AC Electronic Load Circuit

[eric74]

Hi, I am looking to make an AC load 110-240V (0.1A to ~ 10A approx.) at first 1 phase only, then I will upgrade it to a 3 phase if possible.
Basis:
1.Full bridge rectifier
2. N-channel Mosfet for the load.

Problems:
1. Will the current be linear from AC to DC? Reactive power presence?
2. Mosfet heat dissipation controled?  add a temperature sensor on the mosfet to control the heat ramp?

What do you think about the this rough schematic principle?

[exe]

That's 2.4KW for the poor fet and bridge rectifier . I'd say a boiler or water heater with, say, and auto-transformer would be a better solution, but not sure there are auto-trafos for this power.

[dtmouton]

Not a good idea to design your load like this.

Here is the main problem:
You will dissipate a large amount of power in the MOSFET. This will require a whole bank of MOSFETs with a massive heatsink.


The most elegant solution will be:
Two back-to-back connected DC-to-AC converters. One that draws a sinusoidal current (or more or less any waveform you want) on the input and another that injects a sinusoidal current into the grid. By doing this, you don't have to dissipate the power in your system. This type of systems requires a lot of knowledge to design and control.

My recommendation:
A bank of resistors that you switch in and out with relays.

[eric74]

I Have a few bridge rectifiers KBPC5010 that can handle 50A and FCH023N65S3 transistors that can handle 75A load max.

The only problem I can see in that design is that heat ramp can be very fast if it's not well controlled, then of course it must have a huge heatsink with fans to dissipate.

Transformer is not an option.

[samofab]

The biggest problem with proposed circuit is that there's no control of current. Typical schematic for DC loads has a current measurement resistor and opamp feedback. And the second problem is that the mosfet in question must dissipate a lot of power. Depending on the heatsink you'll be able to dissipate less than 100W per mosfet (that will be 100K temperature rise at very optimistic 1K/W and you need to derate mosfet for temperature, too). On the plus side, it seems that the mosfet actually has SOA defined for DC operation (not all of them do).

Then there are "higher level problems" such as how to control the MOSFET in a way to achieve a power factor close to 1.

As always, it's much easier to start playing around with (much) lower power. Take a transformer and see if you can make 1A AC load at 12V.

[magic]

It can take 75A when drain-source voltage is very low.
No TO-247 device will ever handle 10A at 240Vrms - that's 2.4kW of power. You would need tens of them in parallel.

Maybe you could use such FET to PWM some 2.5kW heater, but it would take decent filtering to keep the PWM frequency from leaking to the mains supply (and also preserve close to unity power factor). I wouldn't have a clue how to design that

Another option is a PWM heater with an active PFC circuit and capacitor in front. APFC tends to have naturally "smooth" input current due to it being a boost converter, although all implementations I have seen in computer PSUs and such still employ LC filtering to meet EMI regulations.


If you don't need a lot of load current variability, just take a bunch of high power resistors / heaters of various resistances and switch them into the circuit as needed - that's the only simple solution at such power levels.

[eric74]

Interesting point of view. BTW I was trying to search in google some schematic of this kind of device but no success..
This topic/device seems to be like a some kind of black magic.

[Zero999]

The schematic is not simple.

The power should be dissipated in cheap resistors, not expensive MOSFETs.  If you use the MOSFET to switch a resistor and filter the current with an inductor, like a buck converter, the load resistance can be controlled, using PWM. The MOSFET gate drive circuit needs to be isolated from the control circuitry.

[Jester]

I made a closed loop binary one with many many 100W power resistors. AC and DC current to 30+Amps and 0-250V. I used an old very long HP chassis with the resistors mounted to heatsinks with lots of blow-through airflow. It's a bit like a blow dryer out the back when you crank it up!

It works reasonably well for what it is. Regulating the loop with high currents becomes a bit tricky because of the resistor tolerances, drift with temperature etc. especially when the counter activates a large number of simultaneous resistor changes at the critical transitions.  I used SSR's to switch the low current paths and back to back FET's for the high current paths.

I guess I forgot to take a picture of the finished project with the covers off, here is a really bad image while being developed, the white panel on the left fits in the opening of the chassis on the right, you can see some of the resistors along the LHS of the chassis. (top in image).

Update better image

[dtmouton]

You can also add a "smaller" version of your original MOSFET idea in parallel with the resistor bank to do the fine regulation. One problem is that the relation between the voltage across the MOSFET and the current through it is non linear at low voltages, i.e. near the zero crossing of the voltage.

[Jeroen3]

Hi, I am looking to make an AC load 110-240V (0.1A to ~ 10A approx.) at first 1 phase only, then I will upgrade it to a 3 phase if possible.

At what power factor do you want 10A?

Traditional way:
power - array of power resistors.
inductive - array of inductors, or a locked induction motor with adjustable rotor angle.
capacitive - array of capacitors.

Modern way:
PHIL Power-Hardware-in-the-Loop simulators.
eg: https://www.opal-rt.com/power-hardware-in-the-loop/

[eric74]

The idea of the AC load is to use it as a calibration tool, for any AC project such as energy meter, etc..
For sure the PF must be as close to 1 as possible.
An array of resistors + fan will be easier than control an array of mosfets, also it's simplify the heat dissipation voltage and load control. The challenge here is the load feedback in order to have a very stable thing.

[Yansi]

To calibrate a 10kW power  meter you do not need 10kW load. Use your brain again and think how you can fool the meter to see a bigger load.

[eric74]

Yeah I know.. a simple function generator is enough to do that, but I also like to simulate a real-world application just to confirm that all is correct

[Zero999]

The idea of the AC load is to use it as a calibration tool, for any AC project such as energy meter, etc..
For sure the PF must be as close to 1 as possible.
An array of resistors + fan will be easier than control an array of mosfets, also it's simplify the heat dissipation voltage and load control. The challenge here is the load feedback in order to have a very stable thing.

Why do you need any feedback? What's wrong with open loop?

I doubt the actual load is that critical, as the current and voltage are measured with calibrated meters.

To calibrate a 10kW power  meter you do not need 10kW load. Use your brain again and think how you can fool the meter to see a bigger load.

Yes, if it measures the current using a Hall effect sensor, or current transformer, just wrap extra turns of thinner wire through it to fool it into thinking the current is higher. For example 10 turns will result in it reading 10 times the actual current.

[Yansi]

Yes, if it measures the current using a Hall effect sensor, or current transformer, just wrap extra turns of thinner wire through it to fool it into thinking the current is higher. For example 10 turns will result in it reading 10 times the actual current.

Doesn't matter how it measures the current. Use external current transformer to induce enough current needed for the test loop.

[BFX]

One idea.

What about to use battery charger with charging current regulation as a load?
And probably instead of batteries use supercapacitor bank.

I know this is usable for only limited time after that batteries or capacitors needs to be discharged.
But this is only idea. 

[fourfathom]

And of course in the circuit shown the bridge rectifier is mis-connected, and there is no ground connection to the FET source (relative to the gate), nor is there any isolation.  You might think you can rely on the neutral side of the AC as the ground (if it's single-ended), but this is a good way to kill yourself, or at least burn up your circuit.

[eric74]

That's right, I will try to buy some power resistors to do my first try. Which value do you recommend me, I think it's almost the same because I will drive a Mosfet with PWM.
Of course the mosfet must be isolated with an optocupler.

[Terry Bites]

The source is referrenced to where? Your gate voltage is not defined. Boom! No isolation from mains- death wish!

Use a  "PWM rectifier" built from IGBTs (with isolated gate drive) to adjust the load. You can do smart stuff with this and control PF and simulate reactive loads.

Or for the sake of simplicity just go old school with a manual or motorized Rheostat and a blower. Very reliable, PF certainly 1.0. Not cheap, but it will outlive you.

[fourfathom]

That's right, I will try to buy some power resistors to do my first try. Which value do you recommend me

Well, 120V / 10A = 12 Ohms, 1200W dissipation.  Use a 2400W resistor to not smoke it.

And I'm sort of joking here, just try to find a 2400W resistor and you will see what I mean.  So, perhaps a bank of 100W light bulbs, or a big space -heater?  You will need current-monitoring feedback, since the bulbs or heating filaments have an extremely nonlinear resistance.

[Zero999]

Yes, if it measures the current using a Hall effect sensor, or current transformer, just wrap extra turns of thinner wire through it to fool it into thinking the current is higher. For example 10 turns will result in it reading 10 times the actual current.

Doesn't matter how it measures the current. Use external current transformer to induce enough current needed for the test loop.

Good point. A current transformer run in reverse would do the job.

[shakalnokturn]

I use washing machine heater resistances for my AC dummy loading (most often audio power amplifiers) they're slightly inductive, are designed to handle powers close to your target  and can easily be liquid cooled.
You could experiment with a couple of those in series / single / parallel configurations to set a peak power limit (rather than slaughtering a beefy MOS across the rectified mains) use your MOS for PWM, add current sense and filtering then you can try closed loop.

[eric74]

Heater resistance is a good choice for big loads. What do you use for current sense and filtering?  shunt resistor?

[exe]

I accuracy is required, then I'd say care should be taken with resistive loads as they tend to drift as they heat up.