Adapting battery packs from Cree T6 light to USB charger

For discussions about bikes and equipment.
DoctorRad
Posts: 90
Joined: 10 Nov 2010, 5:48pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby DoctorRad » 29 Apr 2014, 9:07am

Re-reading the YACF forum post, I realised that the crucial phrase is "You could connect only the zener, resistor and SCR across the REEcharge's own bridge rectifier..." (my emphasis). In this case, yes, once the protection circuit kicked in, it would latch on until you stopped.

However, by putting the whole circuit - including an extra rectifier - in front of the Reecharge's own rectifer, the protection circuit will reset every time the dynamo voltage drops to zero, i.e. every half-wave. Basically, in order for the protection circuit to not latch on, you need a diode between the output of the protection circuit and the smoothing capacitor before the regulator; the Reecharge's own rectifier serves this purpose. Such a diode is also included in the Wikibooks 'Crowbar' circuit linked to above as component 'SD1'.

The disadvantage of the extra diode is a small extra voltage drop before you have 'usable' power, but this can be countered to an extent by using a tuning / boost capacitor of the correct value to boost the voltage at lower speeds, as per Martin's Circuits 5 & 6 in the pilom.com link above.

edocaster
Posts: 407
Joined: 10 Apr 2013, 10:43pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby edocaster » 12 May 2014, 11:52pm

Perhaps the first circuit from this page would work?

http://www.maximintegrated.com/app-note ... mvp/id/760

On overvoltage, it seems that Q1 switches off, and that would disconnect the load. Input would be bridge rectifier, load would be regulator. Query which side smoothing cap would be on.

If disconnecting the entire load is annoying, a solid load (e.g. a single power LED) could be connected across the collector-emitter of Q1 to keep output to the load continuous and pull the dynamo voltage back down? Might cause some flip-flopping though.

edocaster
Posts: 407
Joined: 10 Apr 2013, 10:43pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby edocaster » 28 May 2014, 11:26pm

In line with the original post on this thread, I had a go using a Magicshine-style battery to charge a phone.

I used a cigarette lighter socket USB charger I had lying around (I think it was a freebie from one time I upgraded my phone). Pretty much the kind of thing you get in a pound shop. Very temporary bodge to get it connected to the typical 5.5mm DC plug the battery expects.

MC34063 charger-a.jpg


Seems to work fine on an old smartphone I had lying around. So far I've topped up two phones and charged a couple of loose batteries in a USB charger, and the battery pack is at over 7.9V, so plenty of juice left. I'll see how far it goes down while still outputting 5V to gauge how useful this could be for hub dynamo USB charging too (leaving aside the overvoltage protection circuit question for now...).

Technical notes: the cigarette socket USB charger is labelled 12 to 24V input, yet seems to work fine. It's based on a switching regulator (MC34063) which I believe most of the cheapies use.

DoctorRad
Posts: 90
Joined: 10 Nov 2010, 5:48pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby DoctorRad » 29 May 2014, 9:39am

Good work! I've done something similar myself with a 14.8v Li-ion Lumicycle battery back, but had forgotten about it as it worked a little intermittently. I think there was a connection issue between the regulator and the phone - probably due to the micro-USB plug-socket interface - and once the connection was broken, it appeared that the regulator would not 'restart' unless the battery was disconnected and reconnected.

I did some tests with an adjustable power supply when I was building the contraption, and the regulator would start working at around 7v input. A fully charged 7.4v Li-ion or LiPo battery pack would therefore be adequate to power it. Many such regulators would be similar.

As for the issue of overvoltage protection, I was thinking this morning how extraordinarily useful it would be if dynamos came with a crowbar protection circuit set to around 24-30v, and then realised there was probably a gap in the market...

edocaster
Posts: 407
Joined: 10 Apr 2013, 10:43pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby edocaster » 29 May 2014, 11:29pm

DoctorRad wrote:As for the issue of overvoltage protection, I was thinking this morning how extraordinarily useful it would be if dynamos came with a crowbar protection circuit set to around 24-30v, and then realised there was probably a gap in the market...


Clamping at that range would be very welcome. Not something in the dynamo, but just downstream of it. Note I'm talking about voltage clamping and not using a zener to trigger a thyristor to shunt down to near zero voltage.

What I don't understand is if you fit a zener or TVS diode as a shunt regulator to clamp at around 24 to 30V, at the breakdown voltage what current flows through it? Surely not the full 500mA of the dynamo? As that would be an eye-watering 12 to 15W dissipated in the diode. Surely that load would negate the voltage rising that high in the first place?

DoctorRad
Posts: 90
Joined: 10 Nov 2010, 5:48pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby DoctorRad » 30 May 2014, 9:17am

It is possible to use back-to-back zener diodes for overvoltage limiting protection, but from everything I've read, they have to be fairly beefy, e.g.

http://pages.citebite.com/n3m3j0o6q0xfm

"For over-voltage protection, it uses two 7.5 V 5 W zener diodes across the dynamo input. They can get hot when the headlights are turned off and there is no load at the output."

That's quite a low protection voltage, and I would expect the power requirements to rise as the protection voltage rose.

Given that any sensible load drawing at least some current should keep the dynamo output voltage fairly sensible, I think crowbar protection is entirely appropriate. It will short the dynamo, thereby meaning its output voltage (and power) will be minimal, and then reset on the next half-wave. Further research suggests that a simple diac component across the dynamo terminals should provide what's needed, if used before any rectification:

http://en.wikipedia.org/wiki/DIAC

One cost effective option I've found is from Farnell:

http://uk.farnell.com/stmicroelectronic ... dp/1748975

This will protect at a maximum of 34v. One could then likely use a simple rectifier, capacitor, regulator and USB socket to provide a charging solution. The regulator would need to be happy with 34v input (minus the rectifier voltage drop), which this one is if used with an input capacitor:

http://uk.farnell.com/tracopower/tsr-1- ... dp/1696320

I am yet to implement it, but given all the research I've done, this should provide the lowest possible part count for a properly protected dymano regulator, essentially just five components plus the circuit board if one uses a 'single component' rectifier.

edocaster
Posts: 407
Joined: 10 Apr 2013, 10:43pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby edocaster » 1 Jun 2014, 12:59am

Ah that's great, thanks. I'm learning something new every day.

My aversion to a crowbar comes from my experience of charging a smart device which seems to spend too much time negotiating a charge current, with the result that the circuit goes overvoltage before the full load is made available, if accelerating too fast. At least I think that's what's happening with the LM2940, which shuts itself down if it detects a high transient voltage. This problem would be true for any protection circuit, I guess, whether it breaks or crowbars the circuit. I would hypothesise that this is the cause of most of the problems people have had with some dynamo USB chargers (well, those which actually have a protection circuit and so don't otherwise self-destruct).

I think perhaps a two layered protection system would help. The 32V DIAC can be the final line of defence. But to slow the point at which the DIAC will kick in, perhaps two back-to-back zeners of, say 18V, with a current limiting resistor between them? All beefy, 5W parts. The resistor is to prevent the zeners taking the full 500mA and failing, with the compromise that the voltage will still rise (across the resistor) once the zeners are triggered. When the voltage across the resistor and both zeners reaches 32V, the DIAC will trigger. The hope is that with the extra load of the resistor and zeners, the speed at which the DIAC triggers could be higher than normal cycling range?

DoctorRad
Posts: 90
Joined: 10 Nov 2010, 5:48pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby DoctorRad » 2 Jun 2014, 10:13am

I don't think the crowbar would cause too much of a problem. If the regulator is pulling a small enough current to make the diac conduct - i.e. for the smoothing capacitor and reguator to need protection - the smoothing capacitor will be charged up to the diac's breakover voltage before the protection kicks in, typically 32-34v or so. With a small current pull, the capacitor should hold sufficient charge to supply the regulator until the end of the dynamo's half-cycle of output when the diac will reset.

Two things will stop the diac conducting every half-cycle and allow the circuit to enter 'normal' operation. Either

1) the wheel slows down so the dynamo's output is no longer above the diac's breakdown voltage, or
2) the regulator starts to pull current because of an attached device.

In the case of 1), the capacitor will discharge due to the quiescent (residual) current of the regulator until it can be recharged by the now lower output voltage from the dynamo / rectifier. The same will happen in the case of 2), but it will happen more rapidly due to the current draw of the regulator. What is then needed is a smoothing capacitor sufficiently large to keep the voltage across it above the minimum input voltage of the regulator (6.5v for the Traco regulator linked to above) at the desired minimum charging speed. However, neither 1) nor 2) are dependent on the protection circuit / component, as it is not in operation.

edocaster
Posts: 407
Joined: 10 Apr 2013, 10:43pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby edocaster » 22 Jun 2014, 12:32am

I tried a DIAC (a DB3 rather than the DB3TG previously linked to - the only difference is the DB3TG has a more precise breakover voltage). When I saw how tiny the DIAC was, I was a little worried, but the datasheet said it could handle 2A...

Placing it across the dynamo AC with no other load, the voltage rose and then the breakover seemed to occur around 19km/h, with only about 16V across the DIAC.

Two other things happened: 1) The voltage fell to about 5V as speed continued to increase, yet on dialling back speed voltage only really rose again around 11km/h, and 2) smoke came from the component.

I think it's damaged, as the tiny DIAC probably can't handle bike dynamo current. A closer study of the original datasheet shows a current chart where the 2A rating only applies to very short pulses (20 microseconds). Annoyingly, I found another datasheet which confirmed the part is rated 150mW, so can't expect to carry 0.5A across its operating voltages.

I have no idea why the breakover voltage was roughly half what was predicted, but I'll assume this part is now trashed (I have spares, but the low power rating means I can only really use the DIAC as a trigger for something else).

freeflow
Posts: 1191
Joined: 29 Aug 2011, 1:54pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby freeflow » 22 Jun 2014, 4:25pm

'm not an electronic hobbyist so I've just bought one of these which looks as though it will fit in the battery bag.

http://www.amazon.co.uk/gp/product/B00G ... UTF8&psc=1


I've had the battery pack for a month or do now. It has taken a few recharge cycles for it to get upto full charge but is a very adequate performer. It is a little too tall for the flap of the battery bag to close but and elastic band is doing the job until I can persuade Mrs F to extend the velcro or come up with another solution.

Yesterday I did my first 300k audax. I rode with my phone screen permanently on at 75% brightness (sony xperia z ultra 6.4 inch screen) running ipbike (for ant+ sensors) and osmand in the background (for voice navigation) which was a big power drain and the external battery lasted from 7am until just before 5pm (as evidenced by the green charging light going out), leaving me with a fully charged phone (having started with about 75% charge).

I also have on order a two 18650 lithium battery external battery pack which might be a better long term solution as I have a number of 18650 batteries I use in torches when winter riding.

Consequently i think I'm now pretty much sorted for external battery support.

edocaster
Posts: 407
Joined: 10 Apr 2013, 10:43pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby edocaster » 22 Jun 2014, 11:11pm

Looks like I've fried a few other things...

The MC34063-based cigarette socket charger pictured above is now out of action. I tried hooking it up to a dynamo (after the required rectifier and smoothing capacitor). Cranking to about 30km/h it failed in the worst possible failure mode, letting unregulated voltage go straight to the output. Ruined a 4 port unpowered USB hub I had hooked up at the time, although the old smartphone behind that survived.

So, the lesson is don't play around with random switching regulators! Or at least fit a zener as a final line of defence.

I think this will persuade me to buy an off the shelf solution...

DoctorRad
Posts: 90
Joined: 10 Nov 2010, 5:48pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby DoctorRad » 23 Jun 2014, 9:24am

Given that it seems very difficult to find diacs with a higher power rating, it looks as though the zener / thyristor crowbar solution is probably the best fallback. I wasn't aware that the power rating of those diacs was so low, otherwise I would not have suggested their use.

edocaster
Posts: 407
Joined: 10 Apr 2013, 10:43pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby edocaster » 7 Jul 2014, 10:40pm

I tried breadboarding a thyristor-based crowbar, based on this Shimano patent: http://www.google.com/patents/US6788030. The attraction of doing this on the AC side of the circuit is it resets on every half cycle (so you can turn a light on to override the crowbar without stopping).

Playing about with an 18V TVS diode, and then a 22V zener, the results were unexpected. Barring a few false starts when nothing initialised (probably a bad connection in the breadboard), when it worked the crowbar kicked in at a trivially low voltage (about 10VAC). That's probably good enough, but doesn't really match the zener/TVS voltage ratings.

The thyristors are BT151-800R - not sensitive gate devices.

Eventually, using two 18V TVS diodes in a row on each side of the circuit allowed the crowbar to kick in around 15.5VAC. I'll probably build that into an actual circuit, as it's usable. It didn't seem to matter what resistors I used to bias the gate - I settled for around 40 Ohms.

I know TVS diodes aren't precision devices, but I got the same from the zeners. Is there any explanation why the triggers seem to be around half the expected voltage? I got the same 'half voltage' effect during my ill-fated DIAC tests too, IIRC.

Could it be something to do with my (cheap) multimeter? Or are the protection devices somehow kicking on on the peak-to-peak voltages (surely that doesn't make sense?)

DoctorRad
Posts: 90
Joined: 10 Nov 2010, 5:48pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby DoctorRad » 8 Jul 2014, 9:08am

No idea why it's not working as expected, but the circuit I linked to previously on YACF (zener and thyristor after rectifier) will also reset every half-cycle so long as it is followed by a blocking diode before the smoothing capacitor.

edocaster
Posts: 407
Joined: 10 Apr 2013, 10:43pm

Re: Adapting battery packs from Cree T6 light to USB charger

Postby edocaster » 19 Jul 2014, 11:50pm

I put together a circuit using the thyristor on the DC side, with a following diode as DoctorRad indicates, to prevent the smoothing cap keeping the crowbar from resetting.

I think it works well. Using 2 18V TVS diodes in a row, the crowbar triggers at about 19V. Not a precise or easy to predict behaviour, probably due to a mixture of leakage from the TVS diodes, and spiky, unsmoothed voltage exceeding the averaged figure which my multimeter displays. Once triggered, the voltage reading drops and stabilises at around 7V or so, although I'd imagine the actual waveform starts with a spike at 19V and then drops off to almost nothing. Unsurprisingly, this makes the dynamo hum and buzz a bit. But it does prevent overvoltage to any components in the charger circuit, and the thyristor doesn't heat up either.

The crowbar resets every half cycle - i.e. I can turn on the lights without having to stop.

I'm hoping to package this into a thin tube, and might dispense with a switch, now that the overvoltage protection seems to be working.

The only other issue worth noting is that the regulated output voltage does sag under load. All my experiments with linear and switching regulators connected to a dynamo show this behaviour, and it depends on the load. I.e. if it's a light load (somewhat less than the 500mA the dynamo can supply), there's no sag, but certain loads expecting to take 1A or more will cause a sag - about 0.7V in the case of one phone, and a massive 1.1V for one USB battery pack. I worry that this can cause a device to not charge at all, wasting the effort. To offset this slightly, I added a schottky diode to the ground pin of the LM2940 regulator, raising it to between 5.2 and 5.3V unloaded (although it's clearly not quite as regulated any more).