650W test load for current source power supply?

[Flyback]

I must make a resistive SMPS test load which is to be 650W maximum.

The Load will have a 50KHz sinusoidal current in it which is 1.9 Amps RMS.
This is a constant , regulated current.

The voltage will be 342V rms.

The load needs to be comprised of about 13 individual resistors. We need the ability to be able to short out each of these resistors so as to be able to vary the load.
We are thinking of using relays to short out each resistor.

I am told that the load must “stand up” and must have fans.

The resistors should not be inductive, and indeed in the load, we wish the voltage and current to be in phase.
I have been advised to use planar resistors.

https://uk.farnell.com/h-s-marston/cf2-0816-0805-1500na/heat-sink-0-13-c-w/dp/936420

I am thinking of using something like the above to fasten the planar resistors to. Do you know the best way to fasten the resistors to the heatsink, and other details?

 

[alec_t]

The spec specifies the heatsink mounting as 'adhesive', but the channels at the edges look as though they could take self-tap screws so you could use clamping plates screwed on.

 

[ronv]

Yea, I think they mean the fins are epoxied in. It looks like the mounting plate is 1/4 inch thick or there abouts. If you used 15 you could use 2.7k resistors, but the high voltage will force you to large packages for the resistors so they probably won't fit on the heat sink. It's going to be expensive!!!!! The only ones I found with a quick look were 800 watts at 5000 volts @ > $100 each. The smaller TO220 packages don't seem to support the high voltage. Maybe you can find some with higher voltage ratings than I found.
Edit did find this one for $30 each, but they are still over an inch square.

https://www.mouser.com/ProductDetai...1Q0vjWL0FqZSTlV0YI4lM0B%2bowr2g1IDvdQi18iKQ==

 

[Flyback]

the resistors will be 13.85 Ohms (or nearest available value) each and in series. 180 ohms total. This is a current source power supply.

 

[ronv]

Ahh, That removes the voltage problem so you can just use the TO220's in series. So I would just drill & tap the holes into the heat sink.

 

[Flyback]

drill & tap

i cant seem to find heatsinks with 'guttering' so that drilling is not necessary prior to tapping

 

[JoeJester]

I would make the load resistors parallel with you OPENING a relay contact to remove the load. That is normal for higher power loads.

They make wire resistors to test 400 kW generators.

All you need is 13 ... 50 W resistors in parallel. The cooling could be easier as well. Ideally you would have approximately 2.1k resistors. You might have to settle for 2k or 2.2k ... 2k will have power of 52.5W each or a total of 682W. This does not preclude you from using some 2.2k to make the load max out at 650W.

 

[ChrisP58]

I have to agree with Joe. Shorting out resistors in series would make the resistance go down and the power go up.

Also, the resistors don't need to be the same value. I have a box for testing 100W inverters. There are switches for 10W, 20W and 2x 40Watt load values. Decide what power increments you need to test at and choose the values accordingly.

 

[ronv]

Glad I wasn't the only one to miss it. It is a constant current supply that he drives LEDs with I think. So the current is always 1.9 amps and the voltage varies.

 

[Flyback]

Glad I wasn't the only one to miss it. It is a constant current supply that he drives LEDs with I think. So the current is always 1.9 amps and the voltage varies

...Thanks, thats absolutely correct

 

[MrAl]

I must make a resistive SMPS test load which is to be 650W maximum.

The Load will have a 50KHz sinusoidal current in it which is 1.9 Amps RMS.
This is a constant , regulated current.

The voltage will be 342V rms.

The load needs to be comprised of about 13 individual resistors. We need the ability to be able to short out each of these resistors so as to be able to vary the load.
We are thinking of using relays to short out each resistor.

I am told that the load must “stand up” and must have fans.

The resistors should not be inductive, and indeed in the load, we wish the voltage and current to be in phase.
I have been advised to use planar resistors.

https://uk.farnell.com/h-s-marston/cf2-0816-0805-1500na/heat-sink-0-13-c-w/dp/936420

I am thinking of using something like the above to fasten the planar resistors to. Do you know the best way to fasten the resistors to the heatsink, and other details?

Hello there,

What's the budget looking like?

They make non inductive resistors that have two bolt holes. The resistors are attached to the heat sink with bolts through the bolt holes. The resistors also need thermal paste between resistor body and heat sink, and that is on one surface of the resistor only so it's quite easy to apply. If you cant find them let me know and i'll do a search.

The ratings go up pretty high, at least 100 watts for one resistor. I have a few 50 watt types myself that i used for my home made freezer defroster.

 

[Flyback]

The budget fot the whole unit would be about £250 or $379 USD

 

[ChrisP58]

Glad I wasn't the only one to miss it. It is a constant current supply that he drives LEDs with I think. So the current is always 1.9 amps and the voltage varies.

...Thanks, thats absolutely correct

What kind of LED runs directly off of a 50KHz sinewave power source?

 

[MrAl]

The budget fot the whole unit would be about £250 or $379 USD

Hello again,


Chris:
Good question

Flyback:
Well, since your load has to be varied so it can be used to test the power supply at various power outputs, then one idea is to connect all the resistors in parallel and switch them in and out as needed.
For example, if you connect twenty-six 4.7k, 50 watt resistors in parallel, each resistor will dissipate 25 watts which is a safe operating point, and you can switch them in or out to vary the load from 0 to 650 watts in steps of 25 watts each. I would think that is sufficient. So that's 26 resistors, each with a rating of 50 watts, and each with a resistance of 4700 ohms.
Now on a site like Mouser for example, they sell "chassis mount" resistors like this for as low as about $3.50 each (USD) so you could purchase 26 of them for under $100 (USD).
If you would like more steps (like 12.5 watts per step) then you'll need more resistors, and of course more switches too.

Another question is what kind of switches did you intend to use with this to switch the resistors in and out. Most switches are rated for 50/60Hz not 50kHz. At that frequency or 50kHz it might be better to turn the power supply OFF before switching the load. After the load is set to the required test load value, then turn on the power supply.

One note here is that if 25 watts per step is too small, then you can go UP to about 35 watts per step using one 50 watt resistor per step, but you shouldnt really go above about 35 watts for a 50 watt resistor. If you wanted say 50 watts per step (13 steps) then you should go to a higher power resistor like 75 watts or 100 watts.

Another point is that you will probably want to mount the resistors on heat sinks inside the cabinet. You dont have to use only one heat sink for all the resistors, you can use panels of heat sinks with each panel having it's own fan for example. Heat sinks are mandatory for these kind of resistors though, unless you run them at less than 1/2 power (like 1/4 power).

One last note:
Make sure the resistors are non inductive. If not, look for ones that are non inductive. Some of these parts have the same part number for non inductive type, except for a few extra letters like "NS" or something like that. So you'll have to make sure you get that type.

 

[Flyback]

if res's are in parallel then the relays wil need to be much higher voltage rated , which gives a problem.

 

[blueroomelectronics]

What kind of LED runs directly off of a 50KHz sinewave power source?

These things. isotera.com

I guess the OP is copying the design.

 

[ronv]

Take a look at these.

https://www.mouser.com/Passive-Comp...0x88kZ1z0wn7mZ1z0wn9hZ1z0vu8tZ1z0wn8tZ1z0wncf

 

[MrAl]

if res's are in parallel then the relays wil need to be much higher voltage rated , which gives a problem.

Hello again,


Im sorry i dont follow what you mean here. How does the resistors being in parallel vs in series change the voltage rating of the relays?
I guess you are using relays to switch the resistors rather than switches, maybe so it can be an automatically controlled load tester?
If you can turn the relays 'on' before turning the power supply to be tested 'on' you can get away with a cheaper relay.

 

[Flyback]

How does the resistors being in parallel vs in series change the voltage rating of the relays?

because if the resistors are in parallel, then the relays will have the full voltage across them when off...i take it you mean the relays are in series with each of the paralleled resistors?

 

[MrAl]

because if the resistors are in parallel, then the relays will have the full voltage across them when off...i take it you mean the relays are in series with each of the paralleled resistors?

Hello again,


Ok then you will have to show the circuit you are talking about. You'll have to show a diagram about the way you wish to connect the relays to the resistors.
That's because normally it does not matter how you connect the resistors. If you connect them in series and keep them connected, then you still see one of the relays subject to high votlage....maybe not the FULL voltage, but close enough that it doesnt matter that much.
So show your diagram and we'll take it from there.

What model relay did you plan on using anyway?