Jeff Liebermann wrote:

If that's true, then there may be something wrong with the Shimano
dynamo. Got any electronics test equipment, such as an oscilloscope?
Take a look at the dynamo output with no load and with the light
attached. Measure the frequency at various speeds. There may also be
something wrong with the light, such as having one phase (branch) of
the LED string shorted. Either way, I would expect to see an
asymmetry in the AC waveform which would accentuate any flicker. If
one phase in the light were shorted, then it might be triggering some
kind of protection circuit inside the Shimano hub.

Alas, no test equipment beyond a cheap DVM with 400mA AC current range.

I get 28 "notches" per turn on the hub, in a 27" wheel. Current reaches
400 mA at a modest speed, probably 10 mph. Voltage (which I can't measure
simultaneously) reaches 10.3 V at a similar speed and won't exceed 10.5 V
with any effort I can realistically exert.

There is of course no guarantee the meter is true RMS and the wafeforms
are not sinusoidal, but the numbers suggest nothing radically wrong with
either the hub or the LEDs (all six look about the same).


What bothers me is that you say:
"flicker quite noticeably at all speeds".
That's difficult to do when the Shimano hub probably produces a
frequency of 30Hz or greater at operating speeds, which should not be
visible as a flicker. Other than an overcurrent protection circuit
failure in the Shimano hub, I can't imagine how that would be
possible.

Incidentally, if you don't have an oscilloscope, try using a PC and a
sound card:
https://windowsreport.com/oscilloscope-software-pc-laptop/


There's an audio signal processing app for iPads that can do something
crudely similar. It's not an oscilloscope, per se, but does give a
waveform display. If the puzzle gets deeper it might be worth trying.
At the moment I see no reason to suspect malfunction, either in the
dynamo or the LEDs.

Part of the issue can be seen in a darkened garage: Spinning the front
wheel casts double shadows from the two sets of LEDs. Without the shadow
effect the scene looks much less strange. With a shadow, or looking
directly at the LEDs, it's quite conspicuous. I suspect the lateral
"movement" is what catches the eye.

The roller dynamo has 8 poles and runs at roughly 20 times wheel speed.
Even with 28 poles the frequency of the hub dynamo is roughly 6x lower
than the roller setup, which seems enough to account for the flicker,
especially when combined with the alternating groups of LEDs.

It looks like the only defect is my failure to consider the lower
operating frequency of the hub dynamo. Now the question is what, if
anything, to do about it.

Thanks to everybody who replied, this has been a most informative
discussion!

bob prohaska