Looks like I'll be getting a Mega ST4 soon

[rubber_jonnie]

Ok, so a null modem cable is on the way, so at least I can hopefully test out the serial option on another known working machine.

In other news,my electrolube turned up today, but despite spraying the sockets and also scrubbing the contacts with a small nylon brush, the machine still wont power up. It still seems to think there is no CPU.

I will re-check for shorts but for now it looks ok.

The board ran before I removed the CPU to fit a socket. This was necessary as I wanted to fit the IDE adapter, and the CPU needs to be fitted to the adapter itself, and not on the board.

But, I did seem to have some instabilities before too, in that I sometimes had to reseat the Glue chip, otherwise it gave a similar error, in that the machine had just a completely black screen and wouldn't power on, which is what it is currently doing.

I've reflowed the Glue socket, no different either.

Moreover, there are quite a few socketed chips, more so than usual, which is suggestive of repairs, and the soldering underneath which looks different to the original adds to this thought. I'm wondering if it is unstable in some other way and by removing the CPU and socketing it, it has jiggered something that's making stuff worse.

Thinking maybe Deoxit for the sockets, they are still quite oxidised.

Any other troubleshooting tips would be gratefully received taken on board, I can scope it and also have a basic logic analyser, just need some guidance.

CPU is good as tested in another machine, and also tested with the IDE adapter.

Any chance you could knock up a Mega ST version of the Mongrel tonight @exxos andpost a PCB out overnight? No? Worth an ask.

[exxos]

Any chance you could knock up a Mega ST version of the Mongrel tonight @exxos andpost a PCB out overnight? No? Worth an ask.

Yeah, will do it after I am done with the MITX version.

You know MEGA ST's are terrible machines, probably the worst alliteration of the whole ST line.

You could do how I start my repairs in general when this type of oddness happens, and just use sweaty fingers and just push on various CPU pins and keep pressing reset and see if the thing boots. Then go around the CPU pins one by one with X1 probe on the scope and make sure something is going on. The capacitance of the scope and cause the machine to spring into life.. its normally because the clock lines are so bad. One reason I designed the MEGA ST buffer board to help with that.

[exxos]

@rubber_jonnie Also the bus resistors are bad on the MEGA ST as well, its possible the socket just tipped it over the edge into not working. They are normally 4.7K and 10K, they should be 2.2K. That caused myself no end of issues, just like on every other ST.

[rubber_jonnie]

@rubber_jonnie Also the bus resistors are bad on the MEGA ST as well, its possible the socket just tipped it over the edge into not working. They are normally 4.7K and 10K, they should be 2.2K. That caused myself no end of issues, just like on every other ST.

Cool, I will look at the values and see if they need changing. They look original so I suspect they will need to be replaced.

I'll try the other method too with my big sweaty fingers

It's just insane that a supposedly simple change can cause mayhem like this.

I was going to do the same mod to my Mega ST1, as it has a German TOS so I'd potentially like to add in the IDE card with 1.04/2.06 in English. I may just back off that idea for now, though to be fair the Mega ST1 is a very stable machine.

[exxos]

It's just insane that a supposedly simple change can cause mayhem like this.

Welcome to my world

I've known of such things on audio equipment in the past as well. Like when I had a idea to put audio decoder chips in sockets as they were always blowing up.. then old bob said , he tried that years ago, and they just blow up really quick instead.

Its all just borderline working all the time, something goes out of tolerance a tiny bit, and then you end up with DMA issues, crashing space probes.. where will it end...

[rubber_jonnie]

It's just insane that a supposedly simple change can cause mayhem like this.

Welcome to my world

I've known of such things on audio equipment in the past as well. Like when I had a idea to put audio decoder chips in sockets as they were always blowing up.. then old bob said , he tried that years ago, and they just blow up really quick instead.

Its all just borderline working all the time, something goes out of tolerance a tiny bit, and then you end up with DMA issues, crashing space probes.. where will it end...

It makes me wonder how may mainboards across the ST/Mega/STE/Mega STE/TT and Falcon range actually required rework before they made it out of the factory and onto peoples desks.

Such small shifts in tolerance seem like they could easily cause problems on the production line.

[rubber_jonnie]

Null modem cable arrived today, and I've tested with a working STFM, and all is well with the world

I used an old Win 2000 laptop with HyperTerm and I had complete control over the diags from the remote terminal. a worthwhile tool for sure.

I guess I will test the Mega with it and see of something pops up,though I think it unlikely.

Will post any results up here later.

[exxos]

[rubber_jonnie]

Hmmm, hooked up the Mega with the diag cart and serial cable.

Pretty much as I expected, absolutely nothing, the CPU isn't running at all.

Plus, if you boot with the diagnostics cart enabled on a working ST, you get a steady red LED on the cart for ROM access, something that just doesn't happen with the Mega (Or a dead STFM I have).

Off to scope it now, and will also see how different a working machine is, will have another check for shorts too.

I have spare 2.2k SIL resistor packs ordered for the Mongrel, so will probably swap out all of those, and have ordered some more for the Mongrel, which I can't start as a load of resistors I ordered haven't arrived for some reason

[rubber_jonnie]

Ok, so I broke out the scope this evening to see what I could see.

First I scoped all the CPU pins on a good STFM and a good Mega 1 and mapped out what I saw on there against a pinout diagram of a 68k. This showed me what pins had some sort of signal, what had 5v, what had 0v and finally what the clock signal looked like.

As you'd expect, they looked the same (More or less, and to my untrained eye). I then pulled the CPU from the STFM and dropped the Mega 4 CPU into it, and retested. Once again, much the same, and the macine bootede fine, proving the CPU.

Next I pulled the working CPU from the STFM and put it back into the Mega 4. Power on, nothing, as expected.

Scope time again, and the results were quite different. All of the address and data lines showed 5v where previously dome sort of signal was present. The 5v, CLK and GND pins all showed what I expected, and the CLK signal matched what I saw on a working machine.

@exxos Is this perhaps a sign that the bus resistor values are a problem, exacerbated by the socket? What else would likely hold all of the address and data lines high? I can't imagine that I'd screwed up the socket soldering so badly it caused all the address and data lines to go high.

Any advice gratefully received, and I definitely have 2.2k SIL packs I can drop in in place of what's already installed.