excluding the main cover

[Johnex]

If you can find me a BJT that can dissipate around 2000W i can use that How would i control the current to it? I have no idea 

This is the mosfet i planned to use:
http://www.mouser.com/ds/2/149/FDL100N50F-1007460.pdf

I take it you've looked at Figure 9,10 and 11 on page 5 of that datasheet, right???     

Yeah i saw. Each mosfet would not do more than 10A water cooled, so i don't see a problem? First page also says it can do 100A continous, but not using the ohmic region for sure.
I'm not going to use just ONE ofc, but i do want a mosfet that can be rated for such high if properly cooled, increasing my chances of this hack, cos its a hack for sure.
I'm trying to do like this guy:

[Ian.M]

Probably the best option is to switch in paralleled binary weighted resistors for the bulk of the current, and fine trim it with a voltage controlled current sink that only has to handle a couple of Amps.    The voltage drop across the MOSFETs switching in the resistors can be very small compared to the drop across the resistors so they can be passively cooled with a simple heatsink.   The resistors will require active cooling, as will the the current sink.   You'll also need a 200A current sensor to close the loop.

[Cliff Matthews]

Maybe it's my speakers, but the "boiling sound" seemed suspect since it was roughly the same on both videos. On the first it started long before the bubbles appeared (I thought someone was exiting the lou..). Ian.M's idea makes more sense than robbing the bank.

[Johnex]

Thanks for the ideas guys.
Ian, do you have any examples i can take a look at and scale up?

[Cliff Matthews]

That water based cooling reminded me of something King of Random posted a few years ago: The Scariac 

[Kleinstein]

If the battery voltage is approximately known, there is the possibility to use a series resistors to dissipate some of the power. Even without switching different resistors this could about cut the maximum power for the MOSFETs to something like a third. However it still needs high power resistors - though this does not have to be a stable one.

The circuit looked for is a kind of electronic load. It takes a little more than just the Arduino, DAC and MOSFETs. Usually one needs an OP to control the current via feedback from a current shunt. So the MOSFET is not operated in the ohmic region, but more in the "constant current" range. The feedback is usually needed, as the current will change with temperature.
For higher power it is common to use smaller units in parallel.

For a MOSFET of BJT in a TO247 case, something like 100 W is about the practical limit. More is possible, but cooling gets increasingly difficult. The power ratings on modern MOSFETs is more like a theoretical number, not really relevant.

As a beginner, one should start small, so try it first with maybe a single MOSFET for maybe 40 W - so maybe 2 A at up to 20 V.

Even if only in 20 V range, higher power circuits are dangerous. Parts can explode due to to much power and DC arcs don't stop that easy. Overheating batteries are also a danger. So at the very least, don't forget the fuse(s).