[Syzygies]

For your consideration, here is a $1.30 charging circuit to go inside a PIMETA (or anywhere else one sees fit).

My design goal is to use 8 AAA cells, and be able to charge either from an iPod power supply or from a generic 15V wall wart. Short of building a full-blown fast charge controller such as Tangent's NiMh Battery Board, there are two standard ideas for trickle chargers: A three component LM317 charging circuit providing constant current, or a single component resistor charging circuit providing a current that decreases as the battery voltage rises.

With an LM317, there is a minimum voltage overhead of 1.6V for the regulator itself, plus 1.25V for the resistor that follows, and another 0.8V for the protection diode. Trickier-to-use low dropout regulators can reduce the 1.6V, I have designed fancier resistor networks that can reduce the 1.25V, and a Schottky diode can reduce the 0.8V. Nevertheless, it is impossible to charge 8 AAA cells from an iPod power supply using any variation of such a circuit, because we don't have enough voltage overhead.

A resistor circuit poses the opposite problem: They're great if you tune them perfectly for a given power supply, but your batteries will fry if you connect them too long to a higher voltage source. Here, a simple resistor circuit would rule out the option of using a generic 15V wall wart.

I should note that Tangent doesn't like resistor charge circuits. I hang on his every word that I can actually understand, but here I cannot fathom what his objection is. It's not like the batteries are "Yo-Yo diving", the charge curve has all of its derivatives monotone, and is far gentler than a fast charge that suddenly becomes a trickle charge. I've had great luck using resistor charge circuits; their utter simplicity makes me feel like a fool for having played around so long with LM317 resistor networks.

Here is the simplest schematic I can think of that solves the design problem I've posed here:

The resistor values are scaled to end up with a C/12.5 charge rate for AccuPower 1000mAh AAA cells, and would have to be increased to lower the current, for smaller capacity cells.

A representative parts list could be:

Wire Wound Resistors - Cement filled Ceramic
Xicon 5W 5% Cement Power Resistors 5 OHMS 5% TOL
Mouser Part #: 280-CR5-5.0, $0.39

Wire Wound Resistors - Cement filled Ceramic
Xicon 5W 5% Cement Power Resistors 15 OHMS 5% TOL
Mouser Part #: 280-CR5-15, $0.39

Central Semi Diodes*DO-201 12V 5.0W
Mouser Part #: 610-CZ5349B, $0.52

5 watts is of course massive overkill here, but distributes the heat very well. If you lower the watts to save weight, then do the heat and watt calculations.

If you're comfortable just solving the equations, then do so, and please don't let my graphical description of the circuit irritate you. Otherwise, one can understand this circuit in terms of the following graphic, which enables me to solve these equations while in the shower or driving:

In this graphic, width is resistance and height is voltage, so steeper slopes are higher currents.

The top dot on the left is an ideal 15V voltage source. The two lower dots on the left model different iPod power supplies, as ideal voltage sources followed by resistors. If anyone can provide me with more data points here, I'd be very appreciative. Before building this, you should certainly measure your iPod power supply to see if it behaves anything like either of mine.

The scale on the right is the per-cell voltage seen across a charging battery of 8 AAA cells. In my experience at room temperature with trickle charges, they tend to peak at 1.45V per cell. (Tangent models using 1.55V per cell for his fast charging circuit, which is different.)

To understand the current through a resistor charging circuit, stretch a rubber band between the battery voltage and the power supply, and read off the slope. This is exactly what the equations also say (if one takes the trouble to include a model for the power supply, which matters here), choose the approach that you find more comfortable.

Putting a Zener diode in the circuit is like hammering a nail into the middle of the graphic. The rubber band is forced to bend around the nail, protecting the battery from too high a current when using higher voltage power supplies.

As a bonus, the Zener diode serves to protect the amplifier capacitors. If someone takes out the batteries (or one comes loose) and turns the amp on while connected to a power supply, the voltage passed to the amp is limited by the Zener diode. I can safely use 16V caps, and a "15V nominal" power source that could be over 16V.

Here is how I plan to use this circuit, between two RadioShack 4 AAA holders, in a Hammond J16 case sawed down to nearly the size of a J12. The RadioShack battery holders are slightly smaller that any similar holder sold through the usual internet suspects; I doubt anything else will actually fit:

 

[rickcr42]

do a google on the TL431 and battery charging.


same principle but more elegant

nice job BTW-just offering options man.

 

[Syzygies]

Quote:

Originally Posted by rickcr42 do a google on the TL431 and battery charging.

Thanks. Here is the TL431 datasheet. It's an "Adjustable Precision Shunt Regulator", which seems to be a fancy term for an active three pin Zener diode, where the third pin externally sets the threshold voltage. (It appears that one needs an external voltage reference to use this, except in clever circuits that are appropriating its innards for a different purpose.) It's cheap and small, $0.32 at Mouser, a nice jelly bean to keep in mind for future designs.

However, I couldn't find a TL431 charging circuit anywhere near the simplicity of my proposal? Links?

 

[rickcr42]

coming right up if i can find the links

I have some designs on my hard drive i will send you direct for your reading pleasure if i can't turn up the links (been a while

)

 

[rickcr42]

here is something :

http://www.uoguelph.ca/~antoon/gadgets/c72.htm



obvioulsy you can modify it for any voltage up to (i think,from memory) 30 volts at 100mA.

More current with a pass transistor

 

[rickcr42]

as an add-I use shunt regs [tube (see the Tube CAD Journal),discrete and the TL431] for just about all audio circuits except for very high gain low noise circuits such as mic preamps and phono stages.

shunt regs superior sound but usually not the lowest noise levels.At least I can not get them down where they need to be......yet

 

[Syzygies]

Quote:

Originally Posted by rickcr42 here is something : http://www.uoguelph.ca/~antoon/gadgets/c72.htm

Thanks. This is still more complicated than I'm proposing, but it gives me a different idea: I've wanted to design an "active" storage system for my spare NiMh batteries floating around the house, so when I grab a pair in a hurry, they're charged. I had been thinking timer, but a modification of this circuit would be just the ticket for building an array of trickle chargers that tripped on and back off at specified voltages.

Back on topic, it would be nice to design a trickle charge shutoff mechanism that actually fits in the tiny crawl space I illustrated above. Perhaps crossing a voltage trips a TL431 circuit like the above, that sets a timer for later shutdown?

 

[rickcr42]

try to download the TL431 data sheets from each manufacturer.they really are different and have various ciruit ideas/complete ciscuits.

Fairchild,National Semiconductor,Texas Instruments,On-Semi,etc.

the circuit above is an example but as with many such things you can take only the parts of it you need and since the actual theory/description is there far easier than some to decipher.

hope this helps

Shunted and regulated by rick

 

[tangent]

Quote:

Tangent doesn't like resistor charge circuits

Yup. NiMHs want constant current charging for best results, according to their manufacturers. But I think your proposal qualifies, or can be made to qualify. (I'm vague because I haven't run the numbers on the component values you specify.)

Quote:

seems to be a fancy term for an active three pin Zener diode

That's right, it's a glorified zener. The main functional difference between it and the more typical voltage reference is that it goes to higher voltages.

Quote:

Shunted and regulated by rick

Looks like someone's bucking for a new custom title...

 

[@sia@home]

I'd like info on the best way to be able to charge my pimeta with my steps.

The steps is dialed for 32v atm

And my pimeta uses 10V rail caps and has 6 x AAA batteries.

I was thinking, from reading about the PPA BB, I would need 1.5v / cell, + 1.4 for the same diodes, like in the pimeta, but I'm really unsure as to how, to drop the voltage down to 14v or so?

I literally have a bag of 125 jfets, could I use these like some ppl use the jfets in the OPAMP power isolation array so they can still use those ad opamps that max at 20something volts?

 

[Syzygies]

Quote:

Originally Posted by @sia@home The steps is dialed for 32v atm. And my pimeta uses 10V rail caps and has 6 x AAA batteries.

In the "tutorial" thread Standard LM317 Charger for 9V NiMh Batteries, randytsuch wrote:

Quote:

Originally Posted by randytsuch One other thing you could have done is use 2 317's per circuit. Takes a bit more room and parts, but the parts are pretty cheap, so cost is not really a factor. One 317 would be a voltage regulator, and the other a current regulator.

So you don't need the resistor-Zener-resistor combination I'm proposing in this thread, its purpose is to get by with very small voltage overheads, while retaining a tolerance for higher voltage sources.

You need to dump, like, 22V or more, while making sure you never expose your caps to 10V or more. While your battery pack is connected, there's no risk, but if any cell jiggles into open circuit for an instant, a constant current charging circuit could zap your caps. On the other hand, if the first stage of your charging circuit drops the voltage to 10V, you don't have room for a constant current LM317 to work afterwards. So this is a delicate problem.

Off the top of my head, I'd put a 9V Zener diode in series with a Schottky diode, parallel to the battery pack. Then the voltage will never rise above 9.4V, more than enough to charge your battery pack, but not enough to fry your caps. Test this on the bench, with your battery and the caps disconnected, to make sure that it works and that you've figured out which way the diodes need to point. (It's a simple exercise, but you can't phone this one in, unless all diode behavior is obvious to you.)

A potential problem with any kind of safety shunt around the battery is a faint reverse current leak, all the time. NiMh cells lose their charge slowly all the time, you're ok as long as the leakage you introduce is a smaller effect.

Reading from a sample datasheet, say your Schottky diode has a reverse current at 10V of 75 uA, and your AAA cells have a capacity of 750 mA. Then if you use this safety shunt, and do nothing to switch it, it will take over a year to discharge your batteries. However, at 100C the same diode could have a reverse current of 5 mA, discharging your batteries in 150 hours. The truth is undoubtedly somewhere in between, but I'd say if your amp goes long enough between charges for this to be an issue, you've found a new hobby.

Now, how do you dump these 22V? You've got to learn to do heat and watt calculations (I learned off the LM317 datasheet), this is a long free fall. Using a series of LM317's can break up the load. However, over two-thirds of your electricity is being wasted as heat, all the time. I suppose I shouldn't worry about this, the STEPS is a little toaster oven itself, right? But at least it's separate, it's not cooking your amp.

I'd avoid dumping all this heat. I'd use a switching "buck" regulator to drop the voltage to just enough overhead for an LM317 constant current circuit, with the battery pack protected by the Zener/Schottky diode pair. I've never built such a circuit, but it would be fun to learn, tell us how it goes.

 

[Syzygies]

Quote:

Originally Posted by tangent Yup. NiMHs want constant current charging for best results, according to their manufacturers. But I think your proposal qualifies, or can be made to qualify.

My favorite NiMH datasheet is Eveready's NiMH Application Manual, and I've read lots of hearsay elsewhere on the web. I still don't understand the basis for the idea that NiMHs want constant charging current. As opposed to what? All of my schemes for charging current graph as straight lines that change slowly over twelve hours, with perhaps one Zener bend to protect the battery.

I think the "Yo-Yo Diving" analogy is apt. People get the bends by diving a sawtooth pattern while trusting a dive computer. Medical scientists debate the exact mechanism, but divers on a boat say "Doh! It's like shaking a soda bottle!" Wine degrades in a wine cellar if exposed to daily temperature variations, but can tolerate seasonal temperature variations well. Same idea.

Fast charging is a convenience demanded by consumers that NiMH cells tolerate fairly well, but at a cost compared to trickle charging. Fast charging is a series of different current regimes, and is often actually a square wave, where the circuit regularly shuts off for a moment in order to better measure the battery voltage.

I'm of the uninformed opinion that any slow (C/6) to trickle (C/10 to C/20) charging curve, that ends in a trickle and has all of its derivatives monotone, has to be superior to what fast charging does to NiMH internal chemistry. But I'd love to understand the science better.

In any case, using my iPod power supply to charge my amp is a convenience, that I prefer to the convenience of fast charging. Just as people who fast charge are either oblivious to the cost in battery life or accept this cost, I am willing to accept a potential cost in battery life. In my own experience I haven't paid it yet. My primary concern is to not be stranded without music 30,000 feet over Kansas, a fate I've managed to avoid. All of my circuits manage to fill my batteries.

The skeptic in me believes there is no such cost. As you all know, I have an irritating inability to accept advice that comes without empirical support and a scientific explanation. I still think very carefully about your advice, trying to fill in the missing experiments and explanations. What's the science, here?

 

[tangent]

Quote:

any slow (C/6) to trickle (C/10 to C/20) charging curve, that ends in a trickle and has all of its derivatives monotone, has to be superior to what fast charging does to NiMH internal chemistry

That I'll agree with. The only thing that divides us is whether CC trickle charging is better still for the cells.

Quote:

What's the science, here?

Perhaps you can find what you want in "Handbook Of Batteries", ISBN 0071359788. If it wasn't $110, I'd have a copy myself.

I get the impression that you're an EE student, Syzygies, so perhaps you have access to a good technical library, where such a volume might be housed.

 

[@sia@home]

nice big long post/QUOTE]

Thanks mate for some insite.

Ok, I only need to drop 14V to be on the same side

Because, they are 10V caps, 2 per rail, so the rail is divided between the two, and can then therefor tolerate 20V.

LOL,

Ok I have 25 2n5484 jfets and 100 2n5486 jfets.

I was thinking of using jfets for this purpose, I know they cause a voltage drop when they are sourcing currects near their idss.

I guess I'm wondering if its safe to cause them to drop 6v in two steps, I know ppl do this with the PPAs and AD8610 opamps so they can still use highish (32V+) for the buffers.

Yeah, Could I even use the jfets as a current source?

I do have 1 spare lm317 and also one spare lm337 regs here. Many diodes, etc etc.

I'm not really interested in dropping 14V over an lm317 inside my serpac H67 btw

Thanks for your input, I'll probably try what you are suggesting tomorrow

 

[Syzygies]

Quote:

Originally Posted by @sia@home Because, they are 10V caps, 2 per rail, so the rail is divided between the two, and can then therefor tolerate 20V.

In principle, yes, while your railsplitter is working. This is frequently discussed here, there is one suspected fried railsplitter in today's posts, a typical day. If you don't protect against this you could lose your caps. As a habit, don't set up dominos.

Is the jfet voltage drop reliable at low currents? I know that basic diodes drop less at tiny currents, and a high voltage can eventually take out your caps, no matter how small the current. I can imagine various scenarios where a failing charging circuit (again, loose batteries) could yield a tiny current at nearly your source voltage.

Edit: There is nothing mysterious about how an LM317 fails as it loses overhead: It can't help but insert its own voltage drop, the voltage squeeze causes the drop across the current-sensing resistor to go below 1.25V, so the LM317 goes open throttle, but to no avail.

An LM337, meant for negative voltage applications, would have the opposite reaction, shutting down. I don't have time to design this, but I can imagine a chain

* LM317 to set a reliable voltage, e.g. 24V
* LM317 to set a reliable current, staying open if the drop across its resistor falls below 1.25V
* LM337 tuned to not interfere, but closing if the drop across its resistor falls below 1.25V
* A big resistor for the main drop

I'd test such a circuit to see that it provided a constant current to a battery under all battery voltages, but never exceeded 9.5V no matter what resistor load I used to short it, with the batteries removed. The interesting case is a large resistance load, where without the LM337 the above circuit would collapse to a high output voltage, something like 20V. I'm guessing an LM337 could be used to fix this.

Edit: Now I see what Randytsuch might have meant. It would be cleaner and simpler to have a current-limiting LM317 feed a voltage-limiting LM317. Under normal conditions the first LM317 would starve the second, but successfully feed the batteries. Under abnormal conditions the second LM317 would limit the voltage to 9.5V.

We should collectively figure out the drop-dead-simplest version of this idea, because @sia@home's scenario is quite common.