Electronic Load Tester

[Rich D.]

Hi folks, I designed this electronic load device to test various power supply projects I make. Before I built it and blew it up testing it, I thought I'd put it up on here and see if anybody wants to review it, find any engineering problems with it, or just generally make fun of me etc.

A couple of notes on it: The selection of components may be a little odd, but 'twas designed specifically to use components that I have already. For example, the MOSFET, op-amp, zeners, and the pot and analog meters. I plan to heatsink the hell out of the MOSFET, and it will have a small fan. I figured the resistors are overrated and can take care of themselves. Power is supplied by the power source being tested, and I do not plan on testing anything below 5 volts or above 20. I didn't bother to show the exact form of load resistors, only showing the final value and wattage. For example Rload2 will be a series of 1 ohm, 10 watt resistors.

I didn't bother to put an over-temp shutoff on it, figuring I will generally be right there and can feel if the fan exhaust is getting warm, and nothing should be designed to get so hot in the first place.

Thanks in advance for any and all feedback.

 

[k7elp60]

I understand how your circuit works. I question if the fan is going to work on the lower voltages. I would use a separate power source for your control circuitry and the fan. I built mine a number of years ago and only used the power transistor as the variable resistor.

 

[Rich D.]

Oops, with the circuit shown the fan would not turn! That was supposed to be a 50 ohm resistor (Rd) feeding the 5-volt fan. I expected the fan to run near 1/2 speed with a 5 volt source applied, since the fan will try to draw about 100 mA at full power.

 

[Nigel Goodwin]

Presumably the opamp is a special low voltage version?, and (supposedly) 'rail to rail'? - and if so how close to the rails does it manage?.

Also, can it supply enough voltage to feed the FET as required? - I'm with k7elp60, a separate (and split) supply would be better.

 

[Rich D.]

The MCP6002 is spec'd to be within 25mV of both rails.

I don't know exactly how much it will take to turn on the MOSFET fully, and I suspect it may take more than the 3.3v available. So you may have found the weakness of this circuit.

If that's the case and I can't get a full turn-on to a load of about 4 or 5 amps, I can always swap out the zener for a higher one, I recently ordered a bunch of them with a range from 3 to 6 volts, so I hope to be able to tweak the voltage to something acceptable.

But if 5 volts still isn't enough to turn it on to 4 amps, then I'll have to go with a second power supply.
I was just hoping that with test wires all over the bench, I'd like to have something that I DON'T have to plug in!

 

[ronv]

You can power the op amp off the 5.6 volts and it will turn on the FET.
You don't need U1b.
Double check your math on the dropping resistors.
Add a 100 ohm resistor between the op amp output and the gate. This will keep it from oscillating.

 

[Rich D.]

Ronv: I thought the 4.7k resistor to ground would load-down the output capacitance and keep it from oscillating - should I use a series resistor instead?

After playing around with a 5-9V DC-DC converter, I reluctantly decided to add a 2nd power supply. The problem was with the fan. It would work directly plugged into the load at 5V, but with a higher DC supply I would need to drop it down to 5V. I couldn't find a way to make it work for all voltages without having to add a control switch.

So here is a completely revised version, with a power supply providing +15 for the op-amp and +5 for the fan.
I may decide to use a better op-amp since that one, cheap as it is, is optimized for 5V use, and is not so great for offset, drift and speed.
Also, I may use either analog or digital meters, since I have both.

 

[ronv]

Do you have any other FETs? We could make it a lot simpler if you have one with a lower Rds on.
We can worry about compensation later, but here is some "stuff" about the problem of driving FETs with op amps. Fig. 7 is what we are after.
Maybe you have an LM358 or 324 in your parts bin?

 

[ronv]

Take a look at this. It burns most of the power in the FET. But since you have a big heat sink and a fan, why not.
The op amp and FET are common (cheap) ones. With the larger voltage drop across the source resistor the offset of the op amp is less of an error.
I used 12 volts for the op amp thinking you might have a 12 volt fan out of an old PC or something.

 

[Rich D.]

That particular FET (IRF740) has a Rds on of only 0.48 ohms, probably the lowest one I have and certainly the only one that will take 10 Amps. Otherwise I'd have to go with a small bank of MOSFETs. (I did see that old thread of somebody who soldered 8 or 10 of them to a cold water copper pipe and ran water through it to sink upwards of 1500-2000 watts!)
...or are you asking about the gate-on driving voltage which appears pretty high?
I selected the IRF740 because although I do want to use it to throttle the current back, I was hoping the load resistors would bear the brunt of the load (no pun intended), especially at the higher voltages of 10...20v where the square of the voltage determines the power level. That's why the 5 ohm thingies are in there.

As for op-amps, I have a long list of them. Not the LM358 but I have the LM258 flavor. They are about the same as the MCP6002 in cost, speed, offset etc. and in fact they seems so close that they may have been derived from a common design. The 324 is likely a little better though. I can pull out an LF412 or LM308 if necessary, but I don't see this as a circuit that demands great spec's for the op-amp.

I should point out that I didn't really mind making yet another A/C supply for this tester - I tend to make so many of them that it's the reason I want to make a power supply load tester in the first place. What I really wanted to avoid is having yet another device to plug in!

The load1.png picture brings up a lot of questions. Why the 75p for AC feedback? Do I want a series resistor on the gate or a parallel resistor to ground? Is that the "source resistor" you refer to and is op-amp offset going to be an issue? I thought the negative feedback would negate all that "offset crap" in this case. Does a series resistor really do much when it is a high-impedance input? I imagined that the parallel resistor will pull more current from the output, making it a lower Z circuit and therefore making any stray capacitance more insignificant.

I then took another look at my schematic and realized that although I have a similar 0.5 ohm resistor in the load in series with the FET, I have a separate 0.1 ohm used for current measurement that only produces about 1/2 volt output, so why not combine the two resistors and get a larger output voltage? I guess it's back to the CAD software for me.

 

[ronv]

{quote]That particular FET (IRF740) has a Rds on of only 0.48 ohms, probably the lowest one I have and certainly the only one that will take 10 Amps.

OK. We can use the IRF740. I was just trying to get the sense voltage a little higher. But we could make it .3 ohms. You can see in your old circuit why having .5 ohms in the drain side .1 ohms in the source and.5 ohms in the fet you can't get to 5 amps at 5 volts.

I selected the IRF740 because although I do want to use it to throttle the current back, I was hoping the load resistors would bear the brunt of the load (no pun intended), especially at the higher voltages of 10...20v where the square of the voltage determines the power level. That's why the 5 ohm thingies are in there.

Sure, I was just trying to eliminate all the connections for different supplies and all the power resistors.

As for op-amps, I have a long list of them. Not the LM358 but I have the LM258 flavor. They are about the same as the MCP6002 in cost, speed, offset etc. and in fact they seems so close that they may have been derived from a common design.

The 258 is ok. It has the higher voltage rating you need when you change to 12 or 15 volts.

The load1.png picture brings up a lot of questions. Why the 75p for AC feedback? Do I want a series resistor on the gate or a parallel resistor to ground? Is that the "source resistor" you refer to and is op-amp offset going to be an issue? I thought the negative feedback would negate all that "offset crap" in this case. Does a series resistor really do much when it is a high-impedance input? I imagined that the parallel resistor will pull more current from the output, making it a lower Z circuit and therefore making any stray capacitance more insignificant.

You need the gate resistor to isolate the op amp from the high capacitance of the fet. The 75 pf optimizes the frequency response to make sure it does not oscillate.

You could add 5 ohms for the 10 volt and above and that would get the power down in the FET. Or you can just lower the sense resistor and a big heat sink with fan.

 

[Rich D.]

OK, I now have a new rev. Changed the op-amp and surrounding circuitry, and moved the load resistance and re-calculated the values.
At this point I am starting to get confidence that it will work. I have all the parts already except for the A/C transformer, but I'll most likely test this with a bench power supply instead.
The ironically sad part is that I seemed to have misplaced the IRF740 that I designed this whole thing around!
(I harvested it from an old Sony TV). If it doesn't turn up by the time this thing is built I'll probably have to go buy one new.

 

[tomizett]

I sympathise with you there - I designed a similar load around a set of 2SC5200's that I'd been given, and then blew them up in the final testing stage!

The design that I came up with I posted here: https://www.electro-tech-online.com/articles/adjustable-current-load.657/
I certainly found it an interesting design exercise with some unusual challenges.
While my design (and most others I've seen) sink most of the power in a transistor, yours has the advantage of not being so limited in SOA - while mine will sink 20A at low voltage, that goes right down to something like 0.5A at it's maximum. Your design on the other hand looks like it should sink full current right across its input voltage range, which is a nice feature.
The issue of low supply voltage when running off the supply-under-test forced me to go for a discreet solution rather than an op-amp, but of course the op-amp route will buy you stability and precision that my design clearly lacks!

It'll be a handy box to have around the workshop though, when it's finished.

 

[Rich D.]

Wow tomizett, I'm impressed! I'd make that except it would probably take me many months just to find and pay for all those parts. You figured out a way to do it without pluggin in, something I was trying to do but had to give up (because I have a box full of 5V fans lying around).

At first I had to ask myself why you would gang up a .1, 1, and 10 resistor, but then I saw how you cleverly got the user to switch the resistors with a pair of input connectors - sorta what I was doing by having the user plug into different banana jacks depending on the voltage.

And apparently yours actually works. As for mine ...we'll see.

 

[tomizett]

Mine does work, so far... whether it's a "repeatable" design is another matter.

I have every faith that yours will be fine. Keep us posted.

 

[ronv]

The IRF740 has an on resistance of .55 ohms and the sense resistor is .5 ohms at 25C (higher when it's hot) so you will not be able to get 5 amps when testing 5 volts. So if you get another FET get one with a lower Rds on.
You don't need (or want) U1b.

 

[Rich D.]

I also discovered with that MOSFET drive circuit, I wouldn't even get enough drive voltage to turn on the FET all the way! At best the op-amp would only see 2.5 volts on the negative input with 5 or so on the + input, so it would just go up to the + rail trying to get more from the FET without success and it would essentially be out of regulation range.

I have one more revision in which I upped the drive voltage to turn on the MOSFET with up to 5 volts and still be in regulation range. I also gave up trying to find that d%$#ed IRF740 after turning my whole workshop upside down twice.

In it's place I have some IRF820s, but they only handle 4 amps max each and have a Rds on of about 2.5 ohms. These MOSFETs need about 5 volts on the gate to pass over 1 amp, whereas the old '740s needed about 6.5 volts to turn on hard enough to pass 5A. This rev now has 4 of them in parallel with 0.1 ohms in series of each in case of current-sharing variations. This also spreads the power rating out so each FET only dissipates 1/4 of the power.

I also re-calculated all the resistances, currents and power requirements and I think I am ready to start building & testing. The only part I don't have is an 18VAC transformer, so I'll use a bench supply for now. So before I turn on the soldering iron did I miss anything here?

If you don't hear back from me in the next week or two then assume it's because of some horrible tragedy where the combined electromagnetic fields of the IRF820s combined to produce a flux-capacitor effect that drove me back to the age of the dinosaurs, or the distant future where nano-technology has made DC power supplies obsolete.

 

[ronv]

You can usually parallel bipolar junction transistors like that but not FETs. It works ok if they are used as switches, but not as amplifiers. The reason is that the gate threshold voltage can be 2 to 4 volts. So if you get a 2 volt one it will turn on first and take all the current. It works for BJT's because the base emitter voltage doesn't vary much from part to part so the small resistor in the emitter evens them out.

 

[Rich D.]

D$%^& it! What would you do: A) switch over to bipolars and deal with the heavier current demand or B) put in some form of trim resistors to balance them out or C) hand-pick FETs?
... or even D) provide a separate op-amp feedback for each FET?

At this point my usual method of just soldering a motley bunch of power resistors on the bench is starting to look better all the time.

 

[Mosaic]

You can usually parallel bipolar junction transistors like that but not FETs. It works ok if they are used as switches, but not as amplifiers. The reason is that the gate threshold voltage can be 2 to 4 volts. So if you get a 2 volt one it will turn on first and take all the current. It works for BJT's because the base emitter voltage doesn't vary much from part to part so the small resistor in the emitter evens them out.

How does this work if u can't parallel amplifier FETs?
https://frenning.dk/OZ1PIF_HOMEPAGE/50MHz_IRF510.htm