For reference, I used to load crude in the bf at about 60°c and set the mantle at 65°c and it would rise to 90-120°c. Probe in flask, no vac. I’m hoping to alleviate that issue.
Non-Decarbed oleo?
Drop a stir bar in, and load crude. Grease all joints and apply vac. It only ever seemed acceptable between 150-180°c, and even then usually overshoot by ~10°c with 2° increments.
Did that machine ship with an extra RTD?
Decarbed under vac, Etoh crude cleaned to the moon and back. Evap temp loss should be nothing.
The quality of the starting material is completely a irrelevant to the performance of the machine.
That’s what I figured as well. In the 20ish runs on this mantle it drastically overshoots every time. This is the first time attempting ai’s auto calibration. I wont have anything for this machine to run for a week. I’ll post results as I get them.
Which would be why I used (and stated) decarbed crude for my calibration.
It certainly made a huge improvement on the mantle I’m working with.
There may be better points to tune, but until the unit pulls anything other than decarb duty, having tuned it where it’s working has worked remarkably well.
1 Like
I would disagree with this.
yeah, the PT100 RTD are a little slower than thermocouples, but not slow enough that they are contributing to over-shoot (on our distillation controllers) in any meaningful manner.
If you don’t train a PID on your process, it will never do what you want it to. I’d put “user error” higher on the list of possibilities than “slow response from thermosensor”.
Also worth noting that thermocouples require regular recalibration. They can drift upwards of 10 degrees over the course of a couple months. RTDs are the solution for most temperature sensing requirements unless you have a metrology department IMO
1 Like
True dat!
meant to post this link on the subject.
it doesn’t mention how much thermocouples can drift, but does mention that the platinum standard remains stable (which is why you would want your mantle to use a PT100).
1 Like
It’s probably not listed because it varies quite a bit by type. If I’m not mixing it up, type T which we use for VLT applications are the worst (could be talking out my ass if I’m honest).
It’s also probably worth noting that the delay in adjustment with an RTD usually isn’t an issue with a PID control scheme because the derivative “predicts” the change, preventing an overshoot even if the sensor hasn’t caught up.
I have been forced to use thermocouples for PIDs before due to response time but that was a cascade PID scheme operating with a 1 second integral period which is very fast