Cycling UK Forum

Whatever the wisdom of a particular battery scheme, going back to the Velocharger SA_SA_SA linked to, it would work for a light. But aside from being more complex and expensive than necessary it would need a stand light. There's no guarantee it will enter 'charging' mode the moment you set off, so you may end up with no light for several seconds.

A pass-through power bank would work better if it is, itself, more forgiving in accepting inputs that aren't perfectly in spec, than whatever you're trying to charge.

It's a tough life for a power bank, however. So I've wondered whether someone could make a 'buffer' battery using a supercapacitor and an off-the-shelf USB output. It might only have enough energy to output power for under a minute, but it should be more tolerant of brownout conditions and continuous charging and discharging.

Interesting light. It seems to have four identical terminals at the back. It may be easy to mix up front and back terminals as a result. They also specify which is earth and which is live - that will affect things if the dynamo is earthed to the frame, AND the rear light (possibly via a screw).

Is the switch an actual mechanical switch (i.e. it goes in and locks in that position)? Or a momentary switch (i.e. just a click)?

rjb wrote: ↑13 Feb 2024, 1:20pm More lights here from PSW power. Anyone tried these.
"Front 6V-48V 1.8W DC light for bafang BBSHD BBS02B BBS01B TSDZ2 mid drive motor kit" https://www.pswpower.com/products/front ... or-kit-168

Incredible prices. Ironically, the cheapest ones (the QD236) may be better as they at least have lux ratings (30), and are 2.4W rather than 1.8W. The QD197 looks very odd, like the reflector is upside down.

The 80 lux M-Wave one SA_SA_SA listed I've seen under various brands in multiple places. It looks very much like a B&M Ixon IQ-XS.

As previously mentioned, designing the shell has taken the most time. Not because there are any optimistic design or styling goals, but just rather getting ideas into a workable form while learning CAD has been a headache. You can spend ages trying to do something, and then only later learn a better way with a few shortcuts.

It's also very easy to end up designing something which doesn't fit, as something else is blocking access, etc.

The rear of the case was fairly straightforward, as most items were at right angles to each other. The front of the case was another story, as it has to accommodate a lens at an angle.

My first attempt:


This was based on a transparent plastic piece I had from a clock. Making a screw-on retainer was pretty elaborate, but ultimately this just wasted too much space and was fairly ugly.


...and messing up some measurements meant the reflector sat unnecessarily deep.

My current approach is:


...which is a lot shorter. In this photo you can see the rear of the first light and it's attached innards (including the supercapacitors stuffed inside a piece of inner tube), ready to be transferred to the second design. This is plugged into the board, as that is the main way that this light can be assembled and disassembled - a key goal this time, compared to my last lights, was to make something which could be re-opened without too much hassle.

The first incarnation I did actually test out on the road, albeit not at speed, as I was waiting for some copper shims. These are to provide a thermal path from the LED's heatsink to the aluminium case (as the original heatsinks from the Lidl light aren't sized to handle the wattage this light should be able to output). In truth, if I did this design again I probably wouldn't use such a case, as the contact isn't really on a plane that can be tightened. Instead, the heatsink and the case slide in place. Hence the shims have to be sized carefully. In any case, I should be able to get a road test in soon.

SA_SA_SA wrote: ↑12 Feb 2024, 1:18pm I wonder if affordable hi lux Stzvo E bike front lamps aimed high might be better/simpler/more affordable than said narrow MR16 eg
https://www.amazon.co.uk/M-Wave-Unisex_ ... B092M1PY1FA pity the min voltage is usually 6 rather than 5Volts (usb power bank standard volts)...

Stepping up the USB should be easy: https://thepihut.com/products/usb-step- ... -to-9v-12v (as long as about 1 amp is sufficient). Efficiency will be slightly hit of course. Some USB C PD trigger cables can request 12V straight from a power bank if it's USB C and happy to supply 12V (although arguably the conversion is still happening, just in the bank).

Interesting that a loose rear cable could cause this. Presumably the rear light failed to come on, when the problem existed (as it would be open circuit)? Alternatively, if it did come on, that could show that there is an alternative (maybe high resistance) path through the frame.

Another point to note about the light in question, is it seems to have some special low-speed mode ('intelligent beam'), which possibly relies on sensing speed by frequency. That can get messed up somehow too.

An external power bank may also behave the same way as a phone or other device in refusing charging if it doesn't think the available current or voltage is right. That was my experience using a simple linear regulator > USB. Plugging in/switching on while already on the move worked, whereas starting from a standstill didn't.

Chargers like the Sinewave Revolution solve this by only starting charging after 15 seconds of motion.

If this was a normal, bare LED, I suspect the diode would only work if the battery could rise a diode's voltage drop (up to 0.7v) over the LED's voltage. The way to get around this with bare LEDs is to discharge to a lower number of LEDs - e.g charge across 2 and discharge across 1.

With MR16s and their built-in driver circuitry who knows. I would guess that the input voltage, and hence battery voltage, rises depending on the driver's duty cycle. But that would also be bad news for the battery.

Building the light with 3D modelling is probably the biggest time sink so far, but that's largely because of learning, plus the huge range of possibilities leading to a temptation to tinker.

In case it's useful for anyone, here's what I've done, as a complete novice to CAD.

Firstly, I'm using Autodesk Fusion 360, which has a free hobbyist licence. I've not run into any major limitations yet, but it's not the most intuitive program. For a complete beginner, this tutorial playlist has been invaluable: https://www.youtube.com/playlist?list=P ... Y8ipVvFsIp (although sped up by 1.5x, as he's a slow talker). The main thing is that this CAD software is parametric, which means you should try to draw with exact dimensions, and store those dimensions as variables which can be changed later. This way, a model can be tweaked far down the line with less chance of breaking anything.

A set of calipers is super-useful too. E.g. for measuring the dimensions of the aluminium enclosure I'm working around, and various interface elements like switches and a remote jack. Even then, it sometimes takes a revised print if a fit was off, which is where the parametric aspect of CAD really helps.

Design is entirely down to personal preferences and limits in knowledge. In any case, for the rear here is the design in CAD:


And after printing and adding some of the interface parts:


(There is a momentary switch under there, plus a 2.5mm jack - possibly not the most robust or weatherproof option, but that's the current choice.)

And fitted on the enclosure:


The initial mount for the reflector was a pleasant surprise in fitting snugly, as measuring such a curved item was part guesswork:


Things like self-tapping screw holes are the result of experimentation. This is where having a 3D printer rather than relying on an external service has been justified so far.

There seem to be two types of vape li ion cells - cylindrical, and the flat LiPo packs. I tend to believe the former don't have a protection circuit built in, but have no real reason for that guess.

In any case, even with a protection circuit, chaining them in series will likely throw them off balance quickly, as the protection circuit of one will kick in first.

I wonder if the seal ring could be 3D printed? Multiple versions with slight diameter variations could be tested, and then the correct size could be treated as disposable, to be replaced on every service.

My memory of the 3N72 ring is the metal part became rusted, and levering it out with a screwdriver probably deformed it.

Jdsk wrote: ↑6 Jan 2024, 2:36pm

SA_SA_SA wrote: ↑6 Jan 2024, 2:33pm Why would 'dynamo's be less common?
...

Same question.

And why would bikepacking be associated with very expensive headlamps?

Thanks

Jonathan

I would love it if dynamos became more common, but I just can't see the market heading in that direction. Didn't Germany stop mandating dynamos for most bikes a few years back? And, perhaps because they thought sporty cyclists would be put off by drag, Shimano tried 1.5W dynamos, which seemed like a backwards step in terms of market segmentation. Add e-bikes too.

I think both lights target bikepacking as they must think some bikepackers will pay these prices. That's not to say that's the highest spending demographic (when you have MTB'ers, roadies, etc), but they have chosen to pitch these lights very specifically, further giving the impression of dynamo lighting and charging as being niche.

Anyway, fair play to them if they get many customers.

While still finalising my DIY dynamo light detailed elsewhere, I wondered if anyone has any thoughts or experience on several of the new dynamo lights appearing on the market. It seems like this very niche area still generates quite a bit of interest (even if, one might argue, that dynamos are going to be less common over time).

There's the Igaro C1:

https://road.cc/content/tech-news/check ... ted-305815

...and the Llum dynamo light:

https://bikepacking.com/news/introducin ... amo-light/

Both certainly look very snazzy. They are eye-wateringly expensive, but I guess angling towards bikepacking does open a few wallets.

The Igaro may be a little too clever for its own good, having no switches and relying on a phone app (!?!)

The Llum looks very smart and robust, but the connections at the back seem rather exposed to the elements.

Both also do USB output. Beam-wise, Igaro seems to have a cutoff in one mode, but apparently automates switching between modes? Llum is a round beam only.

I had a chance to do most of the items on the bullet point list in my last post.

Modifying the reflector hole for the LEDs was surprisingly easy, as the ABS reflector is quite easy to cut.


The rest of the board has been populated.


Sizing was quite conservative, and I probably could have got away with a smaller enclosure (the current design is a 52*52*100mm enclosure, but when buying I saw 42*42*80mm was available). The three tuning capacitors are smaller than their through-hole equivalents, while still having respectable specifications.

And, finally, when testing by spinning by hand... it worked, but the standlight capacitor (which was crocodile clipped in for this test) didn't seem to be doing anything. It was charging up, as I could see the voltage rising on my multimeter. But when the wheel stopped spinning... no light.

That was frustrating. So I did a careful check for continuity on all connections (I was most worried about the LEDs, as hotplate soldering is a very new process for me, and hard to get feedback on). Ultimately, I tracked down the problem to the current sense resistor.


...which is at the top right of the board here. As can be seen, it's somewhat brown, and was reading an open circuit. Turns out that when I scavenged it from another item, I probably killed it with heat. I do need to work on my desoldering game. (There's also quite a lot of flux on the board which I need to clean off at some point.)

So I replaced it with the 0.25 ohms resistor I used when prototyping (the one in the image was 0.2ohms, to see if a brighter, but less long-lasting standlight would be ideal).

Finally, on testing the 0.25 ohms resistor did indeed work, and the standlight functions. I tested the supercapacitor voltages during run-down, and the 3 farad capacitor bank went from 5V to 1.8V in 2m32s, to 1.0V in 3:39, and to 0.9V in 4m19s. The brightness seemed very good for at least the first 2 minutes. Even at 1.0V it was very serviceable. So I'm happy with a 0.25 ohms sense resistor, and 0.2 ohms probably would have discharged too fast.

Beam-wise, it's certainly slightly more diffuse than the original Crivit's beam. This is no surprise, as the 3.5mm Cree LEDs are noticeably larger than the original single 3.0mm Osram, but at least the overall shape seems to still preserve a cut-off.

I'm now on the last bullet point item, which is the enclosure, and bringing it all together.