The post #13 circuit should have a reverse-biased diode across the relay coil to suppress the voltage spike generated when the relay is de-energised.

 

how would this be implemented with the above circuit

That is what the circuit in post #13 does.

ak

 

The post #13 circuit should have a reverse-biased diode across the relay coil to suppress the voltage spike generated when the relay is de-energised.

Doesn't need one as the voltage across the relay is decreasing as the cap is charging.

 

I'm thinking that no delay is necessary.
Just close the switch immediately and hold it long enough for the PC power to come up, and that would start the computer.
No?

Makes me nervous. It assumes that both the housekeeping and main supplies will come up gracefully. Back when, I interfaced to many different industrial supplies with remote-start, on-standby, whatever inputs. Often they did not play well, and these were supplies from "real" power supply companies. Based on that, I have no faith that a $20 Taiwan wonder will come up first time every time in any protocol that does not emulate the timings of a human interacting with a chassis.

Probably an overreaction. And, the latest version of the ATX spec might cover this situation. Been soo long since I read it.

ak

 

Makes me nervous.

Okay.
The TS can easily test that by holding down the button before and when he applies power to the PC.

 

Here's ak's suggestion of a double timer using two transistors.

EDIT: Correct file attachment.

 

You are getting pretty good at saving me from having to do my own schematics.

FETs are nice because you can use high impedance networks around them. The tradeoff is that the gate voltage range from "off" to "on" is much larger than that of a typical small-signal bipolar transistor such as a 2N2222 or 2N4401. This is compounded by two slow timing ramps in series.

One solution is to change to bipolar transistors, and I might have to do my own work on this one. Another is to add some hysteresis between the Q2 drain and Q1 gate. One added resistor should do it. Still a lot fewer parts than two 555's.

ak

I'm mildly disappointed in the lack of calls from the usual suspects for a microcontroller. I just assumed that Danko would post a half page of psoc code.

 

Those higher gate voltages are what makes the circuit work. Not sure what hysteresis you are speaking of. I'm using the relay lower dropout voltage as part of the delay.