12" Cyclotron RF System


The RF system for the 12-Inch cyclotron was quite straight forward to implement. Most of the questions and problems were resolved with the 9-Inch prototype. There are two "modes" of operation, which are defined by the maximum power used. For student operation, typically no more than 50 watts of RF power is generated, yet results are more than satisfactory. The other "mode" is not so restrictive on RF power limitations. Currently a maximum of 2,500 Watts can be generated, however special cooling procedures must be followed.

The system still uses the MOPA (master oscillator, power amplifier) ideology, at present there is no feedback tuning system. There exists the future possibility of doing so (a potential student project). The precise frequency is generated by an HP8656B signal source, that has a range of 100kHz to 990MHz and a resolution of 10Hz. It's signal (pure sine wave) directly drives an ENI NMR-300L solid state, broad band HF amplifier in CW operation. The amplifier's maximum output is 50 watts CW, which, in the student "mode" is then directly used for acceleration. Otherwise, another amplifier can be brought into operation, the PlasmaTherm HFS2500, which uses three stages of tubes to obtain a full 2.5kW. The HP8656B drives the first tube, a 12BY7A to drive the buffer stage, which consists of one 6146A, which in turn drives the PA stage, an Eimac 5CX1500A pentode. This is unlike the broad band solid state amplifier which does not require tuning. Slight changes in the load values or the cyclotron's resonant frequency will require a manual retuning of the final stage power amplifier. Larger changes in frequency (50 KHz) will definitely require retuning of the first and buffer stages as well.

rf_source

Fig.1 HP8656B RF Signal generator and associated monitoring equipment.

new_rf

Fig.2 The 2500watt amplifier shown next to the cyclotron.

PA_off

Fig.3 The front of the 2.5kW RF generator, unit is OFF.

 

PA_on

Fig. 4.The front of the 2.5kW RF generator, unit is ON.

RF_cabinet_backside

Fig.5 A rear view of the 2.5kW cabinet, the plate supply is housed in the bottom chassis.

PA_backcover_on

Fig.6 The rear of the RF cabinet, note quick releases.

 

PA_backcover_off

Fig.7 Inside of the RF cabinet, DC HV is on green cable, note large output tank coil.

2000watts_small

Fig.8 The record input power level was 2000 watts, about 18kVp-p developed on the dee.

The output power in either mode is run through a Bird dual element directional coupler for power measurements, both forward and reflected. The forward and reflected powers are recorded by the DAQ and displayed at the operators computer. The operator can adjust the frequency at the computer controls to minimize reflected power. This has a significant effect on beam intensity, thus RF power is one of the parameters that can be tuned for maximum beam. Future improvements would most certainly include an automated control base on RF power considerations.

After the final amplification, the RF is brought to the Cyclotron from the control rack through a length of RG-213 coaxial cable. The RF input power has another connection point at the terminal panel on the rear of the cyclotron table. It then proceeds to the RF matching box. The matching box is located at the rear of the magnet, as close as physically possible to the chamber, and houses a large inductor and coupling loop.

rf_sys0

Fig.9 Inductance and RF input coupling box (brass colored with HV warning) mounted at rear of magnet.

The RF matching box houses the lumped inductance which provides the tuning to bring the chamber's capacitance into resonance at the ion revolution frequency. Additionally housed is the coupling, or matching, of the RF input power to the cyclotron resonant circuit. This is completed by the mutual inductance of the cyclotron's inductance in proximity to a coil of larger diameter and of approximately three turns. Electrical taps were used to empirically locate the 50-Ohm loading point of the outer coil. It is interesting to note that the effective length of the outer coil requires no more than one half a turn to properly match.

rf_sys7

Fig.10 A close up of the inductor and coupling loop (outer).

As described in more detail on the cyclotron chamber page, the dee is supported by a 3/8-inch copper rod (the dee stem - Fig.10) that petrudes a vacuum tight ceramic feed-through. The copper dee stem continues into the matching box and is electrically connected to the "HV" side of the coil. The RF current runs over the surface of the dee stem to produce the voltage on the dee. Return RF currents travel inside the chamber lids, chamber walls, and then to an external copper shield the co-axially covers the dee stem and ceramic feed-through. The mixture of aluminum lids, stainless steel chamber and compression type fittings are not the optimal for RF power efficiency. In larger cyclotrons, where space is more permitting, this problem is circumvented by use of copper plates and flexible sheets to provide a less resistive path back to the matching or oscillator box. As mentioned above, the Rutgers cyclotron utilizes at most 2000 watts of RF power, so these inefficiencies can be tolerated. Interestingly, after a lengthy period of running the oscillator, the cyclotron chamber does warm up considerably, there are locations that are unpleasant to touch.

rf_sys3

Fig.11 The dee, portion of the dee stem, and the copper dee stem shield lead to the right.

dee_stem

Fig.12 dee stem, Ceramic Vacuum Feed-Through, and dee. The Copper shield is in background.

rf_sys4

Fig.13 Physical arrangement of RF matching box, dee stem connection, dee Stem shield, and Cyclotron chamber in mounted position. (tube inside box at left is for dee voltage measurements)

rf_sys1

Fig.14 Note connection of nylon tubing (yellowish white) for HV cooling oil.

Since the physical geometry determines the value of the cyclotron inductance, it is important to ensure stability once the desired frequency (shape) is found. The choice of material used to wind the inductance was quite deliberately 1/4-inch copper refrigeration tubing. A large surface area reduces the ac-resistance, and the hollow channel allows a high voltage cooling oil to be pumped through. The need for cooling arises from inefficiencies in the coil where some of the RF power is lost as heat. At low powers (student mode), the heat generated dissipates quickly, however at high powers, (in excess of 100 Watts) the generated heat begins to cause thermal expansion in the copper coil. As a result of this expansion, the shape of the inductor changes, causing a shift in the resonant frequency and many undesirable effects. One can see two white tubes leading to the inner inductor in the RF matching box, these tubes bring the HV oil through without providing a short to ground. Other than the short conductive breaks in the RF matching box, all tubing is of 1/4-inch copper refrigeration tubing.

The cooling circuit is simple: a pump, water-cooled heat exchanger, RF coil, and finally a reservoir before returning to the pump.

rf_sys5

Fig.15 HV oil reservoir tank for RF cooling.

rf_sys6

Fig.16 HV oil heat exchanger.