Reports until 14:15, Tuesday 14 July 2026
H1 CDS
david.barker@LIGO.ORG - posted 14:15, Tuesday 14 July 2026 (91026)
Fixed h1susb2h34 20bit-DACs with kernel boot parameter pcie_bus_safe

Ibrahim, Ryan S, EJ, Erik, Jonathan, Dave:

We chose to fix h1susb2h34's 20bit-DAC issue with the boot parameter pci=pci_bus_safe rather than a BIOS MPS=128 change. We did however make the BIOS change to Enable PERR/SERR support.

1. Control room put BS, MC2, SR2, PR2 into safe. I bypassed the seismic SWWD for BSC2, HAM3,4

2. I powered down the AI chassis which is driven by the two 20bit-DACs. It is in rack SUS-C1 U24. From front, left side is DAC0, right side is DAC1.

3. I verified the IOP shows both 20bit-DACs in AI-WD mode. I turned on DAC-DUOTONE and verified no readback on ADC0-30 (safe to do so now the AI is powered down)

4. EJ configured h1vmboot5-5 to boot h1susb2h34 with pci=pcie_bus_safe and disabled the auto-start of the models.

5. We stopped the models, fenced from Dolphin and rebooted.

6. EJ caught the reboot, went into BIOS, Enabled PERR/SERR Support and reset.

7. After the reboot, we verified the Adnaco cards had MPS=128B and the Dolphin IX had MPS=256B.

8. We started the models. I enabled DAC-DUOTONE (AI still off) and now DAC0-7 is driving correctly.

9. EJ ran his DAC-FIFO-Check, both 20bit-DACs checked out.

10. We started the local_dc and cps_xmit to get DAQ data moving again.

11. I powered the AI chassis on. Unbypassed SWWD and handed system over to the control room.

We will leave this system in no-auto-start state from now until the upgrade at the end of this month.

Instructions to start the DAQ data after models have been started (as root):

 

mbuf_probe list     # Check models are writing

systemctl start rts-local_dc

mbu_probe list    # Check local_dc shows as last item

systemctl start rts-transport@cps_xmit