Ibrahim, Oli, Betsy, Arnaud, Fil
Context: In December ('23) We were having issues confirming that the damping, OSEMs, electronics and model were working (or rather, which wasn't working).
I have more thorough details elsewhere but in short:
- There was initial weirdness in setting offsets and exciting the different degrees of freedom
- Fil and I checked that all OSEMs were healthy by driving excitations to each one at the coil driver level (ensuring that the test stand was working)
- Arnaud was invoked to help with TFs for the BBSS (we couldn't get these to work prior)
- There was a lot more "fixing" and confirming involving excitations, damping, input filters, OSEMs, OLV setting, test stand electronics, and even getting into X1. I have notes on all this elsewhere but this is not the scope of this alog.
Eventually, we were able to go through Jeff and Oli's alog 74142. Here is what was found:
All "crude push around offsets" in the test bank yielded positive drives in the damp channels. These are the ndscope screenshots. Different offsets were needed to make the offset change more apparent in the motion (such as with L). A minimum of 1,000 was arbitrarily chosen and was usually enough.
Transfer Functions: where it gets interesting... (DTT Screenshots)
In these DTTs, each reference (black) are the transfer functions without the damping, while the red traces are with the damping.
All "translation" degrees of freedom (L, V, T) showed correct damping, peak location and resonance
All "rotation" degrees of freedom (P, R, Y) showed completely incorrect damping, usually showing shifted peaks to the right (higher freq).
In trying to figure out why this is, we asked:
- Could it be any culprit OSEMs? If so, then the damping would not have shown as cleanly for L, which uses the same 3 OSEMs (F1, F2, F3) as P.
- Could it be the electronics? Fil and I checked the coil drivers and confirmed through exciting a sine wave that this is unlikely the case. This will be re-confirmed with the new transfer function findings.
- A small idea could indicate satellite box configurations could be misbehaving but again, the satellite boxes which serve certain OSEMs show both working and erroneous behavior.
- E.G. RT and LF show good matching peaks and damping in Vertical but not in Roll, despite both being hooked up to the same electronics throughout.
- A small idea could indicate satellite box configurations could be misbehaving but again, the satellite boxes which serve certain OSEMs show both working and erroneous behavior.
- Could it be the OSEM polarities? These were confirmed to be correct.
- Plus, if so, then we would see more differences with the OSEM offset scopes taken above perhaps?
- E.G. Coil output for SD with positive offset was still positive despite a -1G polarity correction, indicating that the math is right here. Using SD as an example because it's only used in T.
- Could it be that multiple polarities are wrong at the same time in such a was as to distort the results in this Translation vs. Rotation way?
- E.G. LF and RT are set with opposing polarities so the output gains have 1G and -1G respectively. Since V works, but not Roll, I think that this rules out polarity issues?
- Will reconfirm polarities time permitting.
- Plus, if so, then we would see more differences with the OSEM offset scopes taken above perhaps?
- Could it be the model? Some BP filters? I do not know. There have been, as said, a whole host of details left out leading into this transfer function reading, so a thorough troubleshooting, re-testing would need to happen to confirm - the only untouched things are:
- The DTT Template used
- The BS Model provided by Jeff and Oli in alog 74142.
- Confirmed that the matrices in EUL2OSEM are set correctly as far as sign notations and OSEM to DoF combos. Sums are net 0 for rotations and are net positive/negative for translations (which seems intuitive).
(In)conclusion:
It seems that whenever the OSEMs push in the same direction, everything goes as planned, hence why all translation damping works. When we ask the OSEMs to push in opposing directions with respect to one another though, they seem to freak out. This seems to be the prime "discovery" of finally getting the transfer functions.
This is the "for now" update - will keep trying to find out why until expertise becomes available.
Rahul, Ibrahim, Austin
Context: After hopping off the TS call where we decided to try the Pitch TF again and reducing the Damping gain, I met with Rahul and Austin in the control room and we decided to check some more basic OSEM health first.
Something I forgot: When Oli and I were in the control room taking the transfer functions the first time around, I noticed that for the rotational degrees of freedom (P, Y, R), the OSEM outputs were railing immediately (both visibly in number and on the overflow page). I wondered whether I should re-do the TFs without the saturations, by empirically testing the gain until it doesn't overflow. I ultimately kept the nominal -1G in order to report the initial "this is how bad it is" results. This will become relevant later.
Rahul was concerned that the OSEM spectra for the OSEMs that are in M1 were too noisy so we took some spectra measurements of the OSEMs themsevles to see if this was the case ... and it was. These are the screenshots below. We tried them with and without damping to see if damping works, and it doesn't seem like the damping is working exactly as it should be. Additionally, the <10Hz noise is 1-2 orders of magnitude too high according to Rahul. This is a way more "up the chain" (down the chain?) issue and could result in the weirdness we're seeing at the TF level. Why is this?
- Environmental reasons such as dangling ribbon cable or cleanroom turbulence
- When Rahul and I previously were taking AOSEM noise spectra, we ran into a similar issue where the free hanging of the ribbon cables were impeding the cleanliness of the results. I will go in to check if this is the case and if it is, I will secure all cables and re-take the measurements.
- Since the cleanroom cannot be turned off for long, Rahul suggested turning them off briefly to see if the results improve. I will consider this option if spectra are still noisy after everything else. The BBSS is currently surrounded by a protective cover and so it should be a bit more impervious to such turbulence so this may not be relevant.
- Rubbing/Touching
- It is likely that the (now relevant) saturations seen in the rotational modes may have something to do with touching or railing. As Rahul described it to me, if the OSEM is only rubbing when damping is on then it will have a sort of critical failure as it tries to damp something it is touching, causing it to rail almost immediately.
- Rahul and I went into the damping configurations just to see how low the damping needs to be in order for the DAC outputs not to overflow:
- R had to be reduced by 10X
- P had to be reduced by 100X
- Y had to be reduced by 10X
- So again, we can see that the clear culprits are the rotational DoFs but this time, we can see that P in particular is misbehaving. Why?
- P is the only DoF which uses 3 OSEMs (F1, F2, F3) so this could just be as a result of generally exacerbated poor OSEM noise behavior. This would potentially explain why the undamped P reading was terrible combined to the other ones.
- Unhealthy OSEMs
- Rahul also suggested that the OSEMs could just be unhealthy and that we could do Open Light BOSEM Electronic Noise Spectra
- After some discussion we concluded that this is probably not the case because of the clean translational TF results (which showed that a combo of all OSEMs were working well without clear culprits) so it's less likely that all of them are bad. Though this would be the last resort check.
The Plan
Following these quick checks - once I'm out of the Staging Building:
- I will retake noise spectra measurements (same measruements as these screenshots)
- I will maintain course on taking new undamped P TFs followed by empirically determined G damped P, R, Y TFs as discussed with Calum, Betsy, Gabriele.
Minor tasks also include:
- Re-checking electronics with Fil given new results
- Checking with Erik/CDS why the GDS overflow page failed (I only have the state word to check for overflows now)
- Re-checking magnet polarity (though again the offset test kind of shows the minus math is right in the coil output page. This can be done quite easily though if we do end up concluding that some OSEMs are unhealthy and need to be swapped/checked for OL Electronic Noise Spectra
Updates incoming.
