Need method to smoothly warm the system

[ahrendsen]

With the current setup, the rubidium will be heated to much too high of temperatures as the heat slowly dissipates to the location of the thermocouple. There are a number of solutions to this problem.

1. Modify the temperature controller box to have two variac inputs. This will allow me to hold the reservoir variac at a lower voltage which will not be too hot.
2. Play with the PID settings to limit how hot the reservoir can get from the TC controller.
3. Related to the previous point, there's actually a serious issue with the operation of the PID controller. It is set up to allow PROPORTIONAL control, but as we have it configured, it cannot deliver a proportional signal. We're essentially relying on the ID portion of the controller to modulate our temperature. I should have a conversation with @tranthamkw to see what device I should be using rather than the solid state relay to provide proportional control with the temperature controller.
4. Use the Ramp/Soak function of the controller to provide a way to slowly raise the temperature over a period of a few hours.

[ahrendsen]

I will proceed with both option 2 and 4, finding appropriate PID settings for different set points and buffer gas pressures, then implement a ramp/soak procedure which will utilize these different PID settings to efficiently and gradually reach our desired temperatures.

[ahrendsen]

Default PID settings for Omega 7800 controller:

SV=50 deg C
P=47.6
i=260
D=41
i_off=0

PID Settings for collision cell at 150 mTorr:

SV=27->50 deg C

P=18.1
i=432
D=108
i_off=14.8

SV=50->100 deg C

P=19.2
i= 325
D=81
i_off=52

SV=50-> 60 deg C

P=17.3
i= 390
D= 97
i_off= 17.7

[ahrendsen]

I mistakenly tried to go to 50 deg C using the SV=100 deg C training. It overshot the value by 10 deg C! Not ideal at all. I need to try again with the 50 deg C setvalue training, but I can't imagine that it will have a significant improvement.

Took about 10 minutes to go from 27 deg C to 59 deg C. These are important benchmarks for developing my ramp/soak function.

[ahrendsen]

Starting another warming from 27 deg to 50 deg with the SV=50 deg C settings. Started at 10:30 am. After that I'll do auto tune of the PID parameters from 50 to 60 deg C and see if that comes up with different values.

10:45 - Maxed out at 51.8 deg, so a marked improvement from the other set of PID values.

Auto tune from 50 to 60 produced mostly similar values, see updated PID settings post.

Trying out a ramp from 50 to 60 deg C over 2 hours. It reached and went over the setpoint in 30 minutes. I think that this indicates that my PID settings are far too high.

[ahrendsen]

Point 3 in the initial statement was false. I misunderstood how the controller works. Now I understand that pulse width modulation is occurring by changing the length of time of which the 14 V output is applied over the course of 4 s. The length of this duration can be configured to something other than 4 s, but that seems to have worked fine so far.

To solve this problem, I have created a new solution other than the 4 presented. I wrote a C program to check the temperature and set it to a value that is just 5 degrees higher than the previous setting. I then setup a crontab to run this program every 15 minutes, and configure it to run for enough time to reach the desired temperature. In this way, I avoid overheating the reservoir as we are getting to our desired temperature, and I don't have to worry about programming the ramp soak function. I could do the same incremental steps in ramp/soak, but it would involve a lot of steps and take a long time to program, my method is (I think) straightforward to understand and easy to modify to reach a different temperature.

[ahrendsen]

Still have a slight issue in that the temperature is typically overshot with the current PID settings. The distance of the TC from the heating source is surely the cause of this, and I haven't thought of a good way to remedy this problem. We could try to lessen the overshoot by tuning PID parameters, but I don't think it's worth the time.

Changed the program to run every 10 minutes for faster response times.

How much time to warmup formula:
T = period of repeated runs of controlled warmup program (10 min)
dT = temperature step of controlled warmup program (5 degC)
dH = temperature difference between starting point and desired ending point (130 deg C)

n = dH / dT (number of cycles required to reach temperature =26 cycles)

t = n * T (total time to reach temperature = 260 minutes = 4 hours 20 min.)

[ahrendsen]

I modified the position of the thermocouple (TC) to be closer to the heaters. This is nice because it prevents the heaters from glowing as they are straining to reach a temperature. Unfortunately, I think it's now too close because the temperature on the exterior of the reservoir so near to the heaters is not a good reflection of the interior temperature of the reservoir. This results in very slow times to reach a desired density.

The controller is holding the exterior of the reservoir at a given temperature, and then we rely on the heat conductive properties of the metal to transfer that warmth to the inside. Our wait time is now effectively the time for the heat to reach an equilibrium temperature on the interior when the outside is held fixed at a temp.