Amiga ATX power switch

[mikolas]

Hi,

Has anyone tried to implement the ATX power switching circuit here: http://www.ianstedman.co.uk/Amiga/amiga_hacks/Amiga_Power_supplies/body_amiga_power_supplies.html

I did it by the book but for some reason, the schematic simply did not work, ie. pushing the power button did nothing. After checking and double checking with multimeter, I accidentally turned on the PSU and found out that removing the 100k resistor and replacing it with a wire, the circuit started to work.

I'd just like to know if anyone has had success with the original schema, as I did not find any errors on my protoboard connections.

[redrumloa]

Not this one, but I have converted ATX power supplies for Amigas before. This method seems to be a bit excessive.

[Mugo]

The R4 (100k) discharges C1 (100nF) when you press the button. When the Pin3 of IC1A reaches 0V, the Q (Pin5) goes to low level (0V) to turn the PSU on.

If your 100nF capacitor is defective, the circuit can reacts in a wrong way.

Also, note that you should keep the button pressed for some seconds (or some fraction) until the circuit reacts. This time is determined by the C1 and R4 values.

Try to reduce the R4 value to 47k and below, until you find a satisfatory timing.

I don't recomend to use a wire in the place of the R4.

[rkauer]

 It depends on what you want with the ATX power supply:

 - Use the ATX PSU to substitute the original brick?

 If the answer is yes, then you don't even need the circuit! Just carefully open the PSU, locate the green wire on the board and route it to a pole of an on-off switch.

 The other pole must be connected to ground (any black wire) to the PSU start to work.

 [color=ff0000]BEWARE: the innards of the PSU have high voltages! If you don't have experience with switching circuits, at least waits half an hour before opening the PSU case![/color][/b]

- Feed an Amiga inside a tower?

 Like Marcelo told, adjust the resistor (R4) and capacitor (C1) network to match your needs, if you want to use the circuit inside a tower.

 Or to use the circuit with a pico-PSU inside the desktop Amiga.(Hint!)

[mikolas]

Mugo wrote:

If your 100nF capacitor is defective, the circuit can reacts in a wrong way.

Also, note that you should keep the button pressed for some seconds (or some fraction) until the circuit reacts. This time is determined by the C1 and R4 values.

Try to reduce the R4 value to 47k and below, until you find a satisfatory timing.

The capacitor was not at fault as I used multiple different ones and no luck. I also thought the the resistor was at fault. However, one could also simplify the whole schematic if the switch is already debounced as I think ATX power switches are - I might be wrong though.

Mikko.

[mikolas]

Tried different resistor sizes, still no luck.

This makes me crazy as the schematic is really simple and I see no reason why it should not work. The debounce in this case actually should work so that C1 charges fast in order to generate a clock pulse (clk is posedge) and it really does not matter how fast it discharges as negative edge has no function at all. Of course R4 reduces ripple from the button as it reduces the discharge speed so you would not get posedge triggered by accident.

All in all, this still does not work unless R4 is removed altogether and I can assure you the protoboard is correctly wired. Maybe my 7474s are borked and sink too much current.

[motorollin]

I don't understand the need for all this circuitry. Surely you can just take the required voltages from the ATX connector (see here), then connect pin 14 to ground with a latching switch. I did exactly this and it worked perfectly.

--
moto

[mikolas]

@moto: I want to be able to use the original ATX power switch in order to make a perfect case mod :-)

[taunusand]

You can just replace the switch with a regular one, this way you can still keep the original button in the front of the cabinet :-)

[Hans_]

@redrumloa

Why is this circuit excessive? It's a single chip with a few resistors and capacitors.

@motorollin

There isn't much in the circuit. A single d-latch (IC1 a) toggles the ATX power on/off pin whenever someone pushes the button. A capacitor and resistor (C1 & R4) ensure that you can't rapidly switch it on and off due to the switch bouncing.

The second d-latch (IC1 b) is unused, but you have to tie its inputs down, or the latch will switch on and off like crazy and waste power. You can't buy a single-latch chip.

The other capacitor is a decoupling capacitor to ensure a stable power supply to the chip (yes it's necessary) while the other resistors simply pull the input voltages to the chip up to 5V.

Hans

[Hans_]

Looking at that circuit again, I don't think that tying all the +5V pins together is a good idea. It's definitely not necessary for the power switching circuit, and your Amiga won't need that much current either. These pins could be connected to separate +5V outputs from within the power supply and are supposed to supply several amps each (IIRC). Any disparity in +5V from any of these will cause currents to flow between the multiple +5V outputs. At the bare minimum, it's a waste of power.

Hans

[mikolas]

Hans_ wrote:
Looking at that circuit again, I don't think that tying all the +5V pins together is a good idea.

I intend not to tie the 5V lines together but instead use single 5V for the Amiga and power the rest of the system using molex connectors from the PSU. Only exception to this rule is feeding more power to the motherboard (for BPPC) using the floppy connector.

[mikolas]

Well well well. I finally decided to check out the actual board file with Eagle. It seems that the PNG is outdated and the Eagle file has pretty much the same circuit (apart from one 4K7 resistor instead of 10K) that I made. My hunch was that the resistor was not needed as 74LS74 clock is Schmitt-triggered.

[Hans_]

mikolas wrote:
Well well well. I finally decided to check out the actual board file with Eagle. It seems that the PNG is outdated and the Eagle file has pretty much the same circuit (apart from one 4K7 resistor instead of 10K) that I made. My hunch was that the resistor was not needed as 74LS74 clock is Schmitt-triggered.

The resistor on the clock line is needed to pull the input up to 5V when the switch isn't pressed, so it is necessary. It is also part of a low-pass RC filter (R2 and C1) that prevents the switch bouncing from triggering the clock input multiple times for a single press. All switches bounce a little, that is, the contact will bounce up and down for a few milliseconds, causing the signal to fluctuate.

The value of R2 is pretty flexible though, so 10K should still work fine.

Hans

[mikolas]

@Hans_: Yes I meant you don't need a resistor for capacitor discharge, as it is also missing from Ian's proper Eagle file. A pull-up is of course necessary to keep the clock line high when the button is not pressed.