Matching to the impedance of an alternator.

2. John

--
Typical Larkinese, trying to send someone on a wild goose chase when
he can't defend himself against the allegations.

Somewhere in this huge thread, fairly early on, I siggested that both an autotransformer and a fixed-duty-cycle buck switcher are stable current multipiers. It was an original observation, but I can't swear that somebody else here didn't suggest one of them first.

But defend myself? Against... what?

John

(experiment

Output

cycling

You mean like taking 240V and autoxforming it down to 120V to get twice the current ?? (minus xformer losses of course).

Not sure what you mean by stable though.

boB

(experiment

Output

cycling

Since the alternator is acting pretty much like a current source at most speeds, you can get more than normal battery-charging power out of it by increasing the load impedance. A 2:1 step-down transformer (before the rectifier) will do that. An autotransformer may be convenient, some stock part that has dual primaries or a center-tapped secondary.

The other way to do this is to rectify, filter, and drive a step-down switching regulator. The danger there is that the input impedance of most regulating switchers is negative: as the input voltage decreases, the input current increases. There's a potential for latchup states. That is just a gotcha that would have to be considered in designing a switcher.

A non-regulating buck switcher that runs at fixed duty cycle has a positive input impedance and acts like a step-down transformer for DC. That's stable.

In fact, a synchronous buck switcher run at fixed duty cycle acts like a DC transformer in both directions, sort of like a pair of gears with ratio depending on the mosfet drive duty cycle. It's a continuously-variable transmission. Too bad there's no decent mechanical equivalent.

John

(experiment

Output

cycling

the

the

The real snag is that the hub dynamo, if the model paper presented by Marcel is correct, is actually _almost_ a perfect constant POWER source... sure acts like that under a lamp load, see:

Message-ID:

for details.

But I have suspicions about the accuracy of the model. If someone will send me such a hub dynamo I'll characterize it as a _non-linear_ device, which is what I suspect it really is... not a linear lump as presently represented. ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.

(experiment

Output

the

cycling

good

the

the

to

At any given speed, it's a sinewave voltage source in series with some coil resistance and some inductance; Xl is the dominant component of loop impedance at normal speeds. It has a finite short-circuit current and a finite open-circuit voltage, and has the expected curve of load power versus load resistance. That curve peaks well above the usual operating point. Since these things are usually run at a fraction of their open-circuit voltage, practically shorted, they act as current sources, all the better to not blow out the bulbs as the speed increases. Voltage source + R + L, with the source voltage and frequency increasing together, is a perfectly good model.

It doesn't have a lot of reason to be non-linear, and the posted curves that we've seen look pretty close to the simple model. People who design things like that tend to minimize the cost of copper and steel, so at low speeds the resistance matters, and at the higher speeds there might be some eddy-current losses. Actually, the frequencies and field strengths are low enough that cheap laminations will work fine, so there's not going to be much eddy loss.

The curves on page 5 don't suggest any serious non-linearities:

So get on with it and post your switcher design, and quit making excuses. Your latest stall is that you want to wait until someone sends you an actual, working, expensive hub generator. Then you'll have to build a rig to spin it and measure electrical behavior and torque and stuff. By that time, you're hoping we'll have forgotten your many promises to post a switcher circuit "tomorrow."

And JF calls me a fraud!

John

(experiment

Output

the

cycling

good

the

the

to

I think your suspicions are right. From first principles, the model should be an ideal AC voltage source (voltage and freq proportional to hub speed) behind some resistance (constant, assuming no skin effect) and some inductance. Due to the cheap steel used in these dynamos, assume some frequency dependent losses in the inductance. If you want to get fancy, add a term for load current dependent defluxing of the rotor.

(experiment

Output

cycling

the

the

to

You could probably set the buck converter duty cycle based on the incoming AC frequency and still maintain stability, since nothing the converter does is going to change the speed of the bike. That might make your output characteristic more suitable over the speed range of the bike.

[about shorting a generator's terminals]

The generator, operating into a short, cogs. That is, the feel of the rotor as you turn it is alternately a spring-like retarding force and then a spring-like advancing force. A pedaling operator will feel a vibration, more than a rotation resistance.

In the case of a bicycle-type generator, power efficiency is TERRIBLE and you cannot usefully apply an external short to the unit, because there's significant resistance internal to the windings. If any serious attempt is to be made at power efficiency in a bicycle-powered generator, it'll be with an automotive style generator, using a controlled field winding and low-impedance current-output windings. The idea of using permanent magnet dynamos works well for reliability and ease of manufacture, but isn't good for efficient charging of a battery. I'm dubious that electronic regulation, on this small scale, will improve matters.

--
Unfortunately, since a transformer transforms _power_, that 2:1 gain
in current from primary to secondary will necessitate a 2:1 loss in
voltage out of the bridge, lowering the charging current into the
battery.

Right. The force curve on the magnet poles is almost 90 degrees from the velocity curve, so the torque is sinusoidal and averages close to zero. Conservation of Energy survives intact. If the coil has a substantially resistive load, the electrical and mechanical phase angles shift, real power is dissipated, and it takes real torque to do that.

Well, efficiency is always zero into a shorted load. But if the open-circuit voltage is, say, 30 volts RMS, and it's driving a lamp at

6 volts, it's running almost shorted. I conjecture that driving torque increases from a dead short to a lamp-like (low voltage) load. Torque will probably peak around the resistive load that drops the output voltage to half its open-circuit value, because that's about when the most power is generated.

Lots of motorcycles and aircraft generators use PM fields, and have regulators. It's common to short the output with a thyristor to regulate.

John

(experiment

Output

the

cycling

good

the

the

to

Marcel's first paper includes an Rfe resistor in the model, which looks, to me, like the equivalent of the mechanical and eddy/hysteresis losses in the gadget. His voltage/frequency graph is pretty straight, drooping just a little, and his shorted-current graph is pretty much flat, actually sloping up a bit. So the simple model looks plenty good enough.

John

if you only look at a transformer yes, but if the source driving that transformer behaves like a constant current source it can deliver more power at higher voltages

no different that a gearbox in a car if you want maximum power you need to match speed with the rpm where the engine deliver maximum power

-Lasse

(experiment

Output

the

cycling

good

the

the

to

Yes, a smart converter could be done, with a little uP maybe.

You could lightly load the alternator, namely rectify and filter to get lots of voltage, and run a buck regulator off that. That would be the opposite end of the load curve where these things are usually run. You'd get numbers like 100 volts DC, maybe more, screaming down a good hill. If the buck regulator doesn't blow up, that would be stable. The buck still has a negative input impedance, but at high voltage and low current, its magnitude will be small.

It still may have startup issues. An unintelligent regulating buck converter will go 100% duty cycle at startup, and bog down the generator, and never get around to multiplying the available current.

Interesting problem. Almost makes me wish I had a bicycle with lights.

John

What?

If you have an AC current source that will deliver, say, 1 amp RMS, and connect it to a load through a 2:1 step-down transformer, the load current will be 2 amps RMS.

The hub generator is substantially a current source in the normal operating speed range. It's designed to work that way.

John

Blah! Blah! Blah! Schadenfreude, Schadenfreude, dork, dork, dork ;-)

Schematic release at maximum Schadenfreude ;-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.

(experiment

Output

the

cycling

good

the

the

to

Flail away, Schadenfreude, Schadenfreude, Schadenfreude... dork, dork dork :-)

I really don't give a wet fart what some PhD's curves show... where's the meat ?:-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.

(experiment

Output

the

the

cycling

good

shunts the

allow the

7.2V to

And doesn't compute, if you have half a brain. Flail away, Schadenfreude, Schadenfreude, Schadenfreude... dork, dork dork :-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.