7 ms·

Linux on a microwave feels a lot like the NetBSD toaster from all those years ago. Additionally, it doesn't make sense from a technical perspective. People have

Linux on a microwave feels a lot like the NetBSD toaster from all those years ago. Additionally, it doesn't make sense from a technical perspective. People have converted toaster ovens to do smd reflow with an arduino [1], and that requires a lot more delicate control of temperature and timing; hell, Linux is worse for the job because its realtime capabilities are so poor. Also, if you're going to go through the trouble of redesigning the microwave, you'd do a world of wonders to get rid of the magnetron altogether in exchange for a few microwave heating transistors [2]. That way, you can target hot spots better, have better power control, and generally make better food. If you're gonna remake the microwave, do the job right. [1] http://whizoo.com/reflowoven http://whizoo.com/reflowoven [2] http://www.nxp.com/products/rf/rf-power-transistors/rf-cooking/2450-mhz-250-w-cw-32-v-rf-ldmos-transistor-for-consumer-and-commercial-cooking:MHT1003N http://www.nxp.com/products/rf/rf-power-transistors/rf-cooki...

xenonite 11y ago
Hi! Maybe you know of any oven that is using [2] or similar? I would like to buy one. Thank you.
- jfim 11y ago
  It doesn't look there are any available, although there are a multitude of press releases from the last couple of years [0]. [0] https://www.google.com/search?q=solid+state+microwave+oven https://www.google.com/search?q=solid+state+microwave+oven
  - derefr 11y ago
    There are some fascinating things you could do with one of those. You could, for example, make a "stick boiler"—like a "stick blender"—that you stick inside a cup of liquid to heat it up, without any part of the stick itself getting hot. (Presumably there'd be a waffle-grid cross-guard to block reflected RF.) But, in the wrong hands, that's a directed-energy weapon, isn't it? (Hell, stick one of these in the right kind of tube and you've now got a coherent RF beam—a maser pointer. You could probably burn your name into a wooden building from across a football field with one.)
    - toomanybeersies 11y ago
      You can do plenty of damage with a normal microwave oven in the wrong hands, I remember reading a book from the library a few years back that detailed making an EMP from one. Thinking of that book, I don't think it's likely that libraries would carry that kind of book any more, sadly.
      - Grishnakh 11y ago
        Which book was that? A portable EMP could be really useful....
        ccoggins 11y ago
        If I remember correctly, I think it was this one http://www.amazon.com/Build-Laser-Phaser-Working-Projects/dp/0830606041/ref=sr_1_1?ie=UTF8&qid=1455829941&sr=8-1&keywords=build+your+own+laser+phaser+ion+ray+gun http://www.amazon.com/Build-Laser-Phaser-Working-Projects/dp...
- IshKebab 11y ago
  The 250W transistors cost £100 each. You probably need a few of them for a microwave. A microwave with these in would cost at least £1k. Probably more, so I think it will be a while.
  - crispyambulance 11y ago
    It will be a while, yes. The price will plummet as soon as a heavy starts selling products with these components. Looking at NXP's "RF Cooking" page, it appears they're pushing this stuff with some optimism... (http://www.nxp.com/video/sage-freescales-vision-of-an-intelligent-connected-rf-cooking-appliance:RF-SAGE-VIDEO http://www.nxp.com/video/sage-freescales-vision-of-an-intell...)
- Sanddancer 11y ago
  That was the bigger wishlist item, because there aren't any out there at the moment. Unfortunately, the kitchen appliance biz is really conservative in this sector, and even with NXP having introduced the transistors a few years ago, there still aren't any on the consumer market. Freescale showed off the Sage last July [1] to try to encourage companies to move away from magnetrons, but unfortunately, everything on the market at the moment is still oldschool. [1] http://www.nxp.com/products/rf/rf-sage:RF-SAGE-PG http://www.nxp.com/products/rf/rf-sage:RF-SAGE-PG
jahnu 11y ago
Can anyone here explain to me why microwave ovens always have their power settings as multiples of some value, e.g. 200/400/800w but never have the ability to set a specific power output, e.g. 425w.
- trymas 11y ago
  Microwaves with turning knobs (example [0], maybe not the best, but you'll get the point) can have arbitrary power setting. My personal, electronic microwave, has values 125/250/500/750W so not completely exponential (I guess, that's what you meant?). One thing is, you can always (?) set arbitrary time, so you can compensate to gaps in power settings. Other thing, maybe electronics are more simple with power settings like these? When I see exponential settings I would always think of analogy to photography and 'f-stops' [1]. Some devices have exponential settings due to the physical nature of something, but is it the case with microwaves, I don't know. [0] http://www.firstiraq.com/sites/default/files/products/5002-1_neu.jpg http://www.firstiraq.com/sites/default/files/products/5002-1... [1] https://en.wikipedia.org/wiki/F-number https://en.wikipedia.org/wiki/F-number
  - jahnu 11y ago
    Yeah I was wondering if it was analogous to transformers where you have multiple coils all wound together and you combine them in various ways to get different voltages. I have almost no idea how a magnetron actually works.
- rainforest 11y ago
  At a total guess it's down to the lack of precision in the output at anything other than full power. Most (all?) microwaves just pulse the magnetron to achieve an average delivery of the rated power so the difference between 400W and 425W would be dependent on the time the microwave runs, given most tend to use a fairly long time period. You can see this effect if you run a microwave on low power for a short period - sometimes the magnetron will be running for the full duration.
- skykooler 11y ago
  Microwave ovens use magnetrons to produce the microwave radiation. Magnetrons cannot be run at anything less than full power. Some microwaves have additional coils in the magnetron that can be connected to increase the power level, but it's limited to the power levels you can get from the various combinations of coils. For the same reason, running a normal microwave at "50% power" actually means it runs at full power for several seconds, then switches off for several seconds, then repeats. The pulses are so long because starting and stopping are hard on the magnetron.
crispyambulance 11y ago
Yes, it sort of makes me think of the infamous Espresso machine on kickstarter a few years back (https://www.kickstarter.com/projects/zpmespresso/pid-controlled-espresso-machine https://www.kickstarter.com/projects/zpmespresso/pid-control...). They raised a ton of money for a "hackable" PID-contolled espresso machine starting from 5+ years ago, then delay after delay after delay.... The thing is, kitchen appliances are HARD. Appliance designers spend careers doing the stuff and have manufacturing know-how and are backed by serious resources. A dude installing linux in a microwave is going to have a good time and perhaps get other folks excited and I am all for that (and have done ghetto-retrofits myself). But there are good microwaves out there, they just cost more than others. If you just want a good microwave and want to encourage better designs, fork over the money and buy them! On the other hand, it is just a microwave. There are better ways to cook food... don't overthink it!
Grishnakh 11y ago
>People have converted toaster ovens to do smd reflow with an arduino [1], and that requires a lot more delicate control of temperature and timing; hell, Linux is worse for the job because its realtime capabilities are so poor. What the heck are you talking about? Linux has rather decent realtime facilities. It's not a hard real-time, it's soft real-time, but that's good enough for many things. It's used successfully by the LinuxCNC project for controlling CNC machines all the time, and that requires millisecond precision for running stepper motors through a parallel port. There is no way that running a toaster oven is more demanding than that. Hell, a reflow toaster oven can be run with a controller with latencies of a second or more! It's not going to make a difference there; there's far too much latency between activating power to the heating elements and seeing a temperature change for millisecond precision to be necessary. The only problem with Linux on a toaster oven is that it's just plain overkill; an Arduino is simpler and cheaper. SMD reflow does NOT require "delicate control of temperature and timing". Reflow ovens are not that accurate. They just have to follow the recommended profile well enough, and it's not that hard once a controller has learned the characteristics of the oven. I wouldn't run avionics systems needing hard real-time guarantees on Linux, but for soft real-time systems, Linux works well enough. To say that its RT capabilities are "so poor" is just ridiculous, unless you're looking at Linux circa 2001 or something.
SilasX 11y ago
It may not make sense, but in my experience, microwave manufacturers seem to be very bad at making their own device-specific software. Somehow, the software on microwaves has gotten worse over time (even without becoming "smartmicrowaves"), which I didn't think was possible. Limited sample, but I've seen three new microwaves, in '99, '10, and '14. The first one was really easy to use: buttons 1-6 were a quick-start for 1 to 6 minutes. e.g. hit 3 and it immediately starts with a 3-minute timer. If you hit start, it would immediately start with 30 seconds (or add 30 if it was already started). Second one ('10) was basically the same model, but with worse software: if you used the number quickstart, it somehow couldn't accept I/O for the next half second. So if you wanted to do 90 seconds (common operation), you would have to pause after hitting 1, then hit +30. A really annoying inconvenience, and completely unnecessary. The third microwave ('14) was admittedly a different model, but doesn't seem to have learned from previous microwave innovation. The start/+30 button doesn't work for a quick start, only for adding to an active, ticking timer. And cancel no longer deletes previous input, but emits an error sound. How hard is it not to make microwave UI software worse?
- blacksmith_tb 11y ago
  I don't think they have gotten worse, I have always been struck by the inconsistent behavior of microwave keypads. I've used plenty, and each time I have to experiment to see if, like you say, hitting the 'add minute' or '+30 seconds' will cause it to start, or require hitting start afterwards. Changing power levels is also a guessing game, do you set time then power, or the other way around? And on and on. I doubt Linux-enabling one could make this worse...

xenonite 11y ago

Hi! Maybe you know of any oven that is using [2] or similar? I would like to buy one. Thank you.

jfim 11y ago
It doesn't look there are any available, although there are a multitude of press releases from the last couple of years [0]. [0] https://www.google.com/search?q=solid+state+microwave+oven https://www.google.com/search?q=solid+state+microwave+oven
- derefr 11y ago
  There are some fascinating things you could do with one of those. You could, for example, make a "stick boiler"—like a "stick blender"—that you stick inside a cup of liquid to heat it up, without any part of the stick itself getting hot. (Presumably there'd be a waffle-grid cross-guard to block reflected RF.) But, in the wrong hands, that's a directed-energy weapon, isn't it? (Hell, stick one of these in the right kind of tube and you've now got a coherent RF beam—a maser pointer. You could probably burn your name into a wooden building from across a football field with one.)
  - toomanybeersies 11y ago
    You can do plenty of damage with a normal microwave oven in the wrong hands, I remember reading a book from the library a few years back that detailed making an EMP from one. Thinking of that book, I don't think it's likely that libraries would carry that kind of book any more, sadly.
    - Grishnakh 11y ago
      Which book was that? A portable EMP could be really useful....
      - ccoggins 11y ago
        If I remember correctly, I think it was this one http://www.amazon.com/Build-Laser-Phaser-Working-Projects/dp/0830606041/ref=sr_1_1?ie=UTF8&qid=1455829941&sr=8-1&keywords=build+your+own+laser+phaser+ion+ray+gun http://www.amazon.com/Build-Laser-Phaser-Working-Projects/dp...
IshKebab 11y ago
The 250W transistors cost £100 each. You probably need a few of them for a microwave. A microwave with these in would cost at least £1k. Probably more, so I think it will be a while.
- crispyambulance 11y ago
  It will be a while, yes. The price will plummet as soon as a heavy starts selling products with these components. Looking at NXP's "RF Cooking" page, it appears they're pushing this stuff with some optimism... (http://www.nxp.com/video/sage-freescales-vision-of-an-intelligent-connected-rf-cooking-appliance:RF-SAGE-VIDEO http://www.nxp.com/video/sage-freescales-vision-of-an-intell...)
Sanddancer 11y ago
That was the bigger wishlist item, because there aren't any out there at the moment. Unfortunately, the kitchen appliance biz is really conservative in this sector, and even with NXP having introduced the transistors a few years ago, there still aren't any on the consumer market. Freescale showed off the Sage last July [1] to try to encourage companies to move away from magnetrons, but unfortunately, everything on the market at the moment is still oldschool. [1] http://www.nxp.com/products/rf/rf-sage:RF-SAGE-PG http://www.nxp.com/products/rf/rf-sage:RF-SAGE-PG

jahnu 11y ago

Can anyone here explain to me why microwave ovens always have their power settings as multiples of some value, e.g. 200/400/800w but never have the ability to set a specific power output, e.g. 425w.

trymas 11y ago
Microwaves with turning knobs (example [0], maybe not the best, but you'll get the point) can have arbitrary power setting. My personal, electronic microwave, has values 125/250/500/750W so not completely exponential (I guess, that's what you meant?). One thing is, you can always (?) set arbitrary time, so you can compensate to gaps in power settings. Other thing, maybe electronics are more simple with power settings like these? When I see exponential settings I would always think of analogy to photography and 'f-stops' [1]. Some devices have exponential settings due to the physical nature of something, but is it the case with microwaves, I don't know. [0] http://www.firstiraq.com/sites/default/files/products/5002-1_neu.jpg http://www.firstiraq.com/sites/default/files/products/5002-1... [1] https://en.wikipedia.org/wiki/F-number https://en.wikipedia.org/wiki/F-number
- jahnu 11y ago
  Yeah I was wondering if it was analogous to transformers where you have multiple coils all wound together and you combine them in various ways to get different voltages. I have almost no idea how a magnetron actually works.
rainforest 11y ago
At a total guess it's down to the lack of precision in the output at anything other than full power. Most (all?) microwaves just pulse the magnetron to achieve an average delivery of the rated power so the difference between 400W and 425W would be dependent on the time the microwave runs, given most tend to use a fairly long time period. You can see this effect if you run a microwave on low power for a short period - sometimes the magnetron will be running for the full duration.
skykooler 11y ago
Microwave ovens use magnetrons to produce the microwave radiation. Magnetrons cannot be run at anything less than full power. Some microwaves have additional coils in the magnetron that can be connected to increase the power level, but it's limited to the power levels you can get from the various combinations of coils. For the same reason, running a normal microwave at "50% power" actually means it runs at full power for several seconds, then switches off for several seconds, then repeats. The pulses are so long because starting and stopping are hard on the magnetron.

crispyambulance 11y ago

Yes, it sort of makes me think of the infamous Espresso machine on kickstarter a few years back (https://www.kickstarter.com/projects/zpmespresso/pid-controlled-espresso-machine https://www.kickstarter.com/projects/zpmespresso/pid-control...). They raised a ton of money for a "hackable" PID-contolled espresso machine starting from 5+ years ago, then delay after delay after delay.... The thing is, kitchen appliances are HARD. Appliance designers spend careers doing the stuff and have manufacturing know-how and are backed by serious resources. A dude installing linux in a microwave is going to have a good time and perhaps get other folks excited and I am all for that (and have done ghetto-retrofits myself). But there are good microwaves out there, they just cost more than others. If you just want a good microwave and want to encourage better designs, fork over the money and buy them! On the other hand, it is just a microwave. There are better ways to cook food... don't overthink it!

Grishnakh 11y ago

>People have converted toaster ovens to do smd reflow with an arduino [1], and that requires a lot more delicate control of temperature and timing; hell, Linux is worse for the job because its realtime capabilities are so poor. What the heck are you talking about? Linux has rather decent realtime facilities. It's not a hard real-time, it's soft real-time, but that's good enough for many things. It's used successfully by the LinuxCNC project for controlling CNC machines all the time, and that requires millisecond precision for running stepper motors through a parallel port. There is no way that running a toaster oven is more demanding than that. Hell, a reflow toaster oven can be run with a controller with latencies of a second or more! It's not going to make a difference there; there's far too much latency between activating power to the heating elements and seeing a temperature change for millisecond precision to be necessary. The only problem with Linux on a toaster oven is that it's just plain overkill; an Arduino is simpler and cheaper. SMD reflow does NOT require "delicate control of temperature and timing". Reflow ovens are not that accurate. They just have to follow the recommended profile well enough, and it's not that hard once a controller has learned the characteristics of the oven. I wouldn't run avionics systems needing hard real-time guarantees on Linux, but for soft real-time systems, Linux works well enough. To say that its RT capabilities are "so poor" is just ridiculous, unless you're looking at Linux circa 2001 or something.

SilasX 11y ago

It may not make sense, but in my experience, microwave manufacturers seem to be very bad at making their own device-specific software. Somehow, the software on microwaves has gotten worse over time (even without becoming "smartmicrowaves"), which I didn't think was possible. Limited sample, but I've seen three new microwaves, in '99, '10, and '14. The first one was really easy to use: buttons 1-6 were a quick-start for 1 to 6 minutes. e.g. hit 3 and it immediately starts with a 3-minute timer. If you hit start, it would immediately start with 30 seconds (or add 30 if it was already started). Second one ('10) was basically the same model, but with worse software: if you used the number quickstart, it somehow couldn't accept I/O for the next half second. So if you wanted to do 90 seconds (common operation), you would have to pause after hitting 1, then hit +30. A really annoying inconvenience, and completely unnecessary. The third microwave ('14) was admittedly a different model, but doesn't seem to have learned from previous microwave innovation. The start/+30 button doesn't work for a quick start, only for adding to an active, ticking timer. And cancel no longer deletes previous input, but emits an error sound. How hard is it not to make microwave UI software worse?

blacksmith_tb 11y ago
I don't think they have gotten worse, I have always been struck by the inconsistent behavior of microwave keypads. I've used plenty, and each time I have to experiment to see if, like you say, hitting the 'add minute' or '+30 seconds' will cause it to start, or require hitting start afterwards. Changing power levels is also a guessing game, do you set time then power, or the other way around? And on and on. I doubt Linux-enabling one could make this worse...