How to build an adjustable current dummy load?

rg12

100 kW
Joined
Jul 26, 2014
Messages
1,584
I know that for different loads I can play with different resistors with different ohm ratings and using them in parallel or in series.
* Not looking for adjustable bulb load or whatever, just resistors.
I was told it's impossible to control the current with a pot as the pot needs to carry the whole load, BUT I was just discharging a few cells with my Turnigy balancer that can go up to 5A Discharge at 8S and I can digitally select the desired amperage in 0.1A increments.
How does that work?
 
Take a look at this thread: https://endless-sphere.com/forums/viewtopic.php?p=1418180

The circuit there is one way to do what you're asking. I used a variation of that circuit years ago to make a variable load for testing batteries. In my implementation I used a microcontroller to control the load and provide an LCD interface where you could set the desired current. You could also control this circuit manually with a potentiometer.
 
rg12 said:
I know that for different loads I can play with different resistors with different ohm ratings and using them in parallel or in series.
* Not looking for adjustable bulb load or whatever, just resistors.
I was told it's impossible to control the current with a pot as the pot needs to carry the whole load, BUT I was just discharging a few cells with my Turnigy balancer that can go up to 5A Discharge at 8S and I can digitally select the desired amperage in 0.1A increments.
How does that work?
The normal way to do this is with a MOSFET with feedback. (MOSFETs are natural current sinks, but feedback makes them a lot more accurate.) Check out these options:

https://www.edn.com/design/power-management/4321777/Power-MOSFET-is-core-of-regulated-dc-electronic-load

https://www.barbouri.com/2017/01/21/linear-mosfet-electronic-dc-load-part-1/

https://www.codrey.com/electronic-circuits/simple-electronic-dc-load/
 
Ok so I understand that the load is not resistors but transistors so the question is, can I have something that can go up to 150A? not constant but would need to have that for a few seconds on about 100V
Constant drain would be sometimes 100V at 20A
 
The FET ratings will let you calculate how many you need in parallel. You need to stay under the continuous current rating and the dissipation rating. Also good to have about 2x or more headroom in the specs.

The real trick is keeping them from burning up. You need a massively large heat sink and good thermal coupling. For very short durations, just a big block of aluminum will do. For longer runs, you need big fins and a fan.
 
fechter said:
The FET ratings will let you calculate how many you need in parallel. You need to stay under the continuous current rating and the dissipation rating. Also good to have about 2x or more headroom in the specs.

The real trick is keeping them from burning up. You need a massively large heat sink and good thermal coupling. For very short durations, just a big block of aluminum will do. For longer runs, you need big fins and a fan.

Do you know a way how to calculate how many and which fets for my requirements?
I think that the max continues will be around 1500w
 
There are dozens of possible choices. I usually pick a few and figure out which one will be the cheapest.

Since I have the datasheet handy for the IRFB4110, I can use that as an example.

VDSS 100V
RDS(on) typ. 3.7mΩ
max. 4.5mΩ
ID (Silicon Limited) 180A 􀀀
ID (Package Limited) 120A
Maximum Power Dissipation 370W @25C

So from the datasheet specs, you won't come close to the maximum current spec but the maximum power dissipation is rated at 370W.

So to dissipate 1500W with no safety factor, you would need 4 in parallel. With x2 safety factor, 8 parallel.
If you have a 100v source, you could see 1500W dissipation at 15A if the load was a short.

FETs with a larger package can dissipate more. A IRF100P218 comes in a TO247 package and is rated for 556W dissipation.

Since this is a load resistor, you don't need low Rds or really high peak current. In fact, you could also consider IGBTs.
A IXXH110N65C4 is rated for 880W dissipation.

You could also consider IGBT modules pulled from a EV inverter. Something like a 1DI300Z-120 is rated for 2kW dissipation.
 
Something doesn't seem right.
How can those 8 little 4110's dissipate 1500w while 8pcs are half the size of a single 100w resistor which gets hot AF with only 100w applied to it?
1500w is a huge amount of heat and it physically doesn't seem possible to dissipate through those little fets.
I have a 1500w resistor and it weighs 2.5kg and half a meter long...
 
Yes, you are right, the dissipation rating is based on a heat sink temperature of 20C. To get rid of 1500W and maintain a 20C heat sink requires a massively large heat sink with air or water cooling.

I may be missing something in the thermal calculation too. I know the little parts can dissipate a lot for a short burst but for continuous operation they might need to be derated more. The inverters in EV cars throw off a lot of heat like this and generally use water cooling.

Another approach is to split the dissipation between a giant resistor and the FET bank. Put the giant resistor in series with the FETs and the dissipation will be shared between them. Pick the resistance to give the maximum desired load when the FET is fully on.
 
rg12 said:
* Not looking for adjustable bulb load or whatever, just resistors.

You understand that an incandescent light bulb is exactly a resistor, right? Even when it's used at lower than its intended voltage and thus doesn't emit light.

It seems to me that no matter what you use as a resistive load, you could use a DC brushed motor controller to PWM whatever current you like, up to the amount that the load will accept. A big fat capacitor for smoothing would probably help make the current smoother and more measurable between the cap and the load.
 
Chalo said:
rg12 said:
* Not looking for adjustable bulb load or whatever, just resistors.

You understand that an incandescent light bulb is exactly a resistor, right? Even when it's used at lower than its intended voltage and thus doesn't emit light.

It seems to me that no matter what you use as a resistive load, you could use a DC brushed motor controller to PWM whatever current you like, up to the amount that the load will accept. A big fat capacitor for smoothing would probably help make the current smoother and more measurable between the cap and the load.

Yes I guess you are right but I rather not go with lots of bulbs...
Im very interested in the fet solution...
 
Here's an example of a commercial Chinese product:



Discharge current: 0.20-9.99A stepper 0.1A or 0.01A
Discharge Current Maximum Error: 0.7% -0.01A
The Maximum Capacity Test Error: 0.5A 2.5%, 2A 1.5%, 5A and above 1.2%
Offline ( termination ) Voltage Range: 1.0-25.0V stepping 1V or 0.1V
Discharge Voltage: 1.00-30.00V
The Maximum Voltage Measurement Error: 1% + - 0.02V
Maximum Power: 60W super power automatically limits the maximum current (for example, up to 60W when it can open 9.99A 6V and 20V maximum at the only open 3.00A)

You can see a single big transistor on a heat sink with a fan. Says maximum continuous 60W. So imagine using 10 of these or 10 FETs on a much bigger heat sink with a fan and you could run 600W.

It may be possible to just get one of these and graft it onto a much larger power section.

Other large commercial load banks I've seen apparently use big resistors, but they are electronically switched to keep the load constant.

It's not a bad idea to consider using a large motor as a load, especially if you have one lying around. A motor driving a big fan will have a high continuous rating if the fan is blowing on the motor. I know I have a few lying around in the garage. Big brushed motor with a cheap brushed speed control would make a great variable load.
 
Any pc type fan and some finned aluminum dissipates a ton of heat. Easy purchases on ebay, or even your local electronics store.

I wonder if the application required is to save space or more of an interested in building something.
 
I need this for discharging packs and also for stress tests so it will need to be able to run very high amps at very high voltages for periods of 30 seconds or so.
Talking around 100V and 180A for maybe 10 seconds and 100V 100A for 30 seconds.
Constant will be around 1500W
 
rg12 said:
Talking around 100V and 180A for maybe 10 seconds and 100V 100A for 30 seconds.
Constant will be around 1500W

That's a pretty tall order. I don't think transistors are going to be an economic solution at that level. You need a giant resistor.

I found several of these on eBay. $50.
https://www.ebay.com/itm/TH34268-U-...819327?hash=item2c7ffc8fbf:g:pC8AAOxy4YdTPqrN

Load Bank Resistor.jpg
Cat# TH34268
Watts: 3333
Volts: 120
Add. info: 14" U Shaped

These are 120v rated, so will be slightly less power at 100V

I have used 1500W hair dryers for a load and it worked fine except for the switch. The switches are not made for DC and will arc when you try to turn it off, so I just left the switch on and unplugged it to turn it off. Hair dryers can be pretty cheap. You could put a bunch in parallel. You might need a gigantic knife switch or contactor to turn it off.
 
rg12 said:
I need this for discharging packs and also for stress tests so it will need to be able to run very high amps at very high voltages for periods of 30 seconds or so.
Talking around 100V and 180A for maybe 10 seconds and 100V 100A for 30 seconds.
Constant will be around 1500W
Yep. You'll need a very large resistor bank. Light bulbs also work but have a high inrush current which can hose things up a bit.
 
If you want to dissipate 10 or 18kW for short durations for cheap then just coil up some steel or stainless wire and drop it in a bucket of water.

You can calculate what length of wire you need and then if you use an over-length coil and (screw)clamp connections you can fine tune the load if you need.

Commercial load banks are either fan-cooled resistors or plates immersed in water tanks.
 
I already have 10 huge 1500w 5ohm resistors I can play with.
I'm after the digital variable version of it.
How many fets will I need? (the bigger ones not small 4110's...)
 
I'm sure there's a way to calculate it, but I don't know how. The big TO247 packages have dissipation ratings over 500W. But the key is temperature. If you can manage to keep the heat sink at room temperature, then you can run the full rating. Most large EVs use a water cooled heat sink. For this kind of power, it's probably the best. If you waterproofed the wires, you could dunk the heat sink assembly into a big bucket of water and it should not get much hotter than 100C.

Here's a datasheet for a random IGBT, chosen for lowest cost.
https://www.mouser.com/datasheet/2/308/FGH60N60SMD-1306000.pdf

FGH60N60SMD — 600 V, 60A
600W @ 25C
300W @ 100C

This one is a MOSFET
https://www.mouser.com/datasheet/2/205/DS100228A(IXFH160N15T2)-347934.pdf

IXFH160N15T2
150V, 160A
880W @ 25C

In looking at these, it seems the Safe Operating Area graph will be the limitation more than the dissipation rating. The IGBT has a slightly wider safe range.

Here's the graph for the IXFH160N15T2



For DC, you use the lowest line. At 100V, it can barely do 1.5A each. For very short pulses, it can take more.
So to run 100A continuous (or for more than a second), you would need 66 of these in parallel. They are $7.78 each (but less if you buy that many).
 
That sounds crazy, how come my BMS of 100V (fully charged 24S) and 150A has 7 fets (dunno the model, the big ones, not 4110 size).
 
rg12 said:
That sounds crazy, how come my BMS of 100V (fully charged 24S) and 150A has 7 fets (dunno the model, the big ones, not 4110 size).

Because the FETs are either fully on or fully off and never asked to dissipate much. When you run a transistor in the linear region, the heat dissipation gets crazy.

On most commercial load banks, they just use big resistors. Fancy ones have an array of resistors that can switch on/off electronically to maintain a nearly constant load.
 
Back
Top