Debugging tips for the CEBAF Hall A FPP

Introduction

This document is intended to be a debugging aid for the FPP system. Our goal is to keep track of the problems / symptoms that we have encountered, and the solutions to these problems.

Problem: HV sparks and trips when turning up

Possible underlying problems include the following:

• Wire from pin to readout board has come loose, sparks to ground at couple hundred volts and up
• Wire from pin to HV board has come loose, sparks to HV board ground at couple hundred volts and up
• Wire inside straw is broken, shorts to ground at about 0 volts
• Sharp point on HV or readout board sparks to chamber ground - all these should have been fixed

Problem: HV leakage currents large

We have encountered this problem several times, each time from a different cause. The causes have included the following:

• Bad gas or bad gas flow to some chamber section. If some portion of the chamber is not well purged so that it has some ``odd'' mixture of air plus chamber gas, this can lead to difficulty getting the chamber to operating voltage. Usually the difficulty starts at about the voltage at which gain starts. Bad gas pockets / flow is a problem that tends to not go away, and the chamber does not exhibit a training effect. If the chamber has worked previously, this is probably not the problem; nonetheless you might try checking the input / output flow rates, or using a leak detector.
• Also check the gas mix. If there is not enough quenching, as, e.g., if one runs pure Ar, then leakage currents will increase. In particular, light causes a large leakage to develop in Ar, and the straws are not light tight.
• If leakage increases while running chamber at constant voltage, and there has recently been, e.g., rain, there could be a surface discharge on the circuit boards from the humidity. Try blowing dry air over them.
• Alternately, the surface discharge could result from, e.g., finger oils, grease, etc. in the region of the high voltage capacitors. If the chamber has not been trained before, the leakage could be internal. (Always use gloves working with internal gas system components.)
• When the leakage current is erratic for voltages at a couple hundred volts and up, that may also be an indication that there are bad pin to pin connections / loose wires, although the connections are not ``bad enough'' to case a spark.

Problem: Low Voltage does not turn on, or is too low.

Problems we have encountered include the following:

• If the power supply indicates voltage is on, but there is no voltage at the cards / busses, there is a problem with the cable / box connectors. You can confirm by seeing that there is voltage at the back of the box with the cable removed. The metal contacts are spring loaded, but can slip and gradually loosen with time to not make contact. With the cable off the back of the LV box, examine both sets of contacts.
• If the supply is not turning on, the overvoltage protection pot R16 inside may need to be adjusted. Get an expert to do this.
• If box turns on without sense wires, but will not with the sense wires, there are a couple possible causes. BUT NOTE THAT AS OF 10/96, WE HAVE DECIDED TO OPERATE THE BOXES BY SHORTING SENSE AND POWER INTERNALLY AND NOT USING EXTERNAL SENSE CAPABILITY! - RG for FPP working group.
  • The sense wires were connected backward, and the supply is trying to lower the voltage through 0 to negative values. But since it is a differential 0 - ``5'' V supply, 0 is as close as it can get.
  • The sense wires are too far from the power supply, leading to impedance problems. This is a suggested, but unconfirmed possibility. Try mounting the sense wires nearer the supply box.
  • The differential sense wires are mounted in very different places, leading to a time shifting in the sensing between the two. This ``confuses'' and trips the supply. Make the sense wires equal lengths.
• If the output voltage is too low, it may result from current limitations or need to adjust the box output voltage. Get an expert to play with the I-limit pot R6 or the voltage adjustment pot R12.
• The following problem was encountered with LV box 9 35-amp supply. One of the bolts on which the shunt mounted loosened, and made sporadic contact with the chassis of the supply. The supply put out 5 V unloaded, but with any load, even the output cable connected with no cards, the contact with the chassis resulted in varying current draws and a sag in the output voltage. It is FPP group policy to fix / prevent this problem from happening in the future by mounting a thin mylar sheet under the phenolic piece on which the shunt is mounted, when any LV supply box is opened.

Problem: Do not see output signals from a group of eight straws.

Problem: Do not see output signals from a single wire.

• Recall that the number of straws in a plane is not generally a multiple of 8, so we do not expect to see a signal for all readout channels. Does the physical wire actually exist?
• If the straw exists, and other wires in the group of 8 work, then there is a problem with the particular straw. The easiest way to check the electronics for the straw is to run the test pulser, and determine with an oscilloscope what component has an input signal but no output signal. It should also be checked during this that the threshold is set reasonably.
• The usual problem we have discovered in pulser tests is that the high voltage board gives an output, but the readout board does not have an input. The usual reason is that one of the wires from the boards to the straw came loose, most often at the board end.
• Sometimes the signal out of the high voltage board looks much larger than the signal into the readout board. The problem here has usually been bad connections rather than a broken wire. In early testing, in particular, one may find the connector was crimped onto the insulation rather than the wire.
• We have found several boards with good analog inputs but no logic output. The easiest problem to fix has been that the grounding foils have curled and shorted the input capacitor or resistors to ground, so that no signal gets to the amplifier. This usually happens on one of the edge channels of the board. It is also possible that, in the assembly of the stack, careless tool use led to a broken lead on the resistors or diodes, which leads to a much reduced or missing signal at the amplifier input. Most of the problems on the boards have been bad connections with the initial 50 ohm resistor, immediately after the HV capacitor, resulting in an open circuit. This can be easily confirmed with a DVM on either the top or bottom side of the board. In this case, the board is best sent back to the shop for fixing.
• If there is no data from the straw, but the test pulser gives signals, then either the straw has insufficient high voltage or has bad gas / large air contamination.
  • Was the HV turned on?
  • High voltage can be checked with the voltage off by measuring the impedance between the straw and another, eg, on the same high voltage board. If both are connected to the HV trace bus, one should obtain 2 Mohms. If impedance is larger, the straw may not be connected to the HV bus.
  • One can also CAREFULLY supply a very small high voltage, perhaps ~ 10 V, and try to measure voltage directly. WARNING: one chances getting bitten if someone else changes the supply on you.
• If high voltage is okay, and assuming gas was on (other straws worked), then the particular straw has a gas problem.
  • Do the gas supply / exhaust tubes appear to be connected to the straw and the manifold?
  • One can try to use a gas detector to search for hard to see or hear leaks.
  • One may also try to use an ⁵⁵Fe source, which gives signals about four times larger than minimum ionizing particles, to look for a reduced gain that is insufficient to trigger the readout board logic.

Bad width on pulser testing

Sometimes in pulser testing a card, an unexpected width is observed. The expected widths for the eight wires in each board section are 25, 45, 35, 55, 85--90, 65, 100-105, and 75 ns, respectively. Typically, one needs a signal level in the test pulser cable of several tenths of a V to ``fire'' a channel, though attenuation and insensitivity on a channel may require closer to 1.5 V. Bad width can result from several causes.

• Reversed wiring between the HV or RO board and chamber straws will cause the widths of two channels to reverse.
• A disconnected wire to a channel will typically cause no signal to appear until a large voltage is supplied; then noise will cause some channel on the board to fire.
• Insensitivity of the input channel may allow a noise channel to fire instead. Most often -- and most easily fixable -- this has resulted from a bad 50 ohm input resistor, which can simply be replaced. If bad, this input resistor will typically read about 220 k ohms. If one sees no output with an input signal of a few mV to the board, the resistor should be checked. If the resistor is okay, an expert should be consulted to see if the board should be sent to the eshop for further testing / fixing.
• A large voltage may cause a noise channel to trigger before the actual channel. Reduce the voltage until ``just before'' the output signal goes away -- at the fifty percent efficiency level, every channel examined has had the expected width if wiring was correct and intact.
• We have found a few boards on which the redone timing resistors were missing. It is unknown if they were not well put in, and fell out, or if someone accidently / intentionally removed them, and did not tell us. This leads to an output signal that is very long, ~ microseconds.