Hello to everyone,
Thanks to Drs.:
Rich Shoemaker
Ryan McKay
Perttu Permi
Youlin Xia
Robert Harker
Nishida Toshi-Cons
Kirk Marat
Tara Sprules
Eugene DeRose
Hsin Wang
for helping on this problem. Here is the original email I sent to AMMRL:
-----------------------------------------------------------------------------------------------------------------------------
A "FIFO underflow" error (experiment stopped) has been persistent on our
INOVA600 instrument. It first happened to
HNCACB, CBCACONH, and CCONH from a 5-year old ProteinPack package. I
installed new BioPack and find out
that HNCACB was finished without such error, but CBCACONH still got
interrupted by this error. Steve Breeding at
Varian said there are some bugs reported regarding FIFO underflow and
3D/4D biopack sequences. Does anybody
have this problem solved? If so, can you share the solution with me?
Your help is highly appreciated.
---------------------------------------------------------------------------------------------------------------------------
In past week, I basically tried all the suggestions attached below. The
results are mixed, which means it is still a problem. No matter what I
did, the error message came with the CCONH anyway. I did a 3D noesy_CN
(CN-edited, CN-filtered NOESY for a complex) sequence (3days) from Dr.
Kay' libray, it was fine without any problem. I am pretty convinced that
this is a software problem instead of the hardware failure. I will keep
posting on any new development.
Thanks again for help.
Cheers,
Yongbo
-----------------------------------------------------------------------
We've had many of these FIFO underflow problems on many our Inovas off and
on a lot. Sometimes, this is a hardware problem (a timing problem in the
digitizing, processing, and storing the FID). There are so many things that
can cause it that it is sometimes very difficult to nail down. Sometimes it
is a problem in communication between the ADC board and DTM board, but there
are other failures that can generate this error.
Our 600 started doing this a lot after we upgraded to VNMR 6.1C with the
203-206 patches. With the ColdProbe we have to use dsp='i' and qcomp='y',
and if "oversamp" is too large the experiment will crash with FIFO
Underflow. With dsp='i', using a longer d1 delay, or decreasing the value
of "oversamp" before starting the experiment helped with our problem. It
never did this before "upgrading" the software.
This also has happened on a couple of occasions on our new 800 Console.
Recently we were running a Biopack HNCACB (3-days) and the experiment
crashed mid-way through with FIFO Underflow. The new Inova-800 is running
VNMRJ 1.1D in the "classic" interface. There are all kinds of rumors of
VNMRJ being more stable in the native "VJ" user-interface, but we don't know
about that yet.
I believe the new "Varian NMR System" shouldn't have these problems because
the VME bus is gone, and hopefully the Gigabit-ethernet communication being
used in the digital side of things will be more stable. Besides, EVERYONE
buying the new Varian spectrometer will be using the new VNMRJ user
interface because the "classic" mode will no longer be an option.
-------------------------------------------------------------------------------------------------------------------
We had been experiencing this problem on and off for years. Mostly
related to the large 3D/4D data sets. Unfortunately
even with high levels of debugging code set to on, and with Varian
assistance, we have not been able to pin down the problem nor a solution.
The chemistry department here at the University of Alberta made some
software alterations and were able to eliminate their problem,
this fix however did not work for us (specifically Chemistry had made a
change, which they later undid and this seemed to remove the problem.
We had not made this change so this fix did not apply).
Of note, since updating our 800 Inova to VNMRJ and a Sun Blade 1500
(previously VNMR 6.1C with an Ultra 10) we have not had FIFO
errors in our larger experiments. Our smaller instruments are also no
longer experiencing this problem under VNMRJ (1.1D with patches).
This could therefore be a software (fixed in later VNMR software) or
hardware (slow computer), or combination issue.
-----------------------------------------------------------------------------------------------------------
Before you start your 3D or 4D experiment, you may exit your VNMR and
then re-enter it. By this way, some occupied memory will be released.
Since I use this way, the FIFO underflow never happened to me.
-----------------------------------------------------------------------
We had similar problems some years ago. I found a simple, although
not the most optimal solution, for this problem. Have you tried to
run the experiment with bs flag set to 'n'? If that does not help try
turning off the dsp... i.e. dsp='n'.
-----------------------------------------------------------------------
We don't do 3D but we see FIFO (First IN First Out) underflow errors in
regard to normal FT spectra on inova's and mercury's when there is some kind
of mismatch between the acquisition time (at) and 1st delay (d1).
All I really know is that when it begins to show, it will always show with
the same parameter set. In general I need to adjust d1 longer to make it go
away so think about a delay that is not critical and increase that one.
My feeling is that this error is a generic error that indicates the
instrument and the computer are not keeping up with each other. Example:
whenever the at is short, d1 is 0 or short and the np is high.
----------------------------------------------------------------------------
I have just completed a series of BioPack experiments on an INOVA 800.
The sample was a relatively large protein, 2H(partial), 13C, 15N labeled.
The concentration was rather low so many transients (days' of acquisition)
were acquired with 2H decoupling and TROSY selection. The experiments included
gHNCACB, gCBCACONH, gHNCOCA, gHNCACO, etc. During three weeks, I haven't seen
FIFO underflow, not a once.
I am not so sure the FIFO underflow is BioPack pulse sequences specific.
>From time to time we saw FIFO underflow occurred but they were not a machine specific,
nor pulse sequence specific and not reproducible.
I don't have remedy but I just restart experiment when it occurred.
Fortunately I don't see it often (we have an INOVA 400, 500, 600 and 800)
---------------------------------------------------------------------------------------------------------------
I had this happen once (I don't remember what experiment it was with)
and after tearing my hair out for an hour or so,
I exited VNMR, logged out of the account, logged back in and started
VNMR again and the problem was gone.
It hasn't happened again. Not really a solution, and by the sounds of it
you have probably been in and out of accouts and vnmr.
-------------------------------------------------------------------------------------------------------------------
We have seen FIFO underflow at times on our INOVA 600s, during long 3D
and 4D runs. The likelihood of an underflow
appears to increase with the length of the experiment. To alleviate
the problem, install the latest VNMR or VNMRJ patches
and the latest version of BioPack. Note, however, we also have
experienced underflows running non-BioPack sequences.
----------------------------------------------------------------------------
We have always used Lewis Kay's sequences for these experiments and have not
encountered FIFO underflow problems. Have you tried getting sequences from
Lewis Kay and try them out on your instrument?
Received on Tue Jun 21 2005 - 17:52:14 MST