A picture of the Torch Triple X now non-functional due to a poorly limpet board. The top cover is not on and the mouse has been despondently placed inside.

It's official. My Torch Triple X is no more.

There is some fault with the limpet board which I can't track down. All RAM is OK. All tracks are OK (or repaired).

I suspect a failure of one of the PAL chips which did get quite warm. Ultimate unobtanium.

Without it, the Torch won't boot.

Arse. 😢

a close up of stitch from lilo and stitch making a face Alt: a close up of stitch from lilo and stitch making a face

Oh. My. Gosh. The Torch is playing with me now. I had about six hours of flawless operation. Then I started compiling uemacs code. Two minutes later, kernel panic with memory management error.

And I'm back to where it was before. At startup, a single flash from the limpet and then nothing.

Gahh!

A picture of the Torch Triple X display having just booted. A kernel info window is showing in the middle of the screen with some icons to the left.

A picture of the Torch Triple X in 'hardware maintenance mode'. The centre section of the machine is lifted like the bonnet of a car and wedged open with a TV remote. My home made DRAM tester is on the right side of the keyboard.

Annnd were back in the room. I have spent hours going over this bloody computer.

Beeped out every trace on mobo and limpet. Nothing missing. Then I moved the black plastic down a bit on the new long pins to keep them more together.

Booted first time. 🙄 That bloody connector... 🤬

But, yay! 😁

A picture of the defunct Torch Triple X. The limpet board is propped on the open case with the keyboard in front. My tears decorate the carpet. 😢

F*CK. F*CK. F*CK. F*CK. F*CK.

Torch is dead. After battling intermittent RAM issues for months I think the Limpet board has died. My last resort is to de-socket the motherboard RAM and see if it's actually the sockets causing issues there.

#bugger

A picture of the Torch Triple X terminal window. It shows the commands formatting a floppy disk image and the creating a filesystem. Then there's a command to copy all the files from the directory with the uemacs source code to the Gotek for closet analysis on my Linux system. Didn't get a picture but about six seconds after I took this, the system had a stream of fault traps followed by a 'double kernel panic', the screen went green, then white, then blank. I had a double panic myself and then turned it off and on again and it's been ok since.

Couldn't work out why I couldn't mount the disk image on the Gotek once I'd formatted it.

I am an idiot. Should've remembered that 'format' is low-level and I still needed to 'mkfs'. Doh!

But now I can copy files and create disk images til the cows come home. Moo.

I think I used something that sounds just like that in my HNC. I want to say they were called 'Emma' boards? Or something like that? 🤔

A picture of a DRAM chip from the Torch limpet board. Pin 2 is folded up which presumably happened when I was reseating the chips last time. This is why I'm ditching the sockets...

A picture showing the side of the Torch motherboard with the limpet board mounted to it. There are long gold pins connecting the limpet board at the top to the motherboard at the bottom. These pins replaced the original repair which consisted of several turned pin sockets. The issue with the limpet is NOT this connector.

Uemacs porting efforts to the Torch are on hold due to a persistent RAM fault. I am removing the sockets I installed on the limpet and putting the chips directly on the board as they're a liability (the sockets).

Things like this don't help though...

Darnit. I'm 99% certain the Torch issue is the limpet connector.

The question is, do I persevere with the constant kernel panics and crashes or do I install a fixed arrangement that is permanent?

🤔

It would make maintenance a pain in the arse. But, if I'm right, it would solve the RAM issue.

Compiling code on the Torch could be said to be 'torch-ure'. 😁

(I know you're out there, I can hear you breathing.)

A picture of a 20 pin turned pin ic socket on a desk.

A picture of the Torch Triple X with its centre section raised up like the bonnet of a car. It is held up by a white TV remote. The limpet board is on the bottom portion of the case to the left with a lit red LED showing the memory being accessed.

Found the cause of the memory issue In the Torch (more or less). I'd used multiple turned pin sockets stacked to create a connector to the limpet. But with the motherboard and limpet removed it was obvious I'd used one too many.

With it removed, the Torch boots far more reliably...not 100% though.

A picture of a black 3d printed block approximately the size of an IC socket but much taller. Wires extend from both sides of the block.

A picture of the Torch OpenTop environment. A terminal window is open showing several 'make' commands. A kernel window is in front of the terminal window showing a kernel panic due to a memory management failure, almost certainly down to the limpet board connector not making good contact.

Now with added wire..

A picture of a 3d printed block in a similar shape to an IC socket but much taller. A series of holes are drilled in the configuration of a 20 pin ic.

It's extended round pins that fit into a turned pin socket. But the original top connector had two broken pins so I hacked it with a few turned pin sockets. Given the issues I'm working on a soldered version with a 3d printed block. 🙂

a man is hammering a piece of wood with a mallet ALT: a man is hammering a piece of wood with a mallet

Compilation on hold as I've got a bit of a hardware issue. The top connector of the limpet board is just not making good contact so keep getting memory failures. Trying to think of a permanent solution...

A picture of an attempt to compile uemacs on the Torch Triple X. The file named fileio.c has generated a number of errors.

I struggled with vi on the Torch. There's only one thing to do. Compile uemacs on the Torch. In the process of modifying the code to K&R formats I've started to like vi, rendering this task pointless. But now I've started I need to to see if uemacs will work. Why do I do these things to myself?

Stop breaking down already! *Sheesh.* After the debacle with the full height MFM hard disk and the exploding power supply (see here ), I decided to come at the Cifer from the oth...

Bloody hell. The Cifer keeps insisting on breaking.

crashedfiesta.blogspot.com/2026/03/stop...

POP goes the ... ...47 ohm 3 amp current inrush device A short post today.  So, I had a major Code Brown moment with the Cifer. I have been trying to get th...

Another blog. I'm spoiling you this month.

This one contains magic smoke..

crashedfiesta.blogspot.com/2026/03/pop-...

Putting the SCART before the (composite) horse. Torch Triple X - Part 12 (Reposted) One of the major issues of vintage/retro computing as a hobby is that a lot of our beloved equipment is just wearing out. When th...

You wait ages for one blog on an ancient obscure UK Unix workstation from the 1980s and then two turn up at once (sort of).

Parts 12 and 13 of my escapades with the Torch Triple X:

crashedfiesta.blogspot.com/2026/03/putt...

crashedfiesta.blogspot.com/2026/03/gett...

A picture of the inside of the Cifer power supply. A black disc like component is in the centre of the image but to left centre of the disk is whole about 1mm in diameter with a grey residue. I suspect this was the magic smoke generator.

I'm assuming that a 3A current inrush thingamajig shouldn't have a burn mark and a hole in it.... 😬

A picture of my properly dead Cifer computer. There are circuit boards on the desk and the Cifer is on its side. The offending hard disk that I think drew too much current to startup sits at the back of the desk looking smug.

Ah. Well that was bad.

While trying to power up the original HD in the Cifer there was a loud crack, a flash, and the familiar smell of magic smoke...

Bugger.

Also, Brown Alert. 😬

Not even sure if the Xebec is working to be honest but I don't have anything else to put it in to test!

My bluescsi.ini file.

A picture of a Xebec 1410 controller card resting on top of a monster Rodime MFM hard disk.

A picture of a Rodime MFM hard disk. It's very gold with a black front. Surprisingly for a drive this age it still spins up and the light on the front declares "drive ready".

My 2nd (or 3rd?) blue SCSI. A blue circuit board with a black 50 way SCSI style connector on one side and a small green raspberry pi Pico about half way down. Brilliant piece of kit.

Trying to get this ancient SASI drive to read but not sure BlueSCSIv2 supports it in initiator mode.

Rodime RO200 series (10Mb I think) with Xebec 1410 card.

This is from the Cifer 1887 and may have a very early Unix on it. My other HD from it died. 😢 Anyone got any tips?

#bluescsi

A picture of the Cifer on its side but switched on. The screen is displaying the correct bottom options and random text that was typed in to verify the keyboard still works.

A picture of the Cifer computer on its side but switched on. The screen is displaying random characters that have been typed to verify the keyboard.

A corrected picture of the Cifer on its side but switched on. The screen is displaying the correct bottom options and random text that was typed in to verify the keyboard still works.

Hmm. The retro gods have a sense of humour. I tested the Cifer ps and voltages were all ok. So I put the first main board in and, guess what...

It now works again. 🤨🥳

Just need to re-flash the Gotek and get the other boards back in.

Looks like a Meanwell RQ-125B would do the job for Torch and Cifer (probably a bit overkill for the Cifer to be honest). And not too expensive at £33 from Mouser.

Just looking for the right model. 🙂

A picture of the back of the Cifer computer. The top cover has been removed showing the back of the dead hard disk and the CRT.

A picture of the back of the Cifer power supply. The case is black with a large yellow warning triangle showing dangerous voltages inside. The supply provides 1.1 amps (RMS) at 230v.a.c.

A picture of the low voltage side of the power supply. There are eight terminals for various voltages, ground and two not connected.

The Retro gods giveth, and the Retro gods taketh away.

The Cifer is dead. I suspect the power supply. The Gotek wouldn't turn on, the screen stayed resolutely dark, and the power supply was making unexpected noises (no magic smoke though).

Sad face. 😔

A picture of the Torch Triple X OpenTop desktop system in colour. The kernel window is open and displaying the kernel version as 2.11 with 1 megabyte of memory available.

A picture of a small green circuit board. It has two integrated circuits and three cables emerging from the back. These are video in from the Torch, SCART out to the TV and USB power in which drives the chips on the circuit board.

Another Torch Triple X success. Yay! My heath-robinson SCART lead now works properly.

The picture quality is TERRIBLE but that's likely down to me using two CD4011s to combine HSync and VSync. But it's proper colour!

Once I get some more time (2027 maybe?) I'll try and improve it..
#torchtriplex

A picture of the OpenTop windowing system. A terminal window is open showing a series of commands that created a new node for the attached Gotek drive and then mounted the Gotek successfully.

Oh. That was easy.

I took the unix file from the original Torch XXX system (which is a later version) and copied it to the OpenTop image.

Reboot.

Run 'mknod' as per the pic and the Gotek is just there, ready to mount... Crikey.

#torchtriplex

OK. So, I'm going to be a Grandad.

Which is nice. 😁

A picture of Harry, judging me for enjoying a tasty snack while failing to provide him with a tasty snack.

#FeelGoodFriday