Hi,

i plan to build arduino controlled current source using the attached schematic ,the question is ,can i use Lf 347N instead of lm324? the second question ,what if i replace the the mosfet
with IRFZ44N and use only 4 instead of 6 ,the maximum voltage to be used is 14V and MAX current is 40A with large heatsink.

 

Neither LF347 nor LM324 have a good offset voltage. LM324 output can get to zero which guarantees being able to turn the MOSFET off. LF347 can’t.
Whatever op-amp you use, don’t go putting a capacitive load on it unless you want it to go unstable, and reduce the gate resistors to 10Ω as you need to avoid the Delay in the feedback loop made by the gate resistor and the gate capacitance.

Look for an op-amp with low offset voltage and an output range that includes zero.

Maxiumum dissipation is 40A x 14V = 560W, that’s almost 100W per transistor. I go for more transistors rather than fewer.
I made something similar (for discharge testing 12V batteries) using six TO220 transistors on a CPU cooler, but it was only 20A. I was impressed how well the CPU cooler worked.

 

Most common FETs are designed for High-Frequency-Switching,
and may not survive heavy Currents in Linear-Mode.

You will need ~3-FETs that are designed specifically to operate in Linear-Mode.
They will have an "SOA" Curve, ( Safe-Operating-Area ), Graph
that looks something like this one ...........
.

.
These are rather expensive at ~$12.oo each, but 3 of them would comfortably
do the job You expect without being heavily stressed,
provided that You have an adequate Fan-Cooled-Heat-Sink.
DigiKey has over ~1000 of them in stock .............
https://www.digikey.com/en/products/detail/ixys/IXFH340N075T2/2354429

You then need to drive each of them with an Op-Amp that can supply
some reasonably heavy Current because of the ~20nf of Gate-Capacitance that each one has.
This one will probably work just fine with a ~4-Ohm Gate-Resistor ............
https://www.digikey.com/en/products...gZwKYA0IA9lANogAMIAugL73EBMFIAKgDICizAzFQgMgA
It's not in stock right now with DigiKey or Mouser,
but they are expected to be available in a couple of Months.
.
.
.

 

If you buy the Exicon lateral MOSFETs (which are the original Hitachi 2SJ50/2SK135 types designed for audio) they have no secondary breakdown at all (the sloping is the same slope all the way across the SOA graph). The Rds(on) is very poor, but that is of no consequence in this design. The input capacitances are also lower.
Also, after being a premium price over other MOSFETs the price of other MOSFETs has caught up, and they are no longer overpriced.
[Edit] Note that the pinout is different from a normal MOSFET

 

As @LowQCab says, you need something like IXYS' Linear2 MOSFETs designed for this purpose, though he has linked to a different Trench-Gate device from the same stable which is primarily a switching device rather than a linear one, and they are cheaper but more prone to hot-spot failure than the Linear2.

You have to be careful of the SOA diagrams - they are usually the theoretical best case with a maximum junction temperature and a 25degC case temperature. Sadly the laws of physics demand an infinitely large heatsink to get the case to 25degC at any reasonable power output and running at max junction temperature is not a low-stress situation, you need to derate by at least 15degC, say 160 for a 175degC device or 135 for a 150degC one. The IXYS Linear2 datasheets show the SOA for a 75degC case temperature which is more realistic. I use 8 IXTX200N10L2 or IXTX110N20L2 devices at 200W each on two 500mm x 300mm x 25mm water-cooled plates in my 1500W battery tester (30v/50A) at 31GBP each device.

40A @ 14v is 560W. 3 devices would therefore dissipate 187W each. Looking at the thermal data for the TO247 part IXFT340N075T2 you have a total thermal resistance of 0.37decC/W junction to heatsink so at 187W the junction witll be 187 * 0.37 degC = 70degC above the heatsink temperature, so assuming a max safe working junction temperature of 160degC the heatsink would be 90degC and at an ambient of 30degC the heatsink would need to be better than (90-30)/187 or 0.32degC/W per device. This is just about doable with something around 300mm L x 200mm W x 50mm high fins (2 devices per heatsink) and a couple of good 120mm fans. Its probably just over the limit for a PC cooler which are generally around the 100 - 130W mark. I'd probably stick with 4 devices in practice.

As @Ian0 says, the Exicon devices are also designed for linear operation so like the IXYS Linear2 have no secondary breakdown, but sadly they are limited in current carrying and power capacity. The TO264 case gives a much higher thermal resistance (approx 0.7degC/W junction to heatsink) so at 135degC their 250W device is practically limited to around 85W, needing 7 or 8 devices.

 

As @Ian0 says, the Exicon devices are also designed for linear operation so like the IXYS Linear2 have no secondary breakdown, but sadly they are limited in current carrying and power capacity. The TO264 case gives a much higher thermal resistance (approx 0.7degC/W junction to heatsink) so at 135degC their 250W device is practically limited to around 85W, needing 7 or 8 devices.

I was intrigued by that. It doesn’t seem to be the fault of the TO264 case, as the TO247 devices are equally poor.
They do have an advantage on paralleling as the temperature coefficient of Vgs(th) is positive, but in a current sink, there is plenty of scope for a bit of source resistance to deal with that.

 

Paralleling MOSFETs will not work well.
The one with the lower Vgs(thres) will tend to hog all the current.

 

There's no Current Hogging when using individual Op-Amps for each Power-FET.
.
.
.
.
.

 

There's no Current Hogging when using individual Op-Amps for each Power-FET.
.
.
.
View attachment 287063.
.

True. as long as there is a separate op amp and sense resistor for each MOSFET.
I was looking at his schematic and hadn't read all of your post.

 

but in a current sink, there is plenty of scope for a bit of source resistance to deal with that.

True, as long as you don't want to go close to zero volts. A 0.1Ω sense resistor at 5A per device is 0.5v. For general battery testing at min 1V say that's fine but for short-circuit PSU testing that could be a problem,.

 

I was intrigued by that. It doesn’t seem to be the fault of the TO264 case, as the TO247 devices are equally poor.

I should qualify my TO247 point; for the IXYS devices I refer to their PLUS-247 package which is slightly better than the standard 247 or 264 packages.

 

thanks a lot for you help ,i found near by online store this mosfet IXFB100N50P ,i think it should be enough for my application,meanwhile i have some industrial boards for component recovery with handreds of mosfets STB120NF10T4 SUM10N10 ,can i use these for prototyping?

for information the maximum current is 50A.

 

The connections as shown for the load make no sense as to the current path. As shown in the schematic, the active devices are in shunt of the load.
So I suggest that the TS explain the "40 amp" current path relative to the circuit shown.

 

The connections as shown for the load make no sense as to the current path. As shown in the schematic, the active devices are in shunt of the load.
So I suggest that the TS explain the "40 amp" current path relative to the circuit shown.

Makes perfect sense if you consider the 'load' to be a battery or PSU under test at a constant or programmed current. The incorrect assertion in his original post by the TS is that its a current source; of course its properly described as a current sink as the current is created externally.

 

OK, and yes, it certainly fooled me. No way will that circuit correctly source current. This is what happens when we are only told about what others think we need to know.

 

OK, and yes, it certainly fooled me. No way will that circuit correctly source current. This is what happens when we are only told about what others think we need to know.

I think, with our level of expertise & experience, we can give less experienced AAC members a bit of slack - despite the incorrect assertion there's enough info given to make it clear what the TS' intent and thinking was!

 

i have assembled the circuit ,test it for short time it works great ,now im working on 4 wire cooling fans to be drived via pwm from esp32 cause the heatsink go hot (80°C) after 20 minutes at 40A load,by the way esp32 internal ADC and DAC is crap used external DAC MCP4725 and INA219 for current and voltage mesurement.

 

Certainly heat sinks WILL GET HOT!! The function of a heat sink is to absorb heat and transfer it to the air. Heat transfer requires a temperature difference. So the concern is the transistor junction temperature.
Possibly forced air cooling of the heat sinks is advised.

 

Here's how to handle the Heat the old-school way ...........
.
.
.
.

 

Here's how to handle the Heat the old-school way ...........
.
.
.
View attachment 288316.

i love the designe but i have a limited space ,since i have a lot of lm75A temperature sensor,will connect one to the esp32 via I2C bus and will use 20Khz PWM signal to drive the Fans (Delta FFB03812VHN-9C3P ),5 fans actualy.