Variable dummy load design

[cprobertson1]

Ive recently been building a variable dummy load similar to Dave's design from way back when with an op amp driving a mosfet.

I was just wondering; is there any particular reason I couldnt power the control circuitry (two pots and an op amp) from the power supply the load will be used on? (since the programmed current draw will be measured from an ammeter - including the current consumed by the control circuitry?)

Im probably missing something obvious, but figured I'd ask!

Ps - will probably need a voltage reference for the pots/op amp to stop the output scaling with the source voltage, I'll do some proper calculations tomorrow when I'm not lying in bed thinking about dummy loads instead of sleeping

[ZeTeX]

You can power two pots and op amp from the power supply the load will be used. (As long as it above ~5V, as the opamp still needs to work somehow)
As for voltage reference, I just used a trimmer that I adjust with the potentiometer connected to get me the right voltage swing (0-1V for example).

Dave's design is not good, It doesnt not contain and compenstaion at all, it would oscilate like hell under some loads, here is a better one:
http://www.electro-labs.com/diy-30w-adjustable-electrical-load/

[cprobertson1]

Dave's design is not good, It doesnt not contain and compenstaion at all, it would oscilate like hell under some load

My design is similar to Dave's but heavily modified (added thermal/overvoltage/overcurrent protection and a few bypass caps and some very limited compensation amongst other things)

What sort of loads would induce oscillation in the circuit?

Also, a quick question about the article: he has three of the op amps with the output tied to the inverting input with a voltage divider on the non-inverting input. What exactly is that doing/how is that working

What exactly is that doing? (I haven't used nor studied op amps much I'm afraid!)

[rdl]

They just seem to be connected to do nothing. The LM324 is a quad op amp and the circuit only uses one of them. Dave's circuit used one of the extras as a buffer, but I don't think he mentioned how the other two were connected. I didn't see any point in using a quad if two were enough, so I used an LM358.

[cprobertson1]

Aye, its a 358 I have as well: was curious as to why he had resistors on them xD Out of curiosity, is it due to noise reduction that he hasn't just left the unused op amps floating?

[rdl]

I didn't known why it was done like that and was curious, so I googled: unused op amps in quad package


First hit:

https://www.maximintegrated.com/en/app-notes/index.mvp/id/1957

[cprobertson1]

Dang! I had googled the wrong thing (noise reduction (techniques) for op amps and couldn't find much! Found a few examples of the practice but couldn't find the reasoning behind it!)

[EntropyWizard]

There are a number of spin-off designs inspired by Dave's video around the internet which use the unused op amps for additional features. I built a load  which uses one for thermal overload protection and one for a gate input. Details posted here:  https://www.eevblog.com/forum/projects/electronic-load-57509/ . I think the main advantage of a seperate power supply is allowing you to load lower voltage sources. You need to make sure you can maintain the voltage across the sense resistor plus  the necessary Vgs to operate the MOSFET at your max current.

[cprobertson1]

Hello again folks!

Ended up settling for a heavily modified design over my original with a split resistor load (so you can toggle between a higher load for increased accuracy at low currents)

Anyway, having problems with the MOSFET: I've just killed the IRFZ44 I was using: failed open and took the

(and with the bang it made when it happened, I'm just glad I was wearing the brown trousers!)

Anyway - I'm assuming I finally killed my MOSFET from overheating: It's screwed to a 0.8 deg/W forced-air heatsink - so I reckon that/s about 1.5-2 deg/W once you take into account the TO-220 package and the thermal paste and whatnot.

Can anybody suggest a MOSFET to replace it with? As I said, I was using an IRFZ44 (salvage) up until now; which worked fine until it died

I could also do with a thermal cutoff: which I'm pretty sure I can bodge together using a thermal switch in series with the load (and bolted to the heatsink); will that do, or shall I go the route that entwiz took and drive it from one of the op-amps?

[cprobertson1]

Sorry - double-posting!

There's something not right here: the MOSFET keeps failing closed at the higher end of the power ratings: any idea what's going wrong?

The case isn't overheating (that's not to say the silicon die isn't!) - and its below its rated 94W (in fact, it should only be running at 50W!) at 3 amps (which is split between two parallel IRFZ44N).

Even worse, it doesn't seem to fail when its on the breadboard: but it does when it's on the veroboard which is confusing me more than a little: as far as I can tell the wiring's the same in both (at least when I compare my scribbly schematics!)

Any idea what I'm doing wrong or if theres a common cause of failure? (Operation-wise, it's practically the same as Dave's design at it's core; I'll make up a non-scribbly schematic if you want one though!)

[EntropyWizard]

The first thing that comes to mind -- I think the tab is connected to drain. Are you by any chance grounding the drain through the heat sink. For this application the MOSFET needs to be insulated from ground.

[cprobertson1]

Good shout! I'm almost certain it will have a good connection to the heatsink! Will insulate it and get back to you!

[cprobertson1]

Hmm - nae luck!

Shoved in a sil pad with a nylon screw and it still died: confirmed it works on breadboard though: same load (20V @ 3A) - works fine on the board, when its in the veroboard it dies within a few seconds.

Weird - must be something to do with the veroboard prototype........... guess I'll be sketching it all out!

--EDIT--
Quick update:
Just tried starting at no power dissipation and cranking up the current limiting on my power supply, then increase the voltage from 0-30V: increased the current and repeated. Seems to kill the mosfet above 15-20 volts, if its passing more than an amp.

--EDIT--
Added attachment with circuit diagram.

I'm not going to lie: it has me confuddled! It should be within spec; I just don't get why it works on the breadboard but not the veroboard! I dissipated 6A at 30V on the breadboard and it was fine. Running a little hot obviously, but still running.

I take it this shows there must be something up with the wiring on the veroboard version?

Ps - there's also a dropper resistor on the gate - forgot to put it in the schematic!

[cprobertson1]

AHA!

I figured it out a week ago, and finally managed to test my hypothesis today...

The IRFZ44N I had were fake; a genuine one survives 20V at 3A: but just to be on the save side I've also swapped it out with a IRL3705N (higher current rating) and it can run fine fine at 30V/3A (and equilibriated around 80 degrees at the junction after an hour of running - I'm a fan of testing my designs to death )

Also added a 5V relay on the logic side so that it cannot apply a load of the main circuitry isn't on (I was origianlly going to just use a DPST switch but I didn't have any with high enough current ratings) and a thermal cutoff for 100 degrees strapped to the heatsink

It's certainly looking good - just need to get the front panel precision drilled at work (one of the perks of working in a CNC machining plant ) and I'll get the finished pics up!

[langwadt]

AHA!

I figured it out a week ago, and finally managed to test my hypothesis today...

The IRFZ44N I had were fake; a genuine one survives 20V at 3A: but just to be on the save side I've also swapped it out with a IRL3705N (higher current rating) and it can run fine fine at 30V/3A (and equilibriated around 80 degrees at the junction after an hour of running - I'm a fan of testing my designs to death )

Also added a 5V relay on the logic side so that it cannot apply a load of the main circuitry isn't on (I was origianlly going to just use a DPST switch but I didn't have any with high enough current ratings) and a thermal cutoff for 100 degrees strapped to the heatsink

It's certainly looking good - just need to get the front panel precision drilled at work (one of the perks of working in a CNC machining plant ) and I'll get the finished pics up!

if you look at the datasheet for IRFZ44N the SOA is only barely 3A/20V at  Tc=25'C

[cprobertson1]

 

Wish I had noticed that! I had checked that a few times as well; wonder how I missed the temperature rating - let's pretend I was tired at the time and didn't just do something stupid

The pack I got off ebay for £2 for ten were definitely showing some weird problems as well though - certainly compounding the problem of being outside the SOA! They had a habit of randomly saturating and then oscillating back down to where they were set to (I had originally thought it was something wrong with my driver circuit) - problem vanished when I got proper parts of Farnell though!

Oh well - lesson learned! I still don't believe I REPEATEDLY misread that though.... I must have read the Tj temp instead of the case temp and ignored the temp rating when I looked at the chart on subsequent occasions.

Inexcusable! 

[ZeTeX]

Dont buy specific parts from ebay, pay some extra to get all the components from a good distributor.
Also your latest schematic seems kinda weird, you should put a capacitor directly on the 7805 output pins (the more the better, combie electrolytic with ceramic).
C4 should also be increased to 100nF - if you are going to order 100nF ceramics better order 50 or so, they are cheap and used pretty much used for every project, compare to 10nF.
IC1B seems useless, are you doing it because you have an extra op amp and you want to use it? if yes then its fine.
R1 is useless (?).
Also, you should take your compensation *directly* from the output of IC1A, and not from R19.
A pull down resistor (~100k) at the gate to ground might be a good idea, it helps keep the mosfet off.

I can recommend https://www.taydaelectronics.com
they seem to sell original parts, didn't had problem with parts from them.

[cprobertson1]

I usually buy from farnell or RS components - except they charge you extra for orders under £20 - lesson learned!

Also your latest schematic seems kinda weird, you should put a capacitor directly on the 7805 output pins (the more the better, combie electrolytic with ceramic).

Lol - weird in general or just some weird choices here and there?

IC1B seems useless, are you doing it because you have an extra op amp and you want to use it?

Good guess

R1 was taken from Dave's original design - I had assumed it was for current limiting and didn't give it a second glance!

Compensation from IC1A: aye, that was a mistake on my part - will change it round next time I'm in the lab.

I wonder if that had something to do with the ringing that occurred with the ebay junk mosfets... they would abruptly saturate at 4-6 second intervals (but only below 10V) - then they would drop back into the ohmic region and oscillate back down to the whatever set point you had it at.

It looked a bit like it was ringing to be honest; the genuine parts don't do that though (even if you abruptly send it into saturation and release it.


Thanks mate! Much appreciated The problem with being a hobbyist is that there doesn't tend to be a lot of critique of your designs; if you're doing something dumb, unless it stops your design working you're not likely to even notice it let alone learn from it!