SparkFun Forums

[SOI_Sentinel]

Working on a little project with a PIC24 XLP, accelerometer, and an RM22 radio. Now, I'd LIKE to let it packet transmit at full power, but run it off a coin cell.

Under unused conditions, the battery would be disconnected. By applying a 3V voltage between A and B, this would charge the supercap. Once at a good level, some switch would turn on the connection to the coin cell. The coin cell would keep the capacitor topped up. The inductors are meant to be used to limit the current flow from the battery to the maximum specified by the battery manufacturer (10-30mA pulse depending). What I'm trying to do is permit a few packets every 10 seconds or so at up to 100mA (although I'll take what I need), and then back into sleep at nanoamps for most of the time. I'm not planning on using any regulator if possible.

I know I could do it with a bistable relay, but it's far too large. The project will not be much larger than the largest component on a side. I'm thinking a MOSFET, but the only way I can think of doing that is with an IO on the PIC24. That takes power, though. I'd prefer something resettable, but right now I'm interested in any ideas.

I've attached an incomplete illustration. No active parts right now.

BasicPower.JPG

[manton]

There is not much point in using an inductor, as when dealing with steady state DC the limiting factor is the resistance of the inductor. You might as well just use a resistor for this application. If however you wanted to limit the inrush current then an inductor might make sense, but in this case the charge time is likely to be quite long, so it is not really inrush current that you will be dealing with.

Watch out for the ESR value of the capacitor if you want to be able to pull 100mA from them without regulation. The ESR is equivalent to a resistor in series with the ideal capacitor, and there will be a voltage drop across it, depending on the current that is being sourced.

Also note that the leakage current on these types of capacitors can be quite high, so it will self discharge, and if charging, will always require some current to keep it charged (this can be in the microamps range).

[markaren1]

Interesting idea.

I would start by learning more about the duration of, and sizes of current pulses involved.

a) What is the quiescent current of the circuit (always-on current, CPU, accelerometer, Radio in RX)
b) What is the duration of the transmit time
c) What is the current required during transmit
d) What is the duration between packets (transmitter off)

Next step is to understand what the coin cell is capable of supplying. Max cap charge current = max cell - quiescent.

My gut feeling is you will need massive inductors. A constant current charger will be more appropriate.

Finally, you will probably have to drive just the transmitter PA section from the charge storage since it is possibly the high current sink.

This may just be possible, but I think you are going to have to do a lot of characterizing of components in isolation. You might learn a lot, this is very analog.

-Mark

[SOI_Sentinel]

Yeah, that's what I was afraid of. Most coin cell type supercaps I've seen have an ESR of 30 ohms, so they would be problematic. Everything up through startup the coin cell can handle without support. Basing the PIC off a PIC24F32KA301.

Idle long term(let's say 10s cycles):
radio: 0.5uA worst case, registers stored
accelerometer: looks like 14uA with a reduced bandwidth, it may be configured to send an interrupt to the PIC for wakeup duty instead of leaving the PIC active.
PIC24: potentially 2.2uA @ 32khz (internal LRPC)
Total: ~70uA worst case

Startup (probably 1ms, transition to Active if PIC24 determines that accelerometer data is valid):
radio: 8.5mA with regulators enabled, but not TX/RX
accelerometer: looks like 165uA all out
PIC24: potentially 1.8mA @ 8MHz (main clock generated from radio clock output)
Total: ~ 10.5mA worst case

Active(probably 50ms at a per cycle in 5-10ms pulses with intermidiate Startup power levels between packets, immediate drop back to idle):
radio: 19mA at RX or TX +1db, to 85mA for full power TX.
accelerometer: looks like 165uA all out
PIC24: potentially 1.8mA @ 8MHz (main clock generated from radio clock output, probably not changing)
Total: ~ 22ma to 97mA worst case, TX pulse power will come in

I can power the whole thing from a CR2450 easily if I don't go full transmit power, but I know I lose overall capacity if I pull above the rated current draw often. I had wanted to avoid a constant current charge circuit, but I think I may have to consider the capacitor method for V2 of this if I find I need it. I think I still will look for a way to put on an electronic switch to latch the battery to active from an external momentary contact or power source.

I'm going for as small as possible. I want to be able to toss these things (with minimal padding/streamer) for various project ideas.

[SOI_Sentinel]

It's been a few days, so time for an update:

Coin cell on/off switch: I might be able to do something like this AC switch, but without having a 3mA drain from a (rather interesting) opto isolator:

http://www.irf.com/technical-info/appnotes/an-1017.pdf

I also found some useful, simple notes on using a capacitor for pulse power applications when paired with a coin cell.

http://dataweek.co.za/news.aspx?pklnewsid=37640

I'm looking at attacking my project in 3 phases:

Phase 1: Coin cell and external power, so I can develop with a benchtop power supply then test at low transmit powers.
Phase 2: Coin cell, capacitor, and external power: With the coin cell shut off for storage. The capacitor (especially if I still go with a Supercap) will need to be precharged to not stress the coin cell prematurely
Phase 3: Coin cell, capacitor, air core transformer "wireless power": A little wireless power for startup if I find the system small and reasonable. It'll let me seal the unit against the environment.