Feedforward Details

This section can be skipped by the casual reader.

This section details the fast feedforward hardware and explains how it is interfaced to the control system. The hardware lives in two CAMAC modules: one in DR12 and the other in EP01. Figure 3 depicts the logic.

Except for one summary display, the control system has no special purpose software devoted to feedforward. To the control system it looks like a collection of standard hardware devices.

1. There are DAC channels. These are not actual Transiac DAC channels; rather the feedforward hardware emulates several DAC channels. These DACs are used to set various offsets and gains. The control system treats these as AMPLs (magnet-like devices). To get to the touch panels which control these devices, go from INDEX to FEEDBACK SYSTEM INDEX to the FEED FORWRD EP01 (or DR12) panel. From there it is possible to look at and change the DAC settings.
2. For each DAC channel there is a SAM channel. These are true SAMs, not emulated. The feedforward module sends several analog signals to a couple of SAM CAMAC modules. These SAM channels act as the analog read-out that the magnet software needs to go with a DAC to form an AMPL.
3. There are other SAM channels used to read out voltages for diagnostic purposes. For example, the estimate of the total beam loading is available. The control system treats these as analog status (ASTS) devices. From the above touch panels you can hit the ANALOG SYSTEM DISPLAY button on either of the above panels to view the appropriate signals. You can also hit the ANALOG SYSTEM DIAG button to go to that diagnostics panel and get more information (such as history plots).
4. There are signals coming from PDUs which are used to trigger various sample and holds at the right time. These are treated as TRIGs by the control system and can be controlled from the FEED FORWARD DR12 panel.
5. Finally some voltages are fed into Gated ADC (GADC) modules. These modules can be read out with the BPM software in a pulse synchronized manner along with the reading of other BPM-like devices. They are treated as ARRYs by the control system and can be viewed with the GADC display on the BPM panel. These readouts provide a very useful tool for setup and diagnostics. Buffered BPM data acquisition can be used to read the beam intensity from a toroid, the ARRY channel which should correspond to that intensity, and BPM.EP01.185 in the extraction line which indicates the energy error. Looking at the correlations of these signals indicates how well feedforward is performing. In particular there should be essentially no correlation between the energy and intensity if fast feedforward is functionally properly. After all, its sole purpose in life is to remove this correlation.

With the above description of how the input and output is handled, interpretation of the block diagram should be fairly easy. Signals from the damping ring BPMs come in from the left. These enter specially modified BPM Processor (BPMP) modules which are gated about 5.3 msec before beam extraction by a TRIGger from a PDU. Each amplitude signal comes out of the BPMP, has an offset subtracted (this removes various electronic offsets), and is multiplied by a gain. This gain basically contains the conversion from beam intensity to expected beam loading. Because the settings of the offsets and gains is done empirically, the BPMP modules should not typically be swapped except by an expert. The signals for the three beams are then summed. All of the above is done in the DR12 feedforward CAMAC module.

The resulting sum which represents the total expected beam loading is then sent in digital form to the EP01 feedforward CAMAC module. This module then calculates the required phase settings for sectors 17 and 18. The result is used to set the solid-state (fast) phase shifters which are in the 476 MHz line just upstream of the sub-boosters.

The exact calculation of the phases needed to obtain a certain energy gain requires several arc-tangents and is too complex to implement in analog hardware. Instead, the following is done.

1. Feedback (running in the EP01 micro) calculates the phase settings needed to obtain feedback's desired energy contribution, 1 GeV more than that, and 2 Gev more than that. (This is a simplified explanation. It actually uses a more complex algorithm to pick 3 energies.)
2. Feedback then fits a three term polynomial (quadratic) to the phase as a function of energy curve for each of the two sectors.
3. Feedback writes these six fit coefficients to DACs in the EP01 feedforward CAMAC module.
4. The feedforward hardware then uses the coefficients to calculate the phases from its estimate of the beam loading.

The net result of all this is that the sector 17 and 18 phases are set according to the desires of both feedback and fast feedforward. Feedback's input comes via the quadratic coefficients it writes to the DACs. In fact the offset term of this quadratic can be thought of as feedback's phase request. Feedforward's input comes via the sum signal which is used to evaluate the quadratic.