To no surprise the AAA voltage is hardly decreasing now, 1.210V, but the UC is going up at a good rate, 1.254V.

Paul

IST,

Do you have an old PC around. They usually have 30+ amp supplies. That'll get it up fast.

Coleman 5.4V Flashcell Cordless Screwdriver,
Interesting information obtained from my last link, you get two 360F capacitors with this screwdriver, from $35 including shipping (maybe cheaper elsewhere)
http://www.amazon.com/Coleman-5-4V-Flashcell-Cordless-Screwdriver/dp/B001U8FF5Q
Compare the information PDF link to the picture, it looks the same to me.
http://fullpowerinc.com/PDF%20Documents/Ness360.pdf
cat

powercat:

Great info!  Those look to be wired in series...I suppose they need the higher voltage to run the driver motor.

Bill

Truthfinder:

Im afraid i disagree with you about the batteries in parallel, ive done that many a time and there's no issue providing that voltages are the same and the 'ah' are the same. If the 'ah' is too radical in difference then they will normally 'equalize' to the lowest 'ah' rating in the pack and your overall current will be that of the lowest rated 'ah' x 2.

Your comments about the ampere-hours don't make sense.  It doesn't work like that for ampere-hours.  However, you acknowledge the issue of one battery discharging into the other battery, as does Paul.  This is not a healthy situation.  It can also happen towards the end of the life of the two batteries.  The one that starts to die first and becomes the load for the healthier battery.  So at the beginning and at the end, you loose energy and potentially damage the batteries.

Chances are noting is going to happen in most cases with small batteries.  However you never know, a small alkaline battery can source quite a bit of current.  It is simply bad practice and should not be done.

I had a glance at a Magnacoaster instruction manual once.  That fool Richard shows a diagram with big car lead-acid batteries wired in parallel.  That is insane and very dangerous.  If you submitted a product like that to UL for approval they would refuse you and behind your back laugh in your face.

For higher-voltage batteries you can give them a common ground and then use diodes to bridge all of the positive outputs together.  Then you can get your big current and avoid thermonuclear meltdown at the relatively small price of a diode voltage drop.

Albert:

DIITO . . total nonsense . you can put any battery in parallel to get more current if there the same volts

I hope that you studied my mini treatise on battery voltages for you.

Paul:

And if someone's afraid of too much current, then discharge the batteries and parallel them.

Love your swagger.  I find it very ironic that you say "discharge the batteries."

My UC is ~ 550 farads at low voltages (less than 0.4V), and lets just assume it's 650F at 2.7V,

Are you acknowledging that the UC capacitance is a function of voltage, which is what I have been saying the whole time?

As far as I'm concerned, one can prove a JT is cop>1 without doubt by ... by using one 2000mAh to 2500mAh NiMH battery to charge a BCAP0650 4 times.

This is where you loose it.  Forgetting about the JT, all you can say when you run a test like that is you were able to make a measurement of how much energy could be extracted out of the battery under these conditions with a certain error margin.  You can't make any sort of statement about COP.  Plus a certain type of battery will have a certain average energy content with some sort of standard deviation, and if you want to get picky the mean and standard deviation would be set on a batch by batch basis.

Albert:

It is here where the magic really begins and the REAL magic is that an ultracap can convert that into real energy .

Let me briefly describe what's going on when a JT generates a spike.  The transistor switches on and current flows through the coil in the JT "transformer."  When the transistor switches off the energy in the coil (1/2 L i-squared) will become a voltage+current spike that goes through the diode and then into the cap.  It can be a cap or an ultracap, either one will absorb the energy in the spike.  That's it, there is no magic.

I can suggest a little experiment.  Take a standard JT circuit and connect your scope to the LED to see the pulses on your scope when the LED fires.  Then replace the LED with a 50-ohm resistor, and look at the scope.  Do the same for 500 and 5K ohm resistors.  You will see that the larger the resistor value, the higher the voltage in the spike, and the shorter the time of the spike.  This is showing you how a discharging inductor reacts to different loads.

Then change the 5K resistor for a regular 25-volt 20,000 uF electrolytic cap and look at the scope.  The voltage spikes are now gone - completely gone.  However, each time the JT fires the voltage on the cap increases, like a step.  You will also notice that the rate of the voltage increase starts to slow down the higher the voltage on the capacitor.  Do you know why this is happening?

MileHigh

I ended the experiment. AAA is 0.491V, UC is 1.313V. I'll have to wait till tomorrow to see what was wrong with it.

Paul

I ended the experiment. AAA is 0.491V, UC is 1.313V. I'll have to wait till tomorrow to see what was wrong with it.

I am not an expert on Joule Thieves but I can make some generic comments that apply to JTs.

The real way to make sure your JT is running as efficiently as possible is to check the timing with a scope, it's all about the timing.  You don't want the main transistor to be on for too long.  If it's on for too long then you are dissipating energy in the JT firing coil for nothing, just producing heat energy that is lost forever.  You also want the base resistor to be as high as possible so that you don't waste any extra current going through the transistor for nothing.  By the same token you want enough base current going through the transistor to be sure that the transistor collector-emitter junction is saturated and the transistor is really on 100%.

I will assume that it is harder to control the running frequency and you have to start playing with the numbers of turns in the trigger coil, not so easy.  However for the purposes of charging an ultracapacitor, it doesn't really matter.  The higher the running frequency the better in fact.  The real critical issues are making sure the transistor is not on for too long and that the base resistor is the correct value.  That way you will transfer the maximum possible battery energy into the ultracapacitor and loose the minimum amount of battery energy as heat.

Oops one last thing.  The battery output impedance is a pain in the butt.  In theory you want a DC load on a battery to be as high an impedance as possible.  Jenna said it, "the faster you discharge a battery the less energy you get out of it."  That's because the higher the discharge current, the more energy lost in the battery itself due to its internal output impedance.  With a JT it's a pulse load with a fairly high pulse current.  It's hard to gauge how this issue affects the battery, since it's a pulse load, not a DC load.  It could get complicated though and if you were hard core you would investigate this.  One possibility would be to use a high inductance inductor and switch off the transistor before the maximum current is reached.  This implies a longer transistor ON time coupled with a bigger coil.  That way the battery doesn't dissipate too much energy internally because the current is lower.  There is another bonus here, the inductor will be more efficient, burning less energy as heat and having proportionally more energy available for the spike.

I almost forgot ha! ha!  There is something to keep in mind about the inductor.  If you make a larger inductor, then you use more wire, so there is a higher resistance in the wire -> more lost energy.  However, the inductance is proportional to the square of the number of turns in the wire.  So your increase in inductance rises faster than your resistance, which is good news.  If you wanted to "pimp out" your JT and avoid loosing energy in the battery, and in the firing coil, then you want a large inductor with a low resistance and you don't switch on the transistor for too long.  That implies the bigger the inductor and the thicker the wire the better.  I have a 4000 watt water-cooled power supply for my computer, and the whole computer is sitting in an aquarium filled with mineral oil!  IST built it for me.  lol

I also forgot to mention the toroid or ferrite core to increase your inductance - how could I forget that!  I think that different core materials have different properties optimized for different applications.  I am no expert here - but I would think that you would want the lowest loss possible ferrite core.  If you are going for a toroidal core as most are, I would guess that you can get laminated low-loss cores for transformers - the typical square form that you can use to build and wind your own transformer with - an off-the-shelf lowest-loss-as-possible laminated transformer core.  They may actually make special low-loss square/toroidal parts.  They would probably be expensive.

I think it is worth investigating cores because for sure some must be cheap and relatively lossy designed for applications where this is not a critical design issue.

MileHigh