Bit of luck.

[Dave Plowman (News)]

Offered to have a look at a MegaSquirt (aftermarket fuel injection) for a forum member who lives in Doncaster. It's on an Overfinch Range Rover which already has aftermarket injection which was causing problems - and the MS was running alongside it, the original still needed to control the auto and some other things. To add to the complication, the MS was switchable from petrol to LPG.

More or less from the off he's had problems with occasional misfires, and recently it dying all together. But always restarts after left for a while.

Everyone he approached seemed certain it was a software thing, and he spent a fortune on rolling road sessions with tuners.

He was convinced it was a hardware thing, so I offered to have a look as a very much last resort.

The MS arrived by post and worked just fine on test. Left it on all night. Phoned him this morning to say no fault found and I'd send it back Monday.

While it was still running on the bench, I switched on the workshop fluorescents (4x6ft) as it had clouded over. They're pretty ancient switch start types and put splats on the radio when starting.

And the MS died. Switched it off and on and it came good. Switched the lights off and on and it died again.

A quick check with the scope showed the problem. The output from the VR input signal conditioner should be an approx 5v square wave - which it was when running. But dropped to 1v when faulty.

One dry soldered joint later, it was fine and past the florrie test.

All I need now is a florrie replicator for my tool box. ;-)

[Tim Watts]

That has to be one obscure fault!

[Graham.]

No more obscure than my in-laws house alarm that used to go off when a taxi or police car passed.

[Bill]

Intermittent faults are great, once you have found them.

I had 3 visits to Nottingham, from Cambridge, to diagnose an intermittent radio system. Everything checked out OK on every visit. Swapped equipment and the customer still complained it was faulty. During the last visit I was stood outside the comm's building when my colleague shouted out of the door, "it's failed, no it hasn't, yes it has" repeatedly. I happened to look at the aerial, 15' fibre glass tube, on top of the tower, and saw a crow perched on it, as it struggled to balance the fault came and went.

Faulty intermittent aerial, if the crow hadn't landed when it did we were about to leave again with NFF.

I owe that crow a drink!

I just need a tame crow........

[Bill]

In message , Graham. writes

The wonders of RF sensitive sensors. :-)

Not that I would ever admit to why I know about this, but there was an RF sensitive alarm on a bank in a small town near to Cambridge 40 years ago. It took a couple of activations for the penny to drop and after that just to remember not to transmit when driving past the bank. All was quiet for a few months until early one morning, just as the day was dawning, there was a young, not bought his first razor yet, PC stood in the bank doorway eying up the passing traffic and trying to look important. Well, the temptation was just too much. He appeared to jump in 3 different directions all at the same time!

The good news is that, after testing a few weeks later, they had the alarm sorted and it no longer went off when 50 Watts erp of UHF was transmitted near to it.

[dennis]

A spark plug in a metal box connected to an aerial sticking out the box

• suitable HV supply. It will create enough interference for most things.

[Bob Eager]

Rather different, but try this one:

[Graham.]

My mobile 2m setup of the time wouldn't trip that particular alarm , but the Band 3 PMR transceiver I had for work consistently did.

[Bill Wright]

Why have a metal box?

Bill

[Brian Gaff]

Probably dry joint on an earth. One friend some years ago had aBMW that seemed to die on one stretch of road. To cut a very long story short, he told me that was a bad earth and it meant that the electronics picked up the local txi firm who had their aerial on a high tower block near the road in question.

I do sometimes wonder how many pieces of equipment have faults that nobody every finds, and instead just junks the whole unit and puts a new one in. Brian

[Bill]

In message , Graham. writes

Sometimes it can be down to a resonant circuit in the set up.

There was a rather interesting case, allegedly, of a self inflicted wound where a manufacturer fitted a current sensing circuit in a portable radio battery, for checking the charge rate. It consisted of a long zig-zag of track and the small voltage drop was measured across this. It was decoupled with a low value cap' and worked fine. Apart from, that is, when the radio was programmed for TX on a small area of VHF, here the resonance of the track and cap' caused the sensing circuit to fail.

Strangely enough I was only talking about Band 3 last week, happy days!

[Dave Plowman (News)]

The only problem I found by looking carefully was a lack of solder on the output pin of the VR input IC. The PCB is multi-layer through hole type and the solder side looked fine. The track from the IC to processor board is on the component side of the board, and the solder didn't appear to have 'syphoned' through. And all that tied in with the signal level being low at the processor with the fault present.

This particular version is hand assembled and soldered. Some others are surface mount and machine made - but obviously not so easy to modify, which this one was, and heavily. But MS dealer supplied, rather than DIY. In the main the soldering looked fine.

[stuart noble]

And hopes a crow doesn't land on it

[dennis]

To stop you getting a shock and to act as an earth plane.

We used one to find problems on system X a couple of decades ago.

I also found some odd problems when I decided to take some photos. The flash caused one of the cards I designed to crash. EMI from the flash you would think, but no a piece of cardboard over the flash stopped the problem. I found that the light from the flash was entering the microprocessor pack and causing it to crash. Military stuff in ceramic packages!!

[dennis]

We had a really obscure fault in a bubble memory card. The sense amplifier would slowly leak electrons into the substrate and then it wouldn't detect stuff properly. The cards were always no fault found when the cards came back. The electrons would leak out again when the power was off on the way back. It took weeks before the fault re-occurred.

[Bill]

Now that I can sympathise with!

We had just put the finishing touches to GB3BF, an amateur VHF repeater and decided to take a photo of it, while it was running. Flash, crash. Luckily on a power cycle it rebooted correctly. I can't remember what the processor is, but it does have a clear window in the top for erasing, so I assumed that is where the flash got in.

Until your comment above I had never witnessed it, or heard of it, happening before.

>

[dennis]

No windows in the one I had. They used something that looked like glass to seal the top and bottom so it was probably getting in there. It was an 8086 of some sort.

[Andrew Gabriel]

Very common problem with EPROMs (and processors with embedded EPROM). Stray light getting in does shorten the life of the data in them too; you will eventually start seeing some 0 bits changed to 1's (their erased state). The flash is probably not good for this either ;-)

They are supposed to have a metalised sticker placed over the window after programming. If you didn't have the right stickers, a write protect sticker for a floppy disk worked OK. You could also buy them with no window at all, so they can only be programmed once, but then last longer.

I have an early 1980's minicomputer which has lots of EPROMs on the

6800 microprocessor control board. Assured data life of those is 10 years, but 30+ years on, they still work, although they do all have window stickers over them. The minicomputer's main processor microprogram is burned into non-reprogrammable PALs (has to be faster than EPROMs), so that should remain safe, although you wouldn't be able to make it run without the 6800 working too.

I've seen a few incidents where photographers come to photograph some new system going live for the press, and it instantly crashes ;-)

[Dave Liquorice]

Had similar with the outdoor humidity/temp sensor of the AWS under some conditions it would just stop working. Not sure how I twigged it was light penetrating the black plastic blob (or maybe through the board) to the chip but a couple of layers of black insulating tape around the board where the chip is cured the problem.

[Vir Campestris]

I had one once, where we were commissioning a new board. All the initial tests were running, so I tried a real BIOS. All went well, until suddenly it restarted without completing the POST.

A bit more digging, and it turned out that the beep was firing reset. An hour's buzzing about, nothing. Logic probes everywhere - nothing. Then a few scope probes linked in. It turned out that the back EMF from the coil in the speaker was causing a big enough RF spike to cause a transient on the nearby reset line. Without any physical connection at all.

Ah, the old days when I could see the pins...

Andy