3 ms·

100 uW is not a lot. You'd have to collect it and burst to run anything more then a couple of gates.

gh02t 3y ago
It's enough to keep a low power microcontroller and maybe a low power sensor or two alive. You can get chips and sensors that will work with that budget, you just have to be very conservative and sleep aggressively. E.g. there are PIC chips that consume 25 uA/MHz at a couple volts while active (not sleep) and can do useful tasks: https://www.microchip.com/en-us/solutions/low-power https://www.microchip.com/en-us/solutions/low-power
- tempestn 3y ago
  You could also use more than one of these sources.
- shadowpho 3y ago
  Couple of problems here. So let’s say you use one of the chips running at 25 uA (lowest run mode). First of all the does not include any peripherals. UART/i2c/timer/adc/dac don’t run in that example. And god forbid you want to do RF! Then we also need to allocate power to rest of the board. The power supply is going to have quiescent (30uA if not more… probably more as it’s probably a HV source). Then you need to allocate power to sensors and rest of board, and the uA go fast.
  - gh02t 3y ago
    Yes that's what I mean that you have to take great care, I was trying to say that it's not easy to engineer around these, just feasible. You still want some capacitor buffer, work in short bursts, sleep everything aggressively, etc. And while there are definitely useful applications they aren't gonna be as revolutionary as the headlines that always come up about these things seem to imply. But you can do more than power a few gates. Regarding power supply, I'm not sure what these exact ones are, but other betavoltaic batteries are not high voltage. They're in the range of a volt or two and these are probably the same since these work more or less the same way as every other betavoltaic battery with a bit of extra proprietary sauce. Edit: Also regarding RF, you can do passive RF via backscatter or a complementary RF energy harvester. Imagine a temperature sensor that logs data a few times a day powered off the battery, and someone with a handheld scanner comes by and reads it out every so often. Or an active tamper seal that spends most of its life making sure a wire isn't broken. Just brainstorming, but something like an RFID tag that can do stuff while it's not being interrogated is the main proposed use I've seen for these.
jtriangle 3y ago
I mean, it's about 100B of them to power a high end Tesla at peak output. You could likely just use 1 billion of them and couple them to a capacitor bank. That's not quite as outlandish as it sounds, given that they're very small, 15x15x5mm And of course, this is an early gen product, not ridiculous to assume they'd be able to make these smaller, and with higher output, in the future, so we very well might get there in time. In terms of applications, there are a ton of viable applications between "occasionally used low power environmental sensor" and "car" of course.
- Kluggy 3y ago
  1 billion would fit into a cube of 1000 units long, wide, and tall. This would be 15 meters by 15 meters by 5 meters, or 49 feet by 49 feet by 16 feet, or much much larger then the entire car.
  - jtriangle 3y ago
    Yes, but, say you can get 1 watt per module, or 10, suddenly it starts to become more reasonable for many things as long as the modules stay small.
  - actionfromafar 3y ago
    Workable submarine though! Could a submarine get by with only 400 horsepower?
    - jtriangle 3y ago
      Likely, submarines are slow and quiet, and this would be very, very quiet given there wouldn't be pump noise to worry about. That's the big hurdle with the current subs, you have to circulate water in and out of the reactors to keep them cool, and for an attack sub, you don't have enough insulation/mass/space to keep them 100% quiet. The missile subs are much quieter, though, once you cook off a missile...well that's a very loud thing to do in a very quiet place. There is still the problem of the rest of the sub's power needs. I can't imagine they're terribly efficient given they have a reactor available...

It's enough to keep a low power microcontroller and maybe a low power sensor or two alive. You can get chips and sensors that will work with that budget, you just have to be very conservative and sleep aggressively. E.g. there are PIC chips that consume 25 uA/MHz at a couple volts while active (not sleep) and can do useful tasks: https://www.microchip.com/en-us/solutions/low-power https://www.microchip.com/en-us/solutions/low-power

tempestn 3y ago
You could also use more than one of these sources.
shadowpho 3y ago
Couple of problems here. So let’s say you use one of the chips running at 25 uA (lowest run mode). First of all the does not include any peripherals. UART/i2c/timer/adc/dac don’t run in that example. And god forbid you want to do RF! Then we also need to allocate power to rest of the board. The power supply is going to have quiescent (30uA if not more… probably more as it’s probably a HV source). Then you need to allocate power to sensors and rest of board, and the uA go fast.
- gh02t 3y ago
  Yes that's what I mean that you have to take great care, I was trying to say that it's not easy to engineer around these, just feasible. You still want some capacitor buffer, work in short bursts, sleep everything aggressively, etc. And while there are definitely useful applications they aren't gonna be as revolutionary as the headlines that always come up about these things seem to imply. But you can do more than power a few gates. Regarding power supply, I'm not sure what these exact ones are, but other betavoltaic batteries are not high voltage. They're in the range of a volt or two and these are probably the same since these work more or less the same way as every other betavoltaic battery with a bit of extra proprietary sauce. Edit: Also regarding RF, you can do passive RF via backscatter or a complementary RF energy harvester. Imagine a temperature sensor that logs data a few times a day powered off the battery, and someone with a handheld scanner comes by and reads it out every so often. Or an active tamper seal that spends most of its life making sure a wire isn't broken. Just brainstorming, but something like an RFID tag that can do stuff while it's not being interrogated is the main proposed use I've seen for these.

jtriangle 3y ago

I mean, it's about 100B of them to power a high end Tesla at peak output. You could likely just use 1 billion of them and couple them to a capacitor bank. That's not quite as outlandish as it sounds, given that they're very small, 15x15x5mm And of course, this is an early gen product, not ridiculous to assume they'd be able to make these smaller, and with higher output, in the future, so we very well might get there in time. In terms of applications, there are a ton of viable applications between "occasionally used low power environmental sensor" and "car" of course.

Kluggy 3y ago
1 billion would fit into a cube of 1000 units long, wide, and tall. This would be 15 meters by 15 meters by 5 meters, or 49 feet by 49 feet by 16 feet, or much much larger then the entire car.
- jtriangle 3y ago
  Yes, but, say you can get 1 watt per module, or 10, suddenly it starts to become more reasonable for many things as long as the modules stay small.
- actionfromafar 3y ago
  Workable submarine though! Could a submarine get by with only 400 horsepower?
  - jtriangle 3y ago
    Likely, submarines are slow and quiet, and this would be very, very quiet given there wouldn't be pump noise to worry about. That's the big hurdle with the current subs, you have to circulate water in and out of the reactors to keep them cool, and for an attack sub, you don't have enough insulation/mass/space to keep them 100% quiet. The missile subs are much quieter, though, once you cook off a missile...well that's a very loud thing to do in a very quiet place. There is still the problem of the rest of the sub's power needs. I can't imagine they're terribly efficient given they have a reactor available...