Can I borrow yours?

Next time you are passing through Illinois, swing by Peoria and I'll loan you my two channel Tektronix. It's an analog model from... late 80's?? Not really sure. The bandwidth is 100MHz, so it is more than fast enough for the voltages from a hub dynamo. :)

If anyone is curious about what the waveforms look like, let me know and I can try to grab some photos or short videos.

Steve in Peoria
(...but thinking about picking up one of those intriguing cheap Rigol scopes)

Steve, I'm interested in seeing the voltage waveforms (versus wheel speed), ideally for several levels of resistive load (6, 12, 18 and 24 ohms?) and with an actual lamp load. I've always wondered about the Thevenin resistance of a SON / SP, and how close it might be matched by an electronic lamp load.

seek out an oscilloscope yet? I found one decades ago when I was in Eugene.. It was not mine,

decades back , Electronics repair guy.. cleaned up my Sony SW radio that got wet in my camp flooding in Bavaria in 1991..

tested my Sanyo BB dynamo .. sine wave was obvious.. on the screen..







...

I was joking, pointing out that it's not a common household item. But yeah, maybe I should get one. It would be fun.

Well, I took some scope photos with the bike in the workstand and spinning the wheel by hand.
I've taken photos with three different loads... one with no load, one with a 20 ohm load, and one with a 10 ohm load. For reference, the standard incandescent dynamo headlight used a nominal 12 ohm bulb.


For the no-load case, there is nothing connected to the dynamo but the scope leads.
The scope settings are 10 volts per division and 5ms per division.
From this, you can see that the amplitude is 26V peak (i.e. 26V from the zero volt position at the center of the screen to the top of the wave). The most obvious thing about the photo is that the waveform is definitely not sinusoidal! Due to the shape of the wave, it's hard to say what the RMS value is.
The period is about 35ms, yielding a frequency of 29Hz. This is equivalent to just a bit over 10mph, which I've measured as being 27.3Hz.
Of course, the relationship between the speed and dynamo frequency is determined by the wheel diameter. In this case, the wheel is 700C, and the tire was 700 x 28.

https://c1.staticflickr.com/5/4908/4...605c7a_c_d.jpg



For a 20 ohm load, the waveform is closer to being a sinusoid, although it looks a lot like a triangle wave. The amplitude has dropped to 10V peak.
The frequency is 21.7Hz, which is a good deal slower than when it was unloaded. This is mostly because it is harder to spin up the wheel by hand when the dynamo is loaded. The slower speed is part of the reason why the voltage is lower, but primarily it is due to voltage being dropped across the dynamo's internal impedance.
Scope settings: 10v/div. 10ms/div.
amplitude is 1 div, or 10V peak.
period is 4.6 div, or 46ms. As such, the frequency is 21.7Hz.

https://c2.staticflickr.com/8/7820/4...619712_c_d.jpg


For a 10 ohm load, the waveform is much more like a sinusoid. It draws more current from the dynamo, so the voltage is reduced. The peak voltage is 8V.
The frequency is roughly 23Hz.
Scope settings: 10v/div. 5ms/div.

https://c2.staticflickr.com/8/7913/4...69393c_c_d.jpg


So how do you predict what sorts of voltages and currents a dynamo will produce at different speeds? Well, each dynamo will be different, so it is hard to generalize. I've made some measurements on this particular SON28.
I did this with a relatively simple resistor network, a digital meter, and just rode at different speeds with different resistive loads, and took notes of the RMS voltage and the frequency.

Here's a shot of the test rig....

https://c2.staticflickr.com/2/1604/2...23d596_z_d.jpg


and here's the raw test data...

https://c2.staticflickr.com/2/1498/2...4ea639_c_d.jpg


and when it's put into a table, where the voltage, current, and power are calculated, it looks like this:
https://c2.staticflickr.com/8/7863/3...c30b76_c_d.jpg

The highlighted numbers are the max power that can be obtained from the dynamo at each of the speeds. As can be seen, the bike (with 700C wheels) had to be going about 9mph before the dynamo could generate 3 watts into a load that is about 15 ohms.
At 20mph, the dynamo could generate over 7 watts, as long as the load was 40 ohms.

For the sake of circuit analysis and simulation, this data can be used to generate a model for the dynamo. The model is about as simple as possible, so it isn't a perfect match for the test data, but it gets close.
The model is composed of three parts:

1. a voltage source that produces a sine wave whose amplitude is proportional to the dynamo's frequency. This is expressed by the equation Voc = K x f, where Voc is the dynamo's open circuit voltage (i.e. there is no load), K is the scaling constant, and f is the frequency of the AC voltage. I found that K = 0.53 gave me pretty good results.

2. inductance. This represents the inductance of the windings and the voltage drop that occurs when the AC current passes through it. I ended up with 0.145 Henries.

3. resistance. I got the best match between the model and the test data when I used 2.8 ohms for the dynamo's resistance. When I measure the resistance of the windings with a meter, I think it is closer to just 2 ohms.

The model is composed of these three elements wired in series.

This might be the longest post I've written on bikeforums, so I may have overlooked something or committed some other error. Let me know what I've missed or got wrong.


Steve in Peoria

Very good analysis. Thanks. This does help explain in part why the Forumslader USB charger can produce so much more power that other chargers at higher speed, it somehow forces the hub to generate more watts with more resistance at those higher speeds.

I can't understand a lot of this stuff, but the graphs towards the bottom of this page tell you what I mean about different USB chargers.
https://translate.googleusercontent....g9zRzdCuSHvqDA

Tourist, the equation in [MENTION=113466]steelbikeguy[/MENTION]'s paragraph labelled 1. is "Voc = K x f" which states that the output voltage of a generator increases as the speed of the rotation of the generator shaft. The equation can be derived from basic physics by any formal student of electrical engineering, and both steelbikeguy and I earned our EE degrees many years ago. It says that for the voltage to rise as the speed increases is totally natural, it's how generators work. The USB charger does not "forces the hub to generate more watts with more resistance at those higher speeds." The hub generates more voltage just because it is turning faster.

I didn't pay too much attention to the data that compares different USB chargers. I'm not saying there are or are not differences, rather I'm trying to say that if there are performance differences, they might be because of differences between the circuit designs of the various USB chargers under test.

Steelbikeguy, thank you very much for generating and interpreting all this data regarding the SON hubgen.

Thanks. I forgot almost all of my college physics decades ago. As a retired geological engineer I think my training in alternator design was very close to zero, although I do remember attending one 45 minute lecture on electric drive mining trucks decades ago. Most of my knowledge on alternators with permanent magnets is from the electrical system in a 1968 Triumph T100R motorcycle.

Sounds like my "education" in underground construction! I was lead EE on a massive underground instrument for a science experiment, and we all went to a lecture on tunnel boring machines! I can tell you they are big and ... maybe that's it! I COULD throw in f=ma!!

Glad to have helped -- Road Fan.

Your inductance and ohmic resistance sounds to be in the right ballpark, based on small motors and actuators I've been involved with in the automotive sector. Effective resistance (your model value that replicates the observations) nearly 3 ohms sounds high. But the difference could be based on core loss/magnetization effects that cause heating, which dissipate energy and hence detract from power delivered. Whether 0.8 ohms is plausible from a physical point of view, I can't say. But from a designer's point of view, if the core loss is 30 to 40% of the resistive loss, I'd say that was a good day's work! It's probably a plain iron laminated core, rather than any more efficient steel, not being a high-speed machine. I think I'd expect a Velological, with it's much higher shaft speed, to perhaps have thinner lams and maybe some silicon in the steel.

For that matter, I think the oddly peaky curve in the first scope photo is just showing the effect of magnetization, the B-H curve. I think you said it the generator was not loaded.

It also looks (for Duppie's question) like the SON can deliver 7 watts with a load larger than the nominal 12 ohms (your test point was 20 ohms, I think?) so maybe a series connection is a better choice than a parallel, assuming the headlight actually is around 3 watts.

The model has 3 parameters that can be tweaked... the scaling factor for the open circuit voltage, the inductance, and the series resistance. The values that I ended up with for each parameter were adjusted to provide a reasonably good match between the V-I (voltage-current) curves for the model and the test data for the range of speeds and loads that were of interest to me.
The model does have a value of 2.8 ohms for the series resistance, while the winding itself measures 2 ohms (IIRC). Close enough for gov't work? ;)
The fact that the model and the data match pretty well shows that the resistive portion of the dynamo impedance is much smaller than the inductive portion over typical operating speeds.
As far as design choices for a dynamo... that's beyond my expertise! I had a nice class on electric machinery back in my undergrad days, and it was pretty interesting. Never had a chance or reason to get any further experience, beyond some details on sources of losses in the magnetics in a switching power supply. Well, I've spent some time on EMC, where the goal was to add losses at specific frequency bands through the use of ferrites. That doesn't seem to be too relevant to bike dynamos, though. :)

correct.. that was for the unloaded dynamo.
I think I've heard a similar comment regarding the reason for the shape of the waveform, but that's beyond my knowledge of magnetics.
It does serve as a cautionary tale to people who want to design circuits for dynamo lights and assume that the voltage will always be a sine wave.

my guess is that series will probably yield better results, but that does depend on the design of the light. Not sure that I can imagine a light that would actually produce more light when wired in parallel, but I'm not ruling out the possibility.

Steve in Peoria

One way to change the load on the dynamo is through the use of a switching power supply. In a manner of speaking, you can think of it as a transformer, where the power into it is the same as the power out, minus some losses that are due to the conversion itself. The input voltage and current are transformed to a different output voltage and current. Of course, resistance can be calculated by dividing voltage by current, so the resistance that the dynamo sees can be adjusted with the switching power supply.

One of the interesting parts of designing a circuit to extract the maximum power from the dynamo is that the circuit has to evaluate how much power is being drawn. A common method is to continually experiment with small changes to the transformation ratio (or the resistance that it presents to the dynamo). If the power extracted goes up, then that change will be kept. If not, it'll keep making the small changes. Essentially, it should end up at the max power point on the dynamo's voltage-current curve, while continuing to make small excursions to each side of that point as a way of verifying that it is still at the max power point.

This is pretty impressive all by itself, but.... you also have to consider that the dynamo's max power point changes as the speed changes! This means that the circuit will have to chase the max power point constantly. Modern microcontrollers make this much easier, but it still adds cost, size, etc. It also adds drag, because that extra power extracted from the dynamo is coming from the cyclist.
Of course, the same circuit could be used to draw 3 watts from the dynamo, but do it at a different point on the dynamo's V-I curve that would have fewer losses in the dynamo, resulting in less effort required by the cyclist. I wonder if people would pay extra for this feature??

For all of these reasons, it seems relatively uncommon to see circuits that track the max power point on bike lights and battery chargers. It still seems like a good idea to me, and it's nice to see Andreas write up an article to shed some light on their performance.


Steve in Peoria

Supernova clearly has a voltage doubler that increases the light at higher speeds. There is a discrete increase in light and drag at a particular speed. It could be active, because there is a lot of circuitry in there. The issue with a dyno is that if you could increase the wattage put out by the hub, the rider might not appreciate it. I have seen the supernova doubler kick in while I was pedaling.

Voltage doubler circuit represents old school electronics. These days by fast switching you can do crazy things. You can load the dynamo a variable fraction of the time, get any voltage you want etc. I highly doubt they would lower themselves to any conventional textbook voltage doubler.

my opinion of supernova is pretty low, you must be seeing something I'm not. They do make nice cases though.

For a dyno that can truly deliver twice the current needed by the light at the 5 or 6 volt output level, parallel connection might well work. But for the dyno you analyzed, not. If nothing else the dyno internal resistance would have to be about half of what it is.

Fast switching is not new, so there's no "these days." I was designing switching power supplies in the 1980s and well after. Yes, you can do those things, but at a price. You can build one of anything electronic, but can you make a profitable manufactured product out of it?

Yes, all digital electronic relies on fast switching. As to the power circuits, you could not manipulate then high currents with tiny and inexpensive devices that dissipate hardly any power, going immediately between excellent conductor and excellent insulator.

One handy feature of the typical dynamo is that they produced a fairly constant short circuit current of 0.5A over the usual range of speeds. For the sake of the rest of the audience, this means that if you short out the dynamo's output, you'll get 0.5A or so. The resistance of the incandescent bulbs was chosen to be nearly a short circuit; 12 ohms. In the days of incandescent bulbs, which are extremely sensitive to variations in applied voltage or current, this semi-regulation provided fairly long bulb lifetimes and steady illumination levels.
The downside is that the bulb nearly short circuits the dynamo, causing higher losses than would be needed if the dynamo was operated somewhere near the middle of its voltage-current curve.
People have been wiring two incandescent lights in series as a method to increase the light output for a long time. Peter White used to sell a wire harness for this purpose, I believe.
For these reasons, running two lights in parallel is almost guaranteed to not increase the total light output. The lights would just be sharing the half amp.

Steve in Peoria

true... but things have improved since the 80's.
I recall guys designing most switchers around the common Unitrode 3524(?) controller, and the circuit wouldn't be considered small or fast by today's standards. The mosfets have improved quite a bit, and it amazes me how nicely the manufacturers have integrated the mosfets into the controllers, resulting in very small synchronous switchers that are also rather efficient.
There's also the development of the small microcontrollers that can double as a switcher controller in bike lights.
Depending on what is considered to be "fast switching", circuits have been fast for quite a while, but have gotten much faster. What was typical for the 1980's? 50kHz to 100kHz? A lot of the new little switchers are switching at 1MHz and up.
Amazing stuff, and kept me rather busy when I was helping people fix their EMC problems at work. :)

Steve in Peoria

Actually in 1984 I jumped from 20 kHz to 150 kHz and released a product with that. The next year my new group was on the cutting edge of increasing switching frequencies because for space you want everything smaller, lighter and with better dynamics and control. Loop bandwidth was the name of the game. The functional benefits the other writer refers to can be had as you say with megahertz switching and at 20k to 150k to a lesser degree. I didn't design them but I was part of a team that developed them for our market. And my last power electronics product was launched in 2 2005. Did not go above 100kHz, Suitable control chips, magnetics and capacitors to suit my environment, control requirements, EMC, and corporate constraints did not allow otherwise functional chips to be feasible.

Dynamo lights are obsolete. There are numerous reasons to move away from them, but no good reason to purchase a new one. Unless you intention is to recreate a vintage bike?

Thanks for the additional discussion.

Regarding the comment - would people pay for more power, since there are a few people that do bike touring that bring along lots of electronics, there are some that bought the Forumslader (spell?) USB charger. But here in USA they are a small minority. I have found that when bike touring that the Sinewave Revolution with a pass through cache battery was adequate to keep all my toys charged on a two week bike tour. While rolling, I usually have one taillight and one GPS turned on. Phone off, camera frequently used but off when I am riding, headlamp (for my head) in the campsite and tent, etc. And when touring, 99 percent of the time the headlight (on bike) is off, instead just charging batteries. Touring I use battery powered taillights instead of dyno powered in part because during daytime I would prefer a flashing light. I mostly keep internet useage to checking e-mails and weather forecasts, nothing more. I use NiMH rechargeable AA and AAA batteries for GPS, taillight, headlamp (for my head), I charge those off of USB power too.

A few that have used the Forumslader USB charger commented on feeling the additional drag while riding, but I can't recall anyone else commenting that they could feel the additional drag with other chargers. I know I can't feel the additional drag when my USB charger (Sinewave Revolution) is pushing 600 milliamps into one of my devices. And I can't feel any additional drag with a headlight turned on.

I switched to 18650 for everything aside from few minor legacy items. The advantage is that the same batteries can be used in backup power for phone and laptop. They hold far more power per volume or mass, do not lose it and are less sensitive to cold. Some old laptop batteries can be opened to extract 18650 cells from them.