G.C. Harper and T.D. Van Wechel

The 13/12 buncher required the construction of a 13/12 clock generator as well as changes in the way that the LE buncher is phase locked to the linac. The original LE buncher clock generator divided the linac frequency of 148.8 Mhz by three to 49.6 MHz. The LE buncher operates at the linac frequency divided by 12, or 12.4 MHz, producing a bunched beam that harmonically excites the resonant phase detector near the entrance to the HE buncher. The resonant phase detector produces a 49.6 MHz signal. The signal from the resonant phase detector was compared to the LE buncher clock to phase lock the bunched beam from the LE buncher to the linac clock. The frequency of the phase shifted signal was divided by 4 to 12.4 MHz to drive the LE buncher. The problem with this method was that there were 4 phase relationships between the LE buncher and the linac clock to which the LE buncher could phase lock.

The LE buncher clock generator circuit was changed to divide the frequency of the linac clock by 12 so that the LE buncher clock is now 12.4 MHz. The 12.4 MHz clock is now frequency multiplied by 4 to produce a 49.6 MHz reference using two frequency doublers and this signal is compared in phase to the signal from the resonant phase detector. The circuit was modified so that the corrections in the buncher phase are now done by phase shifting the signal at 12.4 MHz instead of 49.6 MHz as was done previously. Now there is only one phase relationship between the LE buncher and the linac where phase lock can occur. There are still 4 phase relationships between the resonant phase detector and the 49.6 MHz reference where lock will occur but the 12.4 MHz buncher signal is not affected by this.

The simplest way to generate the 13/12 clock for the HE buncher is to mix the linac frequency and the LE buncher clock using a double balanced mixer (DBM). This generates frequencies of 11/12 f and 13/12 f, where f is the linac clock frequency, at the output of the DBM. The problem is that since 11/12 f and 13/12 f are relatively close in frequency it is difficult to provide much attenuation of the 11/12 f term without introducing phase shift in the 13/12 f term. The concern here is that the phase shift of the 13/12 f term will not be stable with time or temperature. The solution to this problem is to use an image canceling mixer. The image canceling mixer consists of two DBMs that have their outputs summed together with a power combiner. The RF input of DBM 1 is the linac clock and the RF input of DBM 2 is the linac clock in quadrature phase. The LO input to DBM 2 is the 12.4 MHz clock and the LO input of DBM 1 is the 12.4 MHz clock in quadrature phase. When the outputs of the two DBMs are summed the 11/12 f terms cancel and the 13/12 f terms add constructively.

We investigated the jitter in the LE buncher waveform that has been present for several years. This jitter was synchronous to the reference DACs being updated by the DECK computer. This was traced to the way that the DACs were grounded. The DACs are located in a chassis separate from the controllers, but were grounded only at the controller chassis to avoid ground loops. However, it was found that also grounding the DACs to the DAC chassis eliminated the computer noise that was causing the jitter.