Hi, guys!
I wanted to build an electronic load; the constant current feature will be controlled with an analog circuit, conceptually similar to this one (with different parts) and the feature set will be extended by an ESP32.
EDIT: WITH A DIFFERENT MOSFET THO!
IRFP2907!!
HOWEVER!
It's a school project for my uni and they will actually be using it for testing batteries or something.
If some guy comes and accidentally attaches a car battery to the poor thing, the mosfet is gonna get cooked!
For the mosfet to be safe, it should stay within the region described in the datasheet here:
But... how do I make sure it does that? I cannot just limit the current, because that would mean it wouldn't be able to sink the current even with very low voltage drop, where it is just fine. I can't just limit the voltage for similar reasons.
Handling it in software seems... dangerous... you shouldn't handle critical things like this in software, right? But handling it in hardware seems super complicated! I mean, I'd need to calculate the dissipated power in hardware and then convert it to voltage and feed that voltage to a comparator and... that includes a log/antilog multiplier and that just adds giant complexity to the whole thing!!
I don't think some sort of a ptc or a bimetal or something is fast enough, right? What do I do? Pls help.
thegoodhen:
If some guy comes and accidentally attaches a car battery to the poor thing, the mosfet is gonna get cooked!
Why is the MOSFET gonna get cooked ?
If you are building a constant-current sink, it will just sink that current. The diagram shows a 0.2Ohm load resistor compared to 0.3V so it should sink 1.5A - it doesn't matter if the 1.5A is from a few AA batteries or a huge 24V truck battery.
For the mosfet to be safe, it should stay within the region described in the datasheet snip
Your circuit is limiting the current (not sure what your upper limit will be tho. 20Amps? )
Is the upper voltage limit much of a problem ? could it be connected to more than, say, 24V ?
If you turn up the current, there will be a point where it gets too hot - so you could add a simple temperature sensor (thermistor and comparator) to shut it down.
If you are building a constant-current sink, it will just sink that current. The diagram shows a 0.2Ohm load resistor compared to 0.3V so it should sink 1.5A - it doesn't matter if the 1.5A is from a few AA batteries or a huge 24V truck battery.
Your circuit is limiting the current (not sure what your upper limit will be tho. 20Amps? )
Is the upper voltage limit much of a problem ? could it be connected to more than, say, 24V ?
If you turn up the current, there will be a point where it gets too hot - so you could add a simple temperature sensor (thermistor and comparator) to shut it down.
Yours,
TonyWilk
Damn, I don't know what I was thinking. It seems super obvious! Thank you. Basically, if I stick to under 20V and then limit the current to 20A, I should be OK at all times with my mosfet, right? (That is, until it starts overheating...).
thegoodhen:
Basically, if I stick to under 20V and then limit the current to 20A, I should be OK at all times with my mosfet, right? (That is, until it starts overheating...).
Wishful thinking.
Look at the thermal data of a TO-220 package.
The datasheet of that mosfet estimates 0.95C/watt, junction to heatsink.
With 20volt/5Amp, the chip inside will be 95C hotter than the heatsink if you have used the right thermal grease.
20volt/20Amp (400watt) needs four or five mosfets in parallel, and a massive fan- or water-cooled heatsink.
Leo..
thegoodhen:
[...]if I stick to under 20V and then limit the current to 20A, I should be OK at all times with my mosfet, right?
Uhmmmm, 20A at 20V is a whopping 400W!!! Even with a huge heatsink that will be to much for a common transistor.
You might add a temperature sensor but if someone uses a high voltage the transistor may release it's magic smoke long before the heatsink heats up.
I would add a (simple) shut down above a certain voltage. What that voltage is depends on the max power you can handle with the heatsink you have and the current you want to be able to pass. Let's say ou can handle 100W (which is already pretty large!) and you want 20A that leaves 100W / 20A = 5V.
Easy way to turn off the source is to pull the + input of the opamp to 0V. But if we do that now and the pot is all the way up we have a short circuit. Add a 1k between the pot and the + input. Now you can use a jellybean NPN to pull the line to GND. Drive that (via a resistor) from another opamp that has it's - input connected to 5V and it's + input to the line from the DUT. Aka, that opamp will turn on (thus shutting the system off) when the + input (from the DUT) goes over the - input aka 5V.
You can even make it smarter by using the voltage from the pot (aka, the current setting) instead of the fixed 5V. But that makes it more complex.
And I would add a indication to show when turned off.
Wawa:
Wishful thinking.
Look at the thermal data of a TO-220 package.
The datasheet of that mosfet estimates 0.95C/watt, junction to heatsink.
With 20volt/5Amp, the chip inside will be 95C hotter than the heatsink if you have used the right thermal grease.
20volt/20Amp (400watt) needs four or five mosfets in parallel, and a massive fan- or water-cooled heatsink.
Leo..
Sorry, it turns out I haven't indicated anywhere which mosfet I wanted to use. That schematic I found was just... for illustrative purposes, I didn't realise it had a specific mosfet there... I want to use this one:
The datasheet has those curves; if I am not misreading it, it says 20V @ 20A is the absolute maximum it can take continuously, right?
My heatsink is about 20x20x10 cm alluminium thingy that weights about 2kg.
My original plan was to take 2 of those in parallel (with ballast resistors to balance them) and hope that that would withstand 400W. Is that thinking correct?
If you know the thermal characteristics of your heat sink, you can calculate how hot the spot the transistor is attached to it will be when it's dissipating 200W.
In the data sheet are the part's thermal characteristics, which tell you how much hotter the internal parts of the transistor become for it to conduct out 200W. This shows me numbers of 0.32 °C/W (maximum) for junction to case, and 0.24 °C/W (typical) for case to heat sink, provided excellent contact between the two. So that'd be 112°C for 200W.
I see a maximum junction temperature of 175°C, so your heat sink must be able to dissipate 200W while not getting hotter than 63°C at the contact point (and at that temperature there is little to no margin left).
Another thing that I noticed is the 2-4V VG(TH). That's probably too high for your 5V application - look for the logic level versions of these MOSFETs (the IRL types, instead of the IRF types).
wvmarle:
Another thing that I noticed is the 2-4V VG(TH). That's probably too high for your 5V application - look for the logic level versions of these MOSFETs (the IRL types, instead of the IRF types).
Also note: In that circuit the gate voltage is on top of the sense resistor voltage
e.g. with 0.2 Ohm, 10Amps= 2V, to supply 4V VGS the opamp output has to be 6V
Good one.
At the 20A OP wants to be able to measure that will be 4V. There are logic level MOSFETs that have a VGS(TH) of about 1V, but I have yet to see one that's in TO220 or similar package that allows for high heat dissipation. I only have found SOT-23 packages so far. So you're going to look at MOSFETs that have VGS(TH) of about 2V.
That should work for a 10A load, but not for a 20A load. That OpAmp should be receiving a 9-12V supply, so it can output accordingly.
That also gotta be quite a resistor, as it has to be able to dissipate 80W.
I have designed an electronic load for up to 19V, 5A, and 21W. It has a BTS133 mosfet in TO-220 package. This mosfet is super robust, it has over voltage, over current, over power porotection as well as thermal and ESD protection.
Wawa:
Wishful thinking.
Look at the thermal data of a TO-220 package.
The datasheet of that mosfet estimates 0.95C/watt, junction to heatsink.
With 20volt/5Amp, the chip inside will be 95C hotter than the heatsink if you have used the right thermal grease.
20volt/20Amp (400watt) needs four or five mosfets in parallel, and a massive fan- or water-cooled heatsink.
Leo..
Its a very poor datasheet. The abs max power quoted is not a continuous rating for a start, and the
SOA graph is just plain bogus given the package limit for the current is 90 amps at DC.
I hate it when the marketing department start putting their lies into an engineering document like this,
470W for a TO247 package is just laughable. 50W is more like it.
