Dummy Load II

[()blivion]

If you used 0.1 ohm resistors times 8 on 50A max that is 6.25A through each resistor, max voltage of 625mV, that is enough voltage in a current control loop to be extremely stable. Max dissipation is 3.9W each resistor, using eight 10W resistors and a small fan that would be reliable.

Agreed on the resistors, but then we are back to the concern that you and I brought up about realistic power dissipation for the TO-220 package, even with the amazing cooling we have on it. I don't think we should do less than ten channels at the current power density.

If we do want less than ten channels, we will probably want to go up to a larger FET package. Something like TO-3P, TO-247 or TO-264 maybe. The only problem then is that each unit will be more expensive. It may cost twice as much in FETs for the same amount of power handling. But we would end up with less channels and could rest a little easier knowing that the package can better take the power we are throwing into it.

Unfortunately, this also exacerbates the problem of dissipating more power in the sense resistors. But this all comes down to a balancing reasonable feedback voltage with lowering self heating. Which I think we can solve with better Op-Amps and the lower value tabbed resistors, short wires, regulated Op-Amp supply, good decoupling.

But I don't really know for sure, just throwing out ideas and seeing if something sticks.

 

[Mr RB]

Mr. RB, The only problem I see with the offset is when you have ask for zero with the pot. At that point you can't dial it out. I'm not sure its a problem, but () was a little concerned in one of his simulations.
...

Ahh, thanks I see now.

My dummy load does not have that problem as the pot has a dead band down low, caused conveniently by the biasing of the bipolar trasistors. Also, the load on/off switch is a life saver, it's incredibly handy and allows instant access to zero current.

 

[jocanon]

J, I think the ones you bought need their own dedicated power supply. In other words it's ground can't be tied to the system ground. Remind me what are we running off the 12 volt supply. If I remember right it is just fans and our logic. Maybe we can free it from the logic and go back to powering from the DUT.

I am only powering the logic with the 12v supply. I have 24v fans, so I got a different 24v supply to power them.

 

[jocanon]

I was wondering why the thread went quiet for a while...turns out I wasn't getting the new posts on my iPhone for some reason. I had some catch up reading to do .

ronv - I will lower the DUT voltage to 12.1 which is the rated voltage and test to see if that makes it more stable. I will probably be able to do that sometime tomorrow and will report back. I like the idea of putting more channels in the system. I also like the current sense resistors dusey52 pointed to and then installing circuit breakers. I am not against getting higher quality op-amps either. The dummy load is really dirt cheap to begin with compared to an equivalent unit at retail.

To the point ()blivion made, it will get a little crowded in there. I can do two things, get a larger case to put everything in and just make the copper pipe as long as it needs to be, or, I can put the current sense resistors and FETS closer together, or both A & B (OK, so that's 3 things ) I don't know how much this matters, or if it even does, but I was trying to get as much space between the FETs as possible inside the computer case to assist in cooling. But maybe that won't make a difference on the water cooled pipe since I am pumping so much water through it. Do you think I could safely butt them right up next to each other???

 

[()blivion]

I was wondering why the thread went quiet for a while...turns out I wasn't getting the new posts on my iPhone for some reason. I had some catch up reading to do .

The forums sometimes spontaneously decide to not send post notifications. That or it has something to do with the email system and some kind of spam detection, I'm not 100% sure to tell the truth. In any case, the cure is to make sure to log into you actual forum account and check your threads/posts manually.

Do you think I could safely butt them right up next to each other???

There is no cooling problem. Even at 300 Watts each, with the amount of water being moved, it's specific heat capacity, the thermal conductivity and so on, The temperature is only going to change by a few degrees at most. It really is quite the effective cooling setup. You would probably have to dump 10,000 watts into the thing to get a temperature gradient that was measurable by consumer grade thermometers. Then again this really depends on how fast you are moving the water. But if it's anything like my home water system, then it's moving quite good.

The real problem I would worry about is soldering them on while not shorting out parts to each other. Do you feel confident enough to maybe put some FETs on the back side of the pipe? That is how I have my setup and it works wonders. I have 10 TO-220's on one pipe and it is only 17cm long. There is even enough room for ~8 more. Another thought is to make some point up, and some point down. This lets you put some distance between the leads, while packing the packages really close to each other.

Like so...

 

[jocanon]

Yeah, those are all good ideas. I was trying to figure out how I could solder them on both sides of the pipe but just can't see how I could feasibly do it. Reason being, the only way I can get the copper hot enough to wet the solder was to apply a torch to the under side while the FETs rested on top, gravity holding them in place. I tried soldering them with a high wattage soldering iron but just could not get the large pipe hot enough. I also tried using the torch on top of them pipe but could not do it without burning the FET package, maybe your skills are better than mine . Anyway, maybe you have some tips?

 

[ronv]

(),
Maybe you could run a little experiment on the fuse thing. If you have a good sized FET maybe you could put say a 5 amp fuse and a .1 ohm resistor in series with it and turn the fet on with 12 volts on the gate with one of your PC supplies. Lets see if it protects the FET and resistor.
We could also use hall sensors for the current sense. A little pricey, but almost no power loss.
Did find a little better FET as well.
I'll look for some better op amps as well.
I think if we run the logic supply off the DUT we can make sure the 12 volts comes up after the 24 so we could also use Mr. RB's scheme if you would rather.
We should decide on the voltage and current specs before we go to far.

 

[()blivion]

Maybe you could run a little experiment on the fuse thing. If you have a good sized FET maybe you could put say a 5 amp fuse and a .1 ohm resistor in series with it and turn the fet on with 12 volts on the gate with one of your PC supplies. Lets see if it protects the FET and resistor.

OK, I'm on it. Give me a few...

 

[Dusey52]

OPA2234 looks like it would fit the bill. Available in single, dual and quad packages.

 

[jocanon]

Question, so if the FET goes out could it take the op-amp down with it? Perhaps not if there is a circuit breaker. But if it could and I had an op-amp in a dual or quad pack, could it also take the other op-amps within the same pack down as well? I don't know, but I was just wondering if so if that would be a reason to get them in single packs.

 

[Dusey52]

The circuits in the dual and quad packages are completely independent except for common power and ground. The thing to worry about would be pulling the output pin above the op amps supply rail (or below ground), they don't react well to that. As long as the fuse is on the high side of the FET I think you'll be OK.

I'm not sure what the group thinks about Polyfuse/Polyswitch but that might be an option too. They're like a little circuit breaker that resets itself. The problem would be finding one that can handle the voltage & current needed.

 

[()blivion]

high side of the FET I think you'll be OK.

I would like to do this, but sadly, we can't. The high side (tab) of the FET is soldered to the pipe. We can't insert a fuse in between it. This is part of the reason I wanted to make a functional P-FET dummy load. It would be a high side switch, so common ground, not common hot. My only problem has been the control has to be referenced from hot, not ground. This has slowed my work on it to a halt.

I'm not sure what the group thinks about Polyfuse/Polyswitch but that might be an option too. They're like a little circuit breaker that resets itself. The problem would be finding one that can handle the voltage & current needed.

I love pollyfuses for certain things. But at this power level and this particular application, I think it's a bad idea.

We really really really want an indicator that something has tripped a satiety. And we also want the system to completely shutdown if something does happen. If something fails once, it's likely to fail again, and we really don't want it to keep cycling it through that kind of torment. At these power levels, we probably won't get too many second chances before something actually breaks. Unfortunately, pollyfuses will do exactly the wrong thing with the above in mind. They are much better for devices that are low maintenance, cheep, or are intended to be semi-disposable. This is a tool though, so it's expected that one will always have the will to fix and tune it. Self correctability is not desirable IMO.



Fuse experiment results

With a 220 milliohm wire wound 1 watt resistor and a 120 watt FET...

5 Amp fuse blows instantly.
10 Amp fuse blows after a few seconds, the resistor wire gets red hot.
20 Amp fuse will never blow, the resistor wire glows red hot, and eventually blows.

With a 1 ohm wire wound 1 watt resistor and the same FET...

5 Amp fuse acted like the 10 Amp above.
10 Amp fuse acted like the 20 Amp above.
Didn't need to try the 20 Amps fuse.

The FET survived the entire experiment fully intact and functional. After a while, both resistors will go open circuit if the fuse is large enough. So, it would appear that the FET will take almost any situation thrown at it without getting damaged. And that the resistor will act like a fuse if it's value is high enough with respect to the wattage. Which is all about what I expected.

So, I believe the trick to getting the fuse to blow while sparing the resistor is to make sure the resistor is high enough wattage, and is cooled well. And finally, make sure the fuse is rated low enough to actually blow at a reasonable current level. The tabbed resistors are going to be best for the first part. They will be able to get any heat made off of the resistance element faster. This keeps them from increasing resistance/dissipation do to positive temperature coefficient. Which will in turn keep the current up, making the fuse the most likely thing to blow.

As for fuses. Illuminated blade fuses are the bees knees. Can just use female blade connectors as the sockets, and mount them on the outside of the case if you want. Might get a little jerry rigged, but it can work. Or you could probably buy a 10+ slot blade fuse block just as easily.

 

[Dusey52]

(), yeah you are right of course. I'd actually thought of the tab issue and ruled out the high side fuse myself. When I was responding to j's post I had completely forgotten about it. Brain freeze I guess.

I really didn’t think the polyfuse would fly, I just thought I’d throw it out there.

 

[jocanon]

Alright, here are the results of a couple tests. I connected the sense pins, verified that the voltage was lowered to the rated voltage of 12.15, but it did not have an effect on the oscillating voltage on the dummy load, it still jumped around +/- 1 volt or so, same as before. Second test, I used the same little volt meter as I have hooked up to the dummy load to read voltage from the PSU while it was hooked up to a charger and charging lipos at about 40 amps on 24v, near full capacity. The voltage was pretty stable, it bounced around a little but nothing like when connected to the dummy load. It would be stable for a second or two oscillating only about 1/10 of a volt at every second or so (it went between 24.7v and 24.8v).

 

[ronv]

OK, 1 last try. how about one of your 0.1 ufd caps across the 2 outside legs of the pot. The circuit is sensitive to ripple on the 12 volts. This would filter it a bit.

Interesting info on the fuse (). I have to admit I'm suprised, but it's a good thing.

I tried the new op amp, but spice didn't like it to much. I think it is the drain to gate capacitance getting to it on power up of the 24 volts. I'll play some more and post what I have tomorrow.

I added a 12 volt regulator (actually 10 volts) so the logic could run off the DUT. This eliminates the current spike on power on due to the reference voltage being there with no feedback so the op amp is hard on.

So since that problem was gone I tried mr RB's method. It seems pretty good except there is a lot of ringing (osc.) when there is ripple on the 24 volts.

 

[()blivion]

(), yeah you are right of course. I'd actually thought of the tab issue and ruled out the high side fuse myself. When I was responding to j's post I had completely forgotten about it. Brain freeze I guess.

Yeah, I know what you mean. I keep coming up with "good ideas" only to remember that the tab is hot, not ground. It continues throwing me off.

I added a 12 volt regulator (actually 10 volts) so the logic could run off the DUT.

You sure that is going to work without a boost converter?

I have tried to run everything off of the DUT before and it didn't go well. The FETs stayed off during start up and that kept the power shorted, this kept the Op-Amps off, which kept the FETs short. Positive feedback, eventually leading to tripping overcurrent.

Do you think there is something different when using a regulator? Or is it something else altogether?

 

[Dusey52]

I added a 12 volt regulator (actually 10 volts) so the logic could run off the DUT. This eliminates the current spike on power on due to the reference voltage being there with no feedback so the op amp is hard on.

Another option would be to keep the reference (or even the gate drive) clamped to ground until DUT was connected. Then use DUT power to release the clamp.

 

[()blivion]

Speaking of reference, no matter what I think we should be using a stabilized reference. Be it a three terminal regulator, or a zener, or a settable precision reference. The more accurate and stable the reference is, the better the dummy load will function.

As for powering the dummy load logic off the load, I was thinking if we try it we should do what Mr RB was saying and install a heavy duty switch in series down stream from where the logic gets it's power. Maybe a switch and circuit breaker combo. I have a ~20 Amp 12 Volt power supply that has one of these. Something like **broken link removed** but in higher amperage.

 

[jocanon]

I don't see any benefit in powering it from the DUT. Seems easy to just put a little $10 or $15 regulated 12v power supply in there, or am I missing something.

 

[()blivion]

It's convenient, and if the voltage moves around, the dummy load will follow it (which may or may not be a good thing).

But yes... other than that I don't think there is really too much benefit.