WP13311 Check 20it-DACs
Summary of work done as part of the 20bit-DAC issue discovered last week by Sheila when driving FC1 in HAM7. Not all channels of the first 20bit-DAC have been driving since the 13apr2026 upgrade of h1sush7 timing-card firmware, TC-sync-ribbon and RCG5.5.0 to RCG5.5.2.
We have 4 more frontends with 20bit-DACs in the same configuration as h1sush7, they are: h1susb2h34, h1cdsh8, h1oaf0 and h1iscex.
We have been unable to reproduce this issue on the test stand (x7eetest1).
On Tuesday 09jun2026 we investigated the status of h1oaf0's 20bit-DAC and on Wednesday 10jun2026 h1iscex.
-----
h1oaf0: Tue 09jun2026.
The first 20bit-DAC has all 8 channels driven by the h1pemcs model. We used this model to drive the DAC for our tests. This DAC has a "CAL" Anti-Image chassis, each channel has a single ended BNC and a DB9 differential output.
We labeled and disconnected the field cables from the AI for ch0,1,2,3,4,6 (BNC) and ch7 (DB9). Ch5 was not being used.
We drove all 8 channels to 5.0V (262144 counts) and DVM'ed all the channels, no drive on any channel. Note this differs from h1sush7's test Tue02apr2026 where its ch7 DAC-DUOTONE was working but the others were not.
This gave us the opportunity to carefully test each change made on 13apr2026 to narrow down which change caused the issue.
Replace 20bit-DAC: First we replaced h1oaf0's DAC (S2101135) with the one tested on x7eetest1 (S2101117). BROKEN.
Install original Sync Ribbon Cable: replaced the upgraded sync ribbon (with the 2 wire shift) with the original pin-for-pin cable. BROKEN
Install new Sync Ribbon Cable backwards: we put the new ribbon back in, but connected it backwards. BROKEN
No Sync Ribbon Cable: ran with no ribbon cable. BROKEN.
we put the new ribbon back in correctly. Problem does not appear to be related to this cable in any way.
Shift all Cards Left: We then applied the "h1sush7 fix" which is to move all the cards left one slot, moving the 20bit-DAC from A1-4 to A2-1 in the process. NOW WORKING.
Move 20bit-DAC back: With all the other cards still moved left, we returned the 20bit-DAC back to its original slot (A2-1 to A1-4). BROKEN
Downgrade RCG: Erik downgraded all of h1oaf0's models from RCG5.5.2 to RCG5.5.0. BROKEN
Shuffle cards on Adnaco A1: Next the order of the cards on A1 was shuffled
From
| A1-4 | A1-3 | A1-2 | A1-1 |
| 20bit-DAC | ADC | empty | Timing Card |
To
| A1-4 | A1-3 | A1-2 | A1-1 |
| Timing Card | empty | 20bit-DAC | ADC |
BROKEN
We put the cards back into their original order.
Downgrade timing card firmware: The timing card was replaced with one programmed with an older version of the firmware (S2101153 with 0x2f4). NOW WORKING.
Upgrade timing card firmware: Marc upgraded the timing card using the JTAG port from 0x274 to 0x635. BROKEN
At this point we a definitive proof that it is the timing card firmware upgrade to 0x635 which breaks the 20bit-DAC. As you can imagine, there has been a lot of discussion between CDS and EE on how this could possibly be the case and we are still mystified.
At this point it appears moving the DAC from A1 to A2 fixes it, but with no insight into the mechanism this is purely empirical. It might be better to downgrade the timing card firmware for systems which do not need the new firmware (i.e. LIGO-DAC and OMC) to the O4 version 0x1f0.
-----
h1iscex: Wed10jun2026
the first 20bit-DAC is driven by two models, h1pemex drives ch0-3, ch4 is unused, h1calex drives ch5-6 and ch7 is driven by the IOP with the DAC-DUOTONE (which is permanently enabled).
We can remotely confirm that the DUOTONE signals in ch6,7 are working. Also Tony confirmed the PCAL chn5 is also working.
At EX, I labeled and disconnected the AI cables. I then connected ch0-3 with BNCs to the PEM-AA chassis for ADC4 ch0-3. I was then able to remotely drive the DAC channels and view the output. I drove with a 0.5V signal, since the PEM-AA has a x10 gain. The ADC recorded a 5V signal on all channels.
All channels on this 20bit-DAC are working. Only ch4 was untested since no model drives this channel. We decided no further investigation of h1iscex is needed.
The field cables were restored, h1pemex's filtermodules used for this test were restored to their original settings.