Tuning / Boost capacitor calculation - how to ?

[markwsf]

I have an off-road LED dynamo set-up (SP PD-8 hub dynamo, 3 x cree XP-G, 4 x 4F capacitors to give ~5 mins of standlight capacity @ 3/4 brightness for twisty secitons / short pauses) to which I would like to add some low speed "boost" as per Martins circuits on the "pilom" webpages.

My questions is, can the required tuning capacitors be calculated, and if so how?
If not, is the approach taken one of measuring current or voltage or light output vs speed for several different capacities, plotting graphs, then fine tuning from there?

Currently I've got "some" light from 5MPH, and it's then bright at 6, but I'd like to get this down to 3MPH if possible (- am I expecting too much from too little?)

Thanks all,

Mark

 

[Steve K]

Beats me, but it's good question!

The most difficult part of the problem is just quantifying the dynamo itself. If you can run tests with the dynamo and various loads at various speeds, you can crunch the numbers and figure out what the dynamo source impedance is. With that, you should be able to calculate the dynamo output with a series capacitance.

Using some data generated by Nick Ray for the original Schmidt many years ago, I came up with a source impedance of 0.1H in series with 2 ohms, but I don't assume that there is any precision to these numbers.

If you have access to a well equipped lab, you might be able to measure the source impedance with an impedance analyzer. Otherwise, you might just start with a sine wave source, various series resistors, and use circuit analysis to calculate a source impedance based on the amplitude and phase shift of the signal at the dynamo terminal.

The other way would be to just try out a few different capacitors and measure the dynamo output current at various speeds. This might be the easiest approach. I've got a very small meter that I had strapped to the handlebars when making measurements of the max output voltage of the Schmidt. Once you've got a few data points, you ought to be able to extrapolate the trend for other cap values.

 

[jdp298]

The graphs we've seen here before http://pilom.com/BicycleElectronics/DynamoCircuits.htm
seem to get you more from more, not more from less.

You might want to skew the caps (kind of Circuit 5 with C2 too small) to peak at the lowest you can get, although Steve's idea of driving a lower amount of LEDs at low speed with minimal losses (MOSFET H-bridge etc) might be the pragmatic answer.

The calculations of all this would at one time have been within touching distance, but my grasp of classical control theory, partial differential equations and Bode/Nyqusit plots wasn't that good at its best and was more aerodynamics-related at the time, so I can't really even see where I'd start to get a handle on this.

 

[Bobblehat]

....................... The calculations of all this would at one time have been within touching distance, but my grasp of classical control theory, partial differential equations and Bode/Nyqusit plots wasn't that good at its best and was more aerodynamics-related at the time, so I can't really even see where I'd start to get a handle on this.

:thumbsup: Wot he said!

I'm not entirely sure if you are trying to reduce the point at which you currently produce light without Martin's doubler circuits or with them, as you do refer to the boost circuits in your posting? Are you using the boost circuit currently, but would like to optimise them, or not yet using it? Sorry if I'm being dumb or pedantic! :confused:

Although some say the effect is minimal and others say that they are the bees-knees, MOSFET bridge circuits should allow the speed at which the LEDs starts to produce light at low speed to be minimised. I’m not saying you’ll get lots of light, but you could reduce that 5mph point to nearer 3mph where some light is produced. But LEDs in series, as JDP implies, will need more minimum voltage to be produced before any light is produced than a single LED would need. This is why Martin shows us the voltage doubler circuits for multiple series LEDs to lower the speed at which the LEDs start to produce any light (by doubling the voltage at that speed).

Maybe the problem is also connected to the supercaps. You will be robbing some current, and therefore voltage, from the LEDS as the caps start to charge. This will tend to increase the required speed at which the LEDs start to produce light.

Yep! Different dynamos, different LEDs, different circuits, sounds a good case for practical experimentation rather than theoretical calculation. It’s all a compromise really!

 

[A10K]

My experiences with tuning caps and standlight caps:
You are correct that the LEDs will be dim while the standlight is charging (In practice the larger caps don't ever completely fill up, I can't get an un-boosted circuit go go above .46A with a 4x50F cap bank). However, if your standlight is powerful enough and you don't expect to be stopped for long periods of time, the standlight has the function of smoothing out your stops. When commuting in city traffic my tuning-cap system will have a running power of .75-.85A, and at stop signs will dip down to about .4, and at red lights will drop down to about .10, which is still a respectable amount to be seen by. I don't have problems on hills.

The tuning capacitors basically boost your power at a specific speed, but hurt performance above it. Larger-value tuning caps will dramatically reduce your power at high speeds (on one test I managed to get a boost that peaked at .43A at around 8-10mph, fading down to about .3 above that). The amount of boost you get is also dependent on the operating voltage of the LED system, so it will vary with the number and Vf of LEDs you have hooked up--the more LEDs you have, the less boost you get. I managed to drive a poor XB-D to 1.3A (at under 10mph, no less) with a tuning cap size that was only able to get up to .85 (at higher speeds) when used with a red and white LED in series. I think you may be at the point where your dynamo won't be able to provide much power to all those LEDs no matter what you do with it. I would recommend looking into paring it down to two white LEDs to make it more responsive. I'm not sure exactly how the tuning caps work, but they can't give you more power than the dynamo is capable of producing, which is proportional to speed. At one point I had a 2 front, 2 rear system that was very slow to charge (rather, it only charged once I got up above a certain speed); I prefer having a large standlight with fewer LEDs which takes a small hit in maximum output while having much better low-speed performance.

TL;DR:
Supercaps do lower your light output when you've been stopped for a while, but smooth out your light output in stop-and go situations
Unless you have a switching system, having a tuning capacitor set to give a low-speed boost on a 3-LED system will give you a small boost at low speeds at the cost of severely limiting your high speed performance.
Consider either paring down your system to two LEDs/lower Vf LEDs (like the XM-L), or use a tuning capacitor set for your median speed to allow your supercaps to become more charged, giving you more light for brief stops and slowdowns

P.S. Has anyone played with using a 555 timer to overcharge the supercaps? I think I might be on to something...

 

[Steve K]

<...snip...> I'm not sure exactly how the tuning caps work, but they can't give you more power than the dynamo is capable of producing, which is proportional to speed. At one point I had a 2 front, 2 rear system that was very slow to charge (rather, it only charged once I got up above a certain speed); I prefer having a large standlight with fewer LEDs which takes a small hit in maximum output while having much better low-speed performance. <....snip....>

for the sake of review of EE201, let me comment that the problem of extracting the maximum amount of power from a power source is a classic. The basic answer is that the most power can be extracted from an ideal source when the load matches the source. For a purely resistive source, such as a photovoltaic cell or even a battery, the load resistance should be equal to the source impedance. i.e. if the battery has an internal resistance of 1 ohm, the max power can be extracted when the load is 1 ohm.

For power sources with a complex source impedance, the best load is a bit different. A complex impedance is one that has an inductive or capacitive component (called the "reactive component") as well as a resistive component. This impedance is expressed as Z = R + jX, where R is the resistive component, X is the reactive component, and j is the square root of negative 1 (yeah, this is part of what makes EE classes so much fun).

Anyway.... the important part is this.. when the source has an impedance of R + jX, the max power is obtained when the load has an impedance of R - jX (which is called the complex conjugate). What this means is that if the source is inductive, the load needs to be capacitive. Specifically, the magnitude of the inductance needs to be equal to the magnitude of the capacitance. Don't forget that the resistive portions need to be equal too, which does make life complicated, as there aren't many bike lights that are only 2 ohms or so.

Other complications: nothing is ideal, especially capacitors, coils of wire, and just about everything else. The capacitance of a capacitor varies with frequency, temperature, and applied voltage. It has its own resistance and inductance. Likewise, a coil of wire such as what's in the dynamo has its own resistance and capacitance. An incandescent bulb is very resistive, albeit one that changes a lot with temperature. A LED headlight, on the other hand, is extremely non-linear, so the effective resistance changes greatly depending on the applied voltage. Put all of these things together, and it gets quite difficult to use basic circuit analysis to predict what the appropriate series capacitance should be.

As such, I think the most effective method to identify the best cap is to run some experiments as described in my previous post. Don't forget to pick a suitable capacitor type; it needs to be designed to handle a lot of ripple current without overheating. Stick with something designed for use in switching power supplies. Read the datasheet in detail to check the ripple current rating. Also make sure the cap is rated for the temperature range that you will subject it to. So many details to watch out for! :)