Martin's Switching Circuit Extrapolation of Data for Motorized Bicycle Application

The preliminary graph of power vs speed with load of 190mA, which Martin says gives the maximum wattage. I extrapolated 17 watts for going at a speed of 65kph.


The actual power curves. This was where I started getting confused. I tried to make the 17 Watt power criteria, but couldn't really execute the graphs properly...



I was trying to do some data extrapolation from Martin's power charts. I wanted to figure out how many LEDs I would need to fully load the circuit for a lighting system which is driven by a dynohub going 65kph (or 40mph). I need to set this as my top speed, because that's the fastest my motorized bicycle travels. I do not possess any y(x) functions to make accurate extrapolations of the graphs, but I tried my best to eyeball it. I came up with a minimum of 10 LEDS, where the switch happens around 42.5kph. The brown are the bridge rectifier graphs and the peach lines represent the voltage doubler graphs. They both end at a point (blue for the voltage doubler and red for the bridge rectifier) that represent their maximums. Can somebody help verify that this extrapolated data is somewhat or close to being correct?

Is there any other way I can figure out component values for my application? such as the quantity of LEDs, capacitor values, resistor values, and for all the other components? I am going to use a Sturmey-Archer xl-fdd dyno/drum hub on a 26" wheel.

I'm also a bit concerned that I extrapolated the data where it would require 10 LEDs to fully load the system. That just seems to be too much for bicycle lighting--unless I divide them amongst a head light, tail light, and standing light (that would require more circuitry). Isn't there a way I could just use at most 6 LEDs, and just use some kind of resistor in the input of the rectifier/doubler circuit that would shunt excess voltage (from the dynamo's ac power source) as heat? It's not efficient, but I think it might be a more practical solution, don't you think? Would appreciate your help very much! Thanks!

I am not the expert you need to either refine the charts or tweak the circuit. I do have some questions for clarification.

Given that converting electrical power into motive force back to electrical power loses energy at every step of conversion, I repeat a previous question: Why not just use the same battery that powers the bike? Do you need this 90 MM brake? Sturmey Archer makes the XL-FD sans dyno version if you do. Maybe you wish to pedal home if the battery goes flat and still have lights? You won't be doing 45 klicks unless you have a good hill and a tailwind all the way. So you can design for a lower max speed and accept a plateau say with 6 or 8 total LEDs.

Not sure where your new power graphs end up in total watts but 17 watts at almost 200 mA means a voltage of about 85 volts? That's a lot more than 10 LED Vfs. I may be mistaken, but I thought that dynos like to be current sources of about 500 mA and the voltage doubler just sacrifices current to voltage to drive a larger array of LED's at higher voltage and lower current at a lower speed, then you tune the circuit to switch back at some optimal point for smoothest transistion to the usual current mode with increasing voltage with speed. Assuming 500 mA or so after switching off the voltage doubler, means 17 watts is about 34 volts of about 10 3.4 volt Vf LEDs @ 500 mA. Martin was using Luxeons I think, which would be about 3.7 at max current, and may be about 3.4 at 500 mA. CREE XP-E's are about 3.4 at 0.5 A, XP-G's are about 3.1 volt, so you'd need 11 for 34 volts, XM-Ls are about 2.85 Volts so 12 of them. All component values would change. I think Martin went to 8 maximum with his parts lists. You also have the freedom to accept a plateau of output at some speed, and forget trying to get all the power you can get. Eight Luxeon equivalents or 27 volts of LEDs is enough.

A very bright tailight of about 250 lumens could be made with 3 XP-E orange reds in a tailight at 2.2 Vf each, or even double that for twin triple lights. Each would be brighter than car brake lights. Illegal in a lot of places. Not where I live. Six XP-E Orange-Reds at 2.2 Vf are almost 4 Luxeons in total Vf. Allowing 5 XM-ls at 0.5 amp or about 1000 lumens out the front which is one lesser car headlight. With the right lenses and aiming, to not blind oncoming drivers, it is about what I have in my paired XM-L system. That would be a pretty maiximal system IMHO. More, and you are draining the drive battery for no good purpose.

BrianMc said:

I am not the expert you need to either refine the charts or tweak the circuit. I do have some questions for clarification.

Click to expand...

Yes, your mathematical way (of finding component values) gives me great relief. You give yourself too little credit... My poor excuse for extrapolated power curves was just getting the foot in the door--if anything.

BrianMc said:

Given that converting electrical power into motive force back to electrical power loses energy at every step of conversion, I repeat a previous question: Why not just use the same battery that powers the bike? Do you need this 90 MM brake? Sturmey Archer makes the XL-FD sans dyno version if you do. Maybe you wish to pedal home if the battery goes flat and still have lights? You won't be doing 45 klicks unless you have a good hill and a tailwind all the way. So you can design for a lower max speed and accept a plateau say with 6 or 8 total LEDs.

Click to expand...

Sorry, I should have mentioned that my motor is actually a 4-stroke 49cc gasoline engine. There's a 2-coil magneto in the works for my particular engine, but it's yet to even enter the test phase let alone into the consumer market. It's exigent that I have a power source for lights, because I don't want to be harassed anymore by cops for not being visible in traffic. Yes, I'm aware of the front drum brake w/o dyno, but I figure it's more cost effective to implement the dynohub.

BrianMc said:

Not sure where your new power graphs end up in total watts but 17 watts at almost 200 mA means a voltage of about 85 volts? That's a lot more than 10 LED Vfs. I may be mistaken, but I thought that dynos like to be current sources of about 500 mA and the voltage doubler just sacrifices current to voltage to drive a larger array of LED's at higher voltage and lower current at a lower speed, then you tune the circuit to switch back at some optimal point for smoothest transistion to the usual current mode with increasing voltage with speed. Assuming 500 mA or so after switching off the voltage doubler, means 17 watts is about 34 volts of about 10 3.4 volt Vf LEDs @ 500 mA. Martin was using Luxeons I think, which would be about 3.7 at max current, and may be about 3.4 at 500 mA. CREE XP-E's are about 3.4 at 0.5 A, XP-G's are about 3.1 volt, so you'd need 11 for 34 volts, XM-Ls are about 2.85 Volts so 12 of them. All component values would change. I think Martin went to 8 maximum with his parts lists. You also have the freedom to accept a plateau of output at some speed, and forget trying to get all the power you can get. Eight Luxeon equivalents or 27 volts of LEDs is enough.

Click to expand...

Yes, thanks for harping on the conservation Laws of Physics! P=IxV. That's a great one. Pardon, I don't mean to sound mocking, but I'm just so relieved that at least one of us is standing on some solid ground--no pun intended. Yes, you also restate in your own words very well the function of the circuit. One thing I'm still uncertain about is the amperage rating of the dynohub. Why does Martin state in the first power curve I extrapolated that 190mA is the current where the dynohub delivers its maximum power (I'm guessing because it's the load where the relationship of power and speed is most linear). But for the second curve(s), I extrapolated the load is variable (0-600mA) and therefore the extrapolation being hairy. So you're basically saying for the doubler, less current is used for load hence increasing voltage to power LEDs at low speed. Is it safe to assume that we'll have the same 17Watts of power at 65kph when the load is 500mA? I don't think one can extrapolate that far... I guess the next best thing for me to do is lace up the Sturmey-Archer dynohub and do an engine test run with a voltmeter set at 500mA load to see What volts i arrive at. My guess is somewhere in the low 20s.
Also, since the low pass filter (the resistor in series w/ the load and capacitor in parallel with the load aka passive filter) rejects any frequency above a designated frequency, that would also mean it rejects (or "clips") the respective amplitudes that exceed what the LEDs and all the other components are rated for. Is there any component in the circuit that is susceptible to the high amplitude, or does the low pass filter cover for them all? Sorry i'm still understanding how the components are driven.

BrianMc said:

A very bright tailight of about 250 lumens could be made with 3 XP-E orange reds in a tailight at 2.2 Vf each, or even double that for twin triple lights. Each would be brighter than car brake lights. Illegal in a lot of places. Not where I live. Six XP-E Orange-Reds at 2.2 Vf are almost 4 Luxeons in total Vf. Allowing 5 XM-ls at 0.5 amp or about 1000 lumens out the front which is one lesser car headlight. With the right lenses and aiming, to not blind oncoming drivers, it is about what I have in my paired XM-L system. That would be a pretty maiximal system IMHO. More, and you are draining the drive battery for no good purpose.

Click to expand...

By drive battery, are you talking about the C1-C4 tuning capacitors? Or were you assuming I had a battery on an electric motor, which I don't have?

Hmmm, I thought around 12W was all you could draw from a hub.

Regardless, six LEDs/1200lm should be more than enough to see adequately on the road.

highlander9 said:

.....By drive battery, are you talking about the C1-C4 tuning capacitors? Or were you assuming I had a battery on an electric motor, which I don't have?

Click to expand...

Yes. I also thought someone showed 12-13 watts was all you could get with tthe right load BUT I don't think they were turning it at this speed. It's in a dyno thread here somewhere. Maybe Highlander is up for a search in the forum? Steve K had a dyno light/battery charging system that would be very appropriate for this application, but batteries, LEDs, and circuit components have changed. It might be stage two once you have good lighting. If you had fun with this project and wanted to learn more electronics, it might be an idea to try. You'd have lights standing and could select when to turn on high beams or an extra tail light. Steve's setup is searchable here, too. 1200 lumens out the front is actually hard to get in a friendly beam and znomit is right, more than enough (about what I am getting on the raod with helmet and twin headlamps.).

BrianMc said:

..... Steve K had a dyno light/battery charging system that would be very appropriate for this application, but batteries, LEDs, and circuit components have changed.

Click to expand...

I tend to keep my battery charging circuits pretty simple. Usually just charge a nicad to a fixed voltage and figure it's good enough. My circuits tend to be a bit complex, but physically small, and not appeal to people who don't have a fair bit of experience in electronics already.

The whole business of getting a lot of power out of a dynamo isn't that hard to do..... just find a load with a high resistance and spin the dynamo fast!
The trickier part is to have a circuit that can adjust the load seen by the dynamo so that you can still get some light when moving slow. The easy way to do this is to have a manual switch where you short across all but a couple of leds when you are at low speeds. It's a bit of a nuisance to have to do this if you are constantly speeding up and slowing down, which is why I built a circuit to do this for me.

As Brian said, I think it would be difficult to have more than 4 or 6 leds in the headlight without it being annoying to oncoming traffic. The optics that are available just don't permit the beam shaping that an automotive headlight can do.

but.... if you are so inclined, I'd say just take a bridge rectifier (with or w/o a tuning cap), and start wiring leds in series as the load. You'll figure out how many is too many to properly drive.

..and another comment.... due to the way that the human eye perceives things, you have to roughly double the light output to notice much of a change. If you find that 2 leds is too dim and 4 is just about enough, you'll need to bump it up to 8 leds to get much of an improvement.

good luck,

Steve K.

If I had 8 XM-Ls drievn from 150 to 600 mA+ to tame into a socially acceptable array of road beams, and I could accept a bit of size and weigt to the lights, I would likely use the 35 mm Iris and 35 mm Eva lenses I have now, a pair of each for four. The two Evas aimed with hot spots just in front of the front wheel, the Irises a bit higher, all with hoods to cut off the top part of their beams headed above the horizon into hostile and useless territory. These four will provide adequate spill to be seen by oncoming cars. The other four I'd use aspheric lenses and project the dies out ahead of the Iris beams. The die projections have as severe a cutoff as you get, bright then nada if you get the right lenses at the right focal lengths. These often don't collect wider angle light well from the LED, so the new multi die XM-Ls with 115 degree output might be good, the cross between the four projected dice won't be problem with four beams overlapping. You can also allow the uncolllected light out the sides as a side marker. In fact the lower current demand of the multi die version might be good for the other four lights too. Haven't really considered this new LED much yet. Unfortunately, it will be a bit of experimenting. The bleeding edge can be interesting. I used surplushed to get cheap aspheric lenses of about 13 mm diameter functional part (16 mm with flange) for my helmet light. DX has a 28 mm or so one. Bigger diameter ones tend to make for smaller brighter projections, but not always.

Hope this brain dump is of some help. Oh, and with many lights, you can mount a pair at front axle height to add shadows to potholes, sticks and the like to give you time to evade. Provided your other lights don't wash them away.

Thanks to everyone for your inputs. Now that we figured out how many LEDs are legal and practical to run on my motorized bicycle--I think eight was the figure--how do I design my circuit? Based on Martin's diagram explaining how his switching circuit works, the value for zener diodes to regulate and rectify the high power of the dynohub going 65 kph I imagine is going to be quite different from when the bike is going 40 kph. Likewise, for the low pass filter which I take is going to need a different value for its capacitor and also its resistor value. I imagine then a lot of the discharge resistor values for the capacitors have to be changed as well.

Martin has done all the work for you using six leds. You could extrapolate/guess component values for eight LEDs based on those values but its a very marginal improvement over six and you'll lose light at low speed.

Should have looked up his table. My memory said he did eight, but 6 it is. I would just drop a pair of aspherics then, so 2 aspherics, plus 2 Evas and 2 Irises with hoods.

Okay, I think I'm understanding now. I was just confused about the component values. After reviewing the "BOM," I see where he's coming from. I saw that all the components values (Volts and Amps) were rated fairly well and above for going fast (45-50 mph (>65kph) apparently yields around 52 volts for the Sturmey-Archer), so no need to fret anymore about whether the dynohub is going to produce levels that will blow the components out.

I don't know if I've made you all aware that my application is high speed. The hub can produce ~45 volts at 65kph. That means 45V/6LEDs=7.5V/LED. That's really not possible in my case. That's why I wanted to know how the circuit worked. Whether the AC low pass filter acts to filter out the high frequency remains a mystery to me. Upon further examination, I saw that there was a closed circuit between the rectifier, the LEDs, and the IC which led me to believe that the AC low pass filter only operates for the doubler, converting frequency to voltage, so it can compare that voltage to some reference voltage to know when to make the switch to the doubler. If the rectifier is somehow tied up with the whole mode selection spiel, I'm not aware of how it is. If anyone can enlighten me about how the rectifier undergoes voltage regulation without the aid of the AC low pass filter, then please be so kind as to do so. Otherwise, I don't see how this switching circuit would work with a high voltage application like my motorized bicycle (which BTW has a gasoline engine, NOT an electric motor). Thanks everyone!

Adding more LEDs in series will also raise the minimum voltage required to illuminate the light. So you will probably have problems at low speeds.

But…

On a relatively slow bicycle, the dynamo will put out just enough power to illuminate your light. Also, you don't want the extra drag of the dynamo when light is not required.

On a motorized bicycle, these considerations are irrelevant.

I therefore suggest a different approach; Use the Dynamo hub to charge a (Li-Ion or other) battery and use that battery to power your lights.

You would need some custom electronics to regulate the charging of your battery.
Head and tail light would simply connect to the standard ~12V battery.

This would have the advantage of continuous lighting, also when you are stopped for a traffic light for example, or riding slowly. You will also be able to easily power other gadgets from the battery.

…just my 2 cents.

Sounds like, Pe2er, StreveK, and I think the battery charge dyno system is the best route. But you likely need something yesterday. I think we understand that you are running fast most of the time with out a battery system at this time. It looks like adding a batttery charged by the dyno and not an engine driven magneto or starter/gerenator is worth thinking about.

That said, I may be wrong and defer to SteveK, or some of the long time dyno DIYers, but as I read the LEDs versus kph curves, a dyno levels out in voltage and current = power at some speed based on the load it has. So, you can opt for a lower load. Dyno output just hits maximum for that load and stays there. That is what I think those charts say. The nice feature of Martrin's PCB version 12, is that it boosts low velocity voltage to fire the LEDs sooner than they would otherwise, then switches into 'normal mode'. If you are rarely that slow for long, then it isn't a big deal for you.

A correction: the '6 LED' circuit of Martin's I think was tested for Luxeons which are about 3.4 volt at about 500 mA, so a load of 20.4 volts or thereabouts. So '6 LEDs' is not necessarily 6 of any kind of LED. About 21 volts would be 5 XM-L's plus three red orange XP-E's in a tailight. You would not be pulled over for inadequate lighting. It would let you decide how much you really wanted and let you design a battery based lighting system.

Oh, if you order one or more of Martin's PCB's you will need to PM him to arrange that, so why not pose any questions you have about modding his circuit to him via a PM?

highlander9 said:

I don't know if I've made you all aware that my application is high speed. The hub can produce ~45 volts at 65kph. That means 45V/6LEDs=7.5V/LED. That's really not possible in my case. That's why I wanted to know how the circuit worked.

Click to expand...

first, let me say that I never paid a lot of attention to Martin's circuit, mostly because it was bulkier than I prefer and was fixing problems that I didn't think I had.

Having said that... let me try to explain that even though a dynamo can produce 45v at a high speed, that doesn't mean that it can deliver that voltage into any arbitrary load. The output voltage (without any load) is essentially proportional to the speed. The dynamo does have an internal series impedance that can be considered to be partly resistive and partly inductive. Once you attach a load to the dynamo and current starts flowing, there will be a voltage drop across this series impedance, and that will lower the voltage that is delivered to the load.

For instance.. my dynamo light is very basic right now. A bridge rectifier and two (or four) leds in series. The dynamo is a SON28, and I've measured the the output at 50mph to be about 100v when delivering a few milliamps into a zener diode. I think the zener was 100v, and I was just measuring voltage across a small series resistor.

Anyway.. despite being capable of delivering 100v at that speed, I can still operate it at that speed with just two leds hooked up to it. It's primarily the inductance that limits the current, and which provides the semi-regulation of the output current.

Martin's circuit adds a series capacitance to the dynamo output, which is better in terms of matching the load impedance to the source impedance, but gets rid of this regulation of the output current (to a degree). While I wouldn't worry about voltage too much, it's possible that it may deliver more current to the leds than they are rated for. Without running a bunch of tests to see what the output current will be, the best course of action may be to run a simple circuit composed of bridge rectifier and leds in series.

As far as batteries are concerned... I have mixed feelings.
I tried running a system where the dynamo charged a battery, and the battery fed a headlight. It worked fine in the summer, when the battery charged most of the time and the light didn't get used much. It didn't work well in the winter when the battery was on constantly and was difficult to charge in the cold weather. If you ride in urban areas where you might spend half of your time sitting at intersections, the dynamo would need to produce twice as much power as the headlight consumed.

My solution was to just use a low power standlight when stopped. The circuit is still a bit complicated, but has met my needs for a number of years now.

regards,
Steve K.

Steve K said:

Having said that... let me try to explain that even though a dynamo can produce 45v at a high speed, that doesn't mean that it can deliver that voltage into any arbitrary load. The output voltage (without any load) is essentially proportional to the speed. The dynamo does have an internal series impedance that can be considered to be partly resistive and partly inductive. Once you attach a load to the dynamo and current starts flowing, there will be a voltage drop across this series impedance, and that will lower the voltage that is delivered to the load.

Click to expand...

Yes, I reckon all AC power sources are. But this voltage drop is marginal, is it not?

Steve K said:

For instance.. my dynamo light is very basic right now. A bridge rectifier and two (or four) leds in series. The dynamo is a SON28, and I've measured the the output at 50mph to be about 100v when delivering a few milliamps into a zener diode. I think the zener was 100v, and I was just measuring voltage across a small series resistor.

Click to expand...

Isn't a zener diode in series with a resistor also known as a shunt regulator?

Steve K said:

Anyway.. despite being capable of delivering 100v at that speed, I can still operate it at that speed with just two leds hooked up to it. It's primarily the inductance that limits the current, and which provides the semi-regulation of the output current.

Click to expand...

The inductance will allow slower deliverance of the 45 volts? What if I maintain a pretty high speed for a while? Since time is a factor in inductance, wouldn't that mean I'll eventually start exceeding voltage requirements of the LEDs? And if the volts are high that also means the Watts are high, and I hear LEDs will simply melt when the Watts are high.

Steve K said:

Martin's circuit adds a series capacitance to the dynamo output, which is better in terms of matching the load impedance to the source impedance, but gets rid of this regulation of the output current (to a degree). While I wouldn't worry about voltage too much, it's possible that it may deliver more current to the leds than they are rated for. Without running a bunch of tests to see what the output current will be, the best course of action may be to run a simple circuit composed of bridge rectifier and leds in series.

Click to expand...

Isn't the output current 500mA? Because 6V, 3W and 3/6=.5 or 500mA.

highlander9 said:

Yes, I reckon all AC power sources are. But this voltage drop is marginal, is it not?

Click to expand...

You can get an idea of what the internal impedance is by measuring the short circuit current, Isc, and the open circuit voltage, Voc. The internal impedance isn't a true resistance, and isn't even just an inductance in series with a resistance, but at a given dynamo speed, the internal impedance is calculated as Voc/Isc. For a bike dynamo, it's usually at least 10 or 20 ohms (I need to run through some data taken on a SON).

highlander9 said:

Isn't a zener diode in series with a resistor also known as a shunt regulator?

Click to expand...

Right, sort of. In a shunt regulator, the resistor is in wired between the power source and the zener. In my test, there was no resistor between the dynamo and the zener, and the resistor was in series with the zener, wired from the anode to ground. Instead of using a series resistance to limit current in the zener, I was using the dynamo's internal impedance.

highlander9 said:

The inductance will allow slower deliverance of the 45 volts? What if I maintain a pretty high speed for a while? Since time is a factor in inductance, wouldn't that mean I'll eventually start exceeding voltage requirements of the LEDs? And if the volts are high that also means the Watts are high, and I hear LEDs will simply melt when the Watts are high.

Click to expand...

If you apply a DC voltage across an inductor, then current will increase linearly (V = L di/dt). In the case with dynamos, the voltage is AC, so the inductance acts as a constant impedance (Z= jwL, or the magnitude of the impedance is 2*pi*freq*L). i.e. at a constant speed (frequency), the inductance will have constant impedance, which will result in a constant current.

highlander9 said:

Isn't the output current 500mA? Because 6V, 3W and 3/6=.5 or 500mA.

Click to expand...

Are you asking if the dynamo output is independant of the load? Dynamo outputs are based on the assumption that the load will be the standard 12 ohms. If you change the load, the output voltage, current, and power will change. That's what lets people such as Martin and myself design circuits that extract more than 3 watts of power from bike dynamos.

regards,
Steve K.

Okay, a little update on where all of this is going. I have a dynohub on its way, and meanwhile I plan to devise a little mechanism that will help me get my experiments on the road.

This little mechanism will be a brace to posture the bike such that only that back wheel will be lifted from the ground. I will most likely implement the use of a bicycle repair stand in conjunction w/ another means to stabilize exclusively the front yaw of the steerer tube of the bike frame. Of course at this point, the engine and all of its mounting hardware (ie. jackshaft and shift kit with transmission hub) will be mounted onto the bike, hence acting as the source of power generation for the front dynohub. This is where you might wonder well how exactly will you accomplish this, when the dynohub will be built into the front wheel. This is where the magic of poorly devised contraptions may somehow eek out some rather crude but favorable results. This experiment will be conducted outdoors because of exhaust fumes. Here's a very rough diagram of how all of this "scientific" instrumentation will be put together:


I still need to clean up all the rust and corrosion from the engine and shiftkit hardware, which accumulated tremendously after riding in the winter season. That's what living in Chicago will do to your bicycle--weather being not conducive to riding all year round. This I approximate will take one sunny afternoon where I can get out and enjoy the smell of vinegar working to take off the rust and corrosion. the mounting of the parts onto the bike will probably take another day. The lacing of the dynohub onto the front wheel, another. Then the making of the drum brake arm clip will probably take another. All in all, I don't plan to get into experimental phase for another 2 weeks, all things considering.

The first experiment will be done to derive internal impedance values (VOS/ISC), which I believe will be attained through the multimeter being directly connected to the dynohub leads. This I believe will give us some cold, hard data about the theoretical current that can be delivered to an arbitrary load. or am I mistaken about that?

Steve K said:

You can get an idea of what the internal impedance is by measuring the short circuit current, Isc, and the open circuit voltage, Voc. The internal impedance isn't a true resistance, and isn't even just an inductance in series with a resistance, but at a given dynamo speed, the internal impedance is calculated as Voc/Isc. For a bike dynamo, it's usually at least 10 or 20 ohms (I need to run through some data taken on a SON).

Click to expand...

Then, I will, like Steve suggests, go ahead and use an appropriate zener diode/resister combination to measure for the voltage drop aft this series impedance.

Steve K said:

Having said that... let me try to explain that even though a dynamo can produce 45v at a high speed, that doesn't mean that it can deliver that voltage into any arbitrary load. The output voltage (without any load) is essentially proportional to the speed. The dynamo does have an internal series impedance that can be considered to be partly resistive and partly inductive. Once you attach a load to the dynamo and current starts flowing, there will be a voltage drop across this series impedance, and that will lower the voltage that is delivered to the load.
For instance.. my dynamo light is very basic right now. A bridge rectifier and two (or four) leds in series. The dynamo is a SON28, and I've measured the the output at 50mph to be about 100v when delivering a few milliamps into a zener diode. I think the zener was 100v, and I was just measuring voltage across a small series resistor.

Click to expand...

In the long term, I plan to develop a power curve that reflects my findings based on the use of different ohm-valued resistors at different speeds. This also gleans from Steve's latest replies to my inquiries. Thank Steve, I can't stop flattering you enough!

Steve K said:

It does take a bit of work to really quantify the source impedance of a dynamo. For each increment in speed, you'd need to measure the dynamo output as you vary the load. For instance, if you checked the output by loading the dynamo with resistors varying from 5 ohms to 50 ohms, measuring the voltage and phase angle (hmm... but you can't measure the phase angle if you don't have a reference... I'll have to think about that). Do this for each speed from 3mph to ... 50mph?

Click to expand...

shoot, this is probably a good time to do a data dump....

I did get a bit of data from Nick Ray back in the BikeCurrent days. I think this was from a SON28, but the first generation....

He plotted the output voltage as a function of dynamo frequency and load resistance. I'm not sure exactly how to relate frequency to bike speed, but it shouldn't be rocket science. Just gotta have a way to drive the dynamo at a constant speed and then read the bike speed and dynamo output frequency. Might be nice to have a meter with a frequency readout.

load
freq 10 12 14 16 18 20 25 30 35
10
15
20
25 4.8 5.6 6.2 6.8 7.3 7.7 8.5 9 9.35
30 5 5.9 6.7 7.4 8 8.5 9.6 10.3 10.8
35 5.2 6.1 7 7.8 8.5 9.15 10.5 11.4 12.2
40 5.3 6.3 7.2 8.1 8.9 9.6 11.2 12.4 13.3
45 5.4 6.4 7.4 8.3 9.2 10 11.8 13.2 14.3
50 5.4 6.5 7.6 8.5 9.5 10.3 12.3 13.8 15.2
55 5.5 6.6 7.7 8.7 9.6 10.5 12.6 14.4 15.9
60 5.6 6.7 7.7 8.8 9.8 10.7 12.9 14.8 16.5


From there, you can play with the numbers.

For instance, I ran through a few calculations to see what the open circuit voltage and source resistance was like for the data at freq = 45:

5.4 0.540 2.916 150.00 86.40 10
6.4 0.533 3.413 210.00 118.40 12
7.4 0.529 3.911 91.64 55.84 14
8.3 0.519 4.306 117.82 69.42 16
9.2 0.511 4.702 72.00 46.00 18
10 0.500 5.000 64.29 42.14 20
11.8 0.472 5.570 43.75 32.45 25
13.2 0.440 5.808 35.00 28.60 30
14.3 0.409 5.843 35

rats... I can't get this stuff to format at all....
let me try something else... CSV, maybe?

5.4, 0.540, 2.916, 150.00, 86.40, 10
6.4, 0.533, 3.413, 210.00, 118.40, 12
7.4, 0.529, 3.911, 91.64, 55.84, 14
8.3, 0.519, 4.306, 117.82, 69.42, 16
9.2, 0.511, 4.702, 72.00, 46.00, 18
10, 0.500, 5.000, 64.29, 42.14, 20
11.8, 0.472, 5.570, 43.75, 32.45, 25
13.2, 0.440, 5.808, 35.00, 28.60, 30
14.3, 0.409, 5.843, ----, ----, 35

well, let's see if that makes any sense at all.

regards,
Steve K.

okay, another bit of data from a while ago (although not as far back as I thought... just 2003)

This is a post to the BikeCurrent list that I saved... written by Abram Dancy at MIT


Thanks for the data. A while ago I derived some equations, for a generic
permanent magnet generator. Using the Nick Ray data, I did some fitting to
get the parameters of the SON. Here's what I've got.

Iout = M*Ir*w / R * 1/sqrt (1 + (Ls/R*w)^2)
(You can get V and P curves based on the current)

where:
M is based on the geometry and materials of the generator,
Ir = the strength of the magnet(s)
R = Rload + Rs
Rs = internal resistance of generator
w = frequency of operation, rad/sec = 2*pi*26poles*RPM/60
Ls = internal inductance

Based on the data, a pretty good fit is:
M*Ir = 0.0785
Rs = 4.75 ohms
Ls = 133 mHenries

With these constants, over most of the range of data Nick provided, I get
errors less than 2%. At 10 and 15Hz (<4mph) the error is a larger,
particularly at higher load resistances, with up to about 6% error.

Maximum output power at a given frequency (including power dissipated in
windings)
found at R = Ls * w
Maximum output power at this resistance (including power dissipated in
windings):
P = 1/4 * M^2*Ir^2 / Ls * w


enjoy,
Steve K.