lionfish0
commented
3 weeks ago Oven Experiment
I cut the 4th connector pin between PCBs:
then put both this flash and a new, unmodified flash, into the oven set to between 60C and 70C, for 15 minutes.
Both flashes were able to fire after 15 minutes. It seemed they both struggled to recharge the capacitors - I'm guessing this might be because of the batteries struggling in the temperature, but as far as I could tell, neither stopped due to the temperature.
Fast Firing Test
In a separate experiment (at room temperature 20C). I simply fired them both quickly (80 times in less than a minute). I found both the modified and unmodified flash could fire at 1/16 power 80 times before they both turned off their power lights. I found I could 'reset' them by simply toggling off/on the power supply -- this would let me take another 80 flashes -- I could reset it again, and take another 80. (similarly at 1/8th power, 40 times).
Summary
I wonder if the thermistor actually is used. It seems like my original suspicion, that the 'thermal protect' is actually programmatic counting of flashes. In a previous version I added a relay to temporarily disconnect the power to the flashes to allow them to reset! (Of use to the field-deployments for learning-flights) I wonder if we should incorporate this (we could add a board between the pi and the PCB to breakout a GPIO pin we could send to some relays? Then every 20 flashes it cuts power for 2 seconds?
What to do next?
I've not figured out why the flashes didn't fire at the top of the greenhouse. Possibly the batteries were also struggling in the heat? For the summer the boxes work fine -- but the relay reset might be nice as we could then take longer learning flights, and also have a greater chance of catching the RFs. But this might be an option we can add if we have time, rather than an urgent task.
@FloofyDwagon / @jlwoodgate thoughts?
The situation: It is hot up in the roof of the greenhouse – at least 122 degrees F (50 degrees C). (also the flashes overheat in the field).
The problem:
The solutions: Switch to ABS plastic. Glass Transition Temperatures PLA 55-60 PETG 80-85 ABS 100-110
Considering each component: 1) Boxes etc: I'm not worried about the melting boxes -- even though that did sound very alarming -- a different plastic should solve it. 2) The raspberry pi: There's no official limit given, but informal discussions online seem to suggest it will run ok. 80-85C is a typical limit on a CPU. 3) Power supplies and regulators: I've not noticed them getting warm and 50C should be ok. 4) The camera: The datasheet specifies a working temperature: 0 °C to 50 °C (32 °F to 122 °F). So we are at the upper edge of that. If it gets a bit toasty inside the box we might think that some active cooling (a fan) might be a good idea. 5) The batteries: Again - they don't seem to feel warm after use; 50C is probably ok -- most NiMH are quoted at with operating temperatures of -30C to +85C. 6) The flashes...
I guess switching to another model of flash is an idea -- although I suspect if this is a problem for this flash it is likely to be a problem with other flashes.
The Neewer TT560 Speedlite I took apart the Neewer flash to see how it did the thermal protection.
The thermal fuse
A thermistor An unlabelled SMD component was between the two big transistors also inside the thermal epoxy stuff.
I did a few experiments to see what it is. It seems to be a Negative-temperature-coefficient (NTC) thermistor with a resistance of about 12K at about 19C. So this seems a likely candidate for being responsible for stopping the system overheating.
One leg was to ground, the other went to the 4th pin of the interface to the other PCB. There is formed part of a potential divider with R1. Then to pin 3 on the IC, I think.
My guess is that this is probably being used by the chip to decide if there's overheating taking place. The problem is 50C+ appears to be above that threshold before the flash has even fired.
The options are to: 1) Use alternative flashes [bit awkward as I just bough ~30-40!] Also need testing 2) Remove the thermal shutdown protection somehow. 3) I think adding a fan to the box might be a good idea eitherway.
I guess adding active cooling (a fan) to the box design would be good. I also wonder about the actual flash design - those transistors that presumably get quite hot could really do with a bit of space?: one could remove the battery holder completely (and this gives us then access to the two battery wires that we currently connect to with 'fake batteries') and that void could then be ventilated better. I'm wanting to avoid having to do time-consuming things to the flashes though (as there are going to be ~100 of them). Hence why alternative flashes would be good!