Errata

If we had to redo the clock interface board, there are the following things to improve:

• Think in time of adding bits for shutter (was luckily remembered before scheme had been finalized) and helium (had been forgotten entirely and been added belatedly). For this let pins as given in table 8.3 go to LEMOs.
• Add internal line, that goes directly from the master clock outs to a trigger input on board NI-PCI-6052E. When we designed the board we had not taken into account, that in case of Sidet scans we want to trigger the ADC in a delayed fashion (as controlled by the Quantum Composer Box), while the Voltages are held constant. However for counter and clock we do not want this delay. So the original scheme of triggering all three kinds of data by the same signal was faulty. We sidestepped this problem by externally routing the master clock signal into Connector J20 (see table 8.2) as well and change the software to have clock and counter triggered from this line. Since this line was originally meant as trigger line for the (currently unused) two DAC channels (connectors J21,J22), it would be better to have the master clock signal directly routed into one of the boards free PFI's (Programmable Function Inputs) and adjust the software to use this line as trigger. I would suggest GPCTR1_GATE (PFI4/pin41) since the ADC trigger uses GPCTR0_GATE (PFI9/pin3).
• Actually the previous error turned out to be a blessing in the end. The proportional counter output turned out not to be able to drive the opto-isolator on the clock-interface input. To work around this problem we switched inputs. The input used for triggering the scaler does not have the opto-isolator and the Programmable Function Inputs (PFI) of the National Instruments board (NI PCI-6052E) can luckily be freely assigned to tasks.
• For some reason for all BNCs (J18,J19,J21,J22) signal and GND lines had been mixed up. It is not clear where this error came from, but it was fixed by the electronics shop by physically switching the lines on the back of the connectors.
• Rethink the wiring of connector J17 (see tables 8.2 and 8.4). Do we really need more pins of the connector routed to ADC channels than we have Sidet channels? (Maybe we want more ADC channels available for other purposes.) And if so, make sure that all free channels connect to GND instead of dangling free.
• Make it possible to adjust drift in point spectra for both axes - X AND Y. Currently we can only stream to two axes (16 bit each) of the Delta Tau and usually this is enough (we take 2D scans not 3D). However during point spectra we stream ZSTG positions on one axes and would like to stream drift adjustments on two more axes (XPZT + YPZT). Right now we deal with this shortcoming by only correcting the axis with the larger drift. If the hardware was redone, one could route the higher 16 streaming bits (or at least 8 of them) into the currently free 3rd PMAC encoder. Then the lowest 16 bit would still be streaming ZSTG positions as now, on the next 8 bit we would stream XPZT positions and on the highest 8 bit we would stream YPZT positions. 8 bit result in a fairly small maximum step size ( $128*0.3nm=\pm 38.4nm$), but it will be enough for small drifts and small delta eV's. In the worst case you would have to enforce a certain energy resolution.
• Add a power on light to the interface box.
• The whole design would be more stable, if connectors were mounted to the case instead of to the printed circuit board only. LEMOs etc breack off easily.
• Add jumpers, which let you choose to 50$\Omega$ terminate unused ADC channels. This is debatable though, since external terminators - on accesible lines at least - make for easier removal.
• Adjust for using trigger inputs from the new Newport XPS controller replacing the old MM3000's. Add 470$\Omega$ pull up resistors to trigger inputs for X/YSTG (J6) and X/YDET (J7). These resistors would have to be between pin 1 and 3 and pin 1 and 5 of each of the connectors, or should be implemented directly on the printed circuit board if remade. Currently these resistors sit in an external adapter (see Sec. 11.4 for more details). However if the possbility of switching back to the old MM3000 controllers shall be retained, there should be a way of disconnecting those resistors. (Actually, if compatibility with the MM3000 is not needed anymore, one could do away with the current circuitry behind J6 and J7 and replace them with something similare to the other trigger inputs of J8-J10)