To: Mark III and Mark IV Stations From: Ed Himwich and Brian Corey Re: Mark III/IV VC Levels March 2, 2001 Introduction This information is only relevant for stations with Mark III or IV Video Converters. It does not apply to stations with VLBA racks or VLBA racks modified for Mark IV by adding a Mark IV formatter (so-called VLBA4 systems). The remainder of this memo consists of two sections. The first titled "Operational Procedures" outlines important rules for working with Mark III and IV VCs. This section is somewhat simplified to give a clear set of guidelines for setting levels. The second section is titled "Discussion" and covers more of the details. Operational Procedures 1. Always adjust the signal levels with the VC 10 dB pads in the signal path. 2. The power level in each video sideband that is being recorded should be greater than 0.1 volts on the front panel, but as low as possible above 0.1 volts. Please set the IF attenuators to the maximum value that keeps the levels above 0.1 volts for all sidebands that are being recorded (note: some modes record both sidebands in some or all VCs). This level should be set during dry weather with the antenna at zenith or wherever the system temperature is a minimum. 3. Please avoid changing IF attenuator settings during experiments. If all levels are set as low as possible, but above 0.1 volts, there should be no need to change them under normal conditions, including rain. If it was not possible to set the levels in dry weather before the experiment, it may be necessary to reduce the attenuation when the weather clears. Discussion As a rule, the IF levels for the Mark III and IV VCs should always be set with the 10 dB pad in the VCs switched in. With the pads in, the minimum front panel power level in the sidebands that are being recorded should be 0.1 volts. The values should be less than 0.5 volts if possible and in no case should be above 2.0 volts. You can also adjust the levels using the TPI measurements from the VCs in the FS. A front panel sideband power reading of 1 volt should correspond to about 32768 counts. Therefore 0.1 volts is about 3300 counts. You should correct for the TPZERO (no signal) level of the converter. For example if the zero level is 600 counts, then the specification of 0.1 volts on the front panel would be the same as 3300+600=3900 counts. Please note that the front panel may have a zero offset that should be corrected for if you are using the front panel to determine the level. The next version of the FS will include a new command, ifadjust, which will use the VC TPI counts to automatically determine the best attenuator levels. The specification of 0.1 volts minimum level is necessary to overcome the intrinsic noise in VCs. (In addition for Mark IV VCs the level must be above 0.1 volts to make sure that there is sufficient input power for the AGC circuit to work properly.) The power levels should also be as low as possible, but above 0.1, in order to give as much dynamic range for Tsys measurements as possible. You will also want to make sure that the levels are low enough so that when the noise diode is turned on, the TPI measurements do not saturate under normal conditions. If you are unable to balance all of these conditions, please contact us. Ideally the front panel levels should be as low as possible above 0.1 volts, during normal measurements, through varying ground pick-up and weather. Please note it is very undesirable to change the attenuator settings during an experiment since this will change the delay through the system. This should not be necessary and will cause some correlator diagnostics to be difficult to use (and therefore more difficult for the correlator to give you useful feedback). In the worst case, if there is a problem with your station's phase-cal and it is necessary to use manual phase-cal at the correlator, it will introduce a clock jump, which will significantly degrade the geodetic solution. In particular, this means you should not change the levels at X-band when it starts to rain. Ideally, your attenuator settings should be adjusted for dry weather and left in that state. The signal level will increase during wet weather, but there should be enough dynamic range to prevent VC overflows. You also shouldn't change the IF attenuators for low-elevation observations. However, if the TPI overflows for low elevation observation you should probably inform us so that the elevation mask for your station can be adjusted in SKED. Ideally, the signal levels of the IFs coming into the rack should be such that, when the VC sideband levels are set as specified above, the IF attenuators are in the range of 10-43 dB. If more than 43 dB of attenuation are required, the error in Tsys due to the measurement of TPZERO will start to exceed 1%. On the other hand, if the attenuators are less than 10 dB the signal is weak enough that there is concern about spurious signals in the rack interfering with the phase-cal in IF. This is reduced as the power level is increased. So the best compromise is when the signal is strong enough that correct attenuator settings are relatively high, but below 43 dB. It is relatively easy to get the attenuators settings down to this level if they are too high by adding attenuators before the rack. On the other hand, increasing the settings requires adding amplification external to the rack. There are some problems with the linearity of the output of the power detectors in the VCs. This issue is still being investigated. The nonlinearities cause fractional errors in Tsys measurements of a few percent when TPICAL is about 20,000 counts, and the error varies roughly quadratically with power level. If high accuracy Tsys measurements are needed, the value of TPICAL should be kept as low as possible consistent with the minimum 3300 count level above the TPZERO level for normal operations. Of course there are other calibration errors that can affect Tsys results at the 10% level as well. Please note that the TPZERO reading for each sideband in each VC should be about 300-1000 counts. If it is more than 2000 counts it should certainly be fixed. An additional point is that in the absence of significant RFI and bandpass problems, the power for the two sidebands should be within 1 dB. If it isn't, the VC should be fixed.