entry level lights to see by

On Tue, 1 Jan 2013 17:22:51 -0800 (PST), Frank Krygowski
wrote:

On Jan 1, 5:36*pm, Jeff Liebermann wrote:


What you need is a switching regulator. *Assuming an input of 6VAC at
0.5A (3 watts) and an output of 5V at 0.5A max, something like this
should work:
http://www.ebay.com/itm/251169704387
(Notice... no heat sink). *Add a Shottky diode bridge and filter cap
on the input and you're done. *With a usable input range of 1 to 5V,
it should also work at low speeds.

Disclaimer: *I haven't tried it.

I assume you're talking about a regulator based on pulse width
modulation,

Correct.

and I've wondered about whether such a thing can be used
with bike dynamos. I'm an ME, not an EE, but is it really feasible to
rapidly switch a power source with as much inductance as a dynamo?

- Frank Krygowski

Yes, it can. Once the AC produced by the dynamo is converted to DC by
the bridge and the BFC (big fat capacitor), such things are easy. I
would not run unfiltered pulsed DC (i.e. half wave rectified) into the
DC to DC converter as that would surely cause problems.

The inductance of the power source is not a consideration. Only the
inductance or capacitance of the load is important. At the
frequencies typically used for controlling the brightness of an LED
(about 1KHz) the capacitance of the LED is not significant. There's
no inductance in the LED. At worst, the change in load resistance
during the on/off cycles might produce some ripple current in the BFC
after the diode bridge, which might cause some self heating. I doubt
if it will be significant, but the solution is to use either low-ESR
caps, or just a bigger capacitor.


--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On Tue, 1 Jan 2013 17:40:53 -0800 (PST), Andre Jute
wrote:

On Tuesday, January 1, 2013 10:36:38 PM UTC, Jeff Liebermann wrote:
http://www.ebay.com/itm/251169704387

Thanks, Jeff. That's where I was heading next. If you'd read all of the thread I referred you to, you'd discover that the outline design is not an engineering exercise but a teaching tool for people who know nothing, zilch, nada about electronics. You don't jump right in with switching designs, you first take them by the hand and put something linear they can understand in their hand. Then you can move on to switching designs.

Thanks for finding the right board for me!

I'm having 2nd thoughts about the first board I found. The problem is
the acceptable range of input voltages. According to my crude
testing, I can sometimes expect 10VAC from the dynamo at suicidal
speeds. That's higher than the range of acceptable input voltages.

There are three basic types of such DC to DC converters. Boost, Buck,
and Both. Boost requires that the output voltage is greater than the
input voltage. More simply, it can't output less than the voltage
input. Buck is the opposite, where the input voltage needs to be
greater than the output voltage. Both, can output both higher and
lower than the input, but is generally more complexicated.

I found these, which seem better.
http://www.ebay.com/itm/261097668334
http://www.ebay.com/itm/400375205633
2.5VDC to 6.0VDC input and 4VDC to 12VDC output. I'm not thrilled
with having 12VDC appear on a USB connector output. Careful with this
one.

http://www.ebay.com/itm/180947008880
4.75VDC to 35VDC input, which covers most of the range. Output is
1.25VDC to 26VDC, which should handle most anything.

http://www.ebay.com/itm/390518044035
3 to 5VDC input. 5V output. This has the same problem as the first
one, where it's a boost regulator, where the input voltage cannot be
higher than the 5V output voltage.

I'm sure there are more, but these are what I could find between
irritating phone calls from relatives and customers.






--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On Tue, 01 Jan 2013 18:19:54 -0800, Jeff Liebermann
wrote:

One mo
http://www.ebay.com/itm/310506835676
3.5VDC to 30VDC input.
4-30VDC output at 2.5A max.
From the photos, it appears to be running at 80% efficiency.

So far, this one has the best chance because it can handle input
overvoltage. However, I can't tell what will happen if the output
voltage is set below the input voltage without actually trying it. The
auction title says "step up" (boost), which might be a problem.


--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On 02/01/13 12:25, Jeff Liebermann wrote:
On Wed, 02 Jan 2013 11:04:15 +1100,
wrote:

How low can you get a Schottky diode forward voltage drop, with sensible
forward current and reverse voltage specs for the job?

0.15V to about 0.40V depending on forward current. My guess(tm) is
that this thing will be running at about 0.5A, which is the nominal
rating on dynamo.
http://www.digikey.com/product-detail/en/CDBA340L-G/641-1258-1-ND/1963390?cur=USD
The data sheet shows 0.25V at 0.5A. Two diodes in series for the
bridge would make it 0.5V drop total.

Also, note that there are other switchers for sale on eBay. I did
some more looking and found these that would probably be as good or
better.
http://www.ebay.com/itm/261097668334
http://www.ebay.com/itm/180947008880
Without a real data sheet, I can't tell what these things are doing. I
guess one has to buy an assortment and Learn By Destroying(tm).

Wouldn't it be better still to use a MOSFET bridge? MOSFET on
resistance these days can be pretty low, and a quad opamp IC to drive
them shouldn't be to difficult to arrange for even better performance,
at the cost of a few extra components that is.

Well, yes. A synchronous rectifier would be more efficient. See
circuit #12 at:
http://pilom.com/BicycleElectronics/DynamoCircuits.htm

I was thinking more like
http://www.extremecircuits.net/2010/05/power-mosfet-bridge-rectifier.html

The FETs used there have an Rds on of about 0.02 ohm.

However, we're talking the difference between 0.25V per device for
Schottky versus maybe 0.05V per device for an IR HexFET:
V = 0.5A * Rds = 0.5A * 0.1 ohms = 0.05V
At 0.5A, that's a power loss of:
W = 0.25V * 0.5A = 125 milliwatts

And 25 milliwatts per FET (using your suggestion of 0.1 ohm on
resistance), plus a bit for the control circuit.

Let's not forget there are 2 devices on most of the time, so it's 250 mW
vs 50 mW.

out of a delivered 3000 milliwatts from the dynamo. I think this can
be safely ignored in the name of simplicity.

8.3% vs 1.7%? I know what I'd rather.

If the circuit was well designed and potted, I think it would be quite
reliable and simple enough not to cause issues. Power dissipation would
be so low that heat sinks would be unnecessary I think.

Obviously 4 Schottky diodes will be simpler and cheaper.

Personally, I think the whole dynamo power LED light is done wrong and
reeks of ultra conservative design. The dynamos were originally
designed to power incandescent 6V lamps, which they do reasonably
well. The voltage and power output nicely match several 6V 0.2A bulbs
in parallel. However, nobody but kids and department store bicycle
shoppers buy incandescent bicycle lamps these daze. Time to
re-examine the assumptions.

Some dynamos with appropriate globes obviously used to work ok most of
the time. My dynamo doesn't have very good regulation, and used to blow
globes at the most inopportune times, i.e. going fast down hill.

I find the dynamo output and LED driving requirements to be a match made
in heaven. They love each other and have no need for external bits to
make them work well together - and much better than incandescents!

The requirements for running a 2 watt LED are quite different. It
needs brightness regulation, dimming, and flashing. The circuitry
needed to do all this is inherently lossy when running a low voltages.
The solution is to build a dynamo that produces more voltage and less
current at the same power level. For the 3 watts, instead of 6V at
0.5A, it should produce 24V at 0.125A. This would go to a DC to DC
inverter to run the LED. While the forward voltage drop of the bridge
rectifier and saturation voltages of the switching components
contribute substantial losses in a low voltage system, they are far
less significant in a higher voltage system. If you want efficiency,
you need higher voltages. (As an added bonus, the copper losses in
the wiring will be less with higher voltages).

Flashing and dimming? C'mon. Talk about trying to keep it simple! ;-)

Oh, and I thought we wanted to rectify the dynamo output to charge some
battery?

But anyway, try running a dynamo with less load and you will find the
output voltage is higher. Mine with a constant 14.2 ohm load produces
6.8 V/0.47 A @ 20 km/h. At 30 km/h, it's producing 8.2 V / 0.58 A, and
at 50 km/h, 9.3 V / 0.65 A. (Do some power calcs and see there is far
more than 3 W coming from my 3 W dynamo ;-)

My LED lights are starting to work (producing reasonably bright light)
at just above walking pace (5-6 km/h), with only about 3 V / 0.22 A from
the dynamo.

You want higher voltages? Run at reduced load, or add a transformer!
(I know, the weight weenies will hang me for that one ;-)

But, I think low voltage, low loss designs are not that difficult these
days.

I have some other criticisms of bicycle generator technology, but that
can wait for another rant.


Dynamo-Powered LED Light Circuits for Bicycles
http://pilom.com/BicycleElectronics/DynamoCircuits.htm

Read it already.

Dynamo light testing:
http://www.myra-simon.com/bike/dynotest.html

Been there. Done my own testing ;-)

--
JS.

On 02/01/13 13:19, Jeff Liebermann wrote:
On Tue, 1 Jan 2013 17:40:53 -0800 (PST), Andre Jute
wrote:

On Tuesday, January 1, 2013 10:36:38 PM UTC, Jeff Liebermann wrote:
http://www.ebay.com/itm/251169704387

Thanks, Jeff. That's where I was heading next. If you'd read all of the thread I referred you to, you'd discover that the outline design is not an engineering exercise but a teaching tool for people who know nothing, zilch, nada about electronics. You don't jump right in with switching designs, you first take them by the hand and put something linear they can understand in their hand. Then you can move on to switching designs.

Thanks for finding the right board for me!

I'm having 2nd thoughts about the first board I found. The problem is
the acceptable range of input voltages. According to my crude
testing, I can sometimes expect 10VAC from the dynamo at suicidal
speeds. That's higher than the range of acceptable input voltages.


I was going to say as much.

A buck regulator would be a better choice in my opinion. Forget the
boost, unless you ride at snails pace ( 15 km/h ?) all the time.

--
JS.

On 02/01/13 12:40, Andre Jute wrote:
On Tuesday, January 1, 2013 10:36:38 PM UTC, Jeff Liebermann wrote:
On Mon, 31 Dec 2012 20:35:07 -0800 (PST), Andre Jute wrote:

I posted an outline design for doing exactly this to the Thorn
Forum. As you will see if you visit, an outline design was all
that was necessary because prebuilt modules to perform the
necessary functions are available under ten bucks total from
China. Several parties ordered modules identified by me and we'll
see when they arrive and are constructed how well my design for a
homebrew kit stacks up against the Tout Terrain and BUMMs E-Werk
and the Basta Nano lamp with built-in charger when many of the
tourers on that board already have. The idea is that the tourers
will be able to recharge all their electronic devices (excluding
an iPad, as that may require current which just isn't available,
and control circuits I don't fancy amateurs soldering and then
blaming me for when they inevitably eat an expensive iPad).
http://www.thorncycles.co.uk/forums/...p?topic=5271.0



Andre Jute



The problem with that scheme is that is uses a linear voltage

regulator. That's fine for very light loads, but is seriously

inefficient for anything that draws power. The size of the
heatsink

would give a clue as to where the power is going. It's also not
an

LDO (low dropout) regulator, which means that it stops regulating
when

the input voltage drops below perhaps 2V above the output voltage

setting. The voltage drop through the input diode bridge
certainly

doesn't help.



What you need is a switching regulator. Assuming an input of 6VAC
at

0.5A (3 watts) and an output of 5V at 0.5A max, something like
this

should work:

http://www.ebay.com/itm/251169704387

(Notice... no heat sink). Add a Shottky diode bridge and filter
cap

on the input and you're done. With a usable input range of 1 to
5V,

it should also work at low speeds.

Disclaimer: I haven't tried it.

Thanks, Jeff. That's where I was heading next. If you'd read all of
the thread I referred you to, you'd discover that the outline design
is not an engineering exercise but a teaching tool for people who
know nothing, zilch, nada about electronics. You don't jump right in
with switching designs, you first take them by the hand and put
something linear they can understand in their hand. Then you can move
on to switching designs.

Thanks for finding the right board for me!

I think you may find some USB devices do not charge from a simple 5 volt
supply. The reason being that they do not switch on their charging
circuit unless they can communicate with the USB host device (normally a
PC) and get approval to sink 0.5A.

Incidentally, some plug packs designed for charging USB devices (such as
iPads) have a small microprocessor inside that handles this
communication so that pedantic USB devices will start to charge.

--
JS

On Wednesday, January 2, 2013 2:46:01 AM UTC, Jeff Liebermann wrote:
On Tue, 01 Jan 2013 18:19:54 -0800, Jeff Liebermann

wrote:



One mo

http://www.ebay.com/itm/310506835676

3.5VDC to 30VDC input.

4-30VDC output at 2.5A max.

From the photos, it appears to be running at 80% efficiency.



So far, this one has the best chance because it can handle input

overvoltage. However, I can't tell what will happen if the output

voltage is set below the input voltage without actually trying it. The

auction title says "step up" (boost), which might be a problem.





--

Jeff Liebermann

150 Felker St #D http://www.LearnByDestroying.com

Santa Cruz CA 95060 http://802.11junk.com

Skype: JeffLiebermann AE6KS 831-336-2558

Replying to all your posts, Jeff, this is a superb amount of information, some of which I will lift whole for explanations to my guys.

In fact, I already have an LM2596HVS buck module on my desk, originally intended to power 6V BUMM Cyo and D-Toplight from the battery on my pedelec. But BUMM came out with an E-Fly with electronics up to the peak 42V of my nominally 36V battery built in, so I just bought that instead because I was in a hurry and the heatsink to transfer the heat to the bicycle's steel frame still needed quite a bit of patient filing...

Thanks man, you're a lifesaver.

Andre Jute

On 02/01/13 12:40, Andre Jute wrote:
On Tuesday, January 1, 2013 10:36:38 PM UTC, Jeff Liebermann wrote:
On Mon, 31 Dec 2012 20:35:07 -0800 (PST), Andre Jute wrote:

I posted an outline design for doing exactly this to the Thorn Forum. As you will see if you visit, an outline design was all that was necessary because prebuilt modules to perform the necessary functions are available under ten bucks total from China. Several parties ordered modules identified by me and we'll see when they arrive and are constructed how well my design for a homebrew kit stacks up against the Tout Terrain and BUMMs E-Werk and the Basta Nano lamp with built-in charger when many of the tourers on that board already have. The idea is that the tourers will be able to recharge all their electronic devices (excluding an iPad, as that may require current which just isn't available, and control circuits I don't fancy amateurs soldering and then blaming me for when they inevitably eat an expensive iPad). http://www.thorncycles.co.uk/forums/...p?topic=5271.0



Andre Jute



The problem with that scheme is that is uses a linear voltage

regulator. That's fine for very light loads, but is seriously

inefficient for anything that draws power. The size of the heatsink

would give a clue as to where the power is going. It's also not an

LDO (low dropout) regulator, which means that it stops regulating when

the input voltage drops below perhaps 2V above the output voltage

setting. The voltage drop through the input diode bridge certainly

doesn't help.



What you need is a switching regulator. Assuming an input of 6VAC at

0.5A (3 watts) and an output of 5V at 0.5A max, something like this

should work:

http://www.ebay.com/itm/251169704387

(Notice... no heat sink). Add a Shottky diode bridge and filter cap

on the input and you're done. With a usable input range of 1 to 5V,

it should also work at low speeds.

Disclaimer: I haven't tried it.

Thanks, Jeff. That's where I was heading next. If you'd read all of the thread I referred you to, you'd discover that the outline design is not an engineering exercise but a teaching tool for people who know nothing, zilch, nada about electronics. You don't jump right in with switching designs, you first take them by the hand and put something linear they can understand in their hand. Then you can move on to switching designs.

Thanks for finding the right board for me!

BTW, have you seen this?

http://www.bicycles.net.au/2012/09/l...e-usb-charger/

--
JS

On Jan 1, 8:52*pm, Jeff Liebermann wrote:

Disclaimer: I haven't tried it.
I assume you're talking about a regulator based on pulse width
modulation,
and I've wondered about whether such a thing can be used
with bike dynamos. I'm an ME, not an EE, but is it really feasible to
rapidly switch a power source with as much inductance as a dynamo?

- Frank Krygowski

Yes, it can. *Once the AC produced by the dynamo is converted to DC by
the bridge and the BFC (big fat capacitor), such things are easy. *I
would not run unfiltered pulsed DC (i.e. half wave rectified) into the
DC to DC converter as that would surely cause problems.

OK. I didn't realize you had a capacitor in there. I was imagining
switching the input directly.

The inductance of the power source is not a consideration. *Only the
inductance or capacitance of the load is important.

Well, in general I'm not sure that's true. I'm thinking of a magneto
or points+coil system with the load being the simple resistance of the
spark plug. Suddenly shutting off the coil dumps a lot of energy into
the resistive load. I'd think that switching a dynamo directly
(without the rectifier and capacitor) would blow out the PWM unit.
The rectifier and capacitor make it different, though.

Again, I'm an ME not an EE, so I'll accept some education on this.

- Frank Krygowski

On Wed, 02 Jan 2013 14:21:33 +1100, James
wrote:

I was thinking more like
http://www.extremecircuits.net/2010/05/power-mosfet-bridge-rectifier.html
The FETs used there have an Rds on of about 0.02 ohm.

Retch. 1% resistors? There's a reason the values are so critical. If
the driver phase angles aren't exactly 180 degrees, there's a really
good chance that the pull up and pull down MOSFET's are going to
conduct simultaneously. Shorting the power source for a few degrees
twice every cycle would not be a good thing. Even if the driver op
amps were exactly 180 degrees out of phase, any noise around the zero
crossing from the generator will produce the same result. It needs a
dead band around zero volts (which is a good idea anyway because
there's no output at zero volts), to prevent this from happening. I
also don't see any upper frequency limiting, no op amp compensation,
no oscillation prevention tricks on the MOSFET's, and slew rate
limiting. None of these are really necessary, but they do offer
benefits in real world applications, where the load doesn't quite look
totally resistive.

The original article came from Elektor Magazine:
http://www.elektor.com/magazines/2006/july/power-mosfet-bridge-rectifier.58316.lynkx
(subscription required to view)
Some not very nice comments on the design:
http://www.elektor.com/forum/elektor-forums/general-topics/power-supplies/mosfet-bridge-rectifier-not-working!.612569.lynkx

However, we're talking the difference between 0.25V per device for
Schottky versus maybe 0.05V per device for an IR HexFET:
V = 0.5A * Rds = 0.5A * 0.1 ohms = 0.05V
At 0.5A, that's a power loss of:
W = 0.25V * 0.5A = 125 milliwatts

And 25 milliwatts per FET (using your suggestion of 0.1 ohm on
resistance), plus a bit for the control circuit.

Let's not forget there are 2 devices on most of the time, so it's 250 mW
vs 50 mW.

out of a delivered 3000 milliwatts from the dynamo. I think this can
be safely ignored in the name of simplicity.

8.3% vs 1.7%? I know what I'd rather.

Agreed. 8.3% out of 3 watts would be about 250 milliwatts of
excessive drag (or wasted energy). I think I can handle that.

If the circuit was well designed and potted, I think it would be quite
reliable and simple enough not to cause issues. Power dissipation would
be so low that heat sinks would be unnecessary I think.

The switchers I excavated from eBay were running at about 80%
efficiency (even though they claimed higher). 20% of 3 watts is 600
milliwatts of heat. That seems a bit on the high side for potting in
something that's not thermally conductive. Allowing the chips to
radiate the heat and allowing some air flow might be useful. On the
other foot, I've potted stuff in allegedly thermally conductive epoxy
and gotten good results at 10 watts dissipation in still air.
http://www.mgchemicals.com/products/protective-coatings/epoxy-potting-and-encapsulating-compounds/thermally-conductive-epoxy-832tc/

Obviously 4 Schottky diodes will be simpler and cheaper.

Also more reliable.

Some dynamos with appropriate globes obviously used to work ok most of
the time. My dynamo doesn't have very good regulation, and used to blow
globes at the most inopportune times, i.e. going fast down hill.

What's a globe? A lamp? Yeah, I can see that happening. With an
incandescent lamp, the bulb life goes down 60% with a 5% increase in
applied voltage. It doesn't take much overvoltage to kill an
incandescent lamp. However, it's assumed that LED's have current
regulators, which prevents blowing the LED with too much voltage.

I find the dynamo output and LED driving requirements to be a match made
in heaven. They love each other and have no need for external bits to
make them work well together - and much better than incandescents!

I don't. To get constant output from an LED, with varying input
voltages, you need all those "extra bits" to make it work. Those
"extra bits" tend to involve voltage drops, which make the 6VAC output
far from ideal.

Flashing and dimming? C'mon. Talk about trying to keep it simple! ;-)

The logic is that once the "intelligence" is added to the design, such
software only features are essentially free. I'm thinking in terms of
a PIC controller. Add in a turn signal indicator. It's also free.
Hmmm... with 3 watts, I can almost keep my coffee warm.

Oh, and I thought we wanted to rectify the dynamo output to charge some
battery?

Yep. However, while attempting to achieve that goal, I took the
liberty of changing everything and redesigning the entire system. I
did stop at the generator, although I was tempted to continue by
redesigning the bicycle to make it easier to attach. I get carried
away sometimes.

But anyway, try running a dynamo with less load and you will find the
output voltage is higher. Mine with a constant 14.2 ohm load produces
6.8 V/0.47 A @ 20 km/h. At 30 km/h, it's producing 8.2 V / 0.58 A, and
at 50 km/h, 9.3 V / 0.65 A. (Do some power calcs and see there is far
more than 3 W coming from my 3 W dynamo ;-)

Good to know.
http://www.myra-simon.com/bike/dynotest.html
From the electrical output curve, it looks like it levels off at 4
watts. I think the core saturation flattening out of the curve is
intentional, introduced by the designers to keep from turning your
head lamp into a photo flash bulb. Again, this is legacy rubbish
designed solely for incandescent bulbs. LED lights would require
current regulators and therefore can handle the over-voltage. Instead
of having the curve flatten out as in the above URL, it will simply
continue to increase in a straight line, giving far more power than
todays dynamo. If it puts out 50VAC, that's fine as long as the
regulator(s) can handle it.

My LED lights are starting to work (producing reasonably bright light)
at just above walking pace (5-6 km/h), with only about 3 V / 0.22 A from
the dynamo.

They probably have an internal current regulator. Try a DC power
supply in place of the AC dynamo. Plot the input voltage and current
curves. I think you'll find that as the voltage goes up, the current
goes down.

You want higher voltages? Run at reduced load, or add a transformer!
(I know, the weight weenies will hang me for that one ;-)

Never mind the weight. It's the lousy efficiency of a transformer
that will cause problem. If it were running at one frequency,
preferably high enough that ferrites could be substituted for
laminated iron, I could probably squeeze 90% efficiency out of the
transformer. However, with the wildly varying frequency coming out of
the generator, I would be lucky if I could get 60% efficiency. No
thanks.

But, I think low voltage, low loss designs are not that difficult these
days.

I don't think it's that easy. I won't know until I try it. Things
are never as simple as they first appear.

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558