Attenuation and Level Control
The setting of attenuation and scale factors in the EOVSA system is critical for several competing criteria:
- Keeping components within a safe operating range in order to avoid damage to components
- Keeping the various amplifier stages in both the frontend module (FEM) and downconverter module (DCM) within a linear operating range
- Keeping the two polarization channels (H and V) matched and balanced
- Keeping the optical link within a linear range
- Keeping the 8-bit digitizer from clipping
- Keeping the power and power-squared products output by the correlator in a range appropriate to correct spectral kurtosis calculation
- Keeping the pre-correlation power levels in the appropriate range for 4-bit down-sampling
This memo examines each of these issues in turn, and offers explicit guidelines for attenuation and level control, based on experience with the prototype system. The control system must incorporate these guidelines in a manner that both meets the above criteria and also allows a reasonable scheme for calibration that is as simple and practical as possible.
Keeping Components Safe
When the prototype system was designed, we had little information about the RFI environment, and hence were working with a range of likely 1-18 GHz RF signal strengths without consideration of RFI. Tests with the prototype (see the recent “EOVSA RFI Environment and Polarization” memo) indicate that the signal levels are dominated by the H-polarized PCS signal near 1.9 GHz, which varies greatly with direction of the antenna. The prototype was also constructed with considerably more gain than it was designed for, which has exacerbated the problem of potential damage to components. We have since modified the prototypes to bring the overall gain down closer to design levels, but there still exists the potential for RFI-related damage to components with certain combinations of attenuation. Of course, there is also the possibility of greatly enhanced signals during strong solar flares.
In the regions of the sky with the minimum RFI, our tests indicate that the input power, Pin, is about -62 dBm (based on Christian Holmstedt’s spreadsheet “EOVSA FE Simplified Power Levels and Noise,” i.e., more than a factor of 10 higher than our minimum design expectation of -74 dBm. This is no doubt due to the residual RFI that is present even at this minimum level. At places where the RFI is maximum, the power Pin is at least 10 times higher, or of order -50 dBm. According to Christian’s spreadsheet, damage to the prototypes can occur when Pin = -48 dBm, if both attenuators in the FEM are set to zero. Thus, from this perspective alone the optical link can become damaged at any time by pointing the antennas at the “wrong” location in the sky. At this same setting, the 3rd stage amplifier is overdriven at as low as Pin = -64 dBm (i.e. over the entire sky), while the 2nd stage amplifier is overdriven at Pin = -50 dBm. The situation is slightly worse for the production system. This implies that the attenuation state with both attenuators set to zero must be avoided at all times, and the control system needs to explicitly prevent this. In the absence of solar flares, it is safe for the 2nd stage amplifier and optical transmitter to set the 1st FEM attenuator to 9 dB, so it is recommended that this be the minimum allowed setting. If the 2nd FEM attenuator is set to 0 dB, there is a potential to overdrive the 3rd amplifier in the region of strongest RFI, but not to a damaging level. Still, it is safest to use a non-zero setting at any time the antennas are slewing from one source to another.
For safety reasons also, the frontend attenuation should be under active control at any time the antennas are pointed at the Sun, so that large flares do not occur that can damage the frontend components.
As for damage to the components in the DCM, or further down the chain, we know that this is possible since we actually did damage the digitizer boards in the correlator due to excess IF power. However, this was due to misunderstanding the nature of the 20 dB amplifier on the digitizer boards, and for the prototype systems new fixed attenuators have been added between the DCMs and digitizers that should eliminate the potential for damage. Provided the frontend systems are properly in range, it is not expected that it will be possible to damage any component in the DCM or digitizer boards, regardless of the setting of the DCM attenuator. However, this should be checked for the case of tuning to band 2 (the band with the 1.9 GHz PCS signal). For the production systems (and the prototype systems will also be retrofitted), the final IF amplifier will be eliminated so that the fixed attenuation can be removed without causing damage to the digitizers.