View Full Version : Direct battery power
10-21-2011, 11:51 PM
I have an odd little application that doesn't quite fit any of the available hardware, so I might end up building and stuffing my own. I won't need enough units to justify bothering one of the people who are better at this, but I might need a couple of dozen over the course of a couple of years. The basic app is that the Propeller (equipped with an oversized EEPROM) plugs into the serial port of an embedded device and either saves the device's state info to EE or restores it through an existing and fairly baroque serial protocol. The embedded devices are very expensive so devoting a save/restore unit to each device is economically practical and has certain advantages for the very stupid people who might have to use them. Since I might be building a couple of dozen of these things I am looking to make them as simple as possible.
My question is, what are your experiences with powering the Prop straight from batteries without a voltage regulator? Is it safe to use two AAA cells in series for raw power with no other conditioning? Or will a lithium cell like a CR2032 work to power an entire system (it only has to run for a minute or two at a time, but needs accurate timing for fullduplexserial so it won't be pulling just microamps). I am hoping to get away with using just the Prop and a handful of passive components.
10-22-2011, 12:09 AM
No, a CR2032 will not work bcause it cannot supply enough current. See these specs http://en.wikipedia.org/wiki/List_of_battery_sizes
Two AAA in series would work but I really do like to have a regulator just in case.I am presuming from what you say, it only requires to run for a minute or two, so you could quite easily use rechargeable AAA's. You could certainly get away with minimum components.
Is your interface TTL? Remember, RS232 can require more current!
10-22-2011, 06:50 AM
I have a board I built with a prop, eeprom, pressure sensor, and some LEDs. So far it has worked fine in tests wired directly to 2 AAA batteries. I also have a pads for a CR2450 holder, but I have not tried the CR2450 yet.
Oh, and during construction of the prototype, I soldered the EEPROM 180 degrees rotated. It got really really really hot. I desoldered it, flipped it around, and it worked fine. I was quite surprised.
10-22-2011, 08:16 AM
I know that this should be prefaced with "It is outside of the stated absolutes written in the spec sheets", and has no guarantee of longevity
I was messing around with a CMOS Z80 and rather than do any level conversions I thought I would try the Z80 at 3.3 Volts or a "happy medium" of 4.2 Volts on both. Nothing happened the Prop just got on with it without any warmth at the higher voltage. At 4.2 Volts a LiPo would be directly connectable, or a diode could be put in circuit to get back nearer to the wanted (this would be a protection "feature").
Wasn't there a tread recently about raised clocks and volts where all was well?
FWIW, very low-power MCUs such as MSP430s and XLP PICs are often run from batteries without a regulator, to maximise battery life. I use 2xAA cells in that sort of application. They have ADCs and can check the battery voltage, of course and have a wide voltage range, down to 1.8V, which helps. You should be able to do the same with the Propeller if you add an ADC, and can accept the limited battery life.
You can get voltage booster devices, which would extend the battery life:
It goes down to 0.35V input!
10-22-2011, 01:59 PM
Some more info: Battery life is not a requirement. These devices will be used for a few minutes at a time then sit on a shelf for years in case they are needed to prep a replacement for the expensive embedded controller. I am looking for through-hole and minimum parts count for simple assembly, as I will probably end up building them by hand. Communications will be RS232 but I am thinking the series resistor trick will work. If not I'll have lots of Prop pins available to drive a voltage doubler.
Toby's experience gives me some confidence that I can get away with a 2 AAA battery holder even if fresh Alkalines are installed, which was my worry.
10-22-2011, 02:12 PM
The series resistor trick for serial works on the receive side. On the transmit side, you're using 0 / 3.3V levels and your RS232 inputs may just see that as noise. You may need a MAX3232 with its internal voltage doubler and inverter. It depends on what your embedded controller uses for its RS232 input circuitry.
10-22-2011, 05:09 PM
Consider a pair of Lithium Iron Disulfide cells (Energizer L92 for AAA size).
-- primary cell, 1200 mAh for L92, maximum 2A pulse.
-- 15 year shelf life
-- Practically flat 1.5 volt per cell discharge curve.
-- Very wide temperature range
10-22-2011, 06:33 PM
Mike, the device I'm linking to itself uses a MAX232 (the 5 volt version) for TTL-RS232. The RS232 spec says 3 to 15V is logic 1, so it should work. (Of course the same spec says 0V is invalid but almost everything takes it as logic 0.)
10-22-2011, 07:20 PM
Real RS232 specifically excludes -3V to +3V in order to improve noise margins. If your embedded device uses a MAX232, you'll need to use a MAX3232 for reliability. Since you're using the device so infrequently, you could us a separate battery to provide the negative voltages for the RS232. A couple of watch batteries would easily provide -6V and last for years and years.
10-22-2011, 07:29 PM
Since this thing only operates at 19200 baud the noise margin isn't a big factor. I've driven these inputs directly from Basic Stamps for years at a time with no problem, the only issue is whether they'll see 3v3 as logic 1.
10-22-2011, 09:06 PM
The MAX3232 receiver high-going input threshold is typically 1.5 V, never more than 2.4 V. Hysteresis 0.3 volt. So you should be safe with a 3.3 or even a 3.0 V swing, if those are indeed the thresholds and the distance and noise immunity are not factors. If need be, a transistor circuit can steal power from the external full range RS232 levels, but that is not quite as KISS.
Another direct battery power option is the primary lithium manganese cell, CR123. Very readily available, and also good shelf life and capacity, and 3V steady output.
10-22-2011, 11:07 PM
Cluso, your link wo Wiki actually suggests to me that a CR2032 might work in this application. I don't need to run the Prop at 80 MHz; I need accurate timing for the UART but I can run the chip at PLLX1 and 5 MHz, which is even slower than RCFAST. With only a few cogs active that should give under 10 mA draw, and the EEPROM isn't likely to add much to that. It only needs to run for a couple of minutes and only once every few years, so this might be the way to go. I think some experimentation might be in order.
(The Prop itself is actually something of an overkill for this app, but it's worth spending the extra $6 on the CPU to avoid the weeks of research to find and tool up for the $2 CPU that can do the job. Less than 100 of these will ever be built, and this is a perfect example of where the Prop's extreme versatility and the familiarity you can get with it justfy the extra cost. That and the DIP version.)