MOSFET and OPAMP Load tester / constant current load

[lordvader88]

I have soldered some power resistors to a PCB for load testing, I need to put in in a box with wires/knobs/etc....never ending parts list. Now I what to make something like this

, but I've barely used mosfets, and I don't have any like Dave uses in the video, that work at 5V logic and predictably near zero, or whatever he means.

So I have LM741/358, TL072/082, LT1013, and some IRF540 that should be fine, but how would I go about biasing the fet(s) ? I have a huge 1 Ohm 300W resistor, that will do perfect.
Thanks

[MarkF]

Here you go:   https://www.eevblog.com/forum/beginners/a-load-off-my-mnd/msg2151265/#msg2151265

[lordvader88]

Great, watch that Sullcom guy's video too. Now I'm even more intrigued, at the possibliity of powering this from some analog PWM circuits I could make, or with a PWM chip I have (for SMPS) or whatever else is allowable.

Someday I'll have to get my Adriano working and learn the coding (not not forget it) for going digital too


Is the only real reason to use a fet, is for it's higher effeciency ? I have tons of power BJTs, I should look up old test gear circuits.

[lordvader88]

Ok so if a mosfet is used the opamp is not driving a base current, so thats a big difference.



I have +10V/-5V regulators hooked up to a small centertapped transformer, and IRF540N. LTspice thinks the circuit works well enough, I'm not worried if it doesn't behave at the milli-amp level.

I better be careful if I hook this up to the PSU I built.

[MarkF]

Assuming 20V @ 2.5A load target requirements.
Parts and sources:

• If you follow along with my circuit and use the LM358 or LT1013 op-amp, you will NOT need a negative power supply.
  The +10V will be okay.  A +12V supply would be better since you will need a fan.
  
  
• Use the IRF540.  If you buy something else, get a MOSFET in a bigger TO-247 package.  Better heat dissipation. 
  A IRFP250 or IRFP064 for example.
  
  
• A Bourns 10K ohm 10-turn potentiometer.
  
  
• A 1 ohm 300W is over the top.  A 5W should be big enough.  If you use a 0.200 ohm resistor, you will have a 0 to 2.5A range.
  Vishay LVR05 Series or Vishay AC05 Series.
  
  
• Heatsink I used is a RA-T2X-64E.  You WILL need a fan.
  
  
• Example 40x40x10mm 12VDC Fan.
  
  

[MarkF]

Great, watch that Sullcom guy's video too. Now I'm even more intrigued, at the possibliity of powering this from some analog PWM circuits I could make, or with a PWM chip I have (for SMPS) or whatever else is allowable.

Someday I'll have to get my Adriano working and learn the coding (not not forget it) for going digital too

I'm not big on using a filtered PWM waveform to generate a DC voltage.
Instead I use MCP4802/4812/4822 DACs with a SPI interface.  These are 8/10/12 bit versions.
There are I2C versions also if you prefer.

[David Hess]

Ok so if a mosfet is used the opamp is not driving a base current, so thats a big difference.

The operational amplifier still has to drive the considerably gate capacitance of the power MOSFET without becoming unstable but this is easy enough to solve if it is a problem.

[lordvader88]

Ok I already made , basically as if I had a low-voltage mosfet. Used -5/10V and used a resistor and a pot, to limit the current to around ~2A (I think, I've changed it). I had a low-pass filter on the gate as well from someone else's version.

Its working but not safely, the mosfet would sometimes be at a few mA, then just jump to 0.56A. Somehow I fried the output 2n3055 of a PSU I built. At the time it was around 15V @1-1.2A. The collector/emitter burned out. It was a cheap 2n3055 from ebay, but probably already damaged from before too.

So I also have computer ATX PSU's I'm working on, and what a real high current load tester for them at some point.

So with typical IR540 mosfets, what circuits should I be looking at to make this better. Where is the "linear" region for a typical mosfet? Is it just moved up 1,2,3 volts compared to a Logic-level fet ?. Should I bias the gate 0V point up 1 V ? I'll see in LTSpice.


For low current stuff I should just make an old 2n3055 based emitter-follower load tester circuit.


I don't really have any ADC or DAC chips, I hardly have any CD4000 or 74 series chips either. They've been on the list a long time tho.

[lordvader88]

Here you go:   https://www.eevblog.com/forum/beginners/a-load-off-my-mnd/msg2151265/#msg2151265

Ok I'll try that version, thanks

[OM222O]

Great, watch that Sullcom guy's video too. Now I'm even more intrigued, at the possibliity of powering this from some analog PWM circuits I could make, or with a PWM chip I have (for SMPS) or whatever else is allowable.

Someday I'll have to get my Adriano working and learn the coding (not not forget it) for going digital too


Is the only real reason to use a fet, is for it's higher effeciency ? I have tons of power BJTs, I should look up old test gear circuits.

Scully always uses the weirdest approaches, despite having the right idea. I usually take his videos as guidelines and basic ideas rather than proper tutorials. for example he uses fets in parallel, with some complicated and weird ass transistor arrangement in conjunction with a negative supply to balance the load between them ... how about using a multi channel op amp, or multiple op amps (some of my favorite op amps are only in single package  ) which ends up being cheaper?  also he uses weird, old and out of stock switching mosfets, rather than just going with some linear ones to begin with  again, not sure why.

[lordvader88]

Well I made it with Marks basic design so far. I used a temperature trimmed LM336 2.5V reference circuit, and doubled it up to 5V w/ a resistor and pot,  instead of unity gain on the 1st 1/2 opamp.

 So it's working up to 110mA max so far intentionally, and 0.37A un-intentionally. It's on the breadboard, and it looks clean on the scope overall.

There's some funny business around 1kHz and 10kHz on the mosfet gate, sort of looks like TV remote control style signal. IDK if that's coming from the main PSU, the DUT, or what. It looks cool.



I'll try it on a 5A power brick I have.


For now I using 0-9V for the gate control side, and LM358 and IRF540 with RDs of 0.077mOhm, . I have lower Rds FET's too I think.

But is seems to work good so far. Thanks again. The 1st circuit I tried and hastley made would jump from a few mA to +0.56A for some reason and settings/etc. Not this circuit, and it turns down to 620uA or so.

[MarkF]

I only tested mine up to 1KHz.  If you want to go faster, change the gate resistor to the MOSFET to 100 ohms.
I had 6.2K for op-amp protection from MOSFET failure but it won't allow high switching rates.

Note.  I updated the circuit with a 620 ohm gate resistor.


I tailored the load after what Peter Oakes did.  Be aware his load had a meltdown from exceeding power dissipation limits caused by power supply spikes that he was testing.  So make sure to operate within the safe operating area.

[lordvader88]

I never did anything with AC input yet. Time to get this soldered over the old model.

I'm also looking at this paper on a modular load tester, but it uses 8-9 mosfets, I only have 20-30 total, so screw that.
https://pdfs.semanticscholar.org/319f/3df15130fad0cdc907b1d7a2f67a3ba42b8a.pdf

I want to setup an over-current shut down. So I could use a comparator and or BJT circuit to turn off the gate before OC. Just in case I turn it too high on something I know should not exceed some limit.

I could just add it where I went from 2.5V to 5V, I can use that to set max limit.

[lordvader88]

Ok the load tester is working nicely, however it blew-up my homemade PSU again at around 1A. I suspect it's the cheap 2n3055's I have, I guess for now I'll keep them under 0.7A. I don't need anywhere that much current, and the PSU was doing good at static loads.


So I should start feeding in 0-5V waveforms to the opamp controlling the gate, so I guess start with some DC offset and a small waveform ripple,

[MarkF]

Another option to limit the max current the load could apply is to modify the voltage divider on the input control.
Setting the max current could be a little tricky.

   

[lordvader88]

Hey Mark 1 problem I'm having is on startup. I can have the circuit turned on, with the output set for zero, and the mosfet drain supply is turned off. The LM358 is on max output. Then hitting the power switch for the drain supply, the current often jumps really high, and around a bit in general and doesn't respond at all to the poti. setting. That's alarming, I had my PSU limitied to 0.5A and is jumped right to it.

Once I can get it settled down, then it behaves fine, and seems very steady and low noise, at static loads.

So the last 2 times I started it with the gate unconnected, aka connected to the external input switch. Just about to troubleshoot a 12V 5A SMPS brick for a monitor, so I'm hesitant to connect the load tester, in case it shoots to max.

For now, I only added a current limit of down to 70% max of any setting, just using the other 1/2 the opamp. I should change it down to near zero%.

[lordvader88]

Wow so the IRL540 I'm using is a logic level mosfet, how did I miss that right on the datasheet. Well I burned out 1 so far, IDK how, it wasn't violent or smokey. Either way mine are from cheapest on ebay, so they might be fakes too, it might have burned out around 1-2A for all I know, like my fake 2n3055

[MarkF]

I have lost track of what your circuit looks like now.  An updated circuit diagram would be helpful.

If there is no load connected to the MOSFET drain (i.e. drain supply is OFF), even a small positive voltage on the (+) op-amp input will drive the MOSFET gate to the max supply voltage of the op-amp because the (-) op-amp input will stay at zero volts because there is no current available to flow through the sense resistor.  This may be caused by the input offset voltage of the op-amp.

If your current limiting is done in the 10:1 voltage divider, the pot (R4) may act up at either end stop.  If it opens, there will be no pull down for the (+) input and you will get some unknown gate voltage (probably max turn on).  You could put a 100K resistor across the pot as a protection (100K being high enough not to interfere with operation).

What is the op-amp supply voltage?  The max IRL540 gate/source voltage (V_GS) is 10V.  If your supply voltage is 12V, the V_GS max can be exceeded.

Have you put a scope on the circuit?  Is it oscillating?

Does it act up if you always turn ON the load voltage (i.e. drain voltage) before turning on the op-amp voltage?

Give a read through this long thread for some background:
https://www.eevblog.com/forum/projects/dynamic-electronic-load-project/

[lordvader88]

I have lost track of what your circuit looks like now.  An updated circuit diagram would be helpful.

If there is no load connected to the MOSFET drain (i.e. drain supply is OFF), even a small positive voltage on the (+) op-amp input will drive the MOSFET gate to the max supply voltage of the op-amp because the (-) op-amp input will stay at zero volts because there is no current available to flow through the sense resistor.  This may be caused by the input offset voltage of the op-amp.

If your current limiting is done in the 10:1 voltage divider, the pot (R4) may act up at either end stop.  If it opens, there will be no pull down for the (+) input and you will get some unknown gate voltage (probably max turn on).  You could put a 100K resistor across the pot as a protection (100K being high enough not to interfere with operation).

What is the op-amp supply voltage?  The max IRL540 gate/source voltage (V_GS) is 10V.  If your supply voltage is 12V, the V_GS max can be exceeded.

Have you put a scope on the circuit?  Is it oscillating?

Does it act up if you always turn ON the load voltage (i.e. drain voltage) before turning on the op-amp voltage?

Give a read through this long thread for some background:
https://www.eevblog.com/forum/projects/dynamic-electronic-load-project/

Ok thanks I lost track of this working on other stuff. My circurt basically was like your's, with 10V, except I used a 2.5 volt reference into one 1/2 the op-amp, and with the other side I multiplied it by x1.7 to x2, so up to 5V from there. Then the rest is the same, I made 5600 and 620 values. And I'm using the IRF540. Geez I never imagined they are also only 12V you say...This was probably the project that got me to buy them ages ago, so I'd have logic level  fets for someday.

It seems to work fine, nothing to worry about on the scope when it's working . It's just at the start, where between the op-amp and gate starting, it shoots right up, possibly to max (but my PSU can't do 2.5A so IDK). And the potenitometer pretty much does nothing, so itt's the op-amp gone crazy.

So when you turn on your loadtester,  you always have the DUT already on first  ?

Well regardless I have now the current limiting working on my PSU so I can try the LT again.

[MarkF]

I need to do some more checking on mine to see if I can reproduce your issue.

I seem to remember something strange when I power the Load before the SUT. 
But can't say the current spikes up and not responding to the pot.

I ask @Jay_Diddy_B on his thread and he responded with a change.
His suggestion and his Load control the current via the (-) input.  Opposite from most Loads.
I have not had time to try it yet.
   https://www.eevblog.com/forum/projects/dynamic-electronic-load-project/msg2488077/#msg2488077

[lordvader88]

Ok thanks. I guess normally mosfet's are switches, but this uses them in the more linear region. I never calculated anything for a mosfet before.