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

[Sebastian Kuhn]

Hi Brian,

thanks for responding. I personally believe we have seen this problem pretty much from the beginning, but it has gotten worse. I don’t think there is any gas dynamics reason why the first fill should take longer; in any case, sometimes it’s the 2nd fill as well. For example, see the following 2 purge cycles:
https://logbooks.jlab.org/entry/3805040
(This one was this morning purging with nitrogen, but it doesn’t depend on the gas)
https://logbooks.jlab.org/entry/3805043
This one was yesterday purging with deuterium. As you can see, in this case the first TWO refills took “forever” (16 min) while the third one was normal (about 4 minutes). NOTHING changed in between - SVBT100 was open, the D2 gas pressure was steady, etc.. Obviously, we can survive with this, but if in the worst case we have to wait for 3 of these long fill cycles, we are wasting close to 1 hour of beam (or whatever else we need the purge for). Also, if the flow rate readback is too large, we can trigger automatic shutdown (target empty) in auto mode, so we cannot use auto mode for purging anymore (at least we feel uneasy).

The flow readback I saw at 200 sccm was when I had both SVBT100 and SVBT102 closed, as well as SVBT108. So there really shouldn’t have been any way for flow to occur. More generally, the flowrate readback often reads 200 sccm throughout the whole fill cycle, although it should go down as the target gets close to the fill pressure. Of course, it’s a “flaky” problem, in the sense that sometimes the feedback is perfectly fine - e.g., right now (with no target connected and all valves closed it does read 0 sccm).

If there is a way to check out MFC104 (or swap it out) over the next 2 days, that might help us save some precious beam time going forward. Of course, we do want to avoid that we can’t start the run back up on Friday because the gas panel is not fully ready for operations. Please let me know if you think there is something we can do to fix this?

Thank you so much - Sebastian

On Mar 16, 2020, at 5:00 PM, beng at jlab.org<mailto:beng at jlab.org> wrote:

MFC104 seems flaky<https://logbooks.jlab.org/entry/3803565>

  *   Edit<https://logbooks.jlab.org/entry/3803565/edit?destination=email/send>
  *   Delete<https://logbooks.jlab.org/entry/3803565/delete?destination=email/send>

Lognumber 3803565<https://logbooks.jlab.org/entry/3803565>. Submitted by kuhn<https://logbooks.jlab.org/user/kuhn> on Fri, 03/13/2020 - 18:47<https://logbooks.jlab.org/entries?start_date=1584136024&end_date=1584143224&book=HBRTPC>.
Last updated on Fri, 03/13/2020 - 18:51
PR (OPS)
Update this Record...<https://logbooks.jlab.org/pr/edit/876536>
Needs Attention:
YES
Affected System:
Hall B > Hall B Detectors > Hall B CLAS12 Detector > CLAS12 Gas
Responsible Group:
Hall B Engineering
There is 1 comment...<x-msg://70/#>
MFC104 Flow Rates<https://logbooks.jlab.org/comment/28152#comment-28152>
new
by beng<https://logbooks.jlab.org/user/beng> on Mon, 03/16/2020 - 16:58
Sebastian, do you have a timestamp when you ran into this?
Also, which valves were you closing when you saw 200 sccm of flow?
The initial fill has always taken longer than subsequent ones, I'll leave that to the gas experts to explain.
Logbooks:       HBRTPC<https://logbooks.jlab.org/book/hbrtpc>
Entry Makers:   kuhn

We continue seeing strange behavior when we first fill the target with a new gas. The rise in pressure is VERY slow, although the MFC104 readback claims 200 ccm. Subsequent fills go much faster, with no change in the readback from MFC104. In fact, I have observed MFC104 Flow reading 200 ccm WITH THE VALVES CLOSED! During the second fill, the readback becomes more reasonable, going to lower values as the target fills up (6 ccm with valves closed). Of course if the MFC reads full flow, the valve doesn't open as much... This may also be the root cause for our continuing troubles taking the system into auto mode - as long as MFC104 reads a high flow rate, the fail-safe logic will assume there is a leak and shut all inlet valves. For now, I continue in manual mode.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/dsg-hallb_rtpc/attachments/20200316/b18e6768/attachment-0002.html>