Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Dummy Load II

Status
Not open for further replies.
Hi Dean, those look like a possiblity. I think I will be fine though once I increase the watts and decrease the resistance...actually the old dummy load is still operating so I don't think they got TOO hot, it just always made me a little nervouse. The resistors are rated up to 275C, so they can get hot and still be fine. btw, they were getting close to 275C on my dummy load when I measured the temp!

So Dean, did you ever get around to building a dummy load?
 
MR RB...one other question I thought of, I like to test my power supplies at about 1000 watts for an extended period of time, like about 10-15 minutes at a minimum, sometimes even 1-2 hours for certain purposes...So how long would a dummy load like the one you made be able to safely dissapate 1000 watts, or is it capable of that amount of current for an extended a time?
 
Last edited:
Mr RB said:
For testing DC PSUs I just run a passive dummy load, it's basically a heap of transistors, driven by 2 darlington stages and a pot.

Hi RB, fancy meeting you in this thread.

An all passive setup has it's advantages to be sure. Originally, I was planning to have the first dummy load powered off the supply under test. But I tried it and came to the conclusion that it wouldn't work. When there is no current, the FETs are low-Z(ish), this starves the Op-Amps of power, making them not work. When the Op-Amps are not powered enough, I guess they hold their outputs high or something, I'm not sure exactly. In any case the FETs would never un-stick from low-Z and the Op-Amps would not get power. A vicious circle that tripped the over current protection eventually.

I had some theory's that I wanted to try to rectify the situation, but it was recommended that I not bother. I guess because there were other problems with driving the load off the supply, even if it could be made to work. Mainly I think the concern was that if the supply voltage sagged, then the reference would drop too. Which I think could possibly be fixed with a zener reference or a 78XX regulator. Or something like a boost converter promptly followed by a regulator. Or perhaps a system that opens the load until the circuitry is powered and ready to work. Might be worth investigating. I hadn't gotten around to trying any of it. ~Lazy™~

I also had a plan for a P-FET dummy load. The idea was that the circuit ground and the cooling pipe would be one in the same. This would allow for multiple rails with a single common ground to be attached to the same load. I was thinking of PC power supply's at the time, but really almost all power supply's are common ground with more than one positive rail. The problem I ran into was that the current sense resistors output would be referenced to each rails positive. This complicated not only the feedback, but drastically complicated how to supply stable reference to the op amps. I still haven't quite figured out how to do it yet. It looks pretty much impossible without some super special happy ninja feng shui magic that I can't even explain or spell.
 
Hi Jeremy, I picked up the heatsinks from a scrapyard complete with all the transistors, for a few dollar each.

Yes the current is adjustable, in operation it acts like a bit "adjustable resistor" with a pot and an on/off switch. (Don't forget the on/off switch, it should switch before the gain stage, the last thing you need is to be trying to switch heaps of amps on/off or pulling connectors to break the current!)

As for build costs and max watts, it all comes down to the cost of your main dissipator. I used the power transistors as the dissipator as this makes it useful for many different PSU voltages, ie can use it on 5v, 12v, 36v etc supplies. If always working with 24v PSUs you could do 75% of the dissipation with a cheap resistive load (like a jug element or huge handmade wire wound resistor?) and use the transistors/FETs to control the current through that resistive load.

With your 8 FETs on the water pipe and you said you want 1700 watts I'm worried about reliably dissipating 212W per FET! What package are these? My dummy load uses huge TO-3 aluminium body transistors which are specced at about 115W max dissipation each. Obviously I allowed for a large safety overhead.

If you are using TO-220 FETs even if you cool the tags with water cooled copper pipe the TO-220 pack thermal resistance (chip to tag) still only allows a safe 50-60W dissipation per FET. Or did i misunderstand and your are using some massive industrial package FETs?
 
"The problem I ran into was that the current sense resistors output would be referenced to each rails positive."

Use a high-side current sensor. A simple one that I've used succesfully is the LTC6102, which being from Linear Tech, you can also simulate with LTSpice.

Another solution would be a hall-sensor type, like the Allegro ACS712 which is completely isolated. Sparkfun offers a breakout board.
 
"So how long would a dummy load like the one you made be able to safely dissapate 1000 watts, or is it capable of that amount of current for an extended a time? "

This can be calculated.... but I need to know:
What is the transistor part number?
How many transistors in total for the load?
 
...With your 8 FETs on the water pipe and you said you want 1700 watts I'm worried about reliably dissipating 212W per FET! What package are these? My dummy load uses huge TO-3 aluminium body transistors which are specced at about 115W max dissipation each. Obviously I allowed for a large safety overhead.

If you are using TO-220 FETs even if you cool the tags with water cooled copper pipe the TO-220 pack thermal resistance (chip to tag) still only allows a safe 50-60W dissipation per FET. Or did i misunderstand and your are using some massive industrial package FETs?

I'm using this one:

https://www.mouser.com/Search/Produ...NF12virtualkey51120000virtualkey511-STP80NF12

It's rated at 300 watts a FET. Actually I have never gone above 1100 watts, so that would be 137.5 watts on the 8...keep in mind too that prior to a few weeks ago I always had 10 FETs so it was only 110 watts a FET. I can attest to the fact that it works since I have been using it without problems for many months (until I shorted it with my VM leads, oops). I have let it run for a couple hours at 1100 watts without any problems. I had to flatten the copper pipe and make sure I did not get a cold solder joint when connecting them, that was a little bit of a challenge because the entire copper pipe is one large heat sink. I had to use a torch to get it hot enough.

Jeremy
 
Mr RBs concerns are well placed actually. The package was never ever intended to dissipate more than 70 Watts. We are getting away with it by the sheer force with which we can cool them. Having them directly soldered to a large copper heatsink, cooled with one of the best coolants known to man, that is already precoold several dozen F below room temp, is probably buying us this extra head room. It might be better to aim for no more than 100 Watts per FET instead. If we end up making enough heat in the die(chip) no amount of cooling can stop it from melting. All things have a rate of thermal conductivity, and even with liquid nitrogen cooling, we could still easily exceed the speed that we can take heat from the system.
 
If you think it is needed, we could always run more than 10 FETs, couldn't we? Some of the resistor values for the readouts might need to be tweeked. I don't want to spend time solving a problem that doesn't exist though...not saying it doesn't, but I would just wonder because I have been using it for more than 6 months and nothing has melted, granted, I never went up to 1700 watts before. I think once I ran it for 2 hours at 1200 watts, that would be the max I have tested it at.

edit:
The temperature reading from the copper pipe always stays very cool. I put the probe as close to the FET as I could get it.
 
Last edited:
Well, we can maybe meet in the middle and say definitely no more than 120 Watts per FET. That's about twice the package rated wattage, but about half the computed wattage. Seems a fair compromise. You just need to be aware that FET's will have a lower lifespan and could "die unexpectedly" because of this.

If this was a production product, we probably wouldn't be having this conversation. It would be no more than 50 watts per FET, or no passing the red tape. The industry is so phobic of part tolerance issues that most all things are overrated to some degree, and for good reason. But because of this one can get away with pushing the ratings of a few individual parts once or twice.

Edit: And of course, yes, we can always use more FETs.
 
Well, here's my thinking on it...I am looking at it from a cost/benefit standpoint (I am an accountant after all)...so how difficult would it be to tweek the schematic to include a couple or a few more FETs versus having to replace some components more often? I don't know the answer to that question since I didnt make the schematic...I need ronv's help with that. As far as cost, the parts are cheap enough that I don't see that as any reason not to add more FETs just to be on the safe side.

So, I know you guys know more than I do about these components because you have more experience with them, but I don't get why they are rated at 300 watts and 175C if they can't really handle that. It almost sounds like you are saying the FET can handle it, but not the case...but then again, isn't the case part of the FET. I mean you can't have one without the other...so shouldn't they say it can only go to 50 watts if indeed that is all it can "safely" handle...again, not doubting you, just don't understand.
 
I think you are using this one?

https://www.electro-tech-online.com/custompdfs/2013/01/CD00003180.pdf

Max junction temp = 175C
Junction to case = .5C/watt
300 watts max
120 watts/fet = 60C rise to say 85C junction temp.
Add 45 more for the over temp shut down and it is 105C with no water.
() From your early tests the case temperature rise should be only 12 to 20 C. So I think it is okay. Not much margin or any with 8.
 
So, ronv, what if I were to push it to 1700 watts with just the 10 FETs? If I understand the math correctly, the junction temp would be half the watts per FET or 85C so it's still well below the 175C rating...is there a piece to the equation I am missing?

Also, would it be able to handle higher voltages, like 50v (still not more than 1700 watts, so the amps would go down).
 
Ron actually did the thermal calcs.
As long as the LM35 is properly attached to the transistor's case (to rapidly detect an overtemp condition) you should be fine.

Something that you must add, I didn't see in your schematic, is a fuse.
 
...Something that you must add, I didn't see in your schematic, is a fuse.

If I remember right, we did bounce that idea around for a while...correct me if I am wrong, I think we decided the current sense resistors actually acted as fuses since they would blow first.
 
Max wattage of the FET's is a small thing, the FET's being cheep works both ways. Does it matter if they melt down every so often? If the system will actually shutdown after one blows then we can just replace the ones that blow. Then again, they ARE cheep, so we could just add more from the start, as it takes almost nothing to add more. The total current/split just needs to be recalculated, and then the reference voltage needs to be recalculated. It's all grade school math and a bit more soldering.

So either way I don't think it really matters. Go with what you feel will work for you. The FET's REALLY are cooled quite well.


() From your early tests the case temperature rise should be only 12 to 20 C.

I'm just slightly worried that the thermal impedance might cause an extreme thermal gradient as the watts/cooling go up. In other words, the case is a nice and cool 20c, but the die gets over Tj Max, because the solder/tab/die can't move 200+ watts of heat through to the heatsink that quickly. Then again, I don't actually know this will for sure happen.

It just seems kinda unrealistic to let 10 pieces of glass smaller than my pinky nail produce as much heat as a space heater.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top