Contact bounce for PSU load control

[Deckert]

Hi!

I'm in the final stages of assembling my new compact variable bench supply. I use a bog-standard toggle switch to control the output load, but when examining the output on a scope, there is a fair amount of bounce during switch-on into a 100 ohm load. So I figured, let me use a relay to switch the output load, but that too is rather noisy.

I first tested with a small 150mW 5V relay. The 5V output takes just over 50us to reach steady state:



The switch-off seems to be much better:



I then tested using a bigger 5V relay capable of switching 10A at 30VDC, but the result was a lot worse at over 300us before it reached steady state:



Now, there's a lot of info available on the net on contact (and specifically relay contact) bounce and it seems to be a general problem. My questions then:

1. Are these figures acceptable for the output load control on a bench PSU?
2. What do you guys see in the commercial PSUs? Same bounce on output load switching?

I guess I'm back to just using a toggle switch - the bounce noise looks almost the same.

--deckert

[Paul Price]

If it is acceptable to you, it is acceptable!

The idea of switching on a power supply output with a toggle switch is not a good practice and unless you have a huge "contactor" around, current surges can destroy the contacts of a relay or a toggle switch in a short time.
A output enabling switch is usually turning on the regulating circuit to ramp up to the required output, so as to prevent these effects.

[peter.mitchell]

Have you considered a solid state relay/ mosfet switch?

[Deckert]

If it is acceptable to you, it is acceptable!

Indeed. Currently I "switch on" by plugging the banana plug into the PSU or by attaching the clip-lead to the breadboard contact. So I guess the toggle switch or relay might be acceptable to me after all. The compact supply is 0-15V and 0-2A, although I rarely use more than 300mA in my circuits.

--deckert

[Deckert]

Have you considered a solid state relay/ mosfet switch?

I have not considered either - I recall that solid state relays can be rather pricey. If you say a mosfet switch, you're referring to a normal mosfet transistor? Would a mosfet have low enough losses for a 0-2A, 0-15V output supply? E.g. something like the IRL520N?

http://robotics.org.za/datasheet/IRL520N.pdf

At R16 (US$1.6) for two it's rather affordable! Looks like an interesting idea, thanks peter - will definitely test it as an option later tonight.

--deckert

PS: Simple appnote here:
http://robotics.org.za/index.php?route=product/product&keyword=mosfet&category_id=0&product_id=339

[peter.mitchell]

A single one should be fine for your application; though I would use a slightly more beefier one, like this, that means it won't need a heatsink
Here is a little video on them;

[Jeroen3]

Or combine a relay with a mosfet.
Only a mosfet might be an problem, if you "melt" a mosfet it becomes a short circuit, thus permanently enabling your output.
Which makes the enable/disable button useless if you blow it. And that might create some unexpected smoke some day...

[asgard]

There is an option to have a soft-toggle, like the Powah-1 board I have been pushing around for a bit.  My original device only spec'd logic-level signals both within, and through the Powah-1.  If you use a beefier MOSFET in place of the original you run the risk of not being able to support the switching load at the gate of the switch.  For that you would need a more complex switch that inserts a gate driver between the logic signal and the gate of the load MOSFET (a beefy P-channel can be found in a SO-8 form, as well as many of the TI gate drivers I have used previously.  For a 2A load you won't probably need anything as large as a TO-220 device.  One thing to worry about here, is if your loads are capacitive or inductive then you need to have a barrier ESD diode across the S-D pins to protect the MOSFET (I have used 1N5819).

[asgard]

This might be a start for something like a soft-relay kind of switch, and it certainly can be improved-upon.  Where it might fail is due to D1, which was a cheesy way of producing a 5.1V supply for the inverters.  If you are supplying less than that the whole thing might just turn off.

[Thomas]

You already have a power semiconductor in seres with the output, right?
Can't you control that with the switch, with some filtering maybe?

[Deckert]

You already have a power semiconductor in seres with the output, right?
Can't you control that with the switch, with some filtering maybe?

It would be one way, but then it won't allow me to set voltage and current limits before I turn on the output.

--deckert

[bilko]

You already have a power semiconductor in seres with the output, right?
Can't you control that with the switch, with some filtering maybe?

It would be one way, but then it won't allow me to set voltage and current limits before I turn on the output.

--deckert

Ok, so why not use a combination.
The switch controls sends a signal to turn an output relay on, but before the relay gets the signal, the semis get an 'off' signal, time delay, relay on, semi on.

[Deckert]

Okay, so I decided to go "hybrid". I turn on the output using a switch and a mosfet in parallel and the result is more than satisfactory for me.

I use a DPDT switch - the one pole switches on the output load directly, but the other switches on the mosfet using a small RC filter circuit. Since the mosfet has a resistance of about 150 milli-ohm, it means I loose 150mV at 1A. That's where the second pole of the switch comes in, as it effectively has a zero-ohm resistance. The mostfet just carries the load for the period of the switch settling time.

On the capture below the RC filter output (eg. after de-bounce) is shown by CH1 and the mosfet switching output is shown by CH2.



--deckert

[Christopher]

Fet on the output...

Fet in parallel with the output and 0V is an idea to play with, too