Digital beamline instrumentation small-group discussion, 2022 May 10

Back to Main Page >> Data Acquisition Meetings

previous meeting << >> next meeting

Meeting information

 Meeting time:  10:00am Eastern

Agenda

• Discussion with John and Chad about the JLab digital beamline instrumentation current capabilities and possible upgrades

Additional Resources

• Before the meeting, Bob had put several documents at https://userweb.jlab.org/~rom/digibcm/ to describe the requirements and status of previous work

• Before the meeting Paul had sent the following questions to John and Chad
  1. In order to use the digital information from your receiver to be able to do the helicity gating (helicity flip period is planned to be 10us for MOLLER), we would need to be able to get all of the data at the 1 Msps level. For each of the four wires, there is effectively 21 bits for both I and Q, so that would be 168 Mbps at the densest packing (or 192 Mbps using a 24bit value for each record), not including any timestamps.
     Can we get this data transmit rate on the optical fiber?
  2. The integrating ADC modules have their input bandwidth cutoff at about 1 MHz, and sample the filtered input at about 15 Msps. We would want to have an option for some beamline elements to be readout at a higher rate than 1 Msps, to be able to match the detector bandwidth. I think this could only reasonably be done for channels that are not mulitplexed, and I think that the analysis of I and Q would set the minimum number of 60MHz samples per readout at four, thus allowing a 15Msps readout.
     Does that sound correct, or are there other issues in the signal processing that would impose a lower maximum readout samples per second for non-multiplexed channels?
  3. During your presentation on 18 March, you mentioned that it ought to be possible to add a third channel to the receiver box to allow a cavity beam position monitor to be instrumented without multiplexing the input signals.
     Would it be possible to have four channels in the receiver box to allow a stripline BPM to be instrumented without mulitplexing? We expect that this would require running additional cables from the BPM, and would only want to do this for the most important of the BPMs in our line.
  4. In the case of instrumenting a stripline BPM without multiplexing with a readout rate of 15 MHz (assuming a 24bit word readout), we'd have a data rate of 2880Mbps. Can that data rate be supported from the receiver as it exists now, or would it require further development of the transfer capability?

• After the meeting, Paul had asked about the bandwidth of the existing SEE electronics
  • JLab Technical note [1] indicates that the "linac-style" SEE instrumentation has an effective analog bandwidth of 56kHz, but the outputs from the sample and hold cards have a filter with a 3-dB bandwidth of 26kHz. The linac style SEE have a sample rate of 114 kS/s (corresponding to a sample interval of 8.75 us); each of the two wires is integrated for 2.6 us out of the 8.75 us to give the S/H outputs. The transport-style SEE sample rate is 7.1kS/s, giving an effective analog bandwidth of 3.6 kHz.
  • John had sent a document describing the SEE software requirements; I'll put it onto the docDB shortly.

Minutes

Participants: R. Michaels, P. King, R. Carlini, J. Musson, C. Seaton

• The system that was previously developed by Chad and Randall used data at 10 kS/s and a 1 kHz bandwidth. Just the calculated magnitude was sent from the receiver, as a 28bit data value.
• The current fiber link is capable of sending 5 Mb/s. John thinks a future upgrade to a SFP would be possible to increase data transfer capability
• Paul asked if the plan is to replace all of the SEE instrumentation
  • John's expectation is that the new modules would be used for new channels, and would slowly be used to replace SEE
  • The SEEs in one of the arcs (west?) are planned to be replaced (what was the timescale?); the removed instrumentation would become available as spares
  • There is no plan to replace the SEEs in the injector. If we wanted to do that, we'd have to bring it up as a priority
  • John thinks there would be sufficient SEE spares for the next ~10 years
• John thought that development of a receiver with four channels to avoid multiplexing would be possible, as well as increasing the sampling toward 15 MS/s. I (Paul) think he had mentioned this would require reconsideration of how the processing was done.
• Discussion of the capabilities of the current version of the receiver running at 1 MS/s, other than data transfer
  • John would recommend a digital data stream with both the I and Q data at 1 MS/s. They would be ~21 bits each.
  • At the 1 MS/s, the effective analog bandwidth is in the range 100-200 kHz
• Discussion of the capabilities of the analog outputs
  • There are four 18-bit DACs running at 1 MS/s in the receiver. They can get data from different stages within the digital processing including at the first 1 MS/s stage.
  • The DAC outputs have a four-pole filter with a 100 kHz cutoff (presumably this is the 3-dB cutoff)
• Initial plans to move forward
  • Bob, John, and Chad will set up a receiver test-stand at the TEDF cubicle.
  • Bob and Chad will work on getting the I and Q data for one channel send out on the fiber (at the current 10 kHz) and read by a V1495 or other FPGA readout card.
  • Bob will develop the firmware on the readout card to calculate and accumulate the magnitude and phase information, and compare to the magnitude and phase computed in the receiver's IOC.