Anyone have an estimated emissivity for the IVAC 9000 radiators?

[globoy]

This forum is really fantastic.  I am following in Frasier's 2018 footsteps to try to make some DIY blackbody radiators for learning and then to see if I can characterize and possibly improve the accuracy of a camera I have designed.

I got an IVAC 9000 from ebay and have been playing with it.  It seems to be operable.  It warms up, displays the "Check Ready" status and seems to hold stable temperatures.  I put my Agilent's K-type thermocouple into a screw hole next to the device's temperature sensor and my Lepton 3.5-based camera so the radiator fill's the camera's field of view.  This lead to some questions I am hoping someone here has experience with.

1. The device's instructions say an in-ear temperature probe being calibrated should read 26.0°C (+/- 0.3) for the low side and 38.0°C (+/- 0.1) for the high side.  My DMM shows 24.3°C for the low side and 36.5°C for the high side.  I realize that my meter has an error band and I may not be making ideal contact with its temperature sensor, but should I expect that the metal radiators are 26 and 38°C if the unit is working properly?  Or is it possible that the IVAC thermometer's somehow adjust what they read to really reflect the in-ear temperature.

2. What estimate should I use for the emissivity of the coating on the radiators?  Could I re-coat them with a paint for which I know the emissivity?

Below are images of my setup and the output from the Lepton peering at the low temperature radiator.

Thanks.

[_Wim_]

According to this paper (https://www.tandfonline.com/doi/abs/10.1080/19315775.2014.11721683) the emissivity for such a calibrator was 0.972. The emissivity of the ear canal is apparently 0.9988 according to a reference in this paper.

Tip: use sci hub to access the full paper if desired. 

[globoy]

Thanks for that information! 

[bap2703]

1. The device's instructions say an in-ear temperature probe being calibrated should read 26.0°C (+/- 0.3) for the low side and 38.0°C (+/- 0.1) for the high side.  My DMM shows 24.3°C for the low side and 36.5°C for the high side.  I realize that my meter has an error band and I may not be making ideal contact with its temperature sensor, but should I expect that the metal radiators are 26 and 38°C if the unit is working properly?  Or is it possible that the IVAC thermometer's somehow adjust what they read to really reflect the in-ear temperature.

The ear conduct should be quite close to a blackbody : made of water + it's a cavity.
It means that when you measure the temperature in the ear, you expect a blackbody. So to calibrate an ear thermometer you do it against a blackbody simulator.
Conclusion: your device should be at 26°C and 38°C and the emissivity pretty close to one.
Error most likely will come from the temperature measurement/regulation electronics.

Maybe you can rent a calibrated infrared thermometer to have another measurement and conclude which of your thermocouple measurement or your blackbody simulator isn't accurate.

[globoy]

The ear conduct should be quite close to a blackbody : made of water + it's a cavity.
It means that when you measure the temperature in the ear, you expect a blackbody. So to calibrate an ear thermometer you do it against a blackbody simulator.
Conclusion: your device should be at 26°C and 38°C and the emissivity pretty close to one.
Error most likely will come from the temperature measurement/regulation electronics.

Maybe you can rent a calibrated infrared thermometer to have another measurement and conclude which of your thermocouple measurement or your blackbody simulator isn't accurate.

Ok.  Thanks for this too.  I will look for an accurate IR thermometer - probably see if I can borrow a good in-ear thermometer.

[_Wim_]

After seeing your post, I put in a low (15$ ) offer on one of these Ivac Core Check 9000 calibrators and the seller accepted! Today it arrived. Attached a measurement of the plates with my Flir E30 who thinks he is an E60 

Thermal span set to only 1°C. The 26°C plate seems a little higher in my case, but the Flir is from 2012 and has probably not been calibrated since. Emissivity was set to 0.99.

[_Wim_]

@Fraser, in case you would buy such a toy also and it comes without a power supply  , attached my investigation work to power it up.

It uses normally a 5V 5A power supply (CL40 7605, datasheet attached).

I currently used some clips as shown in the picture attached, but for that to work, you need a bridge at the bottom of the board. This had me looking for a bit, as the device powers correctly up without it, but does not heat up...

[Fraser]

Thank you 

I own three IVAC 9000’s

Post 42 onwards of this thread may be of interest 

https://www.eevblog.com/forum/thermal-imaging/black-body-thermal-references-and-fraser_s-new-black-body-project-)/msg1556140/#msg1556140

I discovered the need to link the ‘on’ pin of the power connector in order to get the unit working. I do not know why that activation pin is needed.

Fraser

[_Wim_]

Thank you 

I own three IVAC 9000’s

Post 42 onwards of this thread may be of interest 

https://www.eevblog.com/forum/thermal-imaging/black-body-thermal-references-and-fraser_s-new-black-body-project-)/msg1556140/#msg1556140

I discovered the need to link the ‘on’ pin of the power connector in order to get the unit working. I do not know why that activation pin is needed.

Fraser

I should have known better 

The linking of the pin in needed because the original adapter has two 5V pins in the DIN connector (pin 1 & 2), and both are used on the board but are not linked together on the pcb. So when working with only a single 5V clip like I did, the board powers up but does not heat up. This got me confused initially.

[_Wim_]

So my best guess would be that they allowed to use a different voltage for powering the peltier elements, hence you need to connect "two" supplies to make the unit work (so not PSU ok). One of the 5V supplies draws only around 100mA, the other one draws the full power.

[Fraser]

Interesting, thanks 

Fraser

[Fraser]

It would be great if there was a way to set the two black bodies to more useful temperatures for 2 point calibrations. 20C and 60C would be useful as huge temperature differences are not normally required for basic 2 point calibrations. Sadly I have not had time to delve into the design of these IVAC units and I suspect the temperature is hard coded into their firmware as they are for a specific use. Modifying the temperature sensor feedback loop to the processor or its ADC might work but that would need some investigation into how best to fool the processor. As we are trying to make fixed temperature references it may be possible, and a lot easier than making a variable temperature reference.

Fraser

[globoy]

What do you think about hooking one of the two radiators up to a Wells CTI FTC100d-based controller?  I have one of those although it's a 12V unit currently.

I did look at the circuitry of the IVAC unit.  I guess one would remove the uC and the ADC chip although perhaps the ADC could still be used.  The temp sensor connections to that chip were interesting.  Instead of using the voltage input, they seemed to be coupled to the ohms attenuator terminals.

[Fraser]

Possibly a Pt100 RTD sensor for accuracy ?

Fraser

[_Wim_]

It would be great if there was a way to set the two black bodies to more useful temperatures for 2 point calibrations. 20C and 60C would be useful as huge temperature differences are not normally required for basic 2 point calibrations. Sadly I have not had time to delve into the design of these IVAC units and I suspect the temperature is hard coded into their firmware as they are for a specific use. Modifying the temperature sensor feedback loop to the processor or its ADC might work but that would need some investigation into how best to fool the processor. As we are trying to make fixed temperature references it may be possible, and a lot easier than making a variable temperature reference.

Fraser

My intention is to have a look at something like this, but no promises. So many projects, so little time! Hopefully something fairly easy can be done be bodging in some resistors in at the right places...

[_Wim_]

I had a quick play in the lab (no time to wait for stable or accurate results, just a quick proof on concept only).

What I found so far, it that the temperature sensor is an NTC. At room temperature (+-22°C) it was around 12.5K, at 38°C is dropped to 5K.

This equates to a beta of around 5300 (https://www.ametherm.com/thermistor/ntc-thermistor-beta)

Knowing this beta we can calculate what the resistance would be at for example 60°C => 1.9K (https://www.electro-tech-online.com/tools/thermistor-resistance-calculator.php)

We know the controller tries to adjust for around 5K, so if we put 3.1K in series with the NTC, the controller should adjust to around 60°C. Stability will be less good, because we are more in the flat region of the NTC, so temperature variation create less resistance change.

My first test with 3.1K in series is attached: rise is however very slow, and not sure 60°C will be feasible, but 50°C is maybe doable this way.  Will try some more later this weekend.

[Fraser]

Excellent work 

Fraser

[globoy]

Very cool!

[_Wim_]

Done some more testing yesterday and today.

First test was to put a 12K resistor in series to test how far it would heat-up without any regulation. Ivac was powered by 5Vdc.

Both voltage and average current to the peltier were logged (red trace= current in A, bleu trace=voltage).

As can be seen, it was struggling to achieve 50°C. Around 14:31 I increased the power supply to 5.5Vdc, and got again more rise in temperature. So clearly a little more voltage would be beneficial.

[_Wim_]

For the next test I removed CR1,CR2 and CR4 (TVS diodes for 5V, shorting to ground in case of excess voltage).

This way I could increase power to the H-bridge powering the peltier elements, and keep still a 5V supply to the microcontroller and ADC.

Attached is the pin out of the original power supply.

pin 1 gives power to the peltier elements, pin 2 gives power to the micro and ADC, 3 to 5 should be tied together to ground.

I tried several runs, first with 6.5V, than with 7V and finally with 8V for the peltier elements. Each time heating up went a little faster, and stability (aka "check ready") was also sooner reached.

Remark: the removal of CR2 is normally not necessary, this one protects the 5V for the micro and ADC, I have later reinstalled it but forgot to take a picture of it.

[_Wim_]

Now being able to heat-up easily to 60°C, I installed a 5K 10-turn pot in series with the NTC.

This way I could adjust the pot until exactly 60°C on average was archived (remark: I had to redo it as I forgot to change the emissivity to 0.99).

I used my flir E60 for the calibration, which is probably not ideal, but at least it allow me to evaluate the effect of placing a lens in front of the flir (which cause a +-2°C lower temperature with my ZnSe lenses).

[_Wim_]

As using 2 power supplies to power this is kind of annoying, and the microcontroller and ADC draw only 50mA, I bodged in an 7805 and replaced the DIN connector by a power connector so I could easily connect a 8V 5A switch mode power supply adaptor (because 7V adaptors are not easily found on Ali-express, I opted to further increase the power to 8V).

Disclaimer: I have no idea how much margin there is, do so at you own risk! 

I also put everything back in the housing (the front panel pcb can easily be installed behind the back body radiators), and closed the openings with some matt black tape.

[_Wim_]

Final test:

During the final heat-up test in its housing I again forgot to change the emissivity from 0.97 to 0.99  , so the little jump around 14.17:40 is me adjusting the emissivity)

I am very satisfied with the end result, it remains quite stable (unsure if the small fluctuations are related to my flir or the black body, but a good chance it is the flir).

Final value of the series resistor (10-turn pot) in my case was 3.35K

[_Wim_]

There is still one on Ebay for only 10$: https://www.ebay.com/itm/Ivac-Core-Check-Calibrator-9000A-15663-E42/253802723165?hash=item3b17d2375d:g:McgAAOSw6sdbadM~

An absolute steal! Thanks to Fraser for pointing this out originally, and thanks to globoy for reminding us again of this excellent tip!

[_Wim_]

Possible future improvements:
- install a selector with to select between different series resistors to have multiple temperatures
- bring a smaller version of the lcd to the front to have an indication when the temperature is reached & stable