| E1 - E8 |  |
E9 - E16 |  |
E17 - E24 |  |
| E25 - E32 |  |
E33 - E40 |  |
E41 - E48 |  |
Here are some E-counter plateau curves taken on Dec. 19, 2002. The conditions were:
| Beam Energy | 2.035 GeV |
| Beam Current | 20 nA |
| Radiator | A (3 x 10-4 R.L.) |
| Discriminator Thresholds | 10 mV |
| Discriminator Delays | 88 ns |
| Discriminator Pulse Widths | 24 ns |
| HV Range | -1200V to -2100V |
The three prototype boards of the new, JLab designed discriminator/scaler VME modules were tested using the first 48 E-counters. The E-counter signals were plugged directly into the new boards (not through any type of fan-out).
The plateau curves are shown below in groups of 8. Click on a thumbnail to see the plots. Each plot contains two curves. The blue (generally higher) curve was made from the ungated scalers. The red (generally lower and flatter) curve was made by gating the module on a subset of T-counters. The tube voltage is plotted on the X-axis while the Y-axis was determined as follows:
The scaler rates were determined by integrating over 3 seconds. They were then normalized to the gate1 rate from module 0 measured over the same amount of time. (i.e. the Y-axis is the scaler rate normalized to the rate of the OR of the first 7 T-counters.)
The 48 E-counters tested can be broken down into the following groups:
| 33 (69%) | Good - Plateaus can be seen from ungated scalers alone |
| 2 ( 4%) | Dead |
| 5 (10%) | Low Gain - Maybe higher voltage will fix |
| 6 (12%) | No clear plateau in ungated scalers curve |
| 2 ( 4%) | Noisy - Ungated rate high, even at low voltage |
| E1 - E8 | |
E9 - E16 | |
E17 - E24 | |
| E25 - E32 | |
E33 - E40 | |
E41 - E48 | |
The discriminators can be triggered by noise on the VME bus during transactions. The level of this problem depends upon the threshold so for thresholds as low as 10mV, the problem can be severe. During preliminary tests of the electronics in the hall prior to the beam test, one module required a threshold of 80mV and the other two 40mV in order to completely supress the erroneous counts. The module requiring 80mV was in the slot closest to the ROC. At the begining of the beam test, we moved this module over one slot and inserted another module next to the ROC to act as a potential shield. We ere then able to reduce the threshold for this module by about 10mV. Additional shielding and position optimization will be investigated.
At least some indication of a problem was seen in 30% of the E-counters. The 2 dead counters should be an easy fix. We have good hypotheses on the cause and cure for those with low gain and no clear, ungated plateau. We'll need to test these. Even as they are though, the "no clear, ungated plateau" counters have clean gated plateaus. In the worst case, these would have higher rates, but still be efficient at detecting good hits. Probably not a show-stopper.
The noisy tubes may necessitate higher discriminator thresholds which, in turn would reuired higher HV for all E-counters. If it's really the tube (and it probably is), then we may just need to replace them.
These new electronics give us a much clearer insight as to the health of the individual E-counters. At this point, none of the problems seem difficult to overcome. The biggest concern I have left is the mean time to failure for these tubes/bases.