My constant current load - 90W , 3A 30V

[Harvs]

But they don't meet your voltage requirement 

Anyway you've bought them now, so just make sure you don't exceed that 55Vds.

I would still use separate control loops and forget about matching FETs.  But that's you're choice, and I guess you've already got that board made and the single high powered resistor.

BTW, I only talked about the IXFK230N20T because it was the cheapest FET that met your package requirements.  By no means was that a one off case, it's just the cheapest I could find.

[Spikee]

@ Harvs :

Sorry for the confusion my specs are very loose ... i just made up some specs .
I was very buzy whit a college project all week but today i had time to lasercut the front panel @ the FabLab in my school.

(still have to sand of the exessive black marker ink. I should have selected higher power to engrave the text so it would be black but ... beginner fault    )

Pic:


I still need to laser out some holes and make some extra markings , but it is a start.


I have learned that seperate control loops and seperate load resistors is the way to go but i now have this pcb and i would like to use it.

[Spikee]

Hi Spikee

Buy 5 a 10 of these Fets:
http://www.dickbest.nl/index.php?_a=viewProd&productId=2164

Select the fet's, see picture


Kind regarts,
Blackdog

The results: (had to put a few 1k pots in series to get 1A accuratly ...)

https://dl.dropbox.com/u/100819959/mosfet%20scatter.pdf

It won't be super accurate because the current would shift about. But i measured as good as i could.

Fet 2 and 4 would be the best match.  I will test the rest of my circuit in a few hours.

gr. spikee

[blackdog]

Hi Spikee,

Fets 2 and 4 looks good :-)

I know its a little drifty, thats wy i told you to mount the Fet on a heatsink.
If i told you use 30V and 1.5A... No way, drift all over the place.
You dit a fine job.

Please let me see your final schematic.

Kind regarts,
Blackdog

[Spikee]

i mounted it on the heatsink but they were still a bit drifting . The fan wasn't hooked up so i did a second test whit short on time and let the heatsink cool to the same temp. I got the same results =) . I just mounted the two fets and need to find a good spot to mount the resistor.
I'm drawing a simple schematic for you.

Output schematic:
the gate resistor is mounted directly on the gate pin of the mosfet.


Thanks spikee

[blackdog]

Hi Spikee,

For the best sharing of the currents give eacht Fet a small Source resistor (0.1 Ohm)
And, don't forget to place the Fets on the good site of the heatsink!

This is the way to do it...


Kind regarts,
Blackdog

[Spikee]

Can i use the wires as source resistor ?
i can 4-wire measure it to be exact.

[robrenz]

Copper has a large resistance variation with temperature change so it would not be as good as a low temperature coefficient resistor

[Spikee]

i have one laying around but another ... guess i will have to order another one from farnell.

[blackdog]

Hi Spikee,

0.1 Ohm wire resistor... Mmmm

~ 9cm, copper wire, 0.1mm diameter.

Try it,

Kind regarts
Blackdog

[Spikee]

Most of the code is uploaded now. I still have to connect the mosfets together and screw the resistor on the heatsink

Pics:
Hello world :
https://www.dropbox.com/s/i4czyuqh3wfhxnm/2012-12-05%2022.08.35.jpg

Mosfets on heatsink:
https://www.dropbox.com/s/xdn9em9bar9b0d5/2012-12-05%2022.09.21.jpg

Mounting of mosfet:
heatsink -> thermal grease -> mica plate -> thermal grease -> mosfet

[robrenz]

0.1 Ohm wire resistor... Mmmm

~ 9cm, copper wire, 0.1mm diameter.

Try it,

That's called a fuse 

[Spikee]

It is all hooked up now =) .
i made a small mistake at the dac (upside down) and after i put it correctly it  shorted somehow after a while.
Since it is verry late and i should go to bed i just put it in pwm mode and play whit it.
I got it around 80-90W and you could feel the mosfets get hot after a few minutes (not to hot .. just hot )
the heatsink temp was climbing resonably fast (fan still on 5v , other people are sleeping) but did not get over 40 deg.c for the time i tested it.

I need to order a new dac and that will take a few days, until then i can just play in pwm mode.

gr spikee

[LEECH666]

Hi Spikee,

For the best sharing of the currents give eacht Fet a small Source resistor (0.1 Ohm)
And, don't forget to place the Fets on the good site of the heatsink!

This is the way to do it...


Kind regarts,
Blackdog

Hi blackdog,

what's the purpose of C1 in this circuit?

Sorry I don't want to high jack this thread, but I am working on my noob attempt of a higher powered dummy load in this thread. Should I add this capacitor in my circuit?

[blackdog]

Hi Leech666

It's for stability, its best mounted close to the Fet end the lower site of the 0.1 resistor.
You can keep it as small as posible if you modulate the current source with relativ high frequenties and low current's.
In every design you make you have to check the values, is my design stable?
In almost every schematic on the Internet I see no or not enough decoupling.
Decoupeling is very important!!!

Some info over decoupling and why
TI
www.ti.com/lit/an/sloa069/sloa069.pdf

Analog Devices
www.bramcam.nl/MT-101.pdf



Kind regarts,
Blackdog

[Spikee]

So i ordered a few parts:

Artic cooling freezer 7 pro rev. 2 :


Artic silver thermal adhesive :

http://www.arcticsilver.com/arctic_silver_thermal_adhesive.htm

50k 10 turn pot
12bit dac MCP4728 (samples)
2 new 100W resistor 1ohm
10 0.1 ohm 3w resistors

im gonna test how much power i can get out of one IRFP064N using a computer heatsink which has heat pipes.
I want to see how close i can get to the 200W (@25deg c. ) value using 2 normal speed fans on the heatsink.

This will be an interesting experiment .

gr. spikee

[blackdog]

Hi Spikee,


I have here almost the same material as on your picture for a test 

One tip, do not use a mica washer, mount the 2 Fets direct on the bottum of the heatsink.

Check the bottum of the heatsink if it's realy flat!!!

Chack the specs of the IRFP064
Junction to Case 0.75C/Watt, Case to Heatsink 0.24C/Watt, total 1C/Watt
Linear Derating factor, 1.3W/C

Respect the datasheet/Fets 

Checkout EEVBLOG #105 so that you NOT blow up your fets...

Kind regarts,
Blackdog

[LEECH666]

Wouldn't that potentially short the drains of the two FETs? I reckon that's a bad thing. How do you ensure insulation if you mount your FETs like that. Thermal grease?

Cheers,
Florian

//EDIT

HERP DERP, confused drain with source. The metal part of the TO-247AC is internally connected to the drain but those are connected together anyway. Pleaase forget that I asked this question.

[Spikee]

Wouldn't that potentially short the drains of the two FETs? I reckon that's a bad thing. How do you ensure insulation if you mount your FETs like that. Thermal grease?

Cheers,
Florian

//EDIT

HERP DERP, confused drain with source. The metal part of the TO-247AC is connected to the drain but those are connected anyway. Forget I asked this question.

If you dont want electrical connection you need to make sure that the past of fet #1 does not hit the metal or paste of fet #2 .
This is discribed in the artic silver guide / manual for this product.

I did some thermal calculations and you can put around 115W in the mosfet and get a junction temp of 170 deg c. (175 max) so one fet would be enough for one cpu cooler.

example:
90W from psu (30V , 3A )
The 1 ohm resistor will drop 3V ( 1 ohm *3 A )
So the mosfet gets 27V and 3A (81W)
Rt j-c = 0.75 max
Rt c-h = 0.24 min
Rt h-a =  0.3 typ. - fan on high Freezer 7 pro
(calculated from some results on the internet , seems to give around the same number)
so Rth (j-a) is ~1.3 deg. C / W
Temp ambient = 21 deg. C/W
This all results in a junction temp of 126.3 deg. C. (175 max)

It all depends on how fast the heat pipes can transfer the heat but i'm guessing its kinda instant.
If my calculations are correct  in real life you would only use one mosfet per cpu heatsink because it can utilize it fully.

gr.spikee

[T4P]

Careful about using it at 175C, 125 is so hot the mosfet can unsolder themselves and seriously running them at the edge is no good for reliability and i suspect even wire insulation melting/board lamination peeling off or a track peeling off
and be careful of the heatsink itself, that heatsink is about 0.21 at OPTIMUM settings with the fan hard on

[Spikee]

Thanks for the info ... it was kinda obvious that insulation would melt but i didn't look at it in that way 
Monday i will test and aim for a 90W total load whit heatsink and present the result here.

[LEECH666]

Ah yes T4P! Good call.

A problem this guy also faced.



Subtitle translation: ... and this is what you get when you let too much current flow. The MOSFET is getting so hot that even solder starts to get soft (again.)

And you can clearly see that he switched to some sort of screw terminal connection in the next two pictures.





Cheers Florian

[blackdog]

Hi Spikee,

Some pictures of my Active Load, i'am testing it at the moment, and adjusting it.

Full Speed!!! ~280 Watts


Top View, yes, when I'm done is the wiring decent 


Side view,


The temperature is just below 85C @ one off the MJ15004 transistors @ 220Watts total dissipation.
Heatsink + 4x MJ15004


Regarts,
Blackdog

[LEECH666]

Hi blackdog,

nice construction!

Is there any benefit in using BJTs instead of FETs?

Cheers,
Florian

[blackdog]

Hi Leech666

I tried to build him with the parts I had, and I had 20 pieces MJ15004.
I Like the TO3 form because you can get good thermal contact.

But i also use Fets a lot in TO247 package.

Regarts,
Blackdog