Poor man's 15 Amp battery charger?

[oldswamm]

Hi Filip, <edited for spelling,, sorry Filip>
Trying to address your points in order.

You can keep it as simple as possible, or anywhere in between.
I'm trying to come up with a replacement for the switched supplies normally used for bulk charging, and providing for some optional 'bells and whistles' in the process, although my board could be used in a simpler configuration as well, just leave out all the parts but the processor and a few resistors.

As to complexity, the most complicated version of what I'm trying to design is FAR simpler than ANY switched charger. In my opinion, at it's 'worst', with 6 different caps, current sensing, and serial ID chips with temp sensors on the batteries, it's STILL quite simple.
This technology allows us to charge a 4s or 30s from the same charger, something no switched supply that I've seen can do.

We can monitor the voltage at the charge connector, so I don't see an advantage in monitoring a subset of cells as I understand you're suggesting. Also the reason for using wireless battery ID would be to keep the wiring/connections simple.

I'm trying to design in 'foolproofedness' (the spell checker doesn't think that's a word! ).
With a switch, it's easy to use the wrong switch position... Also we might want more than 3 settings. You might want to charge the racer to 4.22 and the commuter to 4.15 for example.

If your Arduino setup has an HVC shutdown input, it will shut off the relay and you wouldn't have high voltage across your BMS. I wouldn't plan on using an external HVC shutdown with this setup, complicated or simple.

If you don't want to use a processor, the analog circuit fechter proposed would be fine. If you're using a processor, it would be overkill (IMHO). I guess you could use it for redundancy, if you're worried about processor failure, in which case I would use redundant relays, but I still think that's overkill, (and you would still have to figure out how to program the redundant circuit for different batteries and keep it foolproof).

Bob

[ga2500ev]

I just found this thread yesterday, and after some thought (and reading) decided it’s a viable idea if carefully implemented. First off some comments.

I'm glad you joined. I've been lurking for a couple of months. This (and your subsequent) post actually spurred me to register.

If you are using ‘start’ caps, you are looking for trouble (IMHO). They are NOT continuous rated. They are only connected during the motors startup in their intended application. You should be looking for ‘run’ caps, which are almost never over 100uF, and for a given capacity are physically 4 to 10 times larger than a ‘start’ cap.

Exactly right. I purchased one from a local appliance parts place. 80 MFD 440V that cost $16. Puts out 3A continuous at
120V.

The hair dryer in a link in an earlier post doesn’t have a GFCI in it. It would HAVE to have a ground connection to be one. <edit> Alan B pointed out to me that the last statement is untrue. GFCI can and do work with only 2 wires. <endedit> I would consider GFI to be essential for this circuit, so beware. They are commonly sold for heat tape (for water pipe), as here:
http://www.ebay.com/itm/320861725418

I picked one up at the local Home Depot for $12. Goes in a standard electrical box. I plan on putting mine in with a switch.

Here’s the circuit I’ve come up with to control one of these chargers. The LCD display interface isn’t included in the schematic and isn’t labeled on the board. RC4-7 are the data pins, and RB6&7 are the R/S and CS.

I snipped the pictures. Refer to original post on page 6.

In an effort to make it ‘fool proof’ I intend to glue a pin header, hardwired for the digital representation of the number of cells, to the battery’s charge connector. If I eliminate a cell from a pack I would reprogram (rewire) the header (removing bad cells tend to cause packs to ‘change’ from time to time which could lead we fools to connect them at the wrong voltage). It would allow the processor to determine the charge voltage needed and could also function as the start charge signal. Obviously the charger’s battery connector would have to have a matching connector. I would connect switches to the 2 least significant bits so the human could communicate with the processor (with the battery disconnected).

That was one of the ideas I was thinking about. Instead of switches, I was thinking that a precision resistor interface specifying the High Voltage Cutoff could work. Build a voltage divider using a standard voltage on the PIC. To keep the Vdd out of it, I was thinking of referencing to one of the precision voltage references on the chip to keep it stable.

I worry about safety systems for these types of projects. I really do not trust my own software enough as the sole control mechanism for systems that can blow up or catch fire. My thinking is that non software backup systems should also be in place.

Most of the safety systems I'm thinking about come from a safety list post Lee Hart made last year on another capacitive charger thread on diyelectric car: http://www.diyelectriccar.com/forums/showpost.php?p=257414&postcount=154

in short, what systems should be in place if the micro isn't functioning at all.

To use programmable current I would probably add another couple pins to the header on the charge connector to program the current. 6 bits for 'S', and 2 bits for 'P'.

My choice of processor is based on the fact that I have some, and the software to program it. I have 18F parts that would give more usable pins, but my newer programming SW is corrupted and I don’t feel like fooling with it just now.

Let me proffer that if someone isn't married to the 16F876 that a much better choice today would be the 16F1938 from the Microchip enhanced 16F family. Same footprint, a ton more hardware tools onboard, and cost 1/3 the price of the 16F876. Microchip has them on their http://buy.microchip.com site for $2.12 each in singles. I never could convince myself to make the jump to 18F parts, and with these parts at this price point, I may never will. Other suggestions include 12F1840 (8 pins), 16F1825 (14 pins), 16F1829 (20 pins), and 16F1939 (40 pins). I'm finishing up a bootloader that can program these parts using just a USB serial port and a DIY code dumper to put the bootloader on using the same serial port.

Also note that the PC board is designed to be hand built. If I had boards built I would redesign it (vias are really handy). I might even remember to include mounting holes. For the prototype I'll just etch it on an oversize board and drill holes wherever.

Another item on my list of things to do. But it's like item 5096 on that list. I'd like to switch to surface mount, etch the board, use a stencil to slather solder paste on it in the right places, populate the chips, then throw it on a temp controlled hot plate to solder. No drilling or hand soldering necessary. One day I guess...

I’ll build the first prototype in the next few days and try to figure a way to test it (I haven't found suitable caps), so if you see an obvious error, please let me know. I just drew this this morning, so it’s POSSIBLE there’s a mistake.

Look to appliance/HVAC stores for a source of caps. Each and every outdoor condenser unit has a motor run cap in it.

Here’s a preliminary ‘outline’ of the program. How ‘steps’ are determined would depend on whether all the caps are the same, If you doubled the size of the cap for each SSR, it would be ‘digital’, With 10, 20, 40, and 80uF caps connected to 4 SSRs for example, you could control the current in 15 approximately equal steps:
[/quote]

That's an interesting idea. Maybe an extra 2-4 80 MFD caps can be used for high current charging too with controlled stepdown once the last part of the charge cycle is reached.

Snipping code...

Bob

Thanks for the insight. The simplicity of the charger has spurred me to start working on my build here. I now have a 48 V gel cell test pack and a 72V 35 Ah U1 AGM power scooter battery bank to test with.

I plan to respond to the other subsequent posts a bit later. I too am glad that I found this thread and that you joined it.

ga2500ev

[ga2500ev]

Guys this is somewhat off the subject. This thread is about a charger, not a BMS.

I'm not saying that a BMS isn't needed for this type of charger, but maybe we should make a different thread for it.

The safety parts of the charger needs to be discussed here. Unless one plans to always attend to charging, the safety features of the charger are the most crucial part for unattended operation. This means discussing the inputs to the charger that direct it to turn off and managing how both in code, and in separate hardware, how to take those inputs and use them to cut the charger off.

Bob is on the right track that the pack needs to somehow, even if very simplistically, identify itself to the charger. At a minimum a high voltage cutoff needs to be in place. Ideally maximum current, maximum temperature and charging time should be included also.

The real question is what should the charger do if there is no BMS or if the BMS fails. Pumping maximum power for an indefinite time into a battery pack will at best destroy the pack, and at worst cause a fire or explosion in an unattended setting. Bad. Very bad.

So at some level of minimum safety, we need a way to specify when a pack is connected, what the charger absolutely cannot do.

I don't think it needs to be to terribly complicated. Bob has mentioned switches. I propose that precision resistors or trimpots can do the job equally well functioning as half of a voltage divider to a fixed precision resistor onboard. I think the latter is better because that voltage data can be fed both into a microprocessor via an ADC and into a separate dumb hardware comparitor that can function independently of the microprocessor in case the software wanders off into the weeds. This will give up to 3 levels of safety control: The actual BMS if there is one, which is external to the charger, the microprocessor which is the main level of control circuit, and the backup dumb safety hardware that functions even if both of the other two systems are absent or malfunctioning.

I know this is only my second post to this forum. But I strongly believe that safety is the #1 issue and that the safety discussion should be right here, not in a separate thread.

ga2500ev

[ga2500ev]

I couldn't sleep last night for thinking about this, so I got up and drew it.
See original for diagram

Just want to say for the record that you are reinventing the wheel. Take a look at the EVILBus:

http://casadelgato.com/evilbus

Everything you've been thinking about, adds optoisolation, and has a defined communications protocol.

BTW will a 5.1V zener actually function properly with that high a resistor. I thought they needed a minimum current to work properly. Another circuit I've seen this on is Ken Boak's power supply for his PIC based telephone interfacing projects, called the Rat Ring. The circuit is on page 7 here:

http://www.scribd.com/doc/59104964/An-Introduction-to-Telephone-Line-Interfacing-Using-the-PIC-Micro-Controller

This circuit should work from less than 10s to 24s, and would draw about 60uA from the 24s (obviously the resistor could be changed to optimize it).
Would wake up from sleep mode, transmit a 24(?)bit ID # (plus a sync string, start and stop characters, and CRC) at 100khz or so, then go back to sleep for 1/2 sec or so.
The processor is old and obsolete, but I have a bunch of them, and since the bulk of the power used goes into the transmitted signal, a more efficient one wouldn't reduce consumption much.

Why not simply have a wake up on interrupt signal, and have the interrupt occur when the pack is plugged into the charger? Then it would only have to send an ID once a day or so.

The only problem I have with it is the complexity vs. safety issue. I'd still like to have a dumb backup of switches or resistors that continue to work even if the micro takes a dump.

I haven't gotten any response about the whole battery ID idea. Does that mean I'm the ONLY person who has any fear of plugging a battery into the charger with the wrong setup, and/or wants the convenience of not having to 'tell' the charger the desired parameters every time?

I'm interested, just with a simpler setup. I propose we start with something simple that works with dumb hardware. As far as I can see the charger needs at the very minimum the current and high cutoff voltage (properly scaled), and the maximum current allowed (again properly scaled). BTW you do not want to use Vdd for your ADC reference. The 16F1938 has a precision voltage reference that scales to a max of 1.024, 2.048, and 4.096 volts. If we use the 4.096 and a 10 bit ADC channel, that gives 4 mV/count, which is plenty precise.

Create voltage/current references that scale down into that range and feed into the ADC and comparitors. Use that data to make decisions.

Thoughts?

HealthYoung, I agree with what you say, but I was looking for suggestions specific to this project. I like to use a pot on some projects to adjust parameters 'on the go' for experimentation, but for this project up and down buttons would suffice for setup.

Bob

P. S. I know that's not the symbol for a zener, but didn't feel like drawing my own.
P. P. S. Man I'm tired of winter. -28F right now! Coldest winter I've ever seen!

Wow! It's been in the 80's here in Atlanta for the last week. Mildest winter on record. Hope that it blows over in Alaska soon.

ga2500ev

[Skippic]

I agree a BMS would be beneficial for this setup, just might be less messy to create a separate thread and link it.

As for the HVC, any way it would implemented should stay right in this thread. Since unless we want to burn stuff it's an integral part of the charger.

What would be the simplest way to implement HW HVC?

As for the 15 charging currents selectable by relays, I can say from experience, only three steps are needed:
- off
- max amps
- BMS leveling, topping off

With Lipo you can charge 98% of the capacity at max amps and trickle charge the last 2%.

My setup will be something like 400uF and 12uF.

[oldswamm]

What I get for sleeping. (Lots of post to respond to.)

Let me reiterate, I'm NOT trying to push more sophistication on you than you want. I'm trying to design a circuit with enough options that it will work for ANYONE.
Just because an option is there, doesn't mean YOU have to use it!

I'm surprised people think this is complicated. One processor, one op amp IC, however many shunts you need, and a handful of caps and resistors for the main board, plus a processor and a half dozen peripheral components for each serial ID unit (if used).

Rather than using quotes and responses, I just going to make random responses.

Thanks for the link to the diyelectriccar thread link, I'll pursue it farther, but the safety devices I saw so far have already been discussed here. (fuses, GFCI, HVC/BMS shutoff)

The caps should be modern UL listed RUN caps. They have a switch that opens if they build up pressure.

I have a lot of faith in my programing, especially if checked by members of this forum.

If you want true redundant safety override, it shouldn't share ANY components with the main controller.
A good cheap alternative would be a crowbar (a heavy SCR) across the output, with fuses in front and behind it. Could use a trip voltage higher than any of your packs ever see (could be switch selected, but there's the danger of it tripping when you forget to change it), and it could have a timer that energizes it if the controller doesn't send regular 'watchdog' resets (not to be confused with the processors internal watchdog).

You don't need to use all 6 relay control outputs. Would only be for those who have more than one 'size' battery. You surely aren't going to use the same current on your lite 4Ah battery as on the 40Ah+ motorcycle battery, are you? At an EXTREME it would provide almost infinite current control 'like a commercial charger'. I rerepeat, you don't have to use them if you don't want to! My immediate plans call for 2 significantly different battery packs, so probably 3 caps would do me fine, but other people might have a wider need, so I'm trying to allow for them. Skippic, are you sure your setup would be ideal for both a 4Ah 4s and a 40Ah 24s?

Good point on the internal references. I'll probably build a prototype with the VR reference, and 'allow' for it's variability when setting my limits. (I don't have a 16F1938)

I can't see why we could possibly need a separate thread to include an HVC input, or why it's 'messy'. One pin on the processor, one line of code.

Evibus is an alternative to CANbus. Involves wires and is much more complicated than adding a serial # IC to the battery. Wake on interrupt would require some form of bidirectional com, which would unnecessarily complicate things (I can live with 20 to 60uA).

My thought on the zener is that it wouldn't have to conduct continuously or even accurately (maybe I'll experiment this afternoon). As long as the voltage stayed between 2.5 and 5.5 volts, even the old 12c508 would be happy. <edit> I would want a different processor for battery temperature measurement.

The more I think about it, the more I like the idea of a wireless serial # ID, because it's simple (I've pretty much abandoned the idea of an extra connector at the charge connect as being unnecessarily complicated). Programming at the charger would be simple since the one I build would have up/down buttons and an LCD (I would require that the battery being programed for, be plugged in, so you would KNOW you were programing for the right one). And before you jump on me, YOU don't have to spend the extra $5 for an LCD if you don't 'need' readouts or programability.

The reason the processor has a watchdog timer is in case the micro 'takes a dump'. Like Alan pointed out, we just have to make SURE the reset shuts the relay(s) off.

Anybody who wants to is welcome to build a non computer controlled version, (they already are after all), but it's NOT what I'm designing.

I'm thinking that at most this would end up as bare boards, maybe with a programmed processor. Since different people would want different variations, preassembly would be problematic. It's intended to be easily assembled. The resistors are all large 1206 smt, and the smt caps are 805 or larger, which are fairly easy to work with. If by some miracle someone started selling them by the thousands, for sure, SMT would be the way to go.

Moving slightly off topic, I live in bush Alaska, and can't go to the store to buy anything in this project, except maybe a GFCI outlet (I want a GFCI plug), (and would have to pay 2 to 3 times (or more) as much as the rest of you if they had it). I have to order it, usually pay double the shipping most of you would, then hope it isn't lost (not to mention it taking an average of twice as long for shipping).

Farther off topic, we Alaskans have one of the wonders of political idiocy called 'power cost equalization'. They 'tax' the people with more efficient power production, and use it to supplement those of us with inefficient power (our local power is generated with diesel gensets burning $6+ fuel, then distributed over an obsolete grid). Helps keep us from being 'too' conservative! Did I mention that the Alaska legislature pretty much works for the oil companies?

It's warm out (over +20F!!), I'm going to get dressed and go for a ride, just for the fun of it!

Bob

[ga2500ev]

I agree a BMS would be beneficial for this setup, just might be less messy to create a separate thread and link it.

I'm good with that. All the charger needs from the BMS is a turn off charging input. There probably doesn't need to be a terribly complicated communication system for that information.

As for the HVC, any way it would implemented should stay right in this thread. Since unless we want to burn stuff it's an integral part of the charger.

Agreed again.

What would be the simplest way to implement HW HVC?

The simplest way may not be the safest. The most crucial element to understand is that:

Anything directly connected to the Cap/Bridge is at Mains Potential!

This cannot be stressed enough. That 's the reason that the GFCI is essential and that all the control systems should be isolated from the charger and the pack.

Given that the simplest way, a voltage divider, may not be the safest. The basic objectives is to scale the pack voltage into an isolated input. Then do a comparison of the current pack voltage to a fixed reference that represents the HVC. When the pack voltage exceeds the fixed reference, turn the charger off and leave it off until it is reset.

The two typical ways of having an isolated sensor is to use a current loop, such as an isolated 4-20mA current loop, or to send a PWM signal that represents a voltage over an optocoupler. But neither are a dead dumb simple as a voltage divider.

As for the 15 charging currents selectable by relays, I can say from experience, only three steps are needed:
- off
- max amps
- BMS leveling, topping off

The different charging currents are for different packs. Max amps for one pack is different than Max amps for another. Bob is trying to parameterize what the charger can be used for. That's the same reason he's trying to get pack info into the charger.

He fear, and a valid one, is that a changed pack is stuck on the charger and parameters for the charger overcharges the pack. I agree with him. It's a safety issue.

With Lipo you can charge 98% of the capacity at max amps and trickle charge the last 2%.My setup will be something like 400uF and 12uF.

Is the regardless of battery capacity?

ga2500ev

[oldswamm]

Skippic, after rereading your last post I see you are separating BMS and HVC. As far as I can see, the only part the charger needs to concern itself with is HVC, unless we want to tell it to go to trickle for balancing.

ga2500ev, my board would be isolated from the mains with a transformer, but it's 'ground' would connect directly to the Batt-. Any other setup would make current measurement problematic. In that case, I don't see why a voltage divider wouldn't be fine for B+ measurement.

I suppose an (optional?) optoisolator would be a good idea for the HVC, but most already are isolated, so it would then require another isolated power supply..... Are there BMSs that don't have the HVC either isolated or referenced to the Batt-? I haven't seen schematics for any except the ones on ES.

I suppose I should make a new set of drawings......

Bob

[ga2500ev]

What I get for sleeping. (Lots of post to respond to.)

Just wander by every day or so. Shouldn't be a problem.

Let me reiterate, I'm NOT trying to push more sophistication you than you want. I'm trying to design a circuit with enough options that it will work for ANYONE.
Just because an option is there, doesn't mean YOU have to use it!

I'm surprised people think this is complicated. One processor, one op amp IC, however many shunts you need, and a handful of caps and resistors for the main board, plus a processor and a half dozen peripheral components for each serial ID unit (if used).

Rather than using quotes and responses, I just going to make random responses.

That's just my way of doing it. Feels like a conversation then...

Thanks for the link to the diyelectriccar thread link, I'll pursue it farther, but the safety devices I saw so far have already been discussed here. (fuses, GFCI, HVC/BMS shutoff)

The caps should be modern UL listed RUN caps. They have a switch that opens if they build up pressure.

I have a lot of faith in my programing, especially if checked by members of this forum.

If you want true redundant safety override, it shouldn't share ANY components with the main controller.
A good cheap alternative would be a crowbar (a heavy SCR) across the output, with fuses in front and behind it. Could use a trip voltage higher than any of your packs ever see (could be switch selected, but there's the danger of it tripping when you forget to change it), and it could have a timer that energizes it if the controller doesn't send regular 'watchdog' resets (not to be confused with the processors internal watchdog).

Interesting idea. Is this better because unlike an SCR at the input that this crowbar doesn't turn on during normal usage, so there's no chance of runaway and not being able to turn it off?

You don't need to use all 6 relay control outputs. Would only be for those who have more than one 'size' battery. You surely aren't going to use the same current on your lite 4Ah battery as on the 40Ah+ motorcycle battery, are you? At an EXTREME it would provide almost infinite current control 'like a commercial charger'. I rerepeat, you don't have to use them if you don't want to! My immediate plans call for 2 significantly different battery packs, so probably 3 caps would do me fine, but other people might have a wider need, so I'm trying to allow for them. Skippic, are you sure your setup would be ideal for both a 4Ah 4s and a 40Ah 24s?

I figured this out. I'll probably have at least 3 different voltage, current, and chemistry packs that I'll be working with. So I'm interested.

Good point on the internal references. I'll probably build a prototype with the VR reference, and 'allow' for it's variability when setting my limits. (I don't have a 16F1938)

Easy to get. Just sample them from Microchip and tell them you are prototyping an application. They'll send you up to 12 parts a month directly shipped from Taiwan. May not be necessary for this project, but just a reference for the future. They may or may not charge a nominal shipping fee. http://www.microchip.com/samples

I can't see why we could possibly need a separate thread to include an HVC input, or why it's 'messy'. One pin on the processor, one line of code.

Not the HVC input, the discussion of having instrumented batteries that send information to the charger. That does constitute a BMS.

Evibus is an alternative to CANbus. Involves wires and is much more complicated than adding a serial # IC to the battery. Wake on interrupt would require some form of bidirectional com, which would unnecessarily complicate things (I can live with 20 to 60uA).

Just suggesting some alternatives...

My thought on the zener is that it wouldn't have to conduct continuously or even accurately (maybe I'll experiment this afternoon). As long as the voltage stayed between 2.5 and 5.5 volts, even the old 12c508 would be happy.

Probably right on that...

The more I think about it, the more I like the idea of a wireless serial # ID, because it's simple (I've pretty much abandoned the idea of an extra connector at the charge connect as being unnecessarily complicated). Programming at the charger would be simple since the one I build would have up/down buttons and an LCD (I would require that the battery being programed for, be plugged in, so you would KNOW you were programing for the right one). And before you jump on me, YOU don't have to spend the extra $5 for an LCD if you don't 'need' readouts or programability.

The reason the processor has a watchdog timer is in case the micro 'takes a dump'. Like Alan pointed out, we just have to make SURE the reset shuts the relay(s) off.

Anybody who wants to is welcome to build a non computer controlled version, (they already are after all), but it's NOT what I'm designing.

That was not my objective. I'm a firm believer of uC controlled projects. But I'm equally aware that it's easy to miss something subtle in the code that only rears its head in extreme and unforseen circumstances. I usually don't worry on projects of little consequence. But since I'm planning on building my own charger, motor controller, and DC/DC converter, I've been thinking about backup systems that catch missed failure modes...

I'll have to tackle the rest a bit later...

ga2500ev

[Skippic]

As far as I can see, the only part the charger needs to concern itself with is HVC, unless we want to tell it to go to trickle for balancing.

Would this work as a BMS balancing without feedback?
viewtopic.php?f=2&t=37978

[parabellum]

KISS entry.
How about next best Zener diode to trigger shut down relay for HVC?

[JohnC]

tumblr_m175z3KgeX1qhx5pzo1_500.jpg (115.58 KiB) Viewed 293 times

[oldswamm]

Ga2500ev, A crowbar puts a direct short across the output, blowing the output fuse from the charger and also the one to the battery. Absolute last resort, but DEFINITELY stops the current no matter WHAT the problem.

Skippic, I'm tempted to "just say no", remember, Nancy's answer to the question "should I quit using drugs?".
Actually I see several problems. Not accurate enough. To much current (what if something happens so you don't charge for a few months) (S__t happens). Just get some of these and use them once in a while. (Check them before you depend on them though. Whether you get a good one is apparently a flip of the coin though. There's a thread about them 'somewhere' on ES.)
http://www.ebay.co.uk/itm/260753810568
That's what I intend to do, then a set of methods LVC/HVC boards if I can ever afford them (if HVC triggers, it means you should have balanced). Already isolated, so it's just a matter of connecting them to the processor and ground (and using a pullup on the input pin).
If you have to balance everytime, you should either replace of eliminate the bad cell.

Parabellum, I don't understand what you mean. Where would you put the zener(s), and how would they trip the HVC? I figured on the output of a set of methods HVC for myself.

Finally got a friend to take my picture this afternoon so I could post an avatar!

I see JohnC saw my avatar!

Bob

[parabellum]

Parabellum, I don't understand what you mean. Where would you put the zener(s), and how would they trip the HVC? I figured on the output of a set of methods HVC for myself.

In simple words of electronics noob.
I use 24s system, 100v fully charged. In my case 100v zener on battery side could trigger small transistor at breakdown voltage, transistor switches of my main relay on the AC side.

Edit: 24s LiPo

[oldswamm]

HVC triggers if ANY one cell goes over a set voltage (4.28 in methods setup). Used so you won't dangerously overcharge a cell if there's a balance problem.
What you're talking about could be used to trigger the crowbar using 102v or so (the charger should shut down before it gets there). The problem with that is that it wouldn't trigger in time with 22s, so not good for a universal charger and letting the processor set the crowbar voltage (however) would fail to keep that circuit totally independent.

Has anybody tried to measure the efficiency of these chargers. A watt meter can be mislead by the power factor, but if NO heat is generated, I would have to assume it's 100% efficient. I would like to see some actual temperature/time/ambient measurements so we could calculate heat energy vs. battery energy. Any other energy leakage (RF, light, xray )?

[ga2500ev]

Has anybody tried to measure the efficiency of these chargers. A watt meter can be mislead by the power factor, but if NO heat is generated, I would have to assume it's 100% efficient. I would like to see some actual temperature/time/ambient measurements so we could calculate heat energy vs. battery energy. Any other energy leakage (RF, light, xray )?

Actually with reactance for the capacitors it's simpler than that. Any unused energy flows back to the source. The only problem with this is that there's a meter between the source and the caps. So while a lower percentage of actual energy flows into the load, with the rest flowing back into the grid, one gets charged for all of it. Also because of that lower efficiency it doesn't charge as fast or is as powerful as other types of chargers.

It's a poor man's charger to build because it's cheap and simple with essentially only two core components. However, it isn't the most efficient.

Power Correction Factor circuitry can fix this problem. But it makes it more complicated.

ga2500ev

[oldswamm]

Thanks, I'm good at missing the obvious. <edit> Obvious where the inefficiency is, not it's % value.<endedit> I've been saying for almost 40 years that computers would be a lot easier if we didn't have to deal with this darn analog world.
So what is our efficiency?
Our power factor approaches 1 into a short and 0 into an open circuit, right? So would it be around .3 into 80v? How would this convert to % efficiency? Hopefully not 30%.
I just woke up and saw your response. It's the middle of my night so I'm not at my sharpest, which isn't that sharp at best. (Can't type. I've corrected about 20 typos in this post )
Thanks,
Bob

[parabellum]

Actually with reactance for the capacitors it's simpler than that. Any unused energy flows back to the source. The only problem with this is that there's a meter between the source and the caps.

The efficiency of this charger is at least much higher then resistive charger I built. Every current path (cycle) is going true battery and voltage difference (which is converted in to heat at resistive charge) is going as revers charge in to the cap, starting next cycle (reverse cycle) at lower grid voltage. In resistive charging duty cycle is when grid voltage is higher then battery voltage and exes voltage (* charge current) is transformed to heat. In this case, duty cycle is every time the sum of grid voltage and cap voltage is > then battery voltage and exes is stored as reverse charge to start next duty cycle at lower grid voltage.
Sure we have loses here but I see no way something going back to the grid like if we just plug capacitor in to the wall socket.
Anyway, it will be just empty talking until we see some solid numbers, scoping IN and OUT for full understanding would be awesome as well.

Edit: Corrected wrong word

[parabellum]

Ups Double post

[auraslip]

DONE. This circuit takes probably half an hour to build. All it needs it the proper switch to make the relay latch on with a momentary switch.

[Skippic]

One thing I haven't thought of and had to learn the hard way is setting LVC when charging with this kind of charger. My voltage sensing wire got disconnected and the charger didn't stop charging. I found out when it was at 4.45V.

For people who are using microcontroller it might be advisable to set a minimum and maximum voltage (like 2.7V - 4.1V). If the sensing wire gets disconnected the voltage drops to around 0V and the charger turns off.

[Alan B]

Yes, check everything available to the micro for sanity and protection. Also have tome kind of a backup redundant overvoltage protection shutoff, and a time limit and amp hour limit. Multiple safety nets.

[ga2500ev]

ga2500ev, my board would be isolated from the mains with a transformer, but it's 'ground' would connect directly to the Batt-. Any other setup would make current measurement problematic. In that case, I don't see why a voltage divider wouldn't be fine for B+ measurement.

Live ground? The charger gets plugged in so that the hot line of the wall circuit gets connected to the ground of the charger... Then the charger, the controller, the battery pack and everything else is at mains potential.

My current game plan is to throw a second PIC at the problem that is powered on the mains side and uses optoisolated interfaces to communicate with the control PIC, which is powered from an isolated supply. To keep things simple I'm planning on using PWM to communicate between the interface. All it takes is a RC circuit and a opamp follower on the isolated side to recover the intended voltages via ADC.

ga2500ev

[oldswamm]

I'm confused. What are you going to do with the pic on the mains side? Are you going to measure something there? The solid state relays are already isolated, and that's all I figured on having on the mains side.

My board would be at Batt-, and there's no practical way to isolate it that I can see. I would ground the case, and the only thing not enclosed would be the battery. The battery leads are connected to the mains, but heck, the battery is dangerous at 24s or more even without this.
<edit> Do you have a plan to isolate the battery from the mains? Because the battery is dangerous,it should be totally isolated, and well insulated whether using this charger or not. <endedit>

Throw up a sketch, so we can see what the extra processor is for. It would be nice if we could come up with a cooperative design. The way things usually work on ES is there's discussion, then everybody does their own design. If you have a better way than I can come up with, and it's not to expensive, I'm fine with going with your design.

Bob

P. S. for those of you who want one of these chargers, but don't want to build one, you can buy it here.
http://www.e-volks.com/capacith_charger.html
I don't think I would want to be their insurance provider, though.

[oldswamm]

ga2500ev, I'm setting here and can't figure out your last post.
I can see where my use of the word ground, even in quotes could confuse.
What I meant was Vss, calling it ground is an OLD habit.

Both the processor and the op amp Vss would be at Batt-.

The CASE would be at earth ground, NOT Batt-.

Are you worried about the processor being at Batt- for some reason?
Or do you think the operator is endangered by setting Vss = Batt-?
The only contact the operator has with the processor is the buttons, and I have a number of different economical push buttons that would be safe. I WOULD ground the case and use a GFCI plug.
Sure easier than a second processor. I can't figure out what it would be for.
The display is easy enough to isolate from the case (and human).

Bob