High-Wattage Dummy Load

[gundog48]

Following what I have seen on one of Dave's vlogs about creating dummy loads with MOSFETs I would like to make something similar, although scaled up considerably. I am looking to create a dummy load to stress enthusiast PC power supplies that can go up to about 1200W on the 12V rails! So I am going to need to make an adjustable dummy load similar to the one in Dave's Vlog that can create loads of up to 100A. I have heard that MOSFETs are very scalable and are capable of this with the right about of cooling, however I would be perfectly happy with a dummy load that could do up to 70A. I will also need a unit that can do around 30A for the other power rails. If possible I would like to have test points in it for measuring the voltage of the rail while it is under load, as that is the aim of what I am doing.

Problem is, I am no expert at electronics. I would like to do electronic engineering in the future, but at the moment I am 15, know my basic electronic theory and can't afford a professional PSU loader. So really to help I will need a circuit diagram as I do not understand a lot of the more advanced side of electronics to make adjustments myself. The annoying part is that a PSU can have up to 4x 12V rails which means I may have to make 4 of these! So if you can tell me what I need and give me a digram, I will be extremely happy and able to build it.

Thanks for your help!
Jake

[alm]

Remember that a dummy loads works by converting power to heat. How about starting with designing a cooling solution that can dissipate 1200W and keep the MOSFET temperature below 30°C temperature rise or so? Look at modern high-end CPU coolers, and how they only have to handle about 10% of that. Dave did a video on thermal calculations quite a while back.

The electronics side is probably much simpler than the thermal side, since you can just put a bunch of MOSFETs in parallel to share the current (might have to do some minor changes like driving the gate(s) with more current).

[NiHaoMike]

Use a bunch of cheap car heaters.

[mikeselectricstuff]

If your voltage is reasonably predictable, You can probably dump a lot of it in metal-clad resistors on a fan cooled heatsink to take some of the power away from the MOSFET(s), or just use a bunch of MOSFETs in parallel, spread out and fan cooled.
A powerful fan will save you a lot of heatsink.   

[bilko]

If you're checking 12v power rails you could use 12v halogens arranged in banks and switchable for the load you require.

[Psi]

you can massively overrate resistors if you put them underwater

[mikeselectricstuff]

If you're checking 12v power rails you could use 12v halogens arranged in banks and switchable for the load you require.

Halogens are not ideal dummy loads due to high cold-filament current, which may trip overcurrent protection.

[gundog48]

I've been talking about this on another forum and have decided that this method is not practical. What I didn’t consider is that the simple laws of physics means that the entire capacity of the PSU I am testing (up to 1200W) will be converted into heat which needs to be taken away from the MOSFETs very quickly otherwise they die. I thought it would be easy, but then I thought of it in terms of cooling a CPU or GPU which have considerably less TDP and it would require an outrageous amount of cooling! Not to mention that I would need 5x 300W+ resistors which come in at around £20 each.

I am thinking more on simple terms now of building a dummy load based on 12V car lights in parallel so I can remove or add the bulbs to make up the load I want. I would use 100W or 50W bulbs to make up the load I want without having to worry about heat ruining the circuit! I would probably cool the bulbs in an enclosure with a few fans to make them last longer.

[bilko]

If you're checking 12v power rails you could use 12v halogens arranged in banks and switchable for the load you require.

Halogens are not ideal dummy loads due to high cold-filament current, which may trip overcurrent protection.

They are not ideal but they are very cheap. You can preheat the filaments from the 5volt supply. The OP did say he wanted to 'stress test' the PSU.

[DrGeoff]

A barrel of carbon and two copper rods might do it.

[gobblegobble]

I thought it would be easy, but then I thought of it in terms of cooling a CPU or GPU which have considerably less TDP and it would require an outrageous amount of cooling!

You don't need to keep the MOSFETs below 60 degrees celsius like you need with processors do you?
It is quite easily manageable with a VERY simple watercooling setup as such is evidently capable of cooling a kitchen stove which puts somewhat more heat than that. Check your local scrap yard for almost-free radiators.

With 100 watt lightbulbs you need to be careful of the inrush current, UV emittance and hot surface temperatures: they're not designed to be operated outside of their enclosures and the UV emittance is something to consider if they're not enclosed. Also too sudden cooling will likely crack the casing glass which is something to look out for.

Which ever option you go with, it might be better not to try running it unsupervised for extended periods as the heat output is indeed significant.

[bilko]

Why not use the waste energy to do something useful, see here
http://cgi.ebay.co.uk/12v-Volt-1-WATER-HEATER-IMMERSION-ELEMENT-solar-wind-V-/150363487613?pt=UK_Home_Garden_Hearing_Cooling_Air&hash=item23025cbd7d

Just an example, you could make a water heater

[Reuben]

The simplest way is to use a huge wirewound potentiometer.  Put this in series with an amp meter and you have a variable dummy load. 

[bilko]

Yes, normally you can but 1200w at 12v is 100A.

[timd]

I started with Dave's design and beefed it up with a bigger FET and resistors, just not to 1200W. I have 4 x 1R 25W Al Clad resistors in parallel/series for 1R total. This gives me about 50w max load.

Without doing any calcs (I'm a hobbyist, not at EE), I'd probably approach it with similar idea based on a dummy load design I saw for testing amplifiers (link below).

Start with a 2U rackmount box with mesh or perforated front and back panels.

While looking for resistors for my load, I came across the BI technologies MHP140. They are in a TO-247 package which will handle 140W and they have a 1% tolerance. You could use 10 of these at 1R giving you 1400W 0.1R. They are about $30AUD ea. You could also do similar Al clad resistors but they seem to be $100+ea  for anything over 100W at 1%. You can also pick up 100W al clad and TO-247 products for <$20 ea if you didn't want to go all the way to 1200W.

You'd then mount your FETs and resistors on several dual fan asisted 'tunnel' heat sinks inside the box at < 0.3^oC/W. You can get high current automotive FETs that will handle that current in one go, but I think it'd be better to use a few to spread the heat load around a bit. (alm suggested this already)

Your circuit would be the same as Dave's, but with several FETs in parallel. You could also do what Dave did and use a panel meter. Mount it in the front of the unit reading the feedback into the opamp to determine the current (needs adjustment as Dave used 1R and the above is 0.1R) and also add an additional panel meter so that you can see the voltage. You'd end up with a pretty professional looking unit.

The photo of the dummy load is down the page a bit:
http://www.hux.com.au/Soapbox%20Items/Gallery%204.htm

[johnwa]

you can massively overrate resistors if you put them underwater

yes, I ran some ceramic wirewound resistors this way at about 5x rated power, for an etchant heater. They appeared to handle the power OK, though this project was ultimately unsuccessful due to electrolytic corrosion issues. (I discovered that silicone sealant isn't really that impermeable to water...)

A good dummy load can be made from a discarded electric fan heater. These usually have 2x 1200W elements - about 48 ohms each. Use a croc clip to tap into the element and adjust the resistance. They are good for a couple of amps each in free air, or can take the full 5 amps if fan cooled.

[ZachFlem]

I know this thread hasnt been posted in for a while, but it seems like it's exactly what I'm after.

I'd also be looking at 100a load at 12vdc, but my limited electronics knowledge is fighting against me here =P

Is there anywhere I could get a schematic for this project (does dave need beer?)

Cheers in advance for helping out a n00b.

Zach

[Rerouter]

here is the basic schematic
http://www.ko4bb.com/Test_Equipment/DynamicLoad/ConstantCurrent.png

for higher current loads, you need to use a few beefy mosfets with a low threshold voltage, and run multiple in parrellel,

you take the op amp, mosfet and load resistor, and parrellel as many of those segments as you need to reach your rated current at 12V, (like i said in the chat box)

say you have a mosfet that is specced at being able to dissipate 150W, at 12A @12V, you would run 12 or more of these in parrellel at 120W operation or less, on a bloody good heatsink, firstly to leave some headroom, and secondly to prevent killing expensive mosfets, 

and for the sense resistor you would likely be chasing 100 milli ohms per mosfet, and for 10A per mosfet they would still be dumping 1W so 3 or 5W resistors wouldnt be a bad idea, firstly to keep the temperatures down, and secondly so the expansion doesnt cause your solder joints to crack after 3 years,

[eevblogfan]

hey

sorry  , I didn't red all the thread , but I can tell you that I wound ~1m of steel (or Iron ) wier , thouse for home usage , and he act as 2 ohm ( after heating ) , he culd dissipate above 100W without cooling , I'll recomand you to wind it the same way  heaters are ( thouse 2200W heaters for home ,with internal fan ) , you'll be able to pass about 12-15A with air cooled devise like this , or even more if you'll put all in distilled water ( you don't want HHO generator , don't you ? )  and you'll be able to run this for about 46.6 minutes untill the water will rise to 40C above ambient , with 20 liter of water , (4200*20=84,000-->84,000*40=3,360,000 (J) --> 3,360,000/1200W=2800 SEC --> 2800/60 =46.666666 ( round that to 47 ) ,

(1) thermal capacity of water (for every 1C above ambient ) --> 4200 (J)

(2) 1200W (or Js )

or just put the halogen bulbs inside something ,

or , buy 3 lots from him --> http://www.ebay.com/itm/lot-of-5-five-CERAMIC-RESISTORS-15W-12-Ohm-/360225231912?pt=LH_DefaultDomain_0&hash=item53df18d428#ht_2102wt_902

use 12 resistors in paralel , total cost =50$ ( tallk to him for buying this , he might discout you a bit )

good luck ! 


@Rerouter : why shall he use 120W mosfet ? , it's better use 4X 37.5W (cheaper) mosfet's . each module will contain lm324 ( or op07's ) and each cipable of 150W ,and , he can use 16 CPU's heat sinks (cheaper ,though the shipping can be a-lot of money , better get them from neighborhood or so )  they'll need only 1X80m'm fan per 2 heat sinks , or in other words , 8X80m'm fans , + 32 mosfets (such as this --> http://www.ebay.com/itm/50-Power-Mosfet-IRF1404-IRF-1404-Transistor-TO-220-/140635440566?pt=LH_DefaultDomain_0&hash=item20be8685b6#ht_3385wt_942 ) , it will take some place and "hurd work " , but it can be not bad at all , 

(I'm just suggesting )

or 16X this schem -->  

(Huge thanks +credit for  blackdog )

or go look at this thread(page 2 ) --> https://www.eevblog.com/forum/projects-designs-and-technical-stuff/how-to-read-the-datasheet-correct-(-2uv-op-amp-huh-)/15/

[SeanB]

I had an odd use test load which was a 2N3055 with 3 leads soldered on to it, running in a cup of ice cold water. Worked well to test the 5V rail it was used for, and I could go up to 15A without a heatsink on it, provided I remembered to check the water level and top it up as needed when it was boiling. Base current was from a 100R 20W wirewound pot, tuned for the appropriate current on the inline multimeter. Worked well enough to choose the right value for the SOT current limiting resistors on the power supply, as they were rather dependant on the gain of the output pair, along with the driver.

[eevblogfan]

I will Not reccomend that , but one day , I've chekd 12V at 40A (with quiet huge heatsink )

But only 4 IRFP064N's , and 6m'm2 wier for each one , with providing stable CV voltage from HP-3610A , it was really tuchy , +- few mV's and +- 5A were changed , agine , Not reccomended at all !

[Rerouter]

hmm, 25 of these in parrellel http://au.element14.com/international-rectifier/irlb8743pbf/mosfet-n-ch-30v-78a-to220/dp/1740785
each taking 4A, would dissipate ~47W (11.6x4) means they would be running 74 degrees above ambient (silicon is 1.11c/w and assume 0.5 for heatsink), or 90-100 degrees

your load resistances would be 10 of these in parrellel per mosfet http://www.jaycar.com.au/productView.asp?ID=RR1502&form=CAT2&SUBCATID=968#1
that will dissapate at half the resistors rating, 1.6W with a 0.4V drop

all mosfets must be insulated from each other by a sil-pad or similar, you will want a fan performing active cooling, but without selecting a capable heatsink, cannot calculate air flow rate,
the wiring between mosfets should be 2 heavy bus bars  such as http://australia.rs-online.com/web/p/busbars/4896568/
this particular one can be cut to size, so you can cut it into 4 stips of 15 (13 mosfet legs, 1 tie in and a negative sense line for op amps) and have 2 groups of mosfets, 1 of 13, 1 of 12,

note i realise it would be quite difficult to solder to these, but insulated spade terminals generally fit onto the tabs and a piece of wire connected to them is very easy to solder to your mosfets and resistors,

so far we are up to only
$29.50 (mosfets)
$11.52 (resistors)
$34.70 (busbar)
$75.72 + postage

then comes op amps which isnt all that bad, it would just depends on how fancy you want to get with them,

the rest of your price will be the heatsink (jaycar do have one that forms and air column of sufficient size but isnt on there site, believe $40-70)
the case, the fan and current measurment,

an idea for current measurement would be to use an op amp (your already buy a crap load of them) and set one up as a 25 way adder, this will give you effectivly 0-10V (0-5 if in a voltage divider) for 0-100A, (yes your lowest and highest may get funky due to the op amp, but say you drove it at 7V and used the spec sheet to allow a 0V input / output,) to a standard volt meter

that is my thought on how to approach a 0-100A constant current load in a realistic size footprint,

[T4P]

74C above ambient is running over the edge ... i think requiring 0.2C would be more likely
AND a powerful fan ...
Still it's crazy

[skyler]

What is the point of using MOSFETS if all you are doing is converting energy into heat. Wouldnt you only need a really bigass resistor?

[elgonzo]

What is the point of using MOSFETS if all you are doing is converting energy into heat. Wouldnt you only need a really bigass resistor?

The load will change with the temperature of the resistor. If you want to keep control over the amount of load, you will need some sort of control circuity. If you don't care about about keeping the load under control or constant, you would need no control circuit and using just resistor(s) would be sufficient.