Re: FACSVantage DIVA specs and performance

From: Marty Bigos (mbigos@gladstone.ucsf.edu)
Date: Wed May 02 2001 - 16:55:13 EST


Larry -

I haven't put a DiVa (need to be careful here with "Vantage with
DIVA" - see RAS syndrome at http://www.newscientist.com/feedback/)
through its paces, but have looked at it recently and can answer a
few of your points.

>I may be in a position to purchase a second sorter

Congratulations!

> and will consider
>another MoFlo and the new Vantage with DIVA.  However, I am not yet very
>familiar with the DIVA component of the Vantage and wanted to pose some
>questions and get feedback from others that may have it or know about it (I
>know it is just now being released).
>
>What I saw in Montpilier  last year had 2 computers - a PC running WinNT
>4.0 to drive the digital electronics and a G3/G4 running CellQuest to do
>everything else.  Is that the way the final system has ended up?  If so
>that seems like a lot of overhead.

That system was (whether BD will admit to it or not) a prototype and
not what they plan to sell. The current system does have two
computers as you mention above, but each is a fully functional (data
collection, analysis, instrument control) system, the Mac for the
standard Vantage and the PC for the DiVa. A switch allows you to use
either one or the other (although it was never clear to me what one
gains by doing this).

>As I understand it the DIVA system does not have log amps but uses high
>resolution (what is being used? - I was told 14bit) ADCs and look up tables
>(like the Coulter XL has done for some time) to generate log data.  Given
>ADC characteristics the low end noise will probably drop the resolution to,
>functionally, about 12bits?.  Will there be quantumization errors at the
>low end with this functional level of resolution range? - at the high end?

Good question; this needs to be evaluated closely for effects on data quallity.

>The way the signals are being taken is seems as though "peak" signals will
>be lost but area signals will be available for all parameters.  Is that
>correct?  High rate sampling of the peak will be used to generate area
>signals.  Are there serious problems with this approach and have they been
>solved?

Not necessarily. There was a previous discussion on the cytometry
list in Feb on peak vs. area measurements which I will not repeat
since it was covered better than I can summarize.

>What is the dead time of this system?  It seems likely to be relatively
>long so this would necessarily drop the sort rate.

Not necessarily so. The dead time on an analog system like the
Vantage is mainly due to peak hold time and non-pipeline system
design - you can't measure the next peak until the first is
processed. In a fully digital system, with the ADC's constantly
running, the dead time is the processing time needed to select the
sample measurement out of the data stream. With the high speed of
DSPs  and FPGA (field programmable gate arrays) I don't think that
this necessarily would be a problem.

>The ADCs that I believe are being used are 10megahertz.  Are such ADCs not
>inherently noisy at the current state-of-the-technology?

Whether this has any effect on the final data needs to be determined.

>  Also given all
>the considerations is this type of log data generation compatible with
>truly high speed sorting - i.e. >15K/sec?  It seems unlikely to me.

There is no reason why it shouldn't be all other things being equal.
That is, if it is designed properly it should go at least as fast as
the Vantage SE.

>Is the principal reason for using the digital log values rather than log
>amps due to compensation issues or are there real accuracy (more truly
>logarithmic) benefits as well that affect things beside the compensation?

This is where I can see the big win for the digital system,
especially given the limitations of the Vantage SE. First, it allows
one to go beyond 8 parameters (a fixed limit in the SE design).
Second it allows for full cross-laser compensation (I have not heard
anyone bragging about BD's Omnicomp here). This puts the DiVa (with
the above caveats) in even competition with the Cytomation Moflo on
these two counts. The digital compensation on the Moflo is done by a
DSP chip on log amp data. Provided that has been implemented properly
(e.g it should give the same results as FlowJo - any experience
here?) the compensated results of samples run on the two instruments
should be pretty close. In the digital system the data is truly
logarithmic, but the big limitation in multi-color work is not
inaccuracies in the log amps but relatively low light levels for the
redder dyes. Currently I don't see significant difference between the
MoFlo and Vantage optical designs here (I am still waiting for
someone to come out with high speed sorting that has the light
collection capability of BD FACSCalibur - i.e. a flow cell with a
high numerical aperture immersion collection lens). Full digital
compensation does help a little with data quality, but most samples
will not show any noticeable difference.

Of course, if one compares the DiVa to the MoFlo, there are other
factors which should be taken into account as well.

>I guess the bottom line is does it work and are there real gains that
>justify what likely seems a loss in sort speed and an increase in
>operational overhead?

Other than a few alpha copies, BD, as far as I know, has not shipped
any DiVas. I have ordered one, and I am waiting.



>
>Thanks for joining the discussion.
>
>Larry
>
>
>
>Larry W. Arnold, Ph.D.
>Res. Assoc. Prof.
>Director, Flow Cytometry Facility
>Department of Microbiology and Immunology
>Lineberger Comprehensive Cancer Center
>CB# 7290
>University of North Carolina at Chapel Hill
>Chapel Hill, NC 27599
>Phone: 919-966-1530
>FAX: 919-962-8103

Marty Bigos
Director
The J. David Gladtone Institutes Flow Cytometry Core
mbigos@gladstone.ucsf.edu
(voice) 415-695-3832
(fax) 415-826-8449

Mailing Address:
Gladstone Institutes
P.O. Box 419100
San Francisco, CA  94141-9100

Shipping Address:
Gladstone Institutes
365 Vermont Street
San Francisco, CA  94103

FedEx, UPS, or courier:
San Francisco General Hospital
Building 3, Room 509
1001 Potrero Avenue
San Francisco, CA 94110



This archive was generated by hypermail 2b29 : Sun Jan 05 2003 - 19:01:17 EST