Date: Wed, 4 Jan 2006 05:32:32 -0600 (CST) Subject: electronics and battery mount solution ; urban grand challenge ; undervolting radios X-UID: 117 Content-Type: IMAGE/JPEG; NAME="img2510.jpg" Content-Type: IMAGE/JPEG; NAME="img2511.jpg" Content-Type: IMAGE/JPEG; NAME="img2512.jpg" Content-Type: IMAGE/JPEG; NAME="img2513.jpg" The electronics box and batteries rest on two 20" x 1" x 1/8" 6061-T6 aluminum rectangles bolted through the drive drill motor mount. This means that the drills, batteries, and electronics all float on a rubber bushing between the motor mount and the frame. The aircraft aluminum rectangles also have just about the right amount of spring in them which further insulates the electronics from shock and vibration. Tempered aluminum alloy is sometimes used for crossbow prods (the bow) so leaf springs are a reasonable design decision. Aluminum is prone to catastrophic failure from accumulation of microstructural strain defects (which is why it is not used generally - only for something like a bow where low mass is paramount) but that's not a big deal for this application. I had all kinds of complicated ideas before. They all felt wrong to me. In the end, the mounting solution was just about the simplest thing you could come up with and just required a few minutes with a hacksaw and the milling machine to do. This solution feels right. ----- DARPA is soliciting input for a grand challenge with robots in an urban area. I have no idea how the safety issues will be handled yet while maintaining a realistic challenge. I know that the military has conducted wargames in dense urban areas before. A person I worked with was involved with the "Land Warrior" project back in the early 1990's. He told me that it was sold as turning infantry soldiers into RoboCop. Soldiers had short barreled carbines with lots of electronics on them, Leica Geovid laser rangefinding binoculars, a palmtop computer, GPS and sat-comm telephone. He told me that because he had DOD identification, he went through airports carrying tricked out machine guns without paperwork - weird. The nearest thing in the regular world I remember is a coworker who would smuggle a Zip drive in and out of AMD because the paperwork was so bad. Anyway, if you were special forces with this gear, you could run around in an urban area, see a target, designate it, upload targeting coordinates, and then some kind of precision ordnance would land on the target very quickly. Everyone would know where they were, the tactical command would know what was going on, etc. I think they held the Land Warrior game in Oakland (which is somewhat infamous given the nickname of "Oak Town" as in the wild west). So I have two predictions for the near future, the next ten years or so. There's an official mandate that 30% of ground vehicles will be robots by 2015 in the US military. By then two things will happen. 1. Artificial intelligence engines will be regulated as munitions, kind of like how crytographic and supercomputing technologies are. Export will be controlled. 2. The bolt-on autonomous vehicle kits will be used to make autonomous truck bombs by terrorists. The third world opposition is still many generations behind the state of the art and has only primitive engineering capability. But when it does eventually become a COTS (common off the shelf) technology, then controlling it will become almost impossible. Two years ago, the autonomous ground vehicle was still science fiction. In this short amount of time, it has become viable. Each vehicle might cost a million dollars. The cost will drop to a consumer level as Moore's law continues to hold. When this happens, technology proliferation will become a real issue. ----- The radio power supply issue is also solved. To recap, the problem is that NiMH batteries start at 1.375 volts when freshly charged. As I later discovered, both NiMH and NiCd exhibit a similar discharge curve. They start at around 1.3 to 1.4 volts and then rapidly discharge down to near 1.2 volts. This is held until exhaustion when voltage drops suddenly. The Maxstream XStream 900 MHz radio is specified as requiring 4.75 to 5.25 volts regulated. So with 4 freshly charged AA cells, the voltage is 5.5 volts which is too high. That means some kind of voltage adjustment is necessary. As I later found out, the specification doesn't tell the entire story. The best way to do a DC to DC power supply from a performance standpoint is a switching regulator. The first motor control board used two, one at 5 volts and the other at 12. PROS: high efficiency over wide input voltage range, good regulation CONS: complexity, more parts, especially need relatively large capacitors and inductor along with a diode The other standard way is a linear regulator. The traditional ones for the hobbyist are the 7805 and 317. I think that Radio Shack even sells both of these. PROS: simpler than switching regulator, good efficiency over narrow voltage range, wide input voltage range, good regulation CONS: still needs more parts, no inductor required but do need capacitors and maybe a voltage divider Another way is to use a germanium or Schottky diode with a forward bias voltage of .2 or .3 volts. Unfortunately, these are difficult to find. Conventional silicon and Schottky diodes with a minimum voltage of .7 volts is all I could find. This is too high for my application. PROS: simple, only one part CONS: inefficient, not regulated at all, narrow input voltage range, difficult to source parts The last way is to use a zener diode. A resistor in series with the zener is connected to input power. The output is across the zener. PROS: simple, only two parts CONS: inefficient when regulated over narrow input voltage range, not regulated otherwise I settled on the last way with a 1.2 ohm resistor in series with a 5.1 volt zener diode. Despite using only two parts, this solution is around 90% efficient. Here's why. NiMH and NiCd batteries spend the majority of their discharge near a nominal voltage of 1.2 volts per cell. So for a battery pack of 4 AA cells, perhaps 10% of the discharge life is spent above 5.1 volts. The remaining 90% is below that. When the pack is above 5.1 volts, the resistor/zener solution is very inefficient. Current is about 150 milliamps without the radio connected. This is three times the radio's command/receive mode current of 50 milliamps. During transmission, the radio draws 150 milliamps. As the zener current is 49 milliamps, the current drain is about 200 milliamps all the time. If the radio transmits about 10% of the time, then total efficiency is around 30% when the pack is above 5.1 volts. When the pack is below 5.1 volts, then the zener acts like a normal switching diode and is open. In this case, only the resistor is doing anything. The voltage drop across it is inefficiency. Again assuming a transmit duty cycle of 10%, the efficiency works out to between 96% to 99%. As the pack is above 5.1 volts only for the initial 10% of life and below 5.1 volts for the last 90%, total efficiency is around 90%. This is pretty good. A switching regulator would increase this to perhaps 95% efficiency. In general, power supplies tend to be somewhat inefficient. The warmth in heat they give off - that's wasted power. However, this only works as the radio really doesn't require a fully regulated power supply. Maxstream specified the narrow input range to guarantee the transmitter would operate. While a minimum of 4.75 volts regulated is specified, I found that the radio continued to work even when the input voltage was as low as 4.34 volts. Maxstream wanted to guarantee reliability of their product. And for that, I'd hold the voltage at 5 volts exactly. The only practical way to do this is with a switching or linear regulator. I may do that. But I don't have time to mess with it now as I do have a workable solution. In the worst case, the battery pack is around 4.8 volts and the radio is transmitting with current at 150 milliamps. The voltage drop across the resistor is then 1.2 ohm x .15 amps = .18 volts so the radio voltage will be 4.62 volts. As I found that the radio keeps transmitting down to 4.34 volts, there is a safety margin of at least a quarter volt. I also found that the Netgear WiFi router can be undervolted. An 8 cell pack of AAs has a voltage range from 11 to 9.6 volts. The device is specified as requiring a 12 volt supply. But it seems to run fine on even 7.73 volts. As you would guess, there must be a DC to DC power supply inside the router. Routers are really single board computers running Linux with lots of network interfaces. Computers require much lower voltages like 5, 3.3 and 1.8 which implies an onboard power supply. Probably anything greater than 6 volts works. What I don't know is if undervolting these radios affects ERP (effective radiated power). I may be reducing range by doing this. I can't tell as I've only used both radios from close range inside my apartment.