High Current Switching with little/no contact resistance

Thread Starter

Metalfan1185

Joined Sep 12, 2008
170
I have a problem designing a small project of mine.

I have the need to switch the outputs of a COmputer ATX Power supply. Its about 200W and it's purpose is for testing motherboards.

the problem is, Im using shunts to get a small Voltage the reads back to a Atmel Arduino to scale it and provide a Current Value to display on a Grahing LCD.

If I use a Relay to switch the PSU ouput between the Board under test, or a Dummy Load (a few 5 Ohm, 5 Watt Resistors arranged to draw about 1A from each of the Voltages 3.3, 5, 12). The relay contacts will have some resistance each time they open and close and that will throw off my current readings. I need a Solid-Contact way to switch <=10A on the 12V and 5V side with little/no resistance from contacts.


Any help is appreciated
 

Thread Starter

Metalfan1185

Joined Sep 12, 2008
170
Thanks for the suggestion, but I was thinking of omitting the relay completely because when the contacts close, it might read 75 mOhms, then they open and close again and may read a different value afterwards (Say, 60mOhms) would this change the voltage drop across the shunt (which is in mV) enough to throw the value off?

How about physically bumping the unit? contacts may shift and lessen the resistance?

I say this because the value will be displayed on a graph, so...the relay (if one was used) would have to Ensure good (excellent) contact in the switch.

I know such a thing exists as a Solid State Relay, but ive never used one. Could that be a solution?>


again, thank you for your input, i appreciate it!
 

Thread Starter

Metalfan1185

Joined Sep 12, 2008
170
hmmm, thats an idea.....

does the Rds resistance stay reasonably constant as the Load changes? Going to need some huge Heatsinks...

So i would control the gate with a switch and a resistor to +V and the Drain would go to the 12V on the Output ATX connector (Motherboard/Resistive load Side) and the Source would go to +V on the Internal PSU? ill draw something up and post it when i get home later
 

rjenkins

Joined Nov 6, 2005
1,013
As long as the relay contact is not 'inside' the shunt resistance (ie. between the two points you are reading the shunt voltage, it's resistance will not significantly affect the current readings.

I would be very wary of switching the PSU connection to the motherboard though!
Connecting an already-charged power supply will cause very high current surges as the capacitors on the board charge up, which may blow any series diodes.

Some multi-voltage components can be damaged if ie. 5V appears bore 12V, even of only for milliseconds.

Most computer PSUs also have extremely sensitive overload detection, and you woul proabably find the the connection surge would shut down the supply anyway..

Controlling the power supply itself, so the outputs charge (relatively) slowly is the only safe way to do it.

You should be able to isolate the power-on control line from the motherboard 20/24pin connector fit a relay in line with that to enable power-up or force shutdown.
(The motherboard logic grounds that line to switch the PSU on & releases it to turn off).
 

Thread Starter

Metalfan1185

Joined Sep 12, 2008
170
OK, I kinda like the MOSFET idea, and i have a bunch of these N channel TrenchMOSFETs from Phillips layin around. THanks for the advise about the slow powering on...I would never switch them without disengaging the power from the applicable PSU first, but maybe i can use the Standby wires as an alternative...More research for me i guess :)

I have some question from the data sheet as to what voltage i should use to drive the gates. But other than that, i attached a drawing (from paint lol) of what i would assume is the correct way of using these MOSFETS for my project.

Datasheet for my MOSFET, PSMN035-150p
http://www.nxp.com/documents/data_sheet/PSMN035_150_SERIES.pdf

Datasheet for the TIP122 (or TIP127 if i use N Channel, specs are similar) for driving the gate of the MOSFET from 5V output from microcontroller
http://www.fairchildsemi.com/ds/TI/TIP122.pdf



Finally, the drawing (only showing one voltage, the 12V Rail)
 

Attachments

Thread Starter

Metalfan1185

Joined Sep 12, 2008
170
Looking now i realize that i havent been very clear with what im trying to do..


Im trying to build a Unit that will allow me to test PSU's and Motherboards. Aside from the external PSU, there is an Internal one that allows the testing of a motherboard / peripherals with no PSU present...(Powered from the internal one)

there will also be connectors on the unit for testing PATA, and SATA Power connections, floppy drives (ok, well, maybe not floppies) and a section for 3/4 pin fans that report back the fan speeds.

Id like to be able to watch Voltage and Current via a GLCD that displays a Graph (driven by an arduino and code from yours truly) so i can watch for spikes visually.

The Dummy load resistors have 2 purposes,
1)a load to power on the internal PSU for powering things with no MOBO present
2)a load to test an external PSU that is conencted to an ATX connector on the unit.
 

SgtWookie

Joined Jul 17, 2007
22,230
If you're going to use an N-ch power MOSFET, you will need to get the gate 10V higher than the source terminal to turn it on.

Your current scheme won't work unless you can get the gate higher.

You might consider using a P-channel power MOSFET; it will be easier to turn on/off.

Something like this one:
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=SPD50P03LGINCT-ND
Datasheet: http://www.infineon.com/dgdl/SPD50P...90004&fileId=db3a304412b407950112b42afaac4437

Use a 1k resistor from the gate to the source terminal to keep it turned off.
You can simply use a 2n2222 to turn it on.

Here's what I'm talking about - ignore the part number of the MOSFET through:



The 10 Ohm resistor acts as a "snubber" to keep the gate from ringing due to the C of the gate and the L of the wiring/traces.

That circuit will work as-is for both your 5v and 12v supplies.
 

Attachments

3ldon

Joined Jan 9, 2010
82
My advice is to stick a .001 ohm 1% resistor in the ground leg, then switch out load(s) with relays/fets. Thus meausring current will not require temperature compensation.
 

Thread Starter

Metalfan1185

Joined Sep 12, 2008
170
Ok, well im considering using P CHannels like sgtwookie said, But i would have to order them...which is not to big of a deal, but i have tons of N channels lyin around lol

I also found that my internal PSU also has a +24V output on it. So there is the option to switch the gate with the 24V with a TIP122 Transistor.



Im also looking into the PSU wires to see if it's a possibility to switch between them with the power on signals...


The only problem is, i dont believe it's a good idea to connect the outputs of both PSU's together, because even if only one is on at a time, couldnt it damage the other? if not i think it would at least put an extra load on the PSU that is on at the time.
 

SgtWookie

Joined Jul 17, 2007
22,230
Just because you have stuff lying around doesn't mean it's the best tool for the job; otherwise we wouldn't have anything but big hammers. ;)

You would need more than a TIP122 to turn on the gate(s); you would also need a PNP transistor to pull the gate to +24v. This is do-able because with the drain at 12v, the source would also be at 12v (less the Rds(on) drop) so Vgs would be around 12v.

You would also need a method to turn on the 5v. 24v-5v=19v, and that's too close to the maximum limit of Vgs +/-20v.
12v-5v=7v, which is not enough to fully turn on the MOSFET.

You would really need Zener diodes from gate to source in order to ensure the gates stayed within about 12v of the source terminal; and corresponding current limiting resistors.
 
Last edited:

Thread Starter

Metalfan1185

Joined Sep 12, 2008
170
lol i know you cant always design things around what's lyin around...:rolleyes: lol but i try my best ;)

Anyways...This is getting complicated...maybe I am going about this the wrong way...

(btw...still thinking about the P channel MOSFETs, but looking into other ways too)


Maybe there is a way that i can connect the outputs of the power supplies together and switch them on and off remotely... I would just need a way to isolate the outputs from each other while each one is on individually. (they should never be on at the same time.)

so the +12 on one connects in paralell to the +12 on the other, and then it switched between a load or motherboard. Now I would need to disconnect one PSU while the other is on, and vice versa.

Oh, and i installed Multi SIM on my laptop, so i dont have to make any more awful drawings from Paint :D
 

t06afre

Joined May 11, 2009
5,934
Can you explain more how your measurements is throw off(post examples if you can). Turing on and off equipment will create current surges, as mentioned. This surges are not caused by the the small "on" resistance in a relay.
What you need is to use your Audrino unit in a smarter way. Then you are turning equipment on/off you stop the current measurements for a short moment, until the system has stabilized. You could implement this a simple push button to the Audrino ,one push on, one push off. Then use a output line+transistor to control the relay
 

Thread Starter

Metalfan1185

Joined Sep 12, 2008
170
I do not have examples unfortunately...

Its not the idea of watching currents while switching...The problem lies in comparison

When a relay closes contacts (like a switch) there is a small resistance in the contacts. This resistance is different each time the switch closes (slight differences in force applied to hold pressure to the connection inside ther relay)

The shunt im using to measure the current only has a slight voltage drop (in mV) and this change in the contact resistance of the switch will throw off that voltage, throwing off the current readings. This is what im trying to aviod. I would like to make it as solid-state as possible for reliable current reading comparisons.
 

Bernard

Joined Aug 7, 2008
5,784
I know you do not care for relays but consider mercury wetted reed relay rated for 10A. Contact resistance almost un-measurable & consistant. I have one somewhere. ' have used Hg wetted rotating contacts for mV signals in well logging with no noise introduced.
 

SgtWookie

Joined Jul 17, 2007
22,230
You're going to have a hard time trying to find mercury-wetted switches or relays nowdays.
There is no way to make them RoHS compliant.

If you DO find some, better buy extras; because they'll just get harder to find. There is no longer a commercial market for them.
 
Top