From: Frank Geurts (geurts_at_rice.edu)
Date: Mon Feb 03 2003 - 19:10:35 EST
>> item D -- unresolved yet
>> Ever since the last access I am seeing DMA errors for TOFp-ADC (see
>> below). The logfiles cover the period
>> log2: Jan31,08:19 -- Feb1,08:18
>> log1: -- Feb2,08:18
>> log0: -- Feb3,08:18
>> log: -- Feb3,15:11
>>
>> i don't remember seeing them in earlier logs. Last year we related the
>> TOFp DMA errors to Q-suppresion because of the negative pedestals. Can
>> you see something like that in the localmon TOFp ADC spectra?
>
>
> i've never seen an adc value less than zero in any localmon data...
> remember the "negative pedestals" problems last run were actually
> coming from either too much rack heat and/or the fried TOFp
> comparators. both of these issues are fixed now i think....
>
>>
>> We could consider switching to a slower channel-by-channel readout to
>> identify which of the channels causes this problem.
>
>
if we want to pursue this problem, it might be helpful to have a list of
TOFp adc channels that have 0 pedestal values. A channel-by-channel run
(typically 3.5 - 4x slower than the DMA mode) can then be used to look
for obvious candidates.
>
> sounds draconian - i.e. how would this affect our dead time?
>
The readout takes less than 5ms (1.4) in singlechannel (DMA) mode. With
the L2 returning in order of ~10ms we are still away from affecting
STAR's deadtime but not order-of-magnitude anymore.
> i'm guessing the interest in running (at least temporarily)
> w single channel readout is to locate which ADC (if its
> just 1 doing this) that's giving these DMA errors - right?
>
yep.
> i have to admit though this problem still seems quote
> "rare"/intermittant to me, so i'm not sure if we really
> need to track this down - what is the typical rate of
> these say per 10k
> events according to your logs?
>
these logs cover the last 3-4 days of data taking. The fact that you see
this error only once does not mean it happened only once. The log
facility won't show them anymore until the system is restarted.
From the current localmon data file the DMA problem looks more like
10/36,691 evts. affected. For all the other localmon files 111/664,939
(apparently i did miss a few in previous runs). The ratio increased by
~1.6 ... no reason to worry, but reason enough to keep track of it.
Maybe you could at some counter in a.C to keep track of this ratio (0.027%)
-frank
This archive was generated by hypermail 2.1.4 : Thu Jul 24 2003 - 00:39:34 EDT