EOS 1000D extreme modding

[yesyes]

What sort of temperature sensors are you using, Chris?

The plan is to use 2 DS1820, one each for sensor temperature (on the cold finger as close as possible to the CMOS) and one measuring the temperature of the heat sink on the hot side. Then a DHT11 or DHT22 for ambient temperature and humidity (to calculate the dew point).

This is actually related to a question I was asking myself today...

Does it make sense to calculate the dew point if ambient is below zero degrees? If not, I can use a DHT11 which I already have but it doesn't do below zero degrees. Otherwise I will need to use a DHT22. Or use the DHT11 for humidity only and another DS1820 for ambient temperature.

[yesyes]

Home / DSLR cooling modification | Gallery - synergous.com

Unless you intend installing a dew heater on the front of the low pass filter there doesn't seem much point to the reconstruction of the camera. The cold finger is easily formed to exit the camera body without the removal of any parts - as in my mod linked above and here - Canon 1000D/XS/Kiss F DSLR cooling modification - images and overview FlatPress

Cheers

Rowland.

THAT is you? ;-)

That page is my main reference for the mod. :-)

At first I planned this way. But then I wanted to automate it with an Arduino and add some displays. This would just not fit into the original housing....

[rcheshire]

That's me - incognito, no longer.

At first I planned this way. But then I wanted to automate it with an Arduino and add some displays. This would just not fit into the original housing....

Ahah! I see. In that case yes the housing needs to be sizeable.

I am not happy with the prototype cold finger. It really needs to be fixed with 2mm nylon screws both sides of the sensor - or 2.5mm if you tap the holes in the sensor frame and do away with the nuts. This is a more stable arrangement. You will also need a layer of clear packing tape on the upside of the cold finger to insulate from the back of the sensor PCB, while making for a tight fit. The sides of the finger need to be insulated from the sensor array metal connectors.

I used blutak to pressed onto the top of the sensor electronics board to exclude condensation.

During testing, I've had the temperature down to -13C for 15 minutes with no side effects, except ice on the face of the low pass filter. If the camera sensor assembly is not yet sealed leave it in a sealable plastic lunch bag - GladBag - with a desicant for a few days before. That way you wont get condensation inside the sensor assembly.

I intend fitting a dew heater to the low pass filter at some stage to allow imaging below dew point. And it's a good idea to fit it while you are doing the mod and have the sensor out of the camera. Wish I'd given it more thought at the time.

Not sure what form it will take. There's not much room to play with.

[rcheshire]

During testing, I've had the temperature down to -13C for 15 minutes with no side effects, except ice on the face of the low pass filter. If the camera sensor assembly is not yet sealed leave it in a sealable plastic lunch bag - GladBag - with a desicant for a few days before. That way you wont get condensation inside the sensor assembly.

I mean before sealing the entire sensor assembly with silicon, if you intend to do that.

Attached is a circuit I've just designed to drive the low pass filter dew heater. Ignore the 9kOhm resistor value, it's only intended to make the green dot current flow visible (electron flow, opposite to conventional flow). 9 Ohm and nominally 75mm of Nichrome wire, R value 250mOhms, is the heater side drawing up to 1 amp with 12v and 100% PWM duty cycle. The rest is identical to the Peltier temperature control circuit.

That capacitor should be 22uf BTW.

[yesyes]

I spend all afternoon carefully disassembing the camera. I now have all the bits separated that need separating. I have also removed the mirror. It doesn't seem to have any feedback switches...

A WORD OF WARNING !!

If someone does disassemble the camera as far as taking the front cover off, BE CAREFUL WITH THE POWER BOARD at the front. Can't shout that loud enough...

There is a huge capacitor for the flash. It still had a lot of charge in it, some of it discharged through my body!!! Left hand on the metal frame, other hand fiddling with all-metal screwdriver and accidentally touched the + terminal of that capacitor. I dropped the camera (luckily only a few centimetres) and the screwdriver ended up on the other side of the shed. I have now unsoldered that capacitor as I don't need the flash any longer...

[Gina]

Ah yes, that caught me a bit but nothing like you. There are warnings in some of the modding threads and articles but it can still catch you out. I just got a shock and swore at it - didn't drop anything.

I've been looking further into cooling mine and worked out a much better way of doing it that I had before. Going to make a new cold finger to replace the current one.

I've been running extensive noise tests without cooling and the EXIF readout has been showing 36C with an ambient of about 22C. So I'm looking to cool my sensor by 20-30C which will give a very worthwhile reduction in noise. I'm going back to the 12v 50W Peltier TEC I used with the test box. This is 40mm square and should be OK. I'm considering using the filter wheel I'm building as a heatsink with extra cooling if needed.

[yesyes]

Well, I survived...

I just hope I didn't damage any of the electronics on the power supply board. Anything else was disconnected from the PSU board already, luckily.

I have finished drawing up the schematics for the cooling controller. See attached. Any comments welcome...

EOS Cooling Controller - Schematic.pdf

[Gina]

Well done Chris

[rfdesigner]

Well, I survived...

I just hope I didn't damage any of the electronics on the power supply board. Anything else was disconnected from the PSU board already, luckily.

I have finished drawing up the schematics for the cooling controller. See attached. Any comments welcome...

Primary comment:

I don't think the DS1820 chips are quite up to the mark. In my opinion you need 0.1C resolution in your temperature measurement and these chips are only 0.5C resolution. (I don't think you need 0.1C accuracy, just resolution), a 0.1C difference between dark frames and light frames is a 0.8% difference in dark current, a 0.5C difference is a 5% difference in dark current.

You may want to look at:

http://datasheets.maxim-ic.com/en/ds/DS18B20-PAR.pdf

which is a 12bit version of the same chip, giving better resolution. (as I said I don't think you need accuracy, just consistant performance.. who cares if you always get -21.2C instead of -20C as long as you always get -21.2C)

I'd also want to think long and hard about supply routing. I'd go for a dedicated ground plane and star +ve supplies so all supply voltage drops occur on the positive side and stop strage ground spikes from giving you grief.

Derek

[Gina]

It's the better chips that I'm using in my 1-wire weather station. They were the same price when I bought them so I thought I might as well get the better ones

[ncjunk]

I spend all afternoon carefully disassembing the camera. I now have all the bits separated that need separating. I have also removed the mirror. It doesn't seem to have any feedback switches...

A WORD OF WARNING !!

If someone does disassemble the camera as far as taking the front cover off, BE CAREFUL WITH THE POWER BOARD at the front. Can't shout that loud enough...

There is a huge capacitor for the flash. It still had a lot of charge in it, some of it discharged through my body!!! Left hand on the metal frame, other hand fiddling with all-metal screwdriver and accidentally touched the + terminal of that capacitor. I dropped the camera (luckily only a few centimetres) and the screwdriver ended up on the other side of the shed. I have now unsoldered that capacitor as I don't need the flash any longer...

An old work friend told me how they used to charge up large capacitors and then throw them to apprentices who couldn´t help but automatically catch them....he also used to wire the metal mens toilet up and give people a shock when they had a pee. Apparently they got him back by betting he couldn´t do pull ups on a pipe and then removing the flooring beneath him leaving a two story drop.

Oh how health and safety has ruined all these "Jolly Japes"! (thank god)

very good thread...hope it all works out ok

[yesyes]

Been working on the cold finger mostly today.

For a better weight balance I want to put the heatsink / fan behind the sensor. So the cold finger is U shaped. This will probably be rather fiddly to put everything back together. I think it needs a few more holes that will give access to a few screws...

Another thing came up when I test-fitted the heatsink.

How am I supposed to mount the heatsink to the peltier and cold finger. If I used metal screws, wouldn't that cause a thermal feedback allowing heat from the hot side back to the cold side via the screws?

[yesyes]

Primary comment:

I don't think the DS1820 chips are quite up to the mark. In my opinion you need 0.1C resolution in your temperature measurement and these chips are only 0.5C resolution. (I don't think you need 0.1C accuracy, just resolution), a 0.1C difference between dark frames and light frames is a 0.8% difference in dark current, a 0.5C difference is a 5% difference in dark current.

You may want to look at:

http://datasheets.maxim-ic.com/en/ds/DS18B20-PAR.pdf

which is a 12bit version of the same chip, giving better resolution. (as I said I don't think you need accuracy, just consistant performance.. who cares if you always get -21.2C instead of -20C as long as you always get -21.2C)

I'd also want to think long and hard about supply routing. I'd go for a dedicated ground plane and star +ve supplies so all supply voltage drops occur on the positive side and stop strage ground spikes from giving you grief.

Derek

Thanks for that. I already had the DS1820 but now ordered 3 DS18B20. I never considered repeatability of sensor temperatures. I just though about cooling the sensor as much as possible. But it makes sense as I would need a dark frame library only for a few temperatures.

[Gina]

Been working on the cold finger mostly today.

For a better weight balance I want to put the heatsink / fan behind the sensor. So the cold finger is U shaped. This will probably be rather fiddly to put everything back together. I think it needs a few more holes that will give access to a few screws...

My first cold finger design was U shaped and I was going to fit the cooler and heatsink on the back of the camera in place of the display but I decided the display would be useful (eg. camera won't work unless the date/time is set) so cold finger v2 comes out the side of the camera.

Another thing came up when I test-fitted the heatsink.

How am I supposed to mount the heatsink to the peltier and cold finger. If I used metal screws, wouldn't that cause a thermal feedback allowing heat from the hot side back to the cold side via the screws?

Use nylon screws. Or metal screws in grommets.

[rcheshire]

Primary comment:

I don't think the DS1820 chips are quite up to the mark. In my opinion you need 0.1C resolution in your temperature measurement and these chips are only 0.5C resolution. (I don't think you need 0.1C accuracy, just resolution), a 0.1C difference between dark frames and light frames is a 0.8% difference in dark current, a 0.5C difference is a 5% difference in dark current.

You may want to look at:

http://datasheets.maxim-ic.com/en/ds/DS18B20-PAR.pdf

which is a 12bit version of the same chip, giving better resolution. (as I said I don't think you need accuracy, just consistant performance.. who cares if you always get -21.2C instead of -20C as long as you always get -21.2C)

I'd also want to think long and hard about supply routing. I'd go for a dedicated ground plane and star +ve supplies so all supply voltage drops occur on the positive side and stop strage ground spikes from giving you grief.

Agree with all of that, except that I don't think the resolution of the temp sensors is all that important. I've found that despite read fluctuations, the system doesn't respond quickly to PWM commands. Thermal inertia tends to override wild fluctuations quite well. SBIG make similar statements about their cooling systems, which they say are stable to 0.1C.

When I shut my system down, it takes about 12 seconds for the cold finger temperature to exceed +1C from set point.

That's a nice looking job on the cold finger and much better for balance if you don't need live view. I would have been inclined to wrap it around further and fix it with 2mm screws to the other side of the sensor frame for stability.

That's a small heat sink and I'm interested to see how it performs, because I would like to use a smaller one myself. Heat produced by the Peltier and the energy required to drive the Peltier need to be dissipated effectively, otherwise cooling is degraded.

It would be nice to keep all the electronics camera side, nice and compact, to avoid adding cables.

[rcheshire]

duplicate of previous:icon_scratch: removed text - how do you delete posts

[yesyes]

Agree with all of that, except that I don't think the resolution of the temp sensors is all that important. I've found that despite read fluctuations, the system doesn't respond quickly to PWM commands. Thermal inertia tends to override wild fluctuations quite well. SBIG make similar statements about their cooling systems, which they say are stable to 0.1C.

When I shut my system down, it takes about 12 seconds for the cold finger temperature to exceed +1C from set point.

That's a nice looking job on the cold finger and much better for balance if you don't need live view. I would have been inclined to wrap it around further and fix it with 2mm screws to the other side of the sensor frame for stability.

That's a small heat sink and I'm interested to see how it performs, because I would like to use a smaller one myself. Heat produced by the Peltier and the energy required to drive the Peltier need to be dissipated effectively, otherwise cooling is degraded.

It would be nice to keep all the electronics camera side, nice and compact, to avoid adding cables.

I've now already ordered the higher resolution sensors. So will go with these. Can't be wrong to have better resolution. :-)

Yes, I also think the heatsink is a bit small. I'll see how it goes. This heatsink was on a Pentium 4 CPU on a very small form factor computer (cPCI) without a (direct) fan. I'm hoping it will be good enough. I do have another heatsink with mounting holes at the same spacing that is quite a bit bigger (higher) and solid copper with lots of fins, but also a lot heavier.

Seeing that Gina is considering water cooling I have a feeling that I will have to use that bigger heatsink.

More parts have just arrived

[rfdesigner]

Are you planning to use a fan on the heatsink?

Derek

[yesyes]

Yes, the heatsink will have a fan. The fan will also be controlled from the Arduino depending on heat sink temperature.

I now have a 3rd heat sink option, a much bigger all-copper heat sink from an old Pentium 4 machine that has a plastic retainer frame (ideal to prevent thermal feedback from the hot side back to the cold side). But this one is much heavier...

[rfdesigner]

Good to see you're on top of this. Heat leakage is likely to be a major issue. Certainly a fan will help reduce this by reducing the necessary temperature differential for a given amount of heat dissipation which will help the peliter pumps, and a heatsink with plastic retainer sounds like it might be useful.

How do you plan to prevent the sensor from icing up?

Derek

[yesyes]

I think I'll start testing with the smallest and lightest heatsink and see how hot that gets. If it gets too hot, even with a fan, then I'll try the bigger ones. Still waiting for my peltier to arrive though...

Not sure about the icing up yet. I read in some other thread that this wasn't an issue, only the front LP filter was freezing up occasionally (IIRC). I guess it depends on how well I will be able to seal the new casing to keep moisture out.

Also, the cooling controller will have an option not to cool below the dew point. I think I'll just have to see how that goes....

[Gina]

When I was testing my cold finger cooling I had an EXIF T of 2-3C and had no condensation of the filter or sensor and got clear photos but I was testing indoors so the RH was probably pretty low. I used an 80mm CPU cooler+fan with large thin fins which weighed 300g - a very effective cooler but heavy (which is why I'm looking at water cooling). A smaller, 60mm cooler+fan was insufficient and got too hot to touch while also reducing the chip cooling with EXIF T of 20C (I think). That weighed about 100g.

[rcheshire]

Peltier performance is a function of the hot to cold side ratio, which needs to be as low as practicable. Another consideration is fan failure.My system is automated and I leave it unattended. If the fan does fail there is plenty of mass to prevent heat damage to the Peltier. Perhaps detection of fan failure is a sensible option?

[yesyes]

Sounds like I will have to use that big heatsink then.

Regarding fan failure, I will constantly measure hot side temperature. If that rises too much I will just switch off the peltier. Better let the CMOS sensor return to its normal temperature than risk damaging the peltier or worse....

The fans I ordered are 3-pin fans. So theoretically I could also measure the actual fan speed using an interrupt pin on the Arduino. But several thousand RPMs might slow down the Arduino a bit too much with all the interrupt requests...

[Gina]

I think monitoring the temperatures is the easiest way - that's what I shall be doing.