[mpatterson]

Okay thanks, I will try replacing the cables.

[mpatterson]

No, sorry, I mean we have all of the reactors, control computer, and pumps inside of a Binder cooling incubator. I was asking whether the incubator itself would be enough of an electrical noise source to be causing these errors.

[mpatterson]

Okay, we have everything in a cooling incubator so we can run at low temperature. Do you think that would be enough?

[mpatterson]

As an update, I did just reclean all moisture sensors with alcohol and all USB and power cables and ports with compressed air, but I am still getting the errors.

[mpatterson]

Okay, thanks. Yes I usually disable Windows updates, but I thought I had an extra week left before I had to pause them again- it was an oversight on my part.

[mpatterson]

Okay I will try that- it has been awhile since I’ve replaced the tubing. Thank you!

[mpatterson]

Yes, actually, it looks like there is a liquid stain on top of M1. We are using the three-port caps, one for the media, one for waste, and I have a little foil cap on the third port but it had fallen off. That probably indicates some liquid came out with significant force. I’m not sure why that happened, there’s nothing blocking any of the lines as far as I can tell.

[mpatterson]

So sorry for the late reply! I would be happy to send you some data, would it be okay if I shared the file via email?

[mpatterson]

Thanks! I will adjust that during our next experiment.

Halving the dither during our last experiment does seem to help a bit. We are using an OD of 0.5 as the setpoint, and in shake flasks they have grown above 1 at 18 ˚C, but we have not tried collecting a full growth curve to see where they begin to enter stationary phase.

As for the time averaging, while this isn’t exactly what you suggested, creating a moving average of doubling times does produce a smoother line at 25 ˚C and 20 ˚C. However, there is so much variation at 18 ˚C that even the moving average line jumps around quite a bit. The problem seems to be more pronounced (by that I mean I observe large spikes in growth rate) when the OD is increasing.

[mpatterson]

The coefficient of variation at 25 ˚C ranges from 0.1-0.5, but at both 20 ˚C and 18 ˚C it can get as high as 7 or 8. At 18 ˚C the coefficient is always larger than 1.

You are right that the cells grow very slowly- doubling times are on the order of hours. Ideally, standard deviation wouldn’t be greater than 20-30 minutes. We’re planning to do some experiments in the future where we track changes in growth rate over time, so the standard deviation just needs to be low enough for us to track that process in a statistically significant way.

And thanks for those suggestions! We could try putting a stirrer in the incubator with the media as well, but it would be a little tight. How would one alter the temperature control algorithm? In this copied/pasted section of app.py (“Gain = 2.5” in the Thermostat function), or are there additional things I would need to change?

def Thermostat(M,item):
#Function that implements thermostat temperature control using MPC algorithm.
global sysData
global sysItems
global sysDevices
ON=sysData[M][item][‘ON’]
sysDevices[M][item][‘threadCount’]=(sysDevices[M][item][‘threadCount’]+1)%100
currentThread=sysDevices[M][item][‘threadCount’]

if (ON==0):
SetOutputOn(M,’Heat’,0)
return

MeasureTemp(M,’IR’) #Measures temperature – note that this may be happening DURING stirring.

CurrentTemp=sysData[M][‘ThermometerIR’][‘current’]
TargetTemp=sysData[M][‘Thermostat’][‘target’]
LastTemp=sysData[M][‘Thermostat’][‘last’]

#MPC Controller Component
MediaTemp=sysData[M][‘ThermometerExternal’][‘current’]
MPC=0
if (MediaTemp>0.0):
Tdiff=CurrentTemp-MediaTemp
Pumping=sysData[M][‘Pump1’][‘target’]*float(sysData[M][‘Pump1’][‘ON’])*float(sysData[M][‘OD’][‘ON’])
Gain=2.5
MPC=Gain*Tdiff*Pumping

[mpatterson]

Okay, is it possible to still regulate OD that way? If I’m remembering correctly, when you switch on OD regulation, the program automatically assigns P1 to inlet and P2 to outlet.

[mpatterson]

Great, thank you!

[mpatterson]

Could you be a little more specific about which line(s) to change if we wanted to grow between, say, an OD of 0.4 and 0.6? Is it in line 61 when creating the sysdata structure, or later on in the Zigzag function itself?

[mpatterson]

Ah, okay, the difference is certainly not meters. Right now I just have the pumps on the benchtop with both ends of the tubing going into a flask. I’d estimate there’s only about a 5 cm difference between each inlet and outlet when they are set up this way. But the stands the reactors sit on leave the tops ~ 25 cm above the bench, so it is helpful to know that more than 20 cm could be a problem.

And yes the pumps are definitely still turning, and not struggling as drastically as 10% of the default speed. So that is also good to know, thank you!

[mpatterson]

Hi,

Regarding your question about the air- if the pump is not running, I can still see a bubble of air moving through the tubing, in the opposite direction that the liquid is pumped. Sorry for being vague.

Regarding potential fixes,
For the first, to check tubing ends: maybe I am not understanding what you mean, but doesn’t one end of the tubing always need to be higher than the other? If the reactor and pump are sitting above the level of a media reservoir, the inlet for Pump 1 must be near or at the bottom of the media reservoir, but the outlet will go into the reactor sitting above. Likewise, the inlet of Pump 2 will also be attached to the reactor waste line and then outlet into a waste container (which does not need to extend to the bottom of a waste container, this is true).
For the second fix, checking the level of the tubing in the pump head: I have tried both alining the tubing precisely with the groove and pushing it all the way down into the pump head, it does not help. The only way I have been able to fix this is by applying several layers of tape, which is when I notice the pump turning slower than other pump heads that require either much less tape or no tape at all. If this struggling doesn’t risk damaging the pump long-term, that is fine, but I was concerned about there being a risk of damage if the pump needs to struggle through so much resistance every time it runs.

[Author]

Posts