If the freedom of speech is taken away then dumb and silent we may be led, like sheep to the slaughter.

- George Washington

Thursday, 8 March 2012

Fun with switches

A year ago, I had a bit of a problem with the XT. It would start and run OK, but as soon as I switched on any electrical item (like the headlight or indicators), the engine would die - in fact, all electrics died, including the dash lights. Turning the ignition switch off and on again cured it, but clearly it wasn't a good idea to have this happening in heavy traffic. I bought a new switch from the acknowledged XT guru David Lambeth ('Custardgrub' on eBay), and all was well. I had to destroy the old switch to get it off, and inside it was a wreck, with dirty contacts and as loose and floppy as a nonagenarian's wedding tackle. A year later, almost to the day, it started happening again. Additionally, it was doing some rather silly things like passing current when it was supposed to be 'off'. Taking the key out and having the ignition light still glowing tells even a numpty like me that something is wrong.

My understanding of electrics is fairly limited (I understand the Smoke Theory, and the water-in-pipes bit makes sense, but advanced fault-finding is usually trial and error. Mostly error. In this situation, my guess is that the switch is making poor contact, and so it will allow the small current required by the engine, but when a bigger load is added to the circuit it gives up. Technically speaking.

This time, work patterns meant that I couldn't sort out a new switch straight away, so I decided to make my own. A look at the wiring diagram, couple of lengths of wire, a few connectors, a fuse carrier, some heat-shrink, and I was in business.

The 'ignition key' was a 20A mini-blade fuse. It worked brilliantly, if I say so myself. The system is robust, foolproof and simple. It has two-point-five drawbacks:
  • The 'ignition key' is tiny and easily lost. If you drop it on the floor in the dark, it can take several minutes to find it, even with a torch. Don't ask. And it isn't as easy to find in a jacket pocket as a bunch of keys.
  • Switching on and off is fiddly and time-consuming.
  • The bike is nickable to anyone with a suitable fuse about their persons.
The last one is only a point-five disadvantage because a) who carries a 20A fuse with them on the off-chance? b) the bike is not a desirable target, far from it, and c) you'd have to know the bodge was there in the first place.

Anyways, I reckoned that a year wasn't a very good service life for a thing like an ignition switch, so I contacted David to see if there had been a problem with that batch. He replied that he hadn't had any problems with the switches in 20 years of selling them. However, he offered to send me a new one if I sent the old one back with ten pounds. That seemed an exceedingly fair offer, but I was unwilling to take him up on it, as I was only 99% sure it was the switch. He's a very helpful guy, and I don't want to abuse his good nature. I don't rule out an unrelated gotcha which could account for the symptoms. So I bought a new one at full price, and I am going to send the old one back to him for inspection. If he finds it is faulty, I am sure we will come to some arrangement.

New switch fitted, and the XT is going like a good 'un.

But I am keeping the little harness - just in case.


  1. ".... as loose and floppy as a nonagenarian's wedding tackle."

    "..... it was doing some rather silly things like passing current when it was supposed to be 'off'."

    The technical term is 'incontinence'.

  2. LOL, yup. Have you been speaking to my friends?

  3. Ahh vehicle electrics, I am in awe of your expertise! It is all double Greek to me.

    Although I did change a regulator rectifier, but that was just pull one out and push new one in!

    I impressed myself if no one else........

  4. When you took out the key and the ignition light remained on, you should have investigated the darksucker theory


  5. Conniesdad: That's expertise interpreted very flexibly. I understand circuits and can calculate fuse ratings and such, but when you get into V=IR and all that I am stumped. There are some good tutorials on the web, though, and I am trying to educate myself. I can solder, though. Badly. (You still up for the RoR? I might be nearer your time slot than I thought.)

    Ripper: thanks, hadn't seen that one. Very good!

  6. You're nearly there with V=IR - a good contact will have negligible resistance and will happily pass the required current. But a dirty, corroded contact may easily be several ohms, and once you try and put (say) 10 amps across it, the voltage drop becomes considerable. The heat generated by this voltage drop then quickly makes matters worse, and usually melts the plastic insulation the contacts are embedded in.

    I had similar problems with the ignition and light switches on my old Austin Maxi (and even the Mini I learned to drive in). Replacements didn't last that long, and I solved the issue by fitting relays and separate fused supplies for the heavy loads. A good designer would incorporate these from new, but (as always these days) cost is the primary consideration...

    The ease of stealing with your temporary "bodge" isn't that far removed from the famous "nail & coin" type key used on Jawa/CZ bikes, which you and I both had!

  7. I was hoping someone with more than half a clue would turn up! I tested the switch terminals with a multimeter set to ohms, and got a reading of around 110 kohms with the switch on. I assume that ought to be zero. Equally, with the switch off (and the key out) it was giving a reading in the 100 Mohm range, when I guess it ought to be infinity (or '1'), as my meter calls it. Squirting contact cleaner in the switch and working it for a bit changed these readings, but they were similar.

    I understand about voltage drops due to high resistance, but the puzzle (to my limited understanding) was why it cut the contact completely. Run engine, switch lights on, everything goes out - engine, lights, idiot lights, the lot. Key switch off and on again, and it's back to 'normal'. No evidence of overheating, but they are chunky brass contacts with no insulation nearby, so perhaps it would be unnoticed.

    Thanks for the insight.

  8. And yes, the Jawa/CZ ignition switch was pants, wasn't it?

  9. "With a multimeter set to ohms" Assuming you have the lowest scale selected (which on most digital multimeters is normally 0-200) you wouldn't get readings with a "k" in them. This means "kilo" or x 1000. So I guess you were reading 110 ohms. This is WAY over the mark (even 10 would be)- realistically it should be close to zero, but very few meters are really accurate at this level of measurement. The problem with badly worn/corroded contacts is that the actual behaviour will vary depending on what you are doing with them. A DVM only passes microamps (that's 10 to the minus 6) to get readings, which is why they are essential for work on sensitive electronics.

    A good old fashioned AVO is more suited for simple vehicle faults, and does pass a bit more current. It's also possible that a single ignition coil, which is a "pulse" load, will work whereas a much higher steady load will fail. I'm assuming here that the XT uses points or a simple electronic ignition with an optical or magnetic trigger. If, on the other hand, it's a CDI (capacitor discharge ignition) system these draw a small but steady current which runs an inverter to charge a capacitor to several hundred volts. This is then discharged through the coil. So long as the knackered switch will pass sufficient juice to keep this circuit working the engine will run fine. But as soon as you apply the lights the voltage will drop so much that they won't even show dimly, and the ignition will cut out.

    Undoubtedly (as with my old Maxi, and Panda) there is a single live feed and, as you've found, the risk of loosing everything. A better answer would be to have separate feeds for ignition & lights with dedicated contacts. This would offer some level of redundancy, but as I said earlier, will cost a bit more. It would be straightforward to add a cheap relay once you've replaced the standard switch, to bypass the lighting load. This would, in all probability, solve the issue for good. I will happily advise you further should you be interested.

  10. Sorry, should have been clearer. I meant set to ohms rather than volts or amps. Set to ohms in the 0-200 kohm range, meter read 110. A high resistance joint indeed. And every time I tested it I got a different reading, suggesting that the contact was not being made reliably. The bike has a simple CDI igniter with a pick-up coil, and I have noticed that it is sensitive to low voltage. When the reg/rec packed up, I could cure the resulting misfire (for a while) by turning the lights off.

    The fact that everything runs off a single live feed (with ONE 20A fuse!) makes me nervous, as some of my commute is on an unlit, fast and narrow rat-run. I plan to take it off the road at some point and give it a decent refurb, and part of the plan is to redesign the wiring system so that the ignition and lights are powered and fused separately. I once saved my bacon when a car's headlights failed by using the left indicator to see the nearside verge and come to a safe stop, and I would like that facility on the XT. I'm not too fussed about adding relays, as I prefer to keep things simple, but siting the reg/rec in a cooling airflow and protecting the wiring better from chafing will be a priority.

    The new switch works perfectly, by the way, so as far as I am concerned the problem is solved, notwithstanding future improvements.

    I appreciate your advice, thanks.

  11. 110 kOhms across the switch just sounds plain wrong - you'd get a lower reading than that trying to pass current across your own body! Probably eggs and sucking and all that, but wiring should be somewhere down in the tenths/hundredths of Ohms...it's kind of deliberately made from really conductive stuff and that reading would be high even for resistive components like the HT coil, CDI or reg/rec.

    Good idea to separate the lighting out, though. No help at all if the battery dies, but in all other circumstances it's quite reassuring to know you can still see and be seen!

    1. "110 kOhms across the switch just sounds plain wrong - you'd get a lower reading than that trying to pass current across your own body!"

      Just a thought ... that reading is almost an open circuit, right? Perhaps I just caught the switch when it was in its non-functional phase, i.e. pretending to be 'on' but really almost 'off'.

  12. OK, I've just been out and checked it in daylight :) Set to 0-200 ohms, it's reading between 8 and 12 ohms switched on. Can't think how I got those other readings - possibly I missed a decimal point on the display, although the readings are so variable I wouldn't discount something bizarre. Meter is reading 0.01 ohm with the contacts touching, so it's in the ballpark. I know that wiring *should be* conductive, but presumably that's the problem with a faulty contact.

  13. Absolutely wasn't trying to be patronising or anything there :-), just trying to give an indication of the orders of magnitude involved. 8 - 12 Ohms is about what I'd expect from corroded or otherwise filthy contacts; it'd be quite hard to get much higher than that without wedging something like a dead rodent or lots of cardboard into the switch, or having a filament/coil involved to deliberately introduce high resistance (eg 500-odd Ohms is 240V/100W lightbulb filament territory).

    So it sounds like your multimeter's now reading fairly true and it could well be corrosion or similar in the switch.

  14. I am going to return the switch to David for inspection, otherwise I would dismantle it and find out. Funny how the OEM switch lasted 16 years and its replacement lasted 12 months :)

    Oh yeah, and *which* dead rodent? I can't check them all.

    1. If I had to guess, I'd say you'd get the greatest resistance out of a coypu, but it'd be a bit of a job getting a whole one squeezed into an ignition barrel. Something like a gerbil would probably give the best compromise between effect and ease of insertion...!

    2. Ah...so that's how the Reply function works!

  15. "But it'd be a bit of a job getting a whole one squeezed into an ignition barrel"

    How about a twist of lemming?


Comment is free, according to C P Scott, so go for it. Word verification is turned off for the time being. Play nicely.

Related Posts Plugin for WordPress, Blogger...