I figured that fluctuation is bc it’s gas and not liquid, right? I’m assuming this is his stack temp.
An oil filled sensor might be a better suited for gas, to quiet the noise.
Sure thing, photo enclosed. This is a simple pot still, so no large reflux at the top. The thermometer at the top is one of your one-wire thermometers inserted into the vapor flow - it reaches down well into the gas. There is some filler at the top that is flexible to provide a decent seal. The other item is the condenser. At the base of the still is a 1Kw electric heater being controlled.
As an FYI, if I replace the small column with a large reflux tower and run the still uncontrolled to simply monitor the temperature, the temperature measured is quite stable once the still has heated up and runs at about 78.8 degrees plus 0.5. So in reflux mode, the same thermometer with the same condenser runs in a very stable state.
Interesting and thanks for the thought. My old, broken power controller had a thermometer that fitted into the same spot as the one-wire one and was much the same size, but I have no idea what it was made from.
Software filtering and making it slower by having it in oil should have a similar effect. Can you show the tip of the sensor? How does the inside look?
Here it is. I’m not sure how I’d look at the inside as it appears to be a solid sleeve. The white material is plumbers tape
This has taken too many iterations, so I’d like to approach it differently and get it sorted.
- Can we set up a zoom session with screen sharing where I can watch and tweak the system remotely?
- Can you add a second sensor that’s in the liquid?
I still think that I can get better control with a cascaded setup that also controls the liquid temperature.
I would love to see a wizard for this, both for pot and reflux configurations.
It is great seeing how much has been tested on this so far.
I think it is probobly condensate dripping on the probe that makes the temperature measurement fluctuates.
That’s a very good theory! A small collar, like a roof, could prevent that. Is that something you can test David?
Should be possible to mcgyver one, maybe just a washer resting on top of a small o-ring, just routing the drops around the tip.
Another useful test would be to disable the setpoint (and with it the PID) and just set a constant PWM value that gives you a about the same temperature in steady state.
If the sensor still fluctuates, it is not due to heating and the droplet theory becomes more plausible.
Interesting ideas. I’ve run out of the wash I was distilling, but I can easily fill the still with plain water. It will change the goal posts a bit, but I’m happy to crack on if you are?
The main issue I see with adding a diverter onto the probe for the condensate is that I think I’d need to remove the snappy-bit-of-plastic on the top of the pot column so that I can insert the probe, get to the tip and then shove everything back together. I’m happy to give that a try at some point, but I want to make sure first I can buy replacement bit of plastic just in case in all goes wrong…
On my simple still, I cannot see an easy way of getting a second probe into the boiler unless I take some of my brave pills, cut a hole in the boiler and add in one of your wall mounted temperature sensors. I might be able to drill a small second hole into the plastic cap and thread another one-wire thermometer in that way. I don’t have a spare probe at the moment though.
But maybe someone else reading this thread with a more modern still has an existing way of doing this? Other volunteers welcome 
But I’m more than happy to kick off a fresh batch of hot water, disable the setpoint and set a constant PWM. I did try that before, but it was a while back in this thread. I did get a fair temperature fluctuation, but it would be best to start again with controlled conditions and report back.
If I use my Kp of 15 which seems OK, Ti is currently 5mins and Td is 0, so I’m thinking setting the PWM to 35 and see where we get too in pointing the direction towards the droplet theory or not?
If you use a constant PWM value, any PID settings don’t matter, because they will be disabled.
Just set a fixed PWM value and let it stabilize to whatever the equilibrium is. If it has reached equilibrium and the heat input is constant due to the fixed PWM, but you still get a fluctuating temperature measurement, that would be very revealing.
I will reveal all then :-). ith PWM fixed at 35, the temperature seems to have stabilised at around 50 but with a much smaller degree of variation than before, IMHO. Graph is:
Actually the temperature is still continuing to rise slowly, slowly.
This probably shows better here on the PID graph:
The PID graph is not relevant, only the sensor in this case.
50 is not a good test temperature, I think you’ll want to increase the pwm to get to a temperature where you get some vapor.
OK I’ll keep increasing PWM by 10 points at a time and see what we get. Here is the 45%
The 55% one:
and we boil, so I’m doing a 50% one and I’ll edit that in later …
Also try if you turn off the pwm/ heater whether that immediately stops the fluctuations. In that case it might be EMI.
And perhaps compare the filtered and unfiltered value in the setpoint to select a good filter. Good luck and good night, I’m going to sleep.
If I turn the power off, the temperature drops rapidly and linealy, so yes I guess the fluctuations stop, but not in the way I think you mean.
Thanks Elco chat later. If I’m using a manual PWM, is a setpoint active? The filter is set to be 5mins from a previous discussion.