Coil current monitor testing
A series of brief offline, supplemental experiments conducted to verify the trustworthiness of the field coil current Rogowski coils.
Motivation
It was observed that during the initial post-upgrade run of LTX-β, a significant discrepancy appeared between the top and bottom orange coils, such that the top orange coil appeared to carry only about 2/3 the current in the bottom orange coil (Fig. 1). Since the coils were connected in series, this posed a significant question about the field coil current diagnostics.
The unaccounted-for current amounted to roughly 1kA in some shots. If the flaw were a resistive/partial short, it could not be an internal coil short that would only reduce impedance but maintain current-per-winding; it would have to be a short to ground, somewhere between the two diagnosed orange coil leads. Also, the difference in currents seems to be a scale factor, whereas the relative impedance (and therefore the relative proportions of current shared) between the coil(s) and a simple (i.e. purely resistive but non-inductive) short should be expected to change with frequency. With all this in mind, a real difference in current seemed highly improbable.
In order of decreasing likelihood, this implied the following possible failure sites:
- A bug or typo in the calibration records in the tree
- A change in the effective gain (e.g. due to change in the wrapping) of the Rogowski coils used to measure field coil currents
- A change in the gain of one or both integrator channels
- A measurement error in the digitizer input
- An actual difference in current flowing through the Orange Top and Orange Bottom coil leads
After thoroughly reviewing the integrator calibration records and scripting within the tree, it seemed that the failure was not in site #1, and must be in hardware (#2-5).
Procedure
A length of 14 gauge wire was run from the magnetics rack to the Hoffman boxes and passed through each of the coil current Rogowskis for the Orange Top and Bottom, Internal, Red, and Blue coils. A current was driven through this wire and a 1.25Ω ballast resistor using the Crown XLS 2502 audio amplifier driven by a signal generator, allowing easy control over output current and frequency.
Results
All Rogowskis tested were within 5% of agreement with the Pearson transformer.
Initially, only the coil current Rogowskis were tested, with the Rogowski output directly connected to an oscilloscope, with current measured independently by a Pearson transformer, effectively testing failure site #2 alone. When this showed little or no difference between the Orange Top and Orange Bottom Rogowskis, the signal was routed through the appropriate integrator channels to the oscilloscope to test failure sites #2 and #3 together.
Still showing no significant difference, the integrator outputs were routed to the appropriate digitizer channels in order to fully test failure sites #1-4 concurrently. The signal generator, set in burst mode, was triggered by the same pulse generator which gates the integrators. Current was measured simultaneously on a Pearson transformer connected to Neos channel 70.
Notable shots include:
100354: Wire is passed 4 times through each Hoffman box based Rogowski, current is driven at 1kHz. Shows 2% difference between Orange Top and Bottom.
100382: Wire is passed once through each of the above Rogowskis at driven again at 1kHz, with similar results to shot 100354.
100388: Wire is instead connected to the Orange Top and Bottom coils only, inside the Hoffman box, so that current is driven through the coil and its leads rather than a secondary conductor. This requires a lower frequency (300Hz) to maximize current, due to the coil's inductance. Orange Top and Bottom coil current measurements differ by less than 1%.
Since the matching signals are measured by applying a known, independently-verified current, and are passed through the appropriate integrator and digitizer channels and then processed automatically by the calibration function in the tree, it seems that the difference in currents seen in Fig. 1 was either a transient flaw which was unwittingly fixed, or something else up-stream of the Rogowski coils.
Future Work
As of Jan. 30th, 2019, no shot has been created with Robicon power since the testing was performed and found no difference between the Orange Top and Bottom Rogowskis. If the historical discrepancy (Fig. 1) disappears on all subsequent shots, there may have been a transient error in hardware or code that was unwittingly or quietly fixed. If the discrepancy persists, however, the cause must be understood in order to inform the inclusion of coil currents in equilibrium reconstructions, which will be crucial for future research goals.
Related Pages
Crown XLS 2502 - the audio amplifier used in testing
Coil current monitors - the Rogowski coils tested
Equilibrium field coils - descriptions of the field coils
