Constant current dummy load, ran a gutter...

[bsodmike]

Hi all,

Much of the inspiration for testing this out is Dave's excellent video
I started by breadboarding the LM324N and just testing the 50K 10-turn pot, to see if it would actually give me the proper output, i.e. 0v - 2.5v (due to the double-10k divider).

However, I ran a gutter with some strange behaviour. So what's going on here, let me summarise.

Turning the pot to its maximum, allows me to measure ~4.99vdc at the Op amp non-inverting input pin 5.  This is expected.  However, somewhere in the middle/towards the upper end, the output at the double-10k divider, I get a max output of 1.88v.

Let's take step back, and ignore the output divider, the voltage at pin 7 is 3.76v.  It does not advance even if I go beyond this 'threashold' point, i.e. 4.99vdc at the non-inverting input, and 3.76dc at the output.  This seems odd.  It should be a voltage follower right?

The exact point, as measured is 12.72kohm and 38.61kohm (50k pot wiper position). Anything beyond this, does not change the output.

Do I need to increase the pot input voltage to "cheat" and get pin 7 to match 5vdc?  This doesn't seem right, or elegant though.

Thoughts please?

P.S. included a shot of the PSU just to show that Chan 2 was used.

Cheers!

[bsodmike]

I decided to just move on leaving this strange behaviour to at least see if I can drive the power MOSFET.  I've used an IRL640A as this is what I have on hand.

The PSU is configured to power the MOSFET using Chan 1, Chan 2 powers the LM324N.
The left Fluke measures the voltage at A.  The right Fluke measures the current through Drain-source.

• I first tried a source voltage of 1vdc, this lead to a max current draw of 0.676A
• At 2vdc, 1.419V at A to give 1.303A drawn
• At 3vdc, the MOSFET started to get hot and I could hit 1.5A and it possibly started to thermal throttle at that point

So yeah, I've got a crude load that can draw ~1.5A, but the voltage follower needs sorting out.

[MarkF]

1)  For stability, you need some resistors around the MOSFET.

2)  Using 5V for the power supply is probably too low and causing some of your issues.  For Instance, the op-amp output
     can not go completely to the power supply rail (5V).  Check the datasheet.  Plus, 5V is pretty close to the
     voltage required to turn on the MOSFET.  A power supply between 9V and 12V would be a better choice.

3)  I used a 10-turn linear pot to set the current.  Do you have a linear or log/audio pot?

4)  Here is mine (0 to 2.5A):
     I used the 620Ω and 5.6KΩ around my MOSFET because those were the values I used in the voltage divider.
     The values in 'red' are probably the best choice.

[bsodmike]

Thanks a lot Mark!

I just noticed in the Datasheet: "with V+ from 5V to 30V; and over the full input common-mode range (0V to V+ − 1.5V) for LM2902, V+ from 5V to 26V".

Thanks for the other comments, I will incorporate them into my circuit!!

Cheers M.

[MarkF]

Thanks a lot Mark!

I just noticed in the Datasheet: "with V+ from 5V to 30V; and over the full input common-mode range (0V to V+ − 1.5V) for LM2902, V+ from 5V to 26V".

Thanks for the other comments, I will incorporate them into my circuit!!

Cheers M.

Also notice, the Output Voltage Swing (V_OH) for V+ = 30V that V_OH = 28V.
You can only expect the output voltage to reach Vsupply - 2V.

[MarkF]

You also may be interested in Jay_Diddy_B's Load project:
   https://www.eevblog.com/forum/projects/dynamic-electronic-load-project/

However, notice one major difference to mine:
   His current set point is through the op-amp (-) negative input while I use the (+) positive input.
   This requires that he use a dual power supply for the op-amps.

[rdl]

Be advised, that circuit has a tendency to oscillate.

[bsodmike]

Hi rdl, do you mean Jay_Diddy_B's or MarkF's?

I notice that Mark has added the dampening suggested by Jay_Diddy_B - he's got a 2R2 and 2.2uF on the load input.  Also his use of the 100R/1k + 10nf near the MOSFET looks similar to the implementation by Jay_Diddy_B.

I would assume, MarkF's doesn't oscillate as much as the poor performing examples? Of course, any PCB implementation should be tested for oscillations.

[bsodmike]

Couple questions Mark -

1) Your Ref02, what's the output voltage? I seem to get away with around 6V into the LM329N but I may need to increase this considering Vsupply - 2V.

3)  I used a 10-turn linear pot to set the current.  Do you have a linear or log/audio pot?

Yes, I've replaced my 10k/10k pair with a 10k 10-turn pot for current limit.

I'm now testing against the LOAD input being a 5vdc source, so as to meet the minimum requirements of the IRL640A.  I'll need to take a break till a few more parts arrive, namely a heatsink for the TO220 MOSFET; so far been testing in free-air and it heats up quickly under 5V/1A!

1)  For stability, you need some resistors around the MOSFET.

I will go with 100R/1k and the 10nF plus dampening at the load input and report back.

[ledtester]

Here's a good video exploring how to go about getting rid of the oscillations in a CC regulation circuit:

https://youtu.be/8xoGhZQO29Y

It is an hour long and progresses at a somewhat leisurely pace, but it's got a lot of good info in it.

[MarkF]

• The REF02 is a 5V reference (See attached datasheet).  The input to the 10:1 voltage divider (from the REF02 or Externally) is 5V max.  This gives a control voltage to the (+) input of 0V to 0.5V.  Which yields a maximum voltage of 0.5V across the 0R2Ω sense resistor.  Giving my max current of 2.5A.  You can select your max current by adjusting the value of the sense resistor knowing it will have a max voltage of 0.5V across it.
  
  
• Items of note:
    - I never populated the fine adjustment pot for the REF02.  Never needed to adjust the REF02 output.
    - Also, I don't have the damping RC circuit.  (Next PCB revision)
    - You will NEED a fan for load currents much over 0.5A (with a RA-T2X-64E or FA-T220-64E heatsink).
    - The IRFP250 is in a TO-247 package which is much bigger than a TO-220.  Better heat dissipation in my opinion.
  
  
• Never had problems with oscillations.  The 10nF capacitor in the feedback loop is about twice the minimum value with my op-amp.  Also, the TLC272 op-amp I used can provide +/-30mA output drive current.  Not sure how the LM324 will fare.  I see people also using the LM358.
  
  
• No one has mentioned the 1KΩ resistor to 12V at the sense resistor...
  That's a kludge, pure and simple!
  
  The problem is that without a device-under-test(DUT) connected or the DUT that is powered 'OFF', the op-amp can NOT develop any current through the sense resistor.  Therefore, it will drive V_GATE of the MOSFET to the maximum possible voltage.  The result is that when the DUT is then turned 'ON', the MOSFET will be fully 'ON' and present a brief short to the DUT.
  
  The 1KΩ resistor provides a minimum current through the sense resistor and will allow you to turn 'OFF' the MOSFET in this situation by setting the set current to 0A (i.e. Adjust the 10-turn pot to minimum)
  
  
• I based my load on a circuit from Peter Oakes.  Here is a video where he is doing performance testing:
  
  
  
  

[MarkF]

Couple questions Mark -

3)  I used a 10-turn linear pot to set the current.  Do you have a linear or log/audio pot?

Yes, I've replaced my 10k/10k pair with a 10k 10-turn pot for current limit.

I'm now testing against the LOAD input being a 5vdc source, so as to meet the minimum requirements of the IRL640A.  I'll need to take a break till a few more parts arrive, namely a heatsink for the TO220 MOSFET; so far been testing in free-air and it heats up quickly under 5V/1A!

Also, using a smaller sense resistor means you will have less voltage drop for the same current and will allow you to get a resistor with a smaller power rating.  Therefore, you can test lower voltages.

The 0R2Ω resistor at 0.5V @ 2.5A = 1.25W requirement.
A 3W resistor like this  https://www.digikey.com/product-detail/en/vishay-dale/LVR03R2000FE70/LVRB-20RTR-ND/1166251  would do.

[pqass]

Because you've used a 1ohm shunt, for every amp through the shunt+MOSFET loop the voltage drop across the shunt increases by 1V. So at 1A, Vshunt_drop = 1V; at 2A, Vshunt_drop = 2V, etc.

Your MOSFET gate needs to be raised above this Vshunt_drop the more amps you wish to push through the shunt+MOSFET loop. But your opamp is only being powered by 5V; an opamp that typically only can output approx. Vcc-1.25V at most.  5V - 1.25V = 3.75V at the output as you've witnessed.

The IRL640A has a Vgs(th) between 1V min and 2V max which means that it starts conducting when its gate is at 1V or 2V above its source (min, max, respectively).  But if you want 2A constant current, you have a Vshunt_drop of 2V, plus add the Vgs(th), and you're at 3V or 4V (min, max, respectively).  To fully conduct at 11A continuously, the MOSFET gate needs to be at 5V above source.  But your opamp won't go that high (given Vcc is 5V, the max output of the opamp as witnessed is 3.75V)!

You have 3 options:
1. raise the opamp Vcc to (Vgs+opamp output max to Vcc diff+Vshunt_drop max at 11A) 5V+1.25V+11V=17.25V; enough to allow its output to reach 5V above shunt voltage drop if it needs to.
2. lower the shunt from 1ohm to, say, 0.1ohm where for each amp through it, the Vshunt_drop will be 0.1V. So now at 2A constant current will only drop 0.2V.  At 2A, this will give you back 2V-0.2V or 1.8V to be used above the Vgs(th) to pump more current through the MOSFET.  But you still won't get close to fully on at 11A.
3. if you really really need Vcc=5V then get another op amp that can do rail-to-rail output plus use the 0.1ohm shunt.

What is your maximum amps design requirement? 

Also, this may be of value (scroll down to my design schematic):
https://www.eevblog.com/forum/buysellwanted/wtb-(uk)-electronic-dc-load/msg3140566/#msg3140566

[bsodmike]

To fully conduct at 11A continuously, the MOSFET gate needs to be at 5V above source.  But your opamp won't go that high (given Vcc is 5V, the max output of the opamp as witnessed is 3.75V)!

You have 3 options:
1. raise the opamp Vcc to (Vgs+opamp output max to Vcc diff+Vshunt_drop max at 11A) 5V+1.25V+11V=17.25V; enough to allow its output to reach 5V above shunt voltage drop if it needs to.

Sorry, but can you explain where you got 11A for Drain-source current from?  We are limiting the max Ids to 2.5A?

What I've understood so far (recap): with a 1ohm current sense resistor, at 1A, Vshunt_drop = 1V.

So our requirements:
2.5A for max Ids: Vshunt_drop = 2.5V

Vgs(th) is 1V to 2V (min/max) so (Vgs(th) + Vshunt_drop) = 3.5V to 4.5V (min/max).  So a gate voltage of ~ 4.5V to 5V is needed to allow 2.5A current flow (Ids) - right?

Therefore, option (1) is to raise the opamp Vcc to 5V+1.25V = 6.25V; enough to allow its output to reach 5V above shunt voltage drop if it needs to.

I^2R for (2.5^2*1ohm) = 6.25 watts power draw on the 1ohm current sense resistor.

2. lower the shunt from 1ohm to, say, 0.1ohm where for each amp through it, the Vshunt_drop will be 0.1V. So now at 2A constant current will only drop 0.2V.  At 2A, this will give you back 2V-0.2V or 1.8V to be used above the Vgs(th) to pump more current through the MOSFET.  But you still won't get close to fully on at 11A.

We want Ids = 2.5A so with a 0.1ohm resistor, Vshunt_drop = 0.25V
Vgs(th) is 1V to 2V (min/max) so (Vgs(th) + Vshunt_drop) = 1.25V to 2.25V (min/max).

Opamp Vcc of ~3V is enough to drive this smaller current sense resistor.  Also I^2R for (2.5^2*0.1ohm) = 0.625 watts power draw on the current sense resistor.

Thanks so much for going through this in detail, really appreciate it (as this is mainly a learning exercise on my part).

[bsodmike]

Wow thanks so much, great resource and I've been following along (till around the middle). Will finish later, BUT, I tweaked my circuit to match the schematic by MarkF (well most of it).  But I did not add the dampening on the load input (yet!).  I wanted to watch the Drain-Source voltage across the MOSFET oscillate.

See the bottom part of my sketch; I just added a 10nF between the OpAmp terminals pin 8 & 9, and a 100 ohm + 1kOhm.  The output in this example is Ids = 200mA.  Output oscillating at 175kHz.

[bsodmike]

• No one has mentioned the 1KΩ resistor to 12V at the sense resistor...
  That's a kludge, pure and simple!
  
  The problem is that without a device-under-test(DUT) connected or the DUT that is powered 'OFF', the op-amp can NOT develop any current through the sense resistor.  Therefore, it will drive V_GATE of the MOSFET to the maximum possible voltage.  The result is that when the DUT is then turned 'ON', the MOSFET will be fully 'ON' and present a brief short to the DUT.
  
  The 1KΩ resistor provides a minimum current through the sense resistor and will allow you to turn 'OFF' the MOSFET in this situation by setting the set current to 0A (i.e. Adjust the 10-turn pot to minimum)
  
  

Sweet, I was going to ask you about that 
Thanks for all the details!

I've placed an order for a 0R1Ω sense resistor and will work with that once it arrives.  For now I'm just going to play with the 1R that I have and see if I can get it stable (for kicks!).

[SmokedComponent]

Also, keep your shunt resistor cool. Not just by using smaller value (to generate less own heat) but also where you place it. The dialed-in current will start to drift with resistor's value. This was a minor snag I encountered during my build of a simple CC load I needed to fix in order to justify an expensive but really satisfying 10x wirewound pot. Position it away from transistor's heatsink. Or, if you are using a fan, in front of the fan, so cool air passes through shunt resistor first.

[bsodmike]

Haha YES, I also picked up a 50K 10-turn wire-wound pot.  It is superbly sublime 

Thanks for the awesome tips, and seriously, thanks to everyone here!!

[MarkF]

• No one has mentioned the 1KΩ resistor to 12V at the sense resistor...
  That's a kludge, pure and simple!
  
  The problem is that without a device-under-test(DUT) connected or the DUT that is powered 'OFF', the op-amp can NOT develop any current through the sense resistor.  Therefore, it will drive V_GATE of the MOSFET to the maximum possible voltage.  The result is that when the DUT is then turned 'ON', the MOSFET will be fully 'ON' and present a brief short to the DUT.
  
  The 1KΩ resistor provides a minimum current through the sense resistor and will allow you to turn 'OFF' the MOSFET in this situation by setting the set current to 0A (i.e. Adjust the 10-turn pot to minimum)
  
  

Sweet, I was going to ask you about that 
Thanks for all the details!

I've placed an order for a 0R1Ω sense resistor and will work with that once it arrives.  For now I'm just going to play with the 1R that I have and see if I can get it stable (for kicks!).

I would not push the current levels to its maximum.

Peter also used a 0R1Ω resistor with a larger IRFP064 MOSFET on a large computer heatsink.  While testing a power supply, it spiked and totally destroyed his electronic load.  Total meltdown.

That is part of why I choose the large resistors between the MOSFET and the op-amp.  To limit the current into the op-amp if the MOSFET shorts.  Also, at 15V @ 5A load settings, you WILL NEED A MASSIVE heatsink with active cooling.  I'm not suggesting values other than the 100Ω and 1KΩ resistors to the MOSFET (probably the best performance).  Just be aware that if you push the MOSFET and it shorts, you WILL destroy all of your semiconductors.

You need to carefully watch your power levels and heat dissipation.

I mounted a small 1.5" 12V fan onto my RA-T2X-64E heatsink and I don't have overheating issues with loads up to 15V and 2.5A.  Also, it's a good reason to use a 12V power supply for your load (i.e. 12V fans).


An idea for an Arduino interface to watch power levels and limit them:

[bsodmike]

I would not push the current levels to its maximum.

Peter also used a 0R1Ω resistor with a larger IRFP064 MOSFET on a large computer heatsink.  While testing a power supply, it spiked and totally destroyed his electronic load.  Total meltdown.

That is part of why I choose the large resistors between the MOSFET and the op-amp.  To limit the current into the op-amp if the MOSFET shorts.  Also, at 15V @ 5A load settings, you WILL NEED A MASSIVE heatsink with active cooling.  I'm not suggesting values other than the 100Ω and 1KΩ resistors to the MOSFET (probably the best performance).  Just be aware that if you push the MOSFET and it shorts, you WILL destroy all of your semiconductors.

You need to carefully watch your power levels and heat dissipation.

I mounted a small 1.5" 12V fan onto my RA-T2X-64E heatsink and I don't have overheating issues with loads up to 15V and 2.5A.  Also, it's a good reason to use a 12V power supply for your load (i.e. 12V fans).


An idea for an Arduino interface to watch power levels and limit them:

Thanks for all that info and fair warnings, yes, going with such a low-voltage current sense resistor and a shorted-MOSFET would surely wipe out the DUT and whatever's attached, eek!

I've ordered this heatsink, which is almost identical to yours, just boasts (spec wise) a slightly better thermal performance (pretty much the same) https://export.farnell.com/aavid-thermalloy/6400bg/heat-sink-to-220-218-2-7-c-w/dp/1213475

Just to give you a heads up, I was looking at pushing this to the extreme just for the "engineering" sake to gain more experience in this area.  In reality I doubt I've ever try to push past 15V/2.5A (we'll see!).

I've also already made changes to incorporate the 100Ω and 1KΩ resistors to the MOSFET. Right now I'm just passing 200mA and trying to see if I can stop the oscillations; I also plan to swap the opamp for a better high-precision one similar to the TLC272 and see what improvements I gain.

I've also got one that does rail-to-rail in my parts bin, which has similar specs to the TLC.

TEXAS INSTRUMENTS Operational Amplifier, 1 Amplifier, 5.5 MHz, SOT-23, 5 Pins   3117581RL   OPA376AIDBVTG4

Will report back on any progress made and any further questions!  Thanks a lot for the Arduino idea as well, I'll look into that as well at a later stage once I get to designing the PCB.

[MarkF]

Thanks for all that info and fair warnings, yes, going with such a low-voltage current sense resistor and a shorted-MOSFET would surely wipe out the DUT and whatever's attached, eek!

You miss understand.  It's not the DUT but the Electronic Load that let out the magic smoke (i.e. the MOSFET and op-amp).  The shorted MOSFET fed the high voltage back into the op-amp.  It would also take out an Arduino if you end up going that route.

[bsodmike]

Interesting.  Out of curiosity, when the MOSFET is shorted, what's the path current takes to destroy the Opamp, i.e. through the opamp's output (but I though that's high impedance?) or through the inverting-input (as per both our schematics).  Sounds like the latter given the 5.6K or 1K resistor chosen.

Assuming it is the latter, any ideas on how we can add some protection circuity to handle such a scenario?  Ultimately it would be a 'current-limit' circuit, but the tricky part is "limit on short only"

[MarkF]

Interesting.  Out of curiosity, when the MOSFET is shorted, what's the path current takes to destroy the Opamp, i.e. through the opamp's output (but I though that's high impedance?) or through the inverting-input (as per both our schematics).  Sounds like the latter given the 5.6K or 1K resistor chosen.

Assuming it is the latter, any ideas on how we can add some protection circuity to handle such a scenario?  Ultimately it would be a 'current-limit' circuit, but the tricky part is "limit on short only"

His explanation of what happened at time 12:00 (This is NOT a solution):



He never did finish this design.

[bsodmike]

Ohh gotcha!! Thanks. 

[bsodmike]

I would not push the current levels to its maximum.

Peter also used a 0R1Ω resistor with a larger IRFP064 MOSFET on a large computer heatsink.  While testing a power supply, it spiked and totally destroyed his electronic load.  Total meltdown.

If I'm going to use my IRL640A, 65W @ 11A (Id=max)(Tc=100degC) = LOAD+ voltage of max 5.9Vdc
65W @ 2.5A = LOAD+ voltage of max ~26Vdc (as per pqass).

With regards to the use of 100R/1Kohm vs 620R/5K6, given that I've got a 10K trim-pot, any advantage to using the larger pair of resistors? i.e. they would offer better protection against any potential MOSFET shortage, but other than that I would "assume" loop-stability would be the same?

Again, curiosity on my part, how have you faired with any potential PCBs you may have had fabricated? Did you run in to any issues, especially with the need for thicker traces for Vds/Ids?