Re: For those who believe in electricity

On 4/10/2025 1:02 PM, Jeff Liebermann wrote:

On Sun, 6 Apr 2025 22:22:33 -0400, Frank Krygowski
<frkrygow@sbcglobal.net> wrote:
 I have a some time right now.  I'll try to answer Frank's original
question and avoid the inevitable topic drift.  Please forgive my
being rather terse.

Thanks for your interest. (I was hoping you'd chime in earlier.) But check my posts 4/8 about 9 PM and 4/9 about noon. I think the diagnosis is complete, but I'll comment below anyway.

I think I
mentioned a few days ago that on a recent night ride, my headlight (B&M
Eyc N plus),  powered by my good old sidewall generator, was randomly
and sporadically blinking off.

 It the light this one?
<https://www.rivbike.com/products/b-m-eyc-t-senso-plus-dyno-headlight>

It is they Eyc N Plus shown in the photo, not the Eyc T Senso Plus described in the text. The N Plus seems to be an earlier version, with no "Senso" daylight sensing function, etc.

I couldn't find a "good old sidewall generator".  Could you be a bit
more specific?

Classic Union sidewall generator. Maybe the most common such unit ever made.

I put the bike on the workstand, checked
out the wiring (it’s original and pristine), and spent some time
spinning the front wheel. Sometimes the headlight worked, sometimes not,
and it changed state at random without having been touched. I was
suspicious of the tiny on/off switch hidden behind a rubber plug, in
part because I felt no satisfying click.

 Start by isolating the problem.  Is the intermittent light in the
generator (dynamo) section or in the light?  Find a different dynamo
to power the light or move the light to a different bicycles with a
different power source.  If replacing the dynamo magically fixes the
problem, I suggest that you tear the dynamo apart and clean it out.

These Union generators (and many others) can't be disassembled without destruction. The aluminum body of the "bottle" is crimped or swaged onto the plastic bottom.
But I did test the generator by substituting a different LED dynamo headlight. That light works fine, although I greatly prefer the output and optics of the Eyc.

Next, pull on the wires between the dynamo and light.  Yes, I said
PULL.  If the copper wire falls apart and the insulating jacket
stretches and narrows at the beak point, you've found the problem.  If
the light refuses to work after you've pulled on the wires, you
probably have a broken wire somewhere that's hiding under the
insulation.

Yep. Wires are perfect, or "pristine" as I said.

If it's not in the dynamo or the wiring, then it must be in the light.
Start with an electric tooth brush.  The idea is to use it as a
vibrator to isolate the section that has gone intermittent.  If the
vibrating tooth brush doesn't crack your teeth when used in the
originally intended manner, it's probably safe to use on a bicycle
light.  The most likely areas are loose magnets, corroded switches,
corroded electrical contacts, broken PCB (printed circuit board)
traces, "cold" PCB solder connections, bad soldering, or intermittent
wire bonds inside the IC's (integrated circuits).

Interesting toothbrush application. Instead, I repeatedly tapped the light to jar it.

 

https://www.flickr.com/photos/16972296@N08/54435279894/in/dateposted-public/
https://www.flickr.com/photos/16972296@N08/54435279589/in/dateposted-public/

 Nice photos of the caliper.  However, the PCB is out of focus.  Flickr
reduces the resolution of the photo when upload.  I can see things
that might be a problem, but only if the photo has higher contrast,
more brightness (to improve depth of field, and of course, that it's
properly in focus.  I think you can improve your photos by using
multiple sources of illumination from diffuse sources.  The PCB photos
that I post on YouTube usually use two LED light sources from lamps
surrounding my desk.  With two sources, shadows are reduced.  With a
ring light, they're reduced even more.

I'll keep that in mind. My workbench has a fluorescent ring light as part of a desk magnifier lamp with a ~8" lens. (Jobst was well known for critiquing or complimenting poster's photos.)

With no power attached, I tried checking for continuity across the
switch. It always showed close to zero ohms, never infinity - but the
resistance readings with pinpoint probes were jumpy...

 That's an important clue

Yes, but not definitive. I long ago found it handy to fit small alligator clips to the probes on my meters. For this tiny work, I stuck straight pins into the alligator clips. Problem was, at this point, I couldn't be positive the connection from pin to clip to meter lead was solid and reliable.

I used contact cleaner to blast out the switch and repeatedly clicked
it, including while power was applied. Except for a few occasions, the
light did not respond to the switch, so I still suspect the switch. When
the light was off, I had about 12 VDC. When on, about 4 VDC.

 The voltage across the switch contacts should be zero volts when the
switch is on.  However, if B&M is doing something odd, it could be
something else.  I can't tell without a schematic.

I should have specified, I was measuring input voltage to the light at that point, not voltage across the switch. Yes, it would have made sense to measure across the switch.

It’s still not fixed, but it's getting late here. I would try to simply
short out the switch, but it’s very difficult to identify and reach its
microscopic solder pads and I’m no good at micro soldering.

 Sharpen a paper clip.  Wind it around the soldering iron tip.  You now
have a micro soldering iron.

That sounds useful.

Get some liquid or paste rosin flux (not
acid flux used in plumbing) and slop it on the solder pad before
soldering.  Use copper braid to suck up any excess solder.  Done use a
"solder sucker" on small pads and traces.
<https://www.google.com/search?q=copper%20braid%20soldering&udm=2>

Yes, I've got some of that. Now that I know the light works with switch contacts open, there's no reason to do that soldering. I'll reassemble and reinstall it later today.
Since my last post, I got curious enough to disassemble (AKA destroy) the tiny switch. It features a tiny formed piece of brass pre-stressed in such a way as to "pop" inward and close contacts, similar to most microswitches of my experience. Since the circuit, not the switch, does the toggling, it must have been somehow rattling into random contact. I couldn't see exactly how, but at this point it doesn't matter. But I wonder if they needed better de-bouncing of the relevant part of the circuit.
The tail end of the circuit board ("bottom" in the photo with the heat sink) has a yellow indicator LED that is supposed to illuminate the translucent rubber switch cover when the main light is operating. But it's so very dim that I never noticed it, until long after I had the circuit board out of the headlight. Not that it would have helped, I suppose.
Again, thanks for your interest.
--
- Frank Krygowski