[Dsg-rtpc] [New Comment] MFC104 seems flaky

[Brian Eng]

I didn't bother with labeling those screenshots, but the temp from MFC is in C and outside temp is in F. Also they're not on the same Y-axis since all the PVs in that plot are tiled (the bolded PV in the legend is the one that the axis applies to). The plots were just to show generally the MFC temp follows the outside temperature except when there is gas flowing.

If there was an issue with the communication between the MFC and LabVIEW the cRIO would stop communications and throw an error. It's what drives the initialization count, 3 errors and it gives up trying to communicating and requires resetting the connection. That I'm confident isn't an issue.

Temperature is an easy one since this is a thermal based MFC if the temperature of the unit was wrong the flow reading would be as well. Which is why that temperature reading is even available at all, the factory uses it to calibrate at a known level (I believe it is 40 C). So the flows will be off when the temperature is different from that. I'll check the specs tomorrow regarding what the temperature effect is. In the past we've only cared about the temperature is when the gas itself is temperature sensitive.

Maybe I misunderstood your earlier emails or you put the wrong valve (SVBT 100, 102, 108), but from the ones you listed there's not just a few short pieces of plumbing, there's all the hundreds of the feet between the gas pad and panel in addition to the orifice in the MFC itself, which complicates things. Also, from my previous email I forgot to prefix the valves with MVBT, i.e. the manual valves before and after the MFC.

I'm not saying the MFC isn't the cause, just that none of the signals that are available aren't pointing towards a failure we've seen before.
________________________________________
From: Sebastian Kuhn <kuhn at jlab.org>
Sent: Monday, March 16, 2020 8:24 PM
To: Brian Eng
Cc: dsg-rtpc at jlab.org
Subject: Re: [New Comment] MFC104 seems flaky

Dear Brian,

I totally understand that, given the extraordinary circumstances, you do not want to come in to try to check the MFC yourself. As I said, we can continue as is, although it will cost us a few hours of beam time (if there WILL be any beam time by the time the RTPC switchout is complete ;-).

On the other hand, I do NOT understand why you keep insisting  that
> nothing I've seen in the archived data would lead me to believe anything is wrong with the MFC itself
Yes, there is a long gas line on ONE (outlet) side of the MFC, but only a few short pieces of steel plumbing on the OTHER side. If I close all valves on that side, there should be no way that gas can still flow (FROM this small volume) at 200 sccm for any appreciable time, even if there is a (slight) pressure differential. And I also don’t see how your other ideas for possible culprits (temperature, gas type) can explain why 2 out of 3 fills in immediate succession take a long time, and then the 3rd one only a short time (or sometimes the other way around). In other words, the behavior is NOT consistent from one moment to the next.

I may be misunderstanding something fundamental about the whole setup, but I am pretty sure that either there is some issue with the MFC itself (which, btw, appears for He, H2, D2 and N2 - not just for one specific type of gas), or an issue with communications between the MFC and the Labview program. I’d be happy to be given an explanation that makes sense or instructions on how to FIX it - otherwise we’ll just live with it (there is too much other stuff to do with the BONuS switchover for our group of users to also do some of the tests you suggest).

Best greetings - Sebastian

 P.S.: It’s also not confidence-inspiring that the MFC temperature readback is about 10 or more degrees F lower (even without anything flowing) than the surrounding air.



> On Mar 16, 2020, at 7:38 PM, Brian Eng <beng at jlab.org> wrote:
>
> Hi Sebastian,
>
> A few comments sections below.
>
> When closing only 100, 102 & 108 I'm not surprised there was still flow as that means the entire length of tubing from the gas pad to the gas panel was "seen" by the MFC so any unequal pressure would cause flow. If you closed all the valves on the gas pad (100, 102 & 104) and still saw flow then it would be more cause for concern.
>
>
>
> I'm not sure if pressure systems allows for this, but an easy test would be to install a N2 calibrated rotameter after the MFC with the other side open to atmosphere and adjust the rotameter up/down when flowing and compare that to what the MFC reports back.
>
>
>
> Something that can be checked is to verify the gas selected matches what the gas the MFC thinks it should be flowing. On the same computer that the controls are being run from open a web browser and go to http://hb-mfc-bonus-target.jlab.org
>
> If you see a green bar about the gas being changed near the top just refresh the page. If the gas there doesn't match the actual gas flown that would explain the times not matching up.
>
> However, I'd be extremely surprised at this being the problem as this is most likely to manifest itself when changing gases and ending up with something wrong, but not just changing for no reason if someone isn't messing with it.
>
>
>
> The only even remotely fishy thing I've seen is that on that on some of the fill cycles the temperature of the MFC dropped more than usual when purging. My guess, the outside temp & gas bottle itself was cold so the cold gas caused the MFC temp to drop more. A few things to keep in mind though: 1) generally a drop in temperature is seen during the purge cycle, though not always as big of a dip nor to that level 2) the temp reading by the MFC is normally only used for calibration so it's readings should be taken with a grain of salt 3) it could just be a red herring and have nothing to do with anything. I only bring it up as food for thought.
>
>
>
> I believe there is a MFC available (not sure on the capacity) and hopefully I can remotely walk through the steps with whoever is available to install it on getting the networking setup on one if that ends up being the solution
>
>
>
> Again, nothing I've seen in the archived data would lead me to believe anything is wrong with the MFC itself.<3-15 purge.png>