[Dsg-hallb_magnets] Sept 2017 QD delay test analysis

[Brian Eng]

Some comments...

- The 100 mV threshold was from the DCP tag on the board. It wasn't something I measured/calculated.

- Your assumption about the pulse width is correct. Values close to but not at the always tripped would occasionally trip the QD board and vise-versa with the never tripped.

- The difference between Never/Always trip can basically be considered the noise in the setup. That's why there is less of a difference in pulse width as the amplitude goes up, but generally those values are pretty close.

- No, the 400mV pulse captured couldn't have been big enough to trip the QD (i.e. it was actually a bigger/faster pulse and this was all the DAQ captured). If this was the case QD4 should have been part of the sum as it only has VT19 on it. Also from the earlier testing, even a 3ms 5V (!) pulse wouldn't trip the QDs no matter what the delay was set to.

----- Original Message -----
> From: "Tyler Lemon" <tlemon at jlab.org>
> To: "dsg-hallb magnets" <dsg-hallb_magnets at jlab.org>
> Sent: Monday, October 8, 2018 3:58:46 PM
> Subject: [Dsg-hallb_magnets] Sept 2017 QD delay test analysis

> Hello,
> 
> Since there have been more QD problems lately with the Hall B Solenoid, DSG has
> started to re-look at data from the QD tests Brian performed back in September
> 2017 (original results at https://logbooks.jlab.org/entry/3485488). The one
> caveat with this new analysis, though, is that the results apply only to the QD
> board tested. The ones actually in use in the hall may behave differently.
> 
> For Brian's test, he mentions that all channels were noted to have 100 mV trip
> thresholds. From the previous test's data, we created a new spreadsheet
> (attached). In the spreadsheet, each QD delay setting tested is denoted by the
> color of the columns. The columns with the title "Never Tripped" indicate
> conditions used where the QD channels never tripped. Columns titled "Always
> Tripped" indicate conditions used when the QD channels always tripped. It
> wasn't explicitly said in the original logbook data, but we assumed that
> between the pulse widths for "Always Tripped" and "Never Tripped" there were
> pulse widths where only some of the pulses tripped the QD channels.
> 
> The first row of plots is Brian's original plots reformatted with the colors
> used in the spreadsheet for the QD delay times.
> 
> The extra analysis performed is to see the QD trips during the test in relation
> to the charge of the input pulse. The charge (integral of current over time) of
> the pulse is proportional to the integral of voltage over time by a factor
> determined by the impedance of the circuit and instrumentation (since the
> impedance is unknown, we were only able to calculate the proportional
> voltage-over-time integral). The charge-pulse amplitude relationship is shown
> in the second and third rows of plots.
> 
> From this data, it seems that the QD delay is really increasing the total charge
> needed to trip the channels. It also seems that a voltage spike that is large
> in amplitude and low in duration may still trip the QDs.
> 
> From Probir's slides last Friday, there was a ~400 mV pulse that lasted 1.1 ms
> on VT19. Could that have caused the amplitude of the pulse gave it enough of a
> charge to trip the QD?
> 
> Best regards,
> Tyler