Making a mobile device charger to work with your hub dynamo, part 2

[Edocaster]

This is pretty fascinating.

I note that the KIS-3R33S here is a switching converter (built around this: https://www.monolithicpower.com/Lists/Product_Synchronous%20Switcher_5717/DispForm.aspx?ID=35), while the first thread covered linear regulators. Near the end of the first thread, several links pointed to the most popular DIY dynamo USB charger design, which is based around a LM2940 low dropout regulator - I built one myself too (http://forum.ctc.org.uk/viewtopic.php?f=5&t=79811#p714769).

So what are the advantages and disadvantages of each approach for our application? I can think of:

- The KIS-3R33S needs, on paper, a 2V overhead before it outputs 5V (although Andre's post suggests some riders have had decent results from just a 1V margin). The LM2940 only needs 0.5V.
- The KIS-3R33S is overall more efficient, as it's a switching converter. But that advantage over a linear regulator comes into play if the output voltage diverges from the input voltage. Am I right in thinking that a dynamo voltage climb would only happen if the circuit was unloaded anyway, in which case efficiency would then be moot point?
- How would each handle the 'no load high voltage' problem? Both are rated below 30V, but the LM2940 would be trying to burn off the excess as heat (although some report it does have protections built in which would result in a shutdown).
- The KIS-3R33S seems to eliminate some of the extra capacitors that linear regulators need - has anyone seen if the output is clean, especially as it doesn't seem like it was designed for rectified bike dynamo current?

[Andre Jute]

Thanks for the tip of trying it first on a cheap phone; I was wondering if I would survive blowing up an iPhone!

My current setup is a motor and a 36V (nominal) battery, so I have no voltage-drop problems over the board. Quite the contrary. My fear is that a glitch could deliver the full 42V of a fresh pedelec battery at the USB port. I've therefore ordered a buffer battery with two USB outputs for 1A and 2.1A, and considerable protection circuitry. I'll rig this lot up as 36V battery>regulator>buffer battery>device to be charged.

My remark about 1V headroom being enough is based on a sample of one, albeit an engineer in whom I have full confidence.

In your situation I'd want to clean up the current between the dynamo and the regulator board, as further up this thread. My understanding is that many USB devices and especially phones, and iPhones particularly, are pretty sensitive to the quality of the current you feed on the charge line in the USB port.

Most hub dynamos will give 90% of full power at 10kph or below. With no load, you should certainly see 6V, most likely more, and by 15kph you should see 7-8V even with the load. But what will be interesting to know is how fast you need to go to keep the dynamo/buck board combo putting out enough current to keep an iPhone interested.

Good luck.

[Edocaster]

Bugmenot - were the two data pins (middle pins) soldered together on your module? It should be apparent if you look at the underside of the USB connector, or check for continuity with the two middle pins of the USB socket. Shorting the data pins is (so I hear) supposed to be a de facto standard to tell smartphones you're using a dumb charger, hence to take maximum current. Doesn't apply to Apple, though.

Has anyone come up with a protection circuit for this, to prevent or deal with overvoltage at the input? Maybe a zener, current limiting resistor and power LED at the rectifier (but how do current limiting resistors work with current sources like dynamos? Will they still try to force 500mA through?)

Or perhaps the module has its own protection once voltage hits 24-ish?

[mickeg]

Resurrecting an old thread here.

Thanks to Andre for posting this information on this cheap USB charger device.

I bought one from Ebay.  And from Radio Shack I got a rectifier and capacitor and wired it up.  I do not know if a capacitor is necessary, but it made sense to me to smooth out the current flow out of the rectifier before it goes into the USB board.  It all fit quite nicely into a small pill bottle for prescription medicine.  Works great so far.  I say "so far" because I just tested it today for about a mile and a half.  (My bike computer quit working after 0.55 miles as my computer never likes cold weather.)  But while my bike computer was working, I found that I was getting usable current out of the device at about 6 miles per hour or roughly 9 km/hour.

This link is to an Ebay posting for a very useful device to measure the voltage and amperage coming out of the USB.  It displays voltage for a few seconds and then amperage for a few seconds, then back to voltage, etc.
http://www.ebay.com/itm/141342187660

If the link above to Ebay does not work, go to Ebay and put into the search field USB Power Detector Voltage Current and see what comes up.

I found this volt and amp meter to be extremely useful, it made me confident that I did not screw anything up and I knew instantly if it was working or not.  It however does not have a very bright LED readout, even on a foggy day I had to hold my hand over the display to read it.  

I did not put an on/off switch on my charger like Bugmenot did on his, instead I plan to just unplug and stow it when not in use.

In the photo I have my home made charger in the orange pill bottle, the volt/amp meter is the blue thing plugged into that, and since I did not want to blow up anything expensive during my testing I used a AA battery charger that is powered by USB to charge two AA rechargeable batteries.  All of this is held on top of my handlebar bag with bunge cord.  The display on the volt/amp meter is blank, the bike was not moving when I took the photo.

Thanks again Andre.

[Andre Jute]

I like your proto, Mick! The pill bottle is particularly clever: watertight enough if mounted with the USB port facing downwards, light, small, not obtrusive, can be painted any colour if desired, the works accessible at any time. Perfect.

Thanks for sharing the cheap and handy socketed USB VA meter. I'm ordering two for my bike and two for entirely non-bike purposes.

[mickeg]

An update from my post on November 22, above.

Did a longer test yesterday of that USB buck board that I bought on Ebay.  I used it for about an hour, it gave consistent performance the entire time.  I used the same USB powered AA battery charger to charge two AA NiMh batteries for a consistent load on the USB output.  I had 5.07 volts and 0.29 amps coming out of the USB at about 8 mph (13 km/hr).  At 15 mph (24 km/hr) had 5.07 volts and 0.31 amps.  I consider the difference between 0.29 and 0.31 amps to be not much more than round off error.  Thus, in this case it appears that the limiting factor was the ability of my AA battery charger to accept more current, the USB delivered all it could take at 8 mph.  I had good current at slower speeds, but voltage was below 5.  Voltage did not climb at higher speeds so I am confident that it will not overpower anything that I plug into it, the highest voltage I saw was 5.08.  I am sure that a heavier load would require a greater minimum speed but my smartphone takes about the same current so I do not expect to do any more testing at different loads.  If anyone is interested, I am using an SP Dynamo PV-8 hub for my power source.

Output was actually 0.04 amps more than cited above.  The volt/amp meter consumes 0.04 amps that is not included in the digital reading.

I made two of the USB devices since they are so cheap to make.  I tried running two in series on the dynohub output.  But to make a long story short, I have concluded that running them in series is not a good idea.  I did not try running both USB devices in parallel as I did not have the proper wire connectors for that.  I might do that test some day but am not in any hurry to do it.

The highest speed I got was 15 mph (24 km/hr), I was running 2.0 width Marathon Winter tires that are exceptionally slow.  With occasional snow and ice and slushy patches, even with a good tailwind it was not optimal conditions.  But realistically I rarely exceed 15 mph for more than a few minutes when cycle touring, thus the speeds I cycled at was a good test for touring, most of the time I was about 11 mph (18 km/hr).  I have no plan to try faster speeds later, I consider this test to be adequate.

Plans for phase two.  I noticed on Ebay that they make these USB buck boards that have a built in volt meter.  I have ordered one, plan to wire that up in January.  I am not sure but I think that the volt meter will give me a reading of the input voltage that comes out of the rectifier and into the buck board.  It will be interesting to know what voltage is coming out of the hub.  If however it provides voltage at output, I already have that so nothing is gained and I can quit experimenting.

I plan to wire that USB device (with voltmeter) in parallel with a switched headlight.  I am assuming that when I switch the headlight off, that full power would go to the USB.  But with the headlight on, that power would be split to both headlight and USB device with the most current going to the least resistive load.  And if the USB device is not powering anything, I assume that full power would go to the headlight.

I also am going to reconsider the wire connectors I am using.  Right now I am using 2.8mm spade connectors.  But those are not the most convenient connectors to use if I frequently plug or unplug different things into the wiring.

[Andre Jute]

Thank you for sharing that thorough report with us, Mick.

You're right not to set aside time for a faster run. When charging any device from a dynamo, what is relevant, rather than your top speed, is your median speed, which is the speed you travel most often. It may or may not lie close to your average speed.

Because of the way German bicycle lighting regulations are written, all these German (and Dutch and OEM Japanese/Chinese) dynamos give 90% (or more) of rated output of 3W at 9mph/15kph. By luck or good judgement you chose a speed for your test a fraction over what the designer of your dynamo chose as the speed at which to optimize output. Higher speed is not likely to tell you anything much.

However, to make that 90% requirement at 15kph, all these dynamos are slightly oversize, and all at higher speeds output will rise to an overvoltage when encountering a resistance set to draw half an amp at 6V, i.e. 3W of power. There is therefore theoretical scope for charging two devices, though I imagine one would have to be pretty undemanding and, unless you builld in sophisticated control circuitry, undiscriminating as well — I take it you are aware that many USB devices' first line of defense is to shut themselves down when the line voltage fluctuates more than marginally.

Good luck and please keep us updated.

Any little module board I've ever seen with only one voltmeter measured its own output; if the input measurement was desired, a second meter was added.

[il padrone]

When charging any device from a dynamo, what is relevant, rather than your top speed, is your median speed, which is the speed you travel most often. It may or may not lie close to your average speed.

Gah!!

Sorry, being pedantic but.......  mean, median, mode.

The speed you travel at most often will be the mode. The median is the mid-range speed between the highest and lowest.

Eg. if you accelerated from stop to 30 kmh, then ride at 30 kmh for an hour, then slow to stop.
Mode = 30 kmh
Mean = something slightly less, say 29.9 kmh
Median = 15 kmh

Just for clarification

[JimK]

The median would be the speed x where half the time you went faster than x  and half the time you went slower than x.

Suppose that 40% of the time I was going 5 mph, 30% of the time 10 mph, 20% of the time 20 mph, and 10% of the time 30 mph.

The mode is 5 mph, the speed I was going the longest time.
The median is 10 mph.
The mean is  12 mph.

The mode is tricky for a continuous variable like speed. I guess it's where the slope of the cumulative probability distribution hits a minimum, i.e. the density of observations is a maximum.

[Andre Jute]

Thanks, Il Padrone, Jim. I agree, I stated my choice of the median clumsily. The tutorial is in an outrage, the roadies being particularly bitter at me calling them slow...

You could measure the mode by taking a speed-reading every minute or some other close-ish interval. This is a standard procedure, for instance in GPS which takes a reading every x meters.

But my choice of the median, by Jim's definition of half the time faster, half the time slower, is based on nobody riding at a constant speed all day. Also, as long as the median is above 9mph/15kph, the dynamo is working at close to maximum efficiency, and above 9mph/15kph the improvement in output is marginal (max 11% i.e. 10/90). Since we're talking about tourers, it isn't too much of a stretch to suppose that most of them would be going around the same speed as Mick, and those who go much faster won't be putting much extra charge into the battery for it.

So I think that, while I submit to Il Padrone's pedalgogical (!) point, that in most touring cases the median of speed  would come closer than the mode as a predictor of potential charge.

At road bike speeds, credit card touring, ultralight touring, the mode may theoretically be the one to work with if the measuring difficulties can be overcome, given that those cyclists are interested in carrying the weight of the device and the chargers.

This is all subject to an empirical test. Until we have one, I'm not planning on getting religiously committed to one plan or the other.

[mickeg]

Median, mean, mode, etc., I think unimportant in this case.  

In the present context, I think all that matters is that you have a speed that exceeds the minimum speed necessary to get all the current you need out of the USB.  Too slow and current (amperage) is insufficient to power your device and/or charge the battery.  Too fast and you get no additional benefit, you have dynohub capacity that is not being used.  

That minimum speed will of course depend on the gizmo that you are pushing the current into.  In my case I found that the minimum speed of about 8 mph to be adequate to give me 0.29 amps (plus 0.04 amps to power the ammeter).  But if I was trying to charge my tablet, that pulls over 0.4 amps out of a USB so I will need a higher minimum speed, that speed to be determined later in warmer weather when such testing is more enjoyable.

But, if you want to debate median, mean, mode, feel free to do so in my absence.

This is the board I have ordered for phase two.  Based on one of the photos, it appears to have an input voltage of 38 and an output voltage of 5.059.  The voltmeter on the board reads 38.0.  That is why I assume that the voltmeter will give me the voltage out of the rectifier in my circuit.
http://www.ebay.com/itm/231357363046?_trksid=p2059210.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

I am sure I will have a small voltage loss across the rectifier, but neglecting that the volt meter should give me the voltage out of the hub.  If so, that would mean that this do-it-yourself USB device will offer something that the Plug, E-Werk, USB-Werk, PedalPower, etc. do not offer.  Maybe the typical cyclist has no interest in voltage readings.  But I am an engineer by training and if I am on a bike tour trying to charge a battery I will be interested in things like voltage and amperage.  For example it might let me know if I can charge a battery while also using a headlight.

Since I expect to put this into another transparent pill bottle, I should be able to read the voltmeter through the side of the pill bottle when I have assembled it.

[Edocaster]

Try charging a device that will definitely take as much output as possible, such as a USB power pack or a smartphone. 0.3A is enough for smaller devices, but isn't really a step up from using a linear regulator such as the LM2940 or L7805 - with an LM2940, you should be getting 0.4 to 0.45A.

The most reliable way to measure input voltage is to have the bike stationary (possibly upside down), with the wheel driven by something else (drill motor, another bike's rear wheel, etc), and to have parallel output to both the USB charger and a multimeter. You can rig up something with a breadboard. A cycle computer to read speeds helps too. Then, set up a video camera to record all the readings (speed, AC input voltage, DC output voltage - and current readings if possible) so you can note them down afterwards.

My experience of DC-DC regulators is that they can be flaky on bike dynamos, and if you don't load them properly they can go overvoltage on input quickly - I'm guessing because the duty cycle has decided to runaway in a particular direction. Plus, the main advantage of a switching regulator, namely being able to trade voltage for current rather than burning off the excess, isn't really realised unless the system is built to work from a higher voltage (say, 12V). There's a system kicking around on German forums I believe, called 'forumslader', that seems to do just that: It puts the 5V switching regulator behind a 12V linear regulator, ensuring the switching regulator never sees a wildly fluctuating voltage.

[bugmenot]

I for one would like an update on the DC-DC board here.  Any notable difference?  Better output amperage? or worse?  It claims the 2A output - which could power an ipad or smart-charge cell phone.

The listing isn't particularly clear about the buttons either---- what do the low power & output buttons toggle?

-Bug

This is the board I have ordered for phase two.  Based on one of the photos, it appears to have an input voltage of 38 and an output voltage of 5.059.  The voltmeter on the board reads 38.0.  That is why I assume that the voltmeter will give me the voltage out of the rectifier in my circuit.
http://www.ebay.com/itm/231357363046?_trksid=p2059210.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

I am sure I will have a small voltage loss across the rectifier, but neglecting that the volt meter should give me the voltage out of the hub.  If so, that would mean that this do-it-yourself USB device will offer something that the Plug, E-Werk, USB-Werk, PedalPower, etc. do not offer.  Maybe the typical cyclist has no interest in voltage readings.  But I am an engineer by training and if I am on a bike tour trying to charge a battery I will be interested in things like voltage and amperage.  For example it might let me know if I can charge a battery while also using a headlight.

Since I expect to put this into another transparent pill bottle, I should be able to read the voltmeter through the side of the pill bottle when I have assembled it.

[mickeg]

This one did not work for me.
http://www.ebay.com/itm/231357363046?_trksid=p2059210.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

The specific problem was that I put it with the rectifier and capacitor and that board into a small transparent plastic bottle.  If you look at the photo on the Ebay listing, there are two small white rectangles on the right, they are button switches.  I do not recall what they do (my experiment was months ago, memory fads) but since I was unable to push the buttons when it was inside a bottle, it was a dismal failure.

At your desired current of a 2 amp output, at 5 volts that is 10 watts.  I do not think I will ever tour fast enough to spin my hub at the rpm that I would likely need to generate 10 watts.  So, I would not plan to power a device that needs that wattage.  A device like a storage battery that could charge at variable wattages for variable bike speeds would likely be more successful.  Then charge your Apple from that battery.  I do not have much experience with Li Ion battery packs, others can provide better advice than me.

I did figure out that some of my capacitors blew out, I suspect voltage overload, my rectifiers had a high voltage rating but I think my capacitors were not rated that high.  If nothing pulls power out of a dynohub, the voltage out of the hub can climb quite high and I think that is what happened.  I bought some zener diodes to add to my circuits in the future, I think I bought 20v diodes, hopefully they will keep the voltage low enough so that I do not blow any more circuits.  I just hope my diodes do not overheat when I go fast down hills.

That is really the only update I can offer.  I am off the bike until sometime between mid June and end of July, Doctors orders.  Thus, no testing will be feasible for several more months.

[Danneaux]

Hope you feel better soon, mickeg! Healing wishes your way.

All the best,

Dan.