Scioly.org

MoMoney$$$;)0) wrote: ↑March 19th, 2020, 5:33 pm

Umaroth wrote: ↑March 19th, 2020, 5:21 pm

MoMoney$$$;)0) wrote: ↑March 19th, 2020, 5:11 pm

Can you clarify on "luck based" in this scenario. I feel that this year the rules were pretty straightforwards and easy to follow though on. Opinions?

They were definitely straightforward (aside from ambiguity on whether or not readings had to be blind), but by luck based they mean how the accuracy of the reading oftentimes comes down to how much the calibration thermometer, and not the device, itself fluctuates.

I think that's a fairly relateable factor, but the way to keep this as straightforward as possible is to try to get a better thermometer and try to keep your calibration thermometer in the same area as a your probe, you can do this by just taping them both together and sticking them on one side of the cup, pot, etc. Every time except for one I got to see the thermometer during my reading allowing me to get exact to ±0.1 Celsius. Just a few suggestions.

We were talking about this on this discord... the problem is that getting good thermometers is pay to win for regionals rules and for states/nats, supervisors may not have access to super reliable thermometers. At every competition I've competed at this year, the thermometers have been generally bad

shrewdPanther46 wrote: ↑March 19th, 2020, 6:07 pm

MoMoney$$$;)0) wrote: ↑March 19th, 2020, 5:33 pm

Umaroth wrote: ↑March 19th, 2020, 5:21 pm

They were definitely straightforward (aside from ambiguity on whether or not readings had to be blind), but by luck based they mean how the accuracy of the reading oftentimes comes down to how much the calibration thermometer, and not the device, itself fluctuates.

I think that's a fairly relateable factor, but the way to keep this as straightforward as possible is to try to get a better thermometer and try to keep your calibration thermometer in the same area as a your probe, you can do this by just taping them both together and sticking them on one side of the cup, pot, etc. Every time except for one I got to see the thermometer during my reading allowing me to get exact to ±0.1 Celsius. Just a few suggestions.

We were talking about this on this discord... the problem is that getting good thermometers is pay to win for regionals rules and for states/nats, supervisors may not have access to super reliable thermometers. At every competition I've competed at this year, the thermometers have been generally bad

I don't really see how it's pay to win if you just need your device to be accurate to your thermometer. I used a $25 thermometer the whole year and I never had problems with fluctuation as long as I stirred the water a bit beforehand.

MTV<=>Operator wrote: ↑March 19th, 2020, 6:20 pm

shrewdPanther46 wrote: ↑March 19th, 2020, 6:07 pm

MoMoney$$$;)0) wrote: ↑March 19th, 2020, 5:33 pm

I think that's a fairly relateable factor, but the way to keep this as straightforward as possible is to try to get a better thermometer and try to keep your calibration thermometer in the same area as a your probe, you can do this by just taping them both together and sticking them on one side of the cup, pot, etc. Every time except for one I got to see the thermometer during my reading allowing me to get exact to ±0.1 Celsius. Just a few suggestions.

We were talking about this on this discord... the problem is that getting good thermometers is pay to win for regionals rules and for states/nats, supervisors may not have access to super reliable thermometers. At every competition I've competed at this year, the thermometers have been generally bad

I don't really see how it's pay to win if you just need your device to be accurate to your thermometer. I used a $25 thermometer the whole year and I never had problems with fluctuation as long as I stirred the water a bit beforehand.

Yeah I don't really agree with the regionals one, but the biggest issue is for states and nats level, where the thermometer is provided by the proctors. Coming from a team who mainly prepares for states and nats rules, it's frustrating at competitions that run those rules when the proctor's thermometer fluctuates a lot and thus messes up my score.

MTV<=>Operator wrote: ↑March 19th, 2020, 6:20 pm

shrewdPanther46 wrote: ↑March 19th, 2020, 6:07 pm

MoMoney$$$;)0) wrote: ↑March 19th, 2020, 5:33 pm

I think that's a fairly relateable factor, but the way to keep this as straightforward as possible is to try to get a better thermometer and try to keep your calibration thermometer in the same area as a your probe, you can do this by just taping them both together and sticking them on one side of the cup, pot, etc. Every time except for one I got to see the thermometer during my reading allowing me to get exact to ±0.1 Celsius. Just a few suggestions.

We were talking about this on this discord... the problem is that getting good thermometers is pay to win for regionals rules and for states/nats, supervisors may not have access to super reliable thermometers. At every competition I've competed at this year, the thermometers have been generally bad

I don't really see how it's pay to win if you just need your device to be accurate to your thermometer. I used a $25 thermometer the whole year and I never had problems with fluctuation as long as I stirred the water a bit beforehand.

I've tried a few different thermometers ranging from hazard fraught (Cen-Tech) to fancy flinn stuff at an invitational that ran with states rules. In my experience, the more expensive thermometers are usually more *accurate*, but not always more repeatable -- often the cheap ones are just as repeatable, and that's the only thing that matters if you're calibrating the error out anyway. The worst thing I see is when it's internally rounding and the display is false precision, so you get it jumping up by half a degree each time.

Generally, though, unless you go into the RTD range of expensive, it's easy to build a better detector than calibration thermometer. And that's where the trouble comes in: you're not limited by your own device, you're limited by calibration thermometer.

My suggestion would be to switch the event from temperature to something else entirely (maybe 21-22). "Wavelength" or "color" detector would be a really, really, good choice. Generating a specific monochromatic wavelength of light is easy and cheap (via an LED, or a laser, or a filter...) and extremely repeatable. An ES's setup will likely behave exactly the same all day. Building a color detector is much more difficult, and requires some innovation if you ban ICs that do that already (would you use multiple sensors, sensitive to different wavelengths, like our eyes? or a prism? who knows!). Accurately nailing down the exact wavelength of light would require quite a lot of innovation and calibration, and still be cheap (photoresistors and colored film are both cheap, relatively speaking, as are [plastic] prisms.)

shrewdPanther46 wrote: ↑March 19th, 2020, 6:07 pm

MoMoney$$$;)0) wrote: ↑March 19th, 2020, 5:33 pm

Umaroth wrote: ↑March 19th, 2020, 5:21 pm

They were definitely straightforward (aside from ambiguity on whether or not readings had to be blind), but by luck based they mean how the accuracy of the reading oftentimes comes down to how much the calibration thermometer, and not the device, itself fluctuates.

I think that's a fairly relateable factor, but the way to keep this as straightforward as possible is to try to get a better thermometer and try to keep your calibration thermometer in the same area as a your probe, you can do this by just taping them both together and sticking them on one side of the cup, pot, etc. Every time except for one I got to see the thermometer during my reading allowing me to get exact to ±0.1 Celsius. Just a few suggestions.

We were talking about this on this discord... the problem is that getting good thermometers is pay to win for regionals rules and for states/nats, supervisors may not have access to super reliable thermometers. At every competition I've competed at this year, the thermometers have been generally bad

I guess it depends on your ideal for a "bad thermometer". I know I ended up using a thermometer like this https://www.amazon.com/GDEALER-DT09-Wa ... 443&sr=8-9, or a TiNspire temperature probe, and for confirmation used a secondary thermometer like an infrared thermometer, or this one machine that tests temperature to the thousandth. To be fair I guess you can take in a 500$ Welch Allyn thermometers, that they use in hospitals; and I'm sure there's some elite teams out there that use them. (https://www.amazon.com/Welch-Allyn-0169 ... B002MBHQI4)

waterlubber wrote: ↑March 19th, 2020, 6:58 pm

MTV<=>Operator wrote: ↑March 19th, 2020, 6:20 pm

shrewdPanther46 wrote: ↑March 19th, 2020, 6:07 pm
We were talking about this on this discord... the problem is that getting good thermometers is pay to win for regionals rules and for states/nats, supervisors may not have access to super reliable thermometers. At every competition I've competed at this year, the thermometers have been generally bad

I don't really see how it's pay to win if you just need your device to be accurate to your thermometer. I used a $25 thermometer the whole year and I never had problems with fluctuation as long as I stirred the water a bit beforehand.

I've tried a few different thermometers ranging from hazard fraught (Cen-Tech) to fancy flinn stuff at an invitational that ran with states rules. In my experience, the more expensive thermometers are usually more *accurate*, but not always more repeatable -- often the cheap ones are just as repeatable, and that's the only thing that matters if you're calibrating the error out anyway. The worst thing I see is when it's internally rounding and the display is false precision, so you get it jumping up by half a degree each time.

Generally, though, unless you go into the RTD range of expensive, it's easy to build a better detector than calibration thermometer. And that's where the trouble comes in: you're not limited by your own device, you're limited by calibration thermometer.

My suggestion would be to switch the event from temperature to something else entirely (maybe 21-22). "Wavelength" or "color" detector would be a really, really, good choice. Generating a specific monochromatic wavelength of light is easy and cheap (via an LED, or a laser, or a filter...) and extremely repeatable. An ES's setup will likely behave exactly the same all day. Building a color detector is much more difficult, and requires some innovation if you ban ICs that do that already (would you use multiple sensors, sensitive to different wavelengths, like our eyes? or a prism? who knows!). Accurately nailing down the exact wavelength of light would require quite a lot of innovation and calibration, and still be cheap (photoresistors and colored film are both cheap, relatively speaking, as are [plastic] prisms.)

I find that this "color detector" would actually be a much more interesting topic then what they currently have, but I feel less teams will be able to nail it down, like many with the temperature sensor.

MoMoney$$$;)0) wrote: ↑March 19th, 2020, 7:03 pm

waterlubber wrote: ↑March 19th, 2020, 6:58 pm

MTV<=>Operator wrote: ↑March 19th, 2020, 6:20 pm

I don't really see how it's pay to win if you just need your device to be accurate to your thermometer. I used a $25 thermometer the whole year and I never had problems with fluctuation as long as I stirred the water a bit beforehand.

I've tried a few different thermometers ranging from hazard fraught (Cen-Tech) to fancy flinn stuff at an invitational that ran with states rules. In my experience, the more expensive thermometers are usually more *accurate*, but not always more repeatable -- often the cheap ones are just as repeatable, and that's the only thing that matters if you're calibrating the error out anyway. The worst thing I see is when it's internally rounding and the display is false precision, so you get it jumping up by half a degree each time.

Generally, though, unless you go into the RTD range of expensive, it's easy to build a better detector than calibration thermometer. And that's where the trouble comes in: you're not limited by your own device, you're limited by calibration thermometer.

My suggestion would be to switch the event from temperature to something else entirely (maybe 21-22). "Wavelength" or "color" detector would be a really, really, good choice. Generating a specific monochromatic wavelength of light is easy and cheap (via an LED, or a laser, or a filter...) and extremely repeatable. An ES's setup will likely behave exactly the same all day. Building a color detector is much more difficult, and requires some innovation if you ban ICs that do that already (would you use multiple sensors, sensitive to different wavelengths, like our eyes? or a prism? who knows!). Accurately nailing down the exact wavelength of light would require quite a lot of innovation and calibration, and still be cheap (photoresistors and colored film are both cheap, relatively speaking, as are [plastic] prisms.)

I find that this "color detector" would actually be a much more interesting topic then what they currently have, but I feel less teams will be able to nail it down, like many with the temperature sensor.

That's literally the point lol. The build is too easy right now to the point that distributions are insanely skewed. Now that the Golden Gate tests are public, you can see the score distributions, particularly the detector one.

MoMoney$$$;)0) wrote: ↑March 19th, 2020, 7:03 pm I find that this "color detector" would actually be a much more interesting topic then what they currently have, but I feel less teams will be able to nail it down, like many with the temperature sensor.

A spectrophotometer based on a diffraction or reflection grating is quite buildable. So is a color sensor with RGB filters in front of some kind of photosensor. I think the biggest challenge with "color detector" as a topic is whether the rules are written clearly enough to define the environment that is expected.

Are you given a monochromatic light source and have to measure the wavelength? Can the source contain multiple wavelengths? Do you have to measure the spectrum? Perhaps you have to detect "color" based on some model of how the human eye works? Do you have to provide a source of illumination for a colored reflective surface? Anyone for optical metamerism?

knightmoves wrote: ↑March 20th, 2020, 9:32 am

MoMoney$$$;)0) wrote: ↑March 19th, 2020, 7:03 pm I find that this "color detector" would actually be a much more interesting topic then what they currently have, but I feel less teams will be able to nail it down, like many with the temperature sensor.

A spectrophotometer based on a diffraction or reflection grating is quite buildable. So is a color sensor with RGB filters in front of some kind of photosensor. I think the biggest challenge with "color detector" as a topic is whether the rules are written clearly enough to define the environment that is expected.

Are you given a monochromatic light source and have to measure the wavelength? Can the source contain multiple wavelengths? Do you have to measure the spectrum? Perhaps you have to detect "color" based on some model of how the human eye works? Do you have to provide a source of illumination for a colored reflective surface? Anyone for optical metamerism?

This is why I really like the idea of color -- it's not that hard or expensive to build a sensor. You could use prisms, diffraction grating, colored filters made from old T-shirts, whatever.

I have the time, because of quarantine, so here's what the event was in my mind:

Teams would construct a device to measure the wavelength and color of a monochromatic light source. The device must not include any prefabricated or commercially available color sensors, not may it include any sensors that produce an output unique to the color of input light. Photoresistors, phototransistors, and photovoltaics are all examples of acceptable sensors, provided they are not sold equipped to detect color or wavelength of light. Microcontrollers and other non-sensing electronics are all permitted, provided the device does not have any wireless communication hardware. The sensor should fit with [defined size] and have an entrance aperture for the light source to be measured. Sensors without an enclosure do not need an entrance aperture. The device should display the wavelength of the light source in nm.

The Event Supervisor will provide one LED or Class-I laser inside a darkened box with a removable lid. Participants will place their sensor inside/alongside (?) the box and may orient the light source to hit their sensor. The box will be sealed and the light source activated, and participants have [x] minutes to give the wavelength, in nm, to the event supervisor.


That's a rough draft, and it's not really meant to be a complete set of rules, but it should give an idea of what I had in mind. You could do multiple light stations (like in detector) as well
Problems I might see:
- LEDs drifting in wavelength significantly
- Non-monochromatic light sources (I know LED's aren't truly monochromatic, but they're close enough for this, probably)
- LEDs coming in standard wavelengths and teams clamping to the nearest one

For an added challenge, (eg at nats), you could have two different monochromatic light sources and ask the teams to return two wavelengths. This would kill the filter approach and force people to use diffraction stuff, though.
Either way, it would be MUCH better than the temperature based thing since the event supervisor's job (provide an LED in a box) is much easier

waterlubber wrote: ↑March 27th, 2020, 7:13 am For an added challenge, (eg at nats), you could have two different monochromatic light sources and ask the teams to return two wavelengths. This would kill the filter approach and force people to use diffraction stuff, though.

If you know that you have no more than two monochromatic light sources, you can still do it with multiple filters (it just takes more data analysis).