IBM 5150 Mainboard Reset Circuit

[cr1901]

I have a spare 64k to 256k motherboard that I use for hardware testing courtesy of a kind donation. It would be very convenient for me to add a reset circuit so I don't have to constantly cycle the power supply, when my software/hardware inevitably crashes the 8088 (or rather, just causes it to blindly execute garbage due to lack of illegal instruction vector).

I seem to recall on modem7's website a circuit that one can add to a custom ISA card to get reset-switch functionality on a board without a reset switch. To conserve slots in the PC (which has 5, not 7.5 :/), is it possible to solder wires directly to the mainboard to create an equivalent reset switch circuit? And if so, where is the best place on the mainboard to solder wires, such as to minimize the number of wires I need to add?

 

[Chuck(G)]

1. Just use a momentary pushbutton to ground the "POWER GOOD" line from the PSU.

If you're super-paranoid, you can do two less-brutal things.

2. Insert a diode with the cathode facing the PG line on the PSU and the anode going to the motherboard side of the power connector with a 1K pullup to +5. Ground the anode side momentarily to reset.

3. Go with (1) except instead of going directly to ground from one side of the pushbutton, insert a 100 uF electrolytic shunted by a 2.2K resistor between the pushbutton and ground.

Any of these should work fine.

 

[cr1901]

3. Go with (1) except instead of going directly to ground from one side of the pushbutton, insert a 100 uF electrolytic shunted by a 2.2K resistor between the pushbutton and ground.

You mean like this? Still doesn't really solve the problem of a temporary short when the pushbutton is pressed, but since "power good" is an output from the power supply (as opposed to a similar pushbutton reset input circuit for a microcontroller), I guess that is inevitable. Also completely violates that "voltage can't change across a capacitor instantaneously" rule if I don't take into account the switch's resistance as it closes .

 

[Chuck(G)]

Trust me, it works. Usually, the PG output is just a OC (or transistor) pulled high through a resistor. A momentary short is well within the ratings of just about any gate.

If you can find a SPDT momentary-contact pushbutton, you can also switch the PG input to the board between ground and the PSU PG output.

Your choice.

 

[modem7]

I seem to recall on modem7's website a circuit that one can add to a custom ISA card to get reset-switch functionality on a board without a reset switch.

No, not on my website.

One problem with using the RESET DRV line on the ISA bus is that it would only reset a subset of the chips that get reset via the POWER GOOD line.

 

[Chuck(G)]

Maybe he was thinking of the NMI pushbutton add-on distributed with the IBM Techref kit.

 

[Plasma]

You can reset any 8088 system by shorting pins 20 and 21 on the CPU.

 

[cr1901]

No, not on my website.

One problem with using the RESET DRV line on the ISA bus is that it would only reset a subset of the chips that get reset via the POWER GOOD line.

Well, learn something new everyday... not that I knew where the RESET DRV line went in the first place :D.

Maybe he was thinking of the NMI pushbutton add-on distributed with the IBM Techref kit.

I may very well have been. I seem to recall a thread where I asked if a custom ISA card could do the trick, referencing that page.

You can reset any 8088 system by shorting pins 20 and 21 on the CPU.

Does the CPU enjoy that process? Though I have spares, I'd rather not damage my 8088s...

 

[sergey]

Well, learn something new everyday... not that I knew where the RESET DRV line went in the first place :D.

The 8284 clock generator outputs RESET signal (which is generated from /RES, that on IBM PC has POWER GOOD connected to it). That RESET signal goes directly to 8088 CPU, 8087 FPU (if installed), and 8288 bus controller. It also gets buffered for the rest of the system (ISA bus, various on-board devices), and that buffered signal is called RESET DRV.

Does the CPU enjoy that process? Though I have spares, I'd rather not damage my 8088s...

It might work, and CPU wouldn't mind. But I am not as sure about the 8284, since you'll be short circuiting its output to the ground... 8284 by the way more difficult to get than 8088

 

[Chuck(G)]

The great thing about using the PG reset is that it's conditioned by the 8284 and support circuitry. That saves you a bunch of effort getting a good clean bounceless signal out of a pushbutton.

 

[cr1901]

Reviving this thread again...

Where would be a good place to attach a pushbutton circuit on the IBM 5150 mainboard? My gut feeling is to solder a wire to the pad JUST below the X1 oscillator and above the 8284 (goes to pin 11), and choose a nearby ground pad, such as the one adjacent to pin 7 on U9. That way, in case the 8284 fails, I can replace the new one without needing to do soldering (can't speak for the typical mainboard, but my spare has a dip socket for the 8284).

Two other things:

• Does C7 (top left corner) connect to anything? It seems to only connect to the 12V line and nothing else.
• I accidentally grazed T1 (the potentiometer), and changed its position. I don't remember the original position- is there a way to recalibrate it, or is this nonessential?

 

[Chuck(G)]

I believe that C7 connects to the buried ground plane. Use a continuity tester to verify.

Don't worry about trimmer--the only thing that really used it was the CGA card in composite video mode driving a modulator. You'd adjust it to eliminate any "rainbow" artifacts due to interference with the color subcarrier.

 

[cr1901]

I believe that C7 connects to the buried ground plane.

Does that make the 5150 a 3 or 4 layer board?

 

[Chuck(G)]

I'm not sure that I remember, but I seem to have "4 layer" stuck in my mind.

 

[cr1901]

The mod is done (but not without some damage due to my carelessness- I'll make another thread for that). I have attached some poorly-taken photos, which is par for the course for me showing the mod:

The circuit shown is the same as described and shown using a schematic earlier in this thread. I have also attached a wire to 5V so I could get a test point for my logic probe. It reduces the chance the wires of the logic probe cause a short this way .

When I'm describing the mod in this paragraph, I'm assuming the orientation is such that the power connector is in the upper left corner, and the "64kB-256kB" text is upside down at the bottom of the board. In keeping with consistency with the power supply, the red wire is 5V, Black wire is GND, Orange wire is POWER_GOOD. I attached 5V to a via to next to the upper-left corner of the BIOS chip (U33). I attached GND to a via adjacent to the lower-left corner of U16. Both these vias seem to only attach to the power and ground plane respectively, and to nothing on top or bottom of the board. POWER_GOOD was attached to a via that feeds directly into to Pin 11 (~RES) of the 8284 Clock Generator. This via very conveniently exposes the POWER_GOOD signal on the surface of the board, and makes it easy to find, since ~RES is the final destination of the POWER_GOOD signal.

As described, this mod only works for 256kB boards. I'd be interested to hear about anyone getting it to work on a 16kB-64kB board.