Question on dead times
Tom_Frey at BDIS.Com
Fri Aug 7 17:22:17 EST 1998
Because of the Poisson nature of cell arrival, the answer to your question is a
little complicated. At all event rates a small fraction of the events are lost.
The difference gets appreciable when the cells start arriving with an average
spacing that gets close to the dead time. (You have to define "appreciable" and
"close") The dead time on a FACSCalibur is around 25 microseconds, 500 events
per second is not close to a dead time problem no matter how you define close.
There is a second bottleneck in the system however, the cytometer/computer
interface was designed to handle event rates of say 10,000/sec even though the
cytometer can analyze (and make sort decisions) at higher rate. The counters in
the sort window (if you have the sorting module) reflect the cytometer rates,
the counters in CellQuest indicate the rate events are getting to the computer.
The sort window abort counter can also give you a feeling for the events that
are being dumped due to coincidence. (Other pulse problems can lead to aborts,
but once you get to high event rates they are mostly due to dead time
Note that the experiment proposed by Bob Ashcroft in his message will give
incorrect results due to the bottleneck at the computer/cytometer interface.
Data rates will top out there rather than at the dead-time.
WARNING: Skip the following if you don't have a FACSCalibur User's Guide handy,
it won't be clear enough.
To get a feeling for the cell analysis rate at different event rates you could
find the "expected sort rate" vs "event rate" plot in your Calibur User's Guide
and multiply each axis by 10. Change "event rate" to "true rate" and "sort
rate" to "analysis rate". If you follow one of the curves you will get an idea
of the analysis rate at a give rate of real event (but you have to multiply by a
correction for the percentage that was assumed when generating the curve - eg
follow the 50% curve and multipy by 2).
END OF WARNING
Your other questions are a little less clear to me, perhaps because I am too
familar with the Calibur and the BD way of naming things. First, the question
regarding particles in the same drop doesn't exactly fit the Calibur, as there
are no drops. If you means in something like an analysis window, I think there
are issues with concentration induced dimerization versus coincidence that would
be at least slightly difficult to sort out. You could look at some kind of
doublet signal (say green versus red beads?) as a function of concentration as
you sugest, but you would need to check at both Hi and Lo flow rates to make
sure that these were not true dimers. I, for one, haven't done this - but have
seen abort rate versus concentration effects that looked pretty reasonable
compared to theory.
The second question was regarding gating. For the FACSCalibur, all of what BD
calls gating is done on the computer. Only the threshold (or dual threshold) is
done on the cytometer. A threshold rate from the sort counters or an ungated
event rate from CellQuest should usually give you a realistic idea of the
processing rate on the cytometer (all kinds of caveats apply here when using
high (>5000/sec) event rates but I hesitate to annoy the group with the
details). Generally, rates to several thousand should not be a problem for
cytometer dead time.
Sorry if I have misinterpretted your questions. I'm sure you'll get other
response which may help. Feel free to follow up with me by e-mail for any
The refernece Jim Leary cites in Re: Technical Question.... may also be helpful.
I haven't looked at it but it sounds like the right kind of information to help
with your question.
Subject: Question on dead times
From: "Josep M. Gasol" <pepgasol at cucafera.icm.csic.es> at INTERNET
Date: 8/6/98 10:50 AM
Following some recent comments I would like to ask if anybody knows for
sure what's the "true" rate of particles that the electronics of a
FacsCalibur can see without losing events due to them passing in the dead
time zone. My experience counting natural planktonic bacteria suggests that
a flow rate above 500 particles per second underestimates true
concentration (measured with an alternative method). This can certainly be
due to two bacteria coming in the same drop through the laser, but could
also be due to the electronics not being able to handle that many events
per second. By the way, anyone has an idea of a way of checking for two
particles going in the same drop ? Increasing bead concentration and
relating doublets to singlets could be a way. Anyone ever tried ?
A further related question: If I am gating what I see with a "software
gate", (I mean not an electronic gate), I imagine that I should be
concerned with the "total flow rate" (particles in my gate and those
outside the gate) and not with the "acquired data rate". Am I right ?
Thanks for your help !
Josep "Pep" Gasol
Josep M. Gasol
Institut de Ci,ncies del Mar, CSIC
Pg. Joan de Borbo, s/n
E-08039 Barcelona, CATALUNYA
Phone: 34 93 221 6416
Fax: 34 93 221 7340
email address: pepgasol at icm.csic.es
Department's web page: <http://www.icm.csic.es/bio/index_bio.html>
More information about the Cytometry