[Halld-online] question about scalers in the EPICS archiver

[Richard Jones]

Hello Hovanes and all,

This entails a serious loss of precision for long-running tallies.
Essentially it means that all we have in EPICS are one-second snapshots
every once in a while, between which the values can wander by 3% without
any record of having done so. This makes them useless for any kind of
normalization, which is what I was hoping to get from them. Suppose I
wanted to get the total number of counts accummulated by a scaler over a
several hour period, is there any way to get that information from anywhere
in our online system right now?

-Richard

On Thu, Jan 11, 2018 at 1:28 PM, Hovanes Egiyan <hovanes.egiyan at gmail.com>
wrote:

> Richard,
>
> The discriminator scalers are latched every second and the units are Hz.
> Sergey Furletov
> runs the server that reads the values out into shared memory, so I do not
> know the details. I believe
> that the value is non-integral because it gets normalized to a clock to
> get rates more precisely in Hz. So
> the square root of the value in the DB for the scaler should be a good
> estimate of the statistical error.
>
> Archiving of the scalers in MYA (or other PVs) has a deadband (for this
> particular variable it is
> 3*sqrt(value)+0.03*value), which means that if the next reading (which is
> indicated by a change in
> the value of the EPICS variable) is within that deadband of the latest
> recorded value then it does not get
> recorded. So, if you stripchart an EPICS variable live, you will see a lot
> more frequent value changes
> compared to what you see when you look at the archive chart of the same
> variable.
> This is done to reduce the update rate in the DB, otherwise it would cost
> a lot more money to keep all
> variables in the DB without any deadbands. The deadbnads are adjustable
> though, I can request smaller
> deadbands for a small subset of variables.
>
> Hovanes
>
>
>
> On 01/11/2018 12:41 PM, Richard Jones wrote:
>
> Hello Hovanes and all,
>
> I am accessing the GlueX scalers in the EPICS archive using the MYA
> interface, and have run into a question. I am looking at a variable like TAGM:T:8:scaler_t1
> whose name suggests that it is the discriminator scaler for TAGM column 8.
> When I look at the numbers in that variable, they are obviously not
> scalers. They are non-integer values in the range of 3000-8000 for the runs
> that I am looking at. Are they rates? In what units? kHZ? What is the
> sampling interval? This matters because of statistical error. I see entries
> in there sometimes separated by tens of seconds during a run. In the time
> between the samples, were the scalers free-running so that I have the full
> statistical information in the one reading at the end of the interval, or
> are all of the statistics during the seconds in between the records in the
> DB lost?
>
> -Richard J.
>
>
> _______________________________________________
> Halld-online mailing listHalld-online at jlab.orghttps://mailman.jlab.org/mailman/listinfo/halld-online
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-online/attachments/20180111/3add20f1/attachment.html>