Designing a PSU load tester.

[mpilchfamily]

I've started this new thread on the tester to cover the new design. Now if anyone out there has a good understanding of electronics please check all my information. I want to make sure all of this works correctly. Last thing i need is to build this thing an blow stuff up.

The new design will incorperate 2N3772 Power Transistors set up in parrellel banks similar to the other design. Each of these transistors are able to disapate up to 150W and produce a constant load of about 20A. The system will be set up to plce no more then 10A across each transistor. So having 3 of these will allow me to plase 30A on a single +12v rail. Of cource the 5V and 3.3V rails will only need 1 maybe 2 if i want to have more flexability in the loads. All the rails will have 1 exstra trasistor used to help regulate the actual load being placed on the transistors.

This is all a bit over my head right now so it will take me some time to figure it all out and get it all mapped out into a schematic. I'm just now teaching myself how to use transistors to create a variable load based off a schematic i've found that is designed to be a load tester.

KBT DC Supplies -- Power Supply Load Tester -- For Sale Used Rebuilt -- We take Trade In

As you can see in the diagram in the link the load is controled by a POT allowing you to dial the desired load in. That is then fead through one of the trasistors. As the voltage on that lead of the trasistor icreases the amount of power allowed to flow through the trasistor increase by about 100X. Which in turn feads the gate on all the other trasistors that will actually be handaling the load. All those trasistors have 10A fuses on them to prevent damage if the system is turned up too high. These trasistors are able to handle more then 10A but by keeping the max down we also keep the heat managable.

Now the good news is the HS that Blunc has sent me, from what i can see in the pictures, are already drilled to hold this kind of transistor. Looks like the one pictured can hold 8 transistors so that one HS alone holds both +12v rail loads. This works out perfectly if he's sending me more then 1. With 2 all the transistors for all the rails will fit. With each transistor costing about $2 the cost of this build drops considerably. And with only 18 of them needed i could buy all them myself. Hell its only $40 plus shipping. This allows me to have seperate voltage and amperage meters built onto each rail of the test bed. I think i would first need to buy what i need for a single rail, build it and test it to see if i want to go ahead with this design. At a max of $25 it would take to build the test part its no big deal. If it has the desied affect and works as predicted then we are in buisness.

2N3772

Now the 5VSB and -12V rails will still use a load resistor since there loads will be the same no matter what PSU is being tested. This now makes this entire setup much more compact and easier to build.

I'll have more comming very soon. There is much work to be done.

[mpilchfamily]

OK i threw together a quick little schematic of the new design so you can get a better idea of what is involved in this build. Chances are i will change the design to drop the 3rd rail and add a transistor or 2 to each of the other 12v rails. This will save a couple of transistors and the $26 for a 5th Pot.

18 X transistors: 2N3772 ~$2 Per part

4 or 5 X 1K Potentiometers: 3590S-2-102 ~$26 Per Part

4 or 5 X 100 Ohm Resistors: 291-100-RC $0.40 for a pack of 10

1 X 24 Ohm 12.5W Wire wound Resistor: RH01024R00FC02 ~$3.38

1 X 2.5 Ohm 10W Wire Wound Resistor: RE65G2R50 ~$3.66

Fuses and fuse holders will be purchased at radio shack or other retailer so i can see what I'm getting and know i have the right fuse and fuse holder.

So all of that plus connectors and i estimate total cost is at about $220. This gives allot more flexibility in the loads that can be placed on the test unit and saves almost $100 over the other design. But i also want to have digital panel meters that will display both Voltage and Amperage. I'll need 4 or 5 of those so there is one for each rail. At the very least i need to have an amp meter on each rail otherwise it is difficult to dial in the exact amperage needed for the tests. Voltage can be measured with a single quality Voltmeter. Oh and i almost forgot the .001 Ohm Power shunts I'll need for those amp meters. The $220 estimate is rounded up enough to cover those.

Well that about covers everything for now. As i mentioned before I'll have to build one of the rails and test a couple of different Pots to see which ones i want to use and if we can go with something a bit cheaper without sacrificing anything. It will be a good test to see if it will act as desired and be within tolerance to make for a quality load tester.

[MeltDown]

These are bipolar transistors you're using here. Might as well get the terminology right, the control terminal is the base. The way they work is (assuming linear region operation) that the current flowing in the collector is proportional to the base current multiplied by a gain usually referred to as "Beta" or "Hfe". The first transistor is serving as a buffer so that only a small base current is required in it to drive a large base current into the other transistors. In this way you don't have to use pots which can handle large currents. This configuration is usually referred to as a Darlington configuration.

The biggest problem with this ganged parallel circuit is that the current sharing between the transistors will be controlled by the characteristics of each transistor and the current gain can frequently vary by a factor of 2 from sample to sample (based on my recollections, I haven't used BJTs in years). You can probably handle this by trying to manually match up transistors with similar performance (easiest if you had a clamp-on type ammeter to measure the current through each one) and swap them around to get the best match. Any product based on these transistors would have a feedback system to enforce current sharing but that makes the project complexity shoot up dramatically.

Don't forget that the case of those TO-3 package transistors is also the collector, you will need to make sure not to create short circuits through the heatsink if you put multiple rails on the same heatsink. You can get isolator pads for mounting them.

[mpilchfamily]

As it stands right now do you think that i'll be able to dial in an amperage and it will stay constant or will there be fluxuation in the current? I would hope with an Amp meter on each bank that i can dial in a specific current load and it would be fine.

Now that you mention it i will have to cut those HS in half to avoid shorts. Thanks for pointing that out.

[mpilchfamily]

Also do i have things correctly oriented in my diagram and will those 12v loads actually give me a max of 30A like i hope they will? This is all new teritory for me. Never worked with Transistors before this.

[MeltDown]

It should be reasonably constant although the transistor parameters usually change some with temperature. My concern is more that you'll have one transistor taking a lot of current and overheating. As I recall for BJTs (and it's been a while) the one getting hot tended to take more current so it would get even hotter and take even more current and so on until failure, where FETs would tend to take less current as they got hotter so they were easier to parallel. But I'll look that back up, 99% of the time when I deal with transistors they're either off or being saturated (for switching power supplies) so the characteristics may be different in the linear region where you'll be operating these.

[MeltDown]

It looks like you've got the transistors all oriented the right way. For an NPN the collector (terminal without an arrow) goes to the positive voltage and the emitter goes to common/negative. They actually will work to some extent hooked up backwards but usually the current gain goes down by a factor of 10 or more. You can certainly pull up to 30A through a bank of transistors, I'll look at those specific ones later to see if you have enough connected in parallel. We have some loads at work that can take 6 kW based on this type of circuit but with more complex controls than pots, they're computer controlled for doing load profiles.

[mpilchfamily]

Shouldn't the current be divided equaly among all of them since they are in Parellel?

I just learned about how the basics of Transistors work a few hours ago. So there is allot i don't know here. I'm using the design i found at the site above and got the trasistor basics from this site.

How do Transistors Work

This may be easier being discussed in the Shoutbox.

[mpilchfamily]

Computer controled would be a nice upgrade. It would be nice to have a dynamic load that can fluxuate similar to the load of a typicla PC does and then log the voltages as it goes through the rutien to get the best possible picture of the overall abilities of the PSU.

[Blunc]

many years ago when I was working with machinery that checked circuits for proper load, the digital panel meters got the monitor voltage from shunts. these would have to be added to the circuit (duh...I know) and the panel meter can dictate the necessary ohm rating of the shunt.

[Blunc]

in case the transistor tutorial didn't mention it, you will need lots of arctic silver when you mount all the transistors to the heatsinks.

[mpilchfamily]

I already figured that. I have aquired plenty to go on either the transistors or the reisitor model. But since the transistors will be the best overall option and those HS you sent are ment for holding these trasistors its a perfect fit.

[mpilchfamily]

Well as expected i have redesigned the layout. I have reduced it to have only 2 separate 12V banks but each bank is able to provide a 40A load. On of them may even be upgraded to provide up to 50A. Each of the 3.3V and 5V banks now offer up to 30A each. The HS that Blunc sent me arrived yesterday. Problem was i was asleep and for some reason my wife didn't here the UPS guy knocking on the door. So I'll have them tomorrow.

It just dawned on me a while ago that i will have to take extra care when mounting everything into the case since the HSs them selfs will be the positive lead for the transistors. So I'll need to insulate them from the case so i don't short the PSUs. It would be so much nicer if the Cathode was on a pin and not the housing.

So here we go with the full parts list. I'll be ordering some parts fairly soon to construct one of the rails. I'll need to test it, as mentioned before, to make sure it will function as i plan it too. Meltdown mentioned the possibility for one or 2 of them continuing to take on more and more current until it fails. Hopefully the Pots will allow me to dial in the desired load and hopefully that load will not fluctuate when left for an exceeded period of time.

Parts List:

Transistors:
2N3772 Power Bipolar Transistors NPN Power Switching ~$2 Per
Needed: 18-20

Potentiometers:
3852A-282-102AL Panel Mount Potentiometers 3/4 POT 1/4 SFT 1K ~$9.16 Per
Needed: 4

Resistors:
291-100-RC 1/4W 5% Carbon Film Resistors 100ohms 0.05 ~$0.04 per
Needed: 4

Load Resistors:
RH01024R00FC02 Aluminum Power Resistors 10WATTS 24OHMS 1% ~$3.38
Needed: 1
RE65G2R50 Aluminum Power Resistors 2.5 Ohm 1% ~$3.66
Needed: 1

Shunts:
3020-01096-0 Digital Panel Meter Accessories 50A/50mV DC Shunt ~$20
Needed: 4

Panel Meters: Optional. A single clamp meter can be used and multiple panel meters be left for a future upgrade. May even be able to go with a single one and use a selector switch. Those can run about $100 so the savings won't be that big.
DCA5-20PC-6-DC4-RL-C Digital Ammeters 50mV/100mV input 3 1/2 digit red led ~$59.00
Needed: 4

Fuses:

Connectors:

Wire:

Estimated Cost: $400+ Could be about $200 if the panel meters are taken out.


I think that will cover everything. I'll fill in the rest of the list soon.

Here is the schematic just in case guest can't view it.

[Blunc]

It just dawned on me a while ago that i will have to take extra care when mounting everything into the case since the HSs them selfs will be the positive lead for the transistors. So I'll need to insulate them from the case so i don't short the PSUs. It would be so much nicer if the Cathode was on a pin and not the housing.

would you consider using MJ14002 in place of the 2n3772? I still have the transistors that were in those heatsinks. if you don't want to have to insulate the heatsinks from the casing, you could insulate the transistors from the heatsink and run discreet load wires from the cathode (I still have the screws and nuts that were used to retain and make the cathode contact to the PCB's they were on).

[mpilchfamily]

My god those things can handle a max of 60A. Acording to the data sheet on thos the Collector-Emitter saturation voltage is 2.5VDC Max. Not sure if they will work here. But all the other parts would be good to have. If you need me to i will cover shipping. I can email you a UPS label or somthing. I'm still learning about transistors so i'm not sure if those MJ14001s will work or not.

Data Sheet:
http://www.onsemi.com/pub/Collateral/MJ14001-D.PDF

[Blunc]

My god those things can handle a max of 60A. Acording to the data sheet on thos the Collector-Emitter saturation voltage is 2.5VDC Max. Not sure if they will work here. But all the other parts would be good to have. If you need me to i will cover shipping. I can email you a UPS label or somthing. I'm still learning about transistors so i'm not sure if those MJ14001s will work or not.

Data Sheet:
http://www.onsemi.com/pub/Collateral/MJ14001-D.PDF

The data sheet I have for MJ14002's lists 80VDC max-collector-emitter.

I also have a whole bunch of TIP125, quite a bit less in power handling and they are PNP.

[mpilchfamily]

Well then send me all of it. It will give me parts to exsperament with. I'm not sure if the voltages matter as much as the actual current passing through them.

Just got the sinks in my hands. A little overly packed don't you think. But these things are great. Thicker then i thought. Thanks

[Blunc]

Well then send me all of it. It will give me parts to exsperament with. I'm not sure if the voltages matter as much as the actual current passing through them.

Just got the sinks in my hands. A little overly packed don't you think. But these things are great. Thicker then i thought. Thanks

didn't want them to bust out of the box during transport.

[Krazyups]

After briefly reviewing your schematics (nice by the way!), the one thing that I would suggest right away is for the 12V rail transistors, to get some 5V zener diodes and put them from base to emitter. The transistors that you are using are rated for 5V base to emitter, and with your emitter grounded, that would put 12V across that junction, which is generally not a good idea, you might see some very odd effects from it.

Along those lines, with the 5V zeners, you should have some sort of current limiter, since the zener will shunt any excess voltage above 5V (7V total in this case), which could in turn rapidly exceed the current requirements of the zeners and blow them out.

[mpilchfamily]

Then it may be easier to use load resistors for the +12v rails and keep the 5V and 3.3v with the transistors. Overall costs will be lower then the first design and I'll be given more flexibility in the 5V and 3.3V rails. Sound like a good design?