Another issue with the D3806 power module!

If you read my previous blog, you will see that I found that the D3806 power module produced large over-voltage spikes on its output terminals when you switch on the DC supply to it (even if it's output was programmed to be off on power-up). This could send a surge of up to 10V for approx 10 mS into your nice little 3.3V arduino and blow it up! It also sent spikes when you switched it off too!

I had planned to use this device together with an old ATX power supply to make a nice variable, current-limited bench power supply.

So I decided to try to work around this issue and I designed a delay circuit. The idea is that I would connect the +ve output of the D3806 through the contacts of a Normall-Open 10A-rated relay.

When you switch on the ATX power supply (or any power supply) the relay contacts would remain open for approx. 1 second and then (after the D3806 nasty over-voltage spike had finished) it would close the relay contacts and connect the D3806 ouput to the output terminals.

The delay circuit worked great. I also designed it so that it would open the relay contacts on power-off/decay BEFORE the D3806 over-voltage spike ocurred on it's output too.

Power-on delay and power-off speedup circuit (preliminary)

The above circuit was bread-boarded and appeared to work well (the R1\C1 values needed a bit of tweaking).

1. On 12V power on, C1 will slowly charge until the MOSFET is switched on and the relay is energised.

2. On power-off, as the 12V decays, the +ve rail and the Schootky diode D3 ensure that the MOSFET turns off well before the power rail voltage gets to 4V (which is where the D3806 over-voltage spike occurs).

3. R2 ensures the MOSFET M1 gate is not floating in case the 12V power is just open-circuited instead of slowly decaying.

4. Adjust D1 so that the 5V relay has exactly 5V across it (it should work from a supply range of 11.4V to 12.6V, the  ATX 5% tolerance).

5. D2 ensures any EMF flyback does not cause any harm.

Testing

So this circuit appeared to work great. I even connected a push-button switch across C1 so I could switch on and off the relay whilst the 12V source was still on.

However, when I connected a small load (52R) to the D3806, I noticed that as soon as I enabled the D3806 output (pressing the OK button), the relay immediately clicked off and then in a second or so it clicked on.. and then it clicked off again, and then a second or so later it clicked on and then off again, etc. etc.

After a lot of head scratching, it seemed that the 12V power source from the ATX power supply was dropping as soon as I enabled the D3806 output (38V 730mA 27W) and this caused my delay circuit to reset.

ATX 12V dips when D3806 output is enabled!

The reason for this is that the D3806 does not have 'soft start', so it's MOSFETs turn on really hard and the instantaneous current draw is enormous as soon as you enable the output!

When I increased the load (decreased the value of the load resistors) then it blew the 5A quick-blow fuse that I had connected in circuit with the ATX PSU 12V rail!

To work around this issue, I simply used both of the 12V ATX power supply output rails, one for the D3806 and the other for the delay timer. (see 12V1 and 12V2 in the picture below). I also increased the quick-blow fuse rating to 10A as the ATX PSU will fold if it sees a 'short' anyway.

The only issue with this is that the voltage decay of 12V1 rail will not be the same as 12V2, so I need to ensure that there is a reasonable load across the delay circuit's 12V1 rail. A 10 Ohm 100W resistor should be enough and will draw 1 Amp at 15W.

The relay now energises well after the glitch has finished.

On switch off, the relay contacts open when the voltage gets to approx. 7V so the power-off glitch is not seen on the output.