4

The HRM, GPS, and cadence in my cyclocomputer give perfectly good readings, but the speedometer (and hence the odometer) readings are obviously incorrect.

Because the speed and distance readings seem to be just about twice what they should be, I'm guessing that something other than the dedicated magnet is triggering the extra events.

Presta valve near receiver

Might the Presta valve be magnetized and provide that extraneous trigger? Since the cadence and wheel receiver are in the same piece, there is not a lot of flexibility in moving the dual-receiver far away from the valve.

Have you been there? What are some ideas to debug this issue?

Sam7919
  • 9,112
  • 4
  • 28
  • 92
  • 5
    Just to address the most obvious explanation: have you set the wheel diameter correctly? – Adam Rice Sep 02 '20 at 21:43
  • 1
    @AdamRice I'm pretty sure the wheel diameter is set correctly, but the reason I don't even suspect this as an issue in this case is that there is no way for the 2x factor (after riding what I know from Google Maps is a 40 km roundtrip, the odometer tells me I just did 78 km) to arise from an incorrect diameter for a 700x23c wheel. – Sam7919 Sep 02 '20 at 21:50
  • 3
    You say that the cadence and wheel sensors are in the same spot. Is there any chance the wheel sensor is picking up the crank magnet? (Should be simple to test, just spin the cranks and see if you get ghost speed.) – DavidW Sep 02 '20 at 22:08
  • The wheel circumference setting in the computer goes down extremely low to support things like recumbent trikes... it is entirely possible to be off by a factor of close to two if it were not set. (although one would expect the factory default would be a 'typical' value but I do not know that brand.) (notably most expect circumference and you are referring to diameter....!) – Affe Sep 02 '20 at 22:12
  • 1
    Had you entered 622mm or 700 for a road bike's radius, and the computer is expecting rolling-circumference of 2,100 mm then that would seriously throw the reading. But you're looking for "double" not triple reads. – Criggie Sep 02 '20 at 23:47
  • 5
    Lazy test - remove the valve stem cap magnet completely, and spin the wheel. Does the computer read any forward speed at all ? Remove the front wheel completely, and hand-pedal the bike while its off the ground. Does the speedo show any forward speed ? – Criggie Sep 02 '20 at 23:48
  • 4
    I was under the impression that Presta valves were made of brass, not something ferromagnetic. I just tested this with a magnet I have on hand, and neither the valves nor the locking collars I tested responded to the magnet. So, while there may be a magnet elsewhere on the bike or you accidentally doubled the wheel size measurement, I think that Presta valves in general shouldn't be able to be magnetized. Perhaps you have an unusual steel Presta valve, of course. – Weiwen Ng Sep 02 '20 at 23:59
  • 1
    Assuming it’s just a garden variety sensor, you should be able to listen carefully (maybe ask a kid if your ears aren't great) and hear the reed switch click as the magnet passes. – MaplePanda Sep 03 '20 at 02:48
  • @MaplePanda Perhaps use a stethoscope of one's available, unless you're going for the Van Gogh earless look. TBH the reed switch is so small its inaudible. Most of them are just a reed switch on two wires, so not complex. – Criggie Sep 03 '20 at 03:29
  • @Criggie Maybe mine is particularly loud or there’s a speaker, but I can definitely hear something clicking in there. Spin the wheel slowly, if all you hear is hub noise, it may be time to clean that. – MaplePanda Sep 03 '20 at 03:55
  • Are you cycling with other people who have the same brand of computer, and is it a wireless sensor? I have one where the computer is not specifically linked to the individual sensor but picks up other sensors on other bikes too. Caused some confusion as my speed suddenly jumped to 60 km/h when some random person passed me. – Stephan Matthiesen Sep 03 '20 at 05:57
  • Is it wired or wireless? In the former case I'd cut the wires and fit an led (plus battery and resistor), then splice back together afterwards. But I've got all the kit to do that anyway – Chris H Sep 03 '20 at 11:52
  • 2
    Fun but admittedly tangential fact: standalone tubeless Presta valves are often done in brass, sometimes aluminum (pretty colors, lighter, slightly fragile), more rarely carbon fiber ($$$$$ and fragile), and in one case titanium ($$, swanky, limited colors possible through anodization, but very hard to break). In theory it might be possible to make a steel valve, but the cost/benefit (if any) over brass might not make this a sensible choice. – Weiwen Ng Sep 03 '20 at 12:11
  • Is your sensor wired or wireless? – Criggie Sep 03 '20 at 13:19

3 Answers3

8

The problem went away after a firmware update. The most likely interpretation is that the reed switch is generating two pulses for each pass of the magnet, but now each double-pulse is interpreted in software as a single pulse. It's not ideal (hardware with one defect has a higher chance of also having other issues), but it's accurate.

Sam7919
  • 9,112
  • 4
  • 28
  • 92
  • 2
    The good news is that they probably didn't debounce the switch readout enough for real-world conditions - a fairly minor design fault rather than a build failure. Or perhaps they changed to a bouncier read-switch. Thanks for coming back and answering – Chris H Sep 18 '20 at 14:22
  • in fact they probably forgot to debounce it at all. Switch bounce is going to be much faster than legitimate switching (even an upper limit 20"/ETRTO 406 wheel doing 100km/h as may be seen on a few recumbents, though they tend to use the larger 20" standard) is only rotating at about 20 rev/s so triggering at 20Hz, while switch bounce is usually of the order of kHz – Chris H Sep 18 '20 at 14:29
  • @ChrisH "When the switch is closed, the two contacts actually separate and reconnect, typically 10 to 100 times over a periode of about 1ms." That's fascinating. Might it be that reed switches are hopelessly undesirable, especially when mounted on a device that is itself going over a tarmac that's less than very smooth? – Sam7919 Sep 18 '20 at 14:30
  • @ChrisH I had asked a question about how to determine whether a switch uses reed or Hall effect. The comments suggested that there is no way to tell (short of breaking it and looking), and so I deleted the question. It would be nice to know before buying. (I dislike relying on "it costs more, ergo it's better built" arguments. We really need to know aside from pricing, or else we're rewarding reverse snobbery and noncompetitive, let alone unscrupulous, manufacturers.) – Sam7919 Sep 18 '20 at 14:33
  • For many years they were the only (sensibly-priced) way to do this, and the reason they have to be so close to the magnet is that the spring is reasonably strong. Mounting orientation is also important - by design vertical and even longitudinal vibrations should be at right angles to the switch motion, as the magnet-switch direction is lateral. In the real world perfect mounting doesn't always happen, and vibrations can couple in strange ways – Chris H Sep 18 '20 at 14:34
  • I'm with you on that - cycling accessories are on of many areas in which you often end up paying to have a fancy name on something no better than the cheap stuff. You might be able to hear a reed switch - I can on one I've got here (cheap and about 10 years old), by pressing the switch against the bone in front of my ear and bringing in a magnet. If there are only 2 cores in the wire (a figure-of-8 cross-section) it should be a reed switch as a hall sensor needs 3 - but this reed switch uses a round wire and of course some are wireless. – Chris H Sep 18 '20 at 14:47
  • 1
    ...If the head unit is detachable from its base and there are only 2 contacts - reed switch (almost certainly; with sufficient effort you could run a Hall sensor down 2 cores but it would be unnecessarily expensive). I tested for continuity between the 2 contacts when at first I though I couldn't hear the switch. As a Hall sensor needs voltage applied, this should also indicate a mechanical switch . And now I've answered your deleted question (even if not conclusively)! – Chris H Sep 18 '20 at 14:50
3

I had a problem with a low-end wired cateye computer. It would work fine for a while, and then occasionally my reported speed would be 50% higher, bursting to 100% higher. I even saw 90 km/h once when I was riding at about 30.

Fortunately I have a couple of these, and managed to isolate the problem to the sensor, not the magnet nor the head unit.

Since this one is cheap, it uses a reed switch not a hall-effect sensor. That reed switch is inside a lightweight glass tube, which had broken but the internals were still in place.

At certain road vibrations the sensor would double-bounce and record 2 pulses for each magnet pass. We have a lot of chipseal roads, which have an inherent Buzz while riding, and there was a sympathetic speed/coarseness which could trigger the sensor to double-read.

My fix was to simply buy more reed switches and solder one in. It wasn't difficult, and that bike has been great ever since. Perhaps your sensor is failing - some investigation might help. If you can swap parts somehow with another similar unit, might help identify a root cause.

Criggie
  • 124,066
  • 14
  • 180
  • 423
  • 2
    I’m stunned that there are computers available so cheap they use mechanical reed switches – Argenti Apparatus Sep 03 '20 at 13:55
  • 1
    @Argenti at one time they all did, before Hall sensors were mass market. So an old design, only slightly updated, could still use reed switches. I'm not sure what my <£/$/€10 Bluetooth speed and cadence sensor uses, because it's the Bluetooth that's too buggy for it to be useful – Chris H Sep 03 '20 at 18:53
  • 1
    @ChrisH Are you sure that they all had switches? Not just simple, plain coils? That's what I would use if I were to design a receiver: Maybe a coil delivers a weaker signal, but it's much easier to filter with a simple hysteresis amplifier. And a hall sensor is just complete overenginering: It requires an additional power source (at least three wires) for basically no gain over a coil. – cmaster - reinstate monica Sep 18 '20 at 14:32
  • 1
    @ChrisH Interestingly, the English Wikipedia mentions only Hall sensors and magnetic switches, while the German Wikipedia lists only coils and magnetic switches. I wonder whether the "Hall sensor" is actually a coil... – cmaster - reinstate monica Sep 18 '20 at 14:41
  • 1
    @cmaster-reinstatemonica the cheap one (new old stock almost identical to this one) I have in front of me, and another (more expensive) one I dismantled over 10 years ago both used reed switches. I may even still have the switch out of the older one; that's why I dismantled it when it broke. – Chris H Sep 18 '20 at 14:57
  • 1
    ... I can't rule out coils though I've never heard of them being used and I have a suspicion they'd be more likely to pick up interference - though Hall sensors may need calibration or at least to measure departure from rolling average – Chris H Sep 18 '20 at 14:58
  • 1
    @ChrisH I definitely have to make a mental note to disassemble the next sensor before tossing it away. Thinking about this a little longer, I believe that the main downside of a coil would that it won't register very slow movements as it registers only changing fields, not absolute field strength. So, a coil-based sensor should have a minimum speed below which it won't notice that the bike is moving at all. Neither switch nor hall sensor have this characteristic. I think, I'll do a little test later... – cmaster - reinstate monica Sep 18 '20 at 15:10
  • 1
    @cmaster-reinstatemonica of course. So the pulse magnitude (and slope) would depend on the rotation speed, making thresholds to reject spurious triggers very hard to implement. I haven't done much with coils (though a drop of water charged to 2kV passing through one shows up nicely - I don't reckon we'd be allowed to give that to our undergrads any more) – Chris H Sep 18 '20 at 15:22
  • 1
    @ChrisH Just made a little test, my bike computer definitely uses a switch, not a coil. Simply connecting the two contacts with a wire is indeed interpreted as a pulse from the sensor. This definitely rules out the coil, the hall sensor is already ruled out by the low price and the missing third contact. – cmaster - reinstate monica Sep 18 '20 at 15:52
  • 1
    @cmaster-reinstatemonica now I wonder where my cheap speed/cadence sensor is. I'd like to see what that has (I should also have a wireless sender somewhere but suspect it's on a bike I no longer have) – Chris H Sep 18 '20 at 15:54
  • 1
    Yet another reason to hate chipseal. – Marc Bernier Oct 15 '20 at 21:02
  • I've had wireless bike computers that used reedswitches but they showed crazy values at two different locations: a cycleway along a 25kV16.3Hz railline and in the vicinity of the 2000kW long-wave radio transmitter of Radio Luxembourg. – Carel Nov 20 '20 at 19:36
2

There are two causes of extra counts:

  1. Reed switch double closure
  2. Radio interference (in an analog radio based wireless cyclocomputer)

The source (1) can happen all the time or rarely. If the magnet is incorrectly oriented, it may happen all the time in such great proportions that the speedometer and thus odometer reads incorrect values. In this case, the reading is almost always double of the true reading. If the magnet is nearly correctly oriented, it may happen only when riding at high speed over bumps (due to momentary vibration during reed switch closure), causing a very momentary error practically only in the maximum speed reading of the speedometer and not any odometer errors.

The source (1) can be avoided by using a high quality brand name (example: Sigma) cyclocomputer that is resistant to double closure and by orienting the magnet and the reed switch sensor correctly. See the instructions of the cyclocomputer for the correct gap between the magnet and the reed switch sensor. Be also sure to use the correct magnet! Some cyclocomputers have a weak magnet and others a strong magnet. If you use a weak magnet for a cyclocomputer requiring a strong magnet, you run into problems.

Also be sure to mount the magnet and sensor at the correct height. If the cyclocomputer manual says down near the hub is preferable, do so. If the cyclocomputer manual says high up the fork leg near the rim is preferable, do so.

Cheap cyclocomputers may not have good adjustment options for different spoke - fork leg spacings in different bicycles. At least the Sigma ones I have come with excellent adjustment options, allowing the reed switch sensor surface to be parallel to the spokes.

The source (2) can be avoided by only using digital radio. For example if using Sigma, select STS and not ATS. The source (2) with analog radio (not recommended) can be reduced by having the cyclocomputer and the sensor at the same side, and by minimizing the distance between the cyclocomputer and the sensor (thus mounting the sensor high up in the fork).

juhist
  • 18,668
  • 3
  • 27
  • 49