Robots charging robots

I've had a recent glut of "good ideas" recently and one might help along my swarm dream. I always wanted a bunch of little robots that could assist each other in some way. The aforementioned idea would take a Joule Thief running off of a fully-charged robot's battery pack and use it to give a dead battery on another robot a quick boost to help it get out of the dark and back to a better lit area (bots would normally recharge in the sun).

For those not into chasing links around the 'net, here's my Ctrl-C/Crtl-V version of what a Joule Thief is:

"'Joule thief' is a nickname for a simple switched-mode power supply that is easy to build and can use nearly all of the energy in a battery for driving light loads, even far beyond the point where other circuits consider the battery "dead". The circuit uses a variant of the blocking oscillator as an unregulated boost mode switching power supply."

For instance, a Joule thief (from hereon refered to as "JT") with only one AA cell with, say, 0.25V, can power the white LED in this cicuit for hours:

 

joule1.jpg

Another interesting example, and semi-pertanent to this post, is this JT Charger on Make online.

The model I chose uses a ferrite toroid in the transformer design. Some have made JT's with only coiled wire but I choose something more robust. There are a million Instructables and blogs of people who've made them but isn't a lot of data about them that is meaningful (to me). I plan to use 4x1.2V NiMh cells on the robots, probably four from a NiMH 9V with ~350mAh. Maybe button cells, I don't know. Space is key. Anyhow, I know that to charge each 1.2V NiMH cell you need to supply 1.5V+ per cell or recharging just won't happen. To acheive full-charge on the cell (or the whole battery pack when cells are connected in series), you need to be putting in at least C/10 (Capacity of one cell divided by 10) amps into it or it will never reach its peak. I also know that when briefly exposed to a fast charge, the dead cell(s) will act charged for a moment. But just a moment.

 

Robot application notes:

  • Robots will "die" around 90-95% of the full charge voltage. Enough to where extended searching for light could potentially render them helpless but not enough to where they couldn't hold still and signal for help for a meaningful duration of time.
  • The "reviving" robot will do the ever-difficult-to-logistically-figure-out docking maneuvers to charge and switch on the JT for awhile. Enough to impart the memory charge but not enough to drain the responding robot into dying itself.

 

My presumptions are:

  • With a dead 1.5V battery a JT can put out 50V. Voltage is not a question for charging it seems with this circuit.
  • Although there doesn't seem to be any real detriment to going far above the required 1.5V per NiMH cell, I feel less is more in this situation (read, closer to 1.5V the better).

 

My questions are:

  • If I strapped a JT to the underside of a small robot would EMF make it bug out? The circuit uses a basic transformer by my elementary undstanding. Noise is inherent in that beasty, yes? Something about it being common noise, the unshieldable kind?
  • Is there a way to wind the JT to make its output more predictable in terms of voltage/amps?  If 0.25V going in gets >50V coming out, what does 4.2V going in do? Again, my feeling that less is more here. Maybe digging for some hard transformer math would be good for me, but to be honest, I'm hoping one of the Big Brains here knows off the top of their surely enormous head.
  • I'm also hoping to have the PICAXE on the bots to stay powered up while the charging sequence is in process. Is this an unreasonable request?

Sorry if these questions are retarded or if the idea is old news. I think I'm going to show off my electronic inexperience here...

TTTThanks for the

TTT

Thanks for the suggestion, OddBot. I went to town, I hope it inspires.

What if the Joule Theif

What if the Joule Theif charged a supercap? Although the JT by itself wouldn’t be able to supply the power required, a supercap charged by a JT may be just enough.

That’s a great angle. I’m

That’s a great angle. I’m looking to do away with the JT idea itself I think, but I like the supercap idea. I could have the solar panel be setup so that when idle (not recharging the battery) it could be used to keep the supercap topped-off at max voltage. Then when the help signal is received it will have a good power shot to give the dead bot. Hmm. It would probably have to be a pretty large cap to hold a meaningful amount of energy. Hmm. The supercap would be faster to recharge as well. Maybe write a CPR routine of “zap dead friend, go to light, charge cap, go back to dead friend, zap again, etc.” It would mimic those crazy battlefield medic types with the crazy drive to make things live.

Hmm.

Seems a bit excessive

One could simply make up a resistor bridge (before your 5v regulator) to be connected to your battery and be adjusted so a full charge appears as 5v (255 or 1024 on an ADC). Boom, done. Just do a battery check once and a while during the main routine and when it is low, but long before a brown out, go charge yourself. I use a 15-turn trimmer for my “resistor bridge” which allowed me to fine tune the system on the fly --I was able to “tune in” a nice 250 on my ADC when the batts were full --I left a little room when adjusting to be sure I wouldn’t send more than 5v to the adc.

New plan…

I’m looking at a simple charge-pump that will double the 4AA battery pack’s voltage. Then run that through a MAX16910 LDO regulator set at ~6V. Use the 6V supply to charge other robots with. The mA’s involved will be low so it will be slow going. Just pondering the concept with all y’all.

Beauty. I was half-assed

Beauty. I was half-assed working on using ADC channels as a voltage monitor. I think you just gave me a freebie. 

The plan you stated is the general goal: Monitor voltage intermittently on a loop, if it drops below the threshold, go charge. In the state of utopia there will be ample light nearby and bots will easily top off their batteries. A second lower voltage threshold will be used as well, set somewhere near the brownout point. This is a failsafe that aims to insure the processor will have stay-alive power as well as enough power run a help signal for a chunk of time. When that second threshold is crossed the bot will go into low power mode and call for help (in this case, a recharge) from another bot. I’m not married to recharging as the only option as far as the assistance idea goes. If a healthy bot could push a tired bot into the light, that would work too. Hence my post elsewhere looking for omniwheels. But I like the idea of bots “feeding” other bots.

Valve

All I can tell you about zis next procedure is zat it will be…excruciating!

-TF2 Medic 

 

Sparkfun sells supercapacitors for like $5. I lost the link, but just search for supercaps.

Go for it

F*ck efficiency! Make it work first. Then measure your performance. Then consider if you’re still interested. Then make it better.

If you want to give a quick

If you want to give a quick helping hand to a 6v battery, skip the regulator, and give it 9 or 12v, otherwise the current is not going to flow quick enough.    you would not want to do that all the time however.   you can use a trim-pot to dial it back later.