Proposal for a 5kW, 0

1 downloads 0 Views 182KB Size Report
Oct 4, 1999 - FOR THE PS TRANSVERSAL DAMPER. M. Paoluzzi ... An important amplifier characteristic is the harmonic distortion. If a careful design and.
1

EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH ORGANISATION EUROPEENNE POUR LA RECHERCHE NUCLEAIRE CERN - PS DIVISION

PS/RF/ Note 99-08

PROPOSAL FOR A 5 KW, 0.1-50 MHZ AMPLIFIER FOR THE PS TRANSVERSAL DAMPER M. Paoluzzi

Geneva, Switzerland 4 October 1999

2

Proposal for a 5 kW, 0.1-50 MHz amplifier for the PS Transversal Damper 1. Introduction In the context of the PS upgrade for LHC a new damping system is foreseen to reduce the beam transverse emittance in both the horizontal and vertical plane. New wide-bandwidth (~15 MHz), single ended, low characteristic impedance (6.25 Ω) kickers are being designed1 for this purpose. Two of these devices will be used on each plane. The driving power required for each kicker is 5 kW and, since the amplifiers will be included in a feedback loop, the bandwidth should extend well beyond the working range of 0.4 – 15 MHz. In this paper a scheme, capable of providing the required power over the frequency range 0.1 – 50 MHz, is proposed together with the results obtained on some basic building blocks. 2. Basic configuration Rf power coupling, splitting and impedance transformation techniques, used in wide-band rf power amplifiers, usually imply the use of transmission lines to improve the high frequency behaviour, and ferrites to extend the low frequency response. Ferrite losses vs induction is a strongly non-linear characteristic so that to limit them, one would tend to increase either the core dimensions or the number of transmission line turns. Unfortunately, this would result in a deterioration of the high frequency response and a compromise between power handling capabilities and frequency response is usually required. For a given power, since induction is proportional to the rf voltage, a wider frequency response can be obtained on a low load impedance. In this respect, the 6.25 Ω characteristic impedance of the kicker under development is a good value and, among other reasons, it has been chosen to simplify the amplifier development. It allows to obtain the required 5kW rf power by coupling eight 50 Ω, 625 W modules which, at this power level, can have a frequency response spanning from less than 100 kHz to close to 100 MHz. The 625 W modules, can in turn, be made by coupling four 160 W units built with modern rf power mosfets. The proposed configuration is shown below.



1

Under the responsibility of J.L.Gonzalez



3

3. The 160 W module The 160 W module is the basic building block composing the 5 kW unit. It is a two-stage amplifier which provides 27 dB gain over a frequency range spanning from 40 kHz to 100 MHz. Both stages use rf power mosfets mounted in push-pull configuration. The devices are over-dimensioned for improved reliability. This is an important parameter because of the large number of units required to obtain the design output power. Water cooling of the active devices, as well as the output transformers, has been preferred to air cooling because of the best efficiency and to limit the module dimensions. The circuit diagram is shown below.

The following plots show the measured transfer functions at low and full power as well as the group delay, which is almost independent from output power.

160W Module - Pout=0.25W Phase

32.0

250

30.0

200

28.0

150

26.0

100

24.0

50

22.0

0

20.0

-50

18.0

-100

16.0

-150

14.0

-200

12.0 0.010

0.100

1.000

10.000 MHz

100.000

-250 1000.000

Deg

dB

Gain

4

160W Module - Pout=160W Phase

32.0

250

30.0

200

28.0

150

26.0

100

24.0

50

22.0

0

20.0

-50

18.0

-100

16.0

-150

14.0

-200

12.0

Deg

dB

Gain

-250

0.010

0.100

1.000

10.000

100.000

1000.000

MHz

160W Module Group Delay 10000

ns

1000

100

10

1 0.010

0.100

1.000

10.000

100.000

1000.000

MHz

An important amplifier characteristic is the harmonic distortion. If a careful design and construction of the transformers can achieve good cancellation of even harmonics, odd harmonics depend on the device forward characteristic and the selected working point. Harmonic distortion has been measured at different bias currents; data are plotted below and will help in the choice of the working point.

3rd Harmonic

2nd Harmonic 0.5A

1A

2.5A

5A

0.5A

1A

2.5A

5A

0

0

-5 -10 -10 -15 dB

dB

-20 -30

-20 -25 -30

-40

-35 -50 -60 0.10

-40

1.00

10.00 MHz

100.00

-45 0.10

1.00

10.00 MHz

100.00

5

Although adequate cooling of the output transformer is provided, due to the ferrite losses, the amplifier can be operated in CW only between 300 kHz and 30 MHz. At frequencies below 300 kHz, CW operation must be limited to 5-15 minutes depending on the frequency. Similarly, at frequencies above 30 MHz the time limit lies between 15 and 30 minutes. 4. The 4-ways splitter and combiner A four-port hybrid configuration has been chosen for both the splitter and the combiner of the 625 W module as shown below. For the splitter, ports 1 to 4 are the outputs and port 5 is the input. The opposite obviously applies for the combiner. For the input splitter T1 to T4 are made of 6 turns of 50 Ω semi-rigid coaxial cable (UT34) wound across a 3E5 ferrite ring (16.5x9.0x6.7 mm - ODxIDxH). The 1 to 4 transformer uses the same ferrite ring and has the same number of turns but uses a 25 Ω UT34 semi-rigid coaxial cable. For the output combiner T1 to T4 are made of 4 turns of 50 Ω coaxial cable (RG316) wound across a N30 ferrite ring (41.8x22.5x17.2 mm - ODxIDxH). The 1 to 4 transformer uses the same ferrite ring with two times six turns of RG316 coaxial cables wound in parallel. The Bmax in the transformer is 300 mT at 50 kHz for 625 W. In both cases good splitting/combining response is obtained and the ports isolation is such that operation of the three 160 W Modules out of four should be possible. All measured data are plotted below. Splitter Transfer Function Phase

-4.0

250

-5.0

200

-6.0

150

-7.0

100

-8.0

50

-9.0

0

-10.0

-50

-11.0

-100

-12.0

-150

-13.0

-200

-14.0 0.010

0.100

1.000

10.000

100.000

Deg

dB

Ampl.

-250 1000.000

MHz

S p litte r P o rts Is o la tio n -1 0 -1 2 -1 4

dB

-1 6 -1 8 -2 0 -2 2 -2 4 0 .0 1 0

0 .1 0 0

1 .0 0 0

1 0 .0 0 0 M Hz

1 0 0 .0 0 0

1 0 0 0 .0 0 0

6

C o m b in e r P o r t s Is o la t io n 0 -5

dB

-1 0 -1 5 -2 0 -2 5 -3 0 0 .0 1 0

0 .1 0 0

1 .0 0 0

1 0 .0 0 0

1 0 0 .0 0 0

1 0 0 0 .0 0 0

M Hz

C o m b in e r T ra n s fe r F u n c tio n Phas e

1 0 .0

250

8 .0

200

6 .0

150

4 .0

100

2 .0

50

0 .0

0

Deg

dB

Am p l.

-2 .0

-5 0

-4 .0

-1 0 0

-6 .0

-1 5 0

-8 .0

-2 0 0

-1 0 .0 0 .0 1 0

-2 5 0 0 .1 0 0

1 .0 0 0

1 0 .0 0 0

1 0 0 .0 0 0

1 0 0 0 .0 0 0

M Hz

5. The 625 W Module The 625 W module has been assembled as shown in the proposed configuration. A fifth 160 W module has been used as driver. Although only a fraction of the power capabilities are used, it provides the required gain and allows only one kind of amplification module to be used. The overall behaviour of the assembly meets the expected performances but requires a frequency compensation network to improve the response at high frequency. It is composed of a simple R-C high-pass network inserted between the driver stage and the splitter. The results are shown below for a final stage bias current of 3 A per mosfet and a supply voltage of 40 V. 625W M odule - P out=1W P hase

56.0

180

54.0

135

52.0

90

50.0

45

48.0

0

46.0

-45

44.0

-90

42.0

-135

40.0 0.010

0.100

1.000

10.000 MHz

100.000

-180 1000.000

Deg

dB

G ain

7

625W Module - Pout=625W Phase

56.0

180

54.0

135

52.0

90

50.0

45

48.0

0

46.0

-45

44.0

-90

42.0

-135

40.0 0.010

0.100

1.000

10.000

100.000

Deg

dB

Gain

-180 1000.000

MHz

625W Module Group Delay 100000

ns

10000

1000

100

10 0.010

0.100

1.000

10.000

100.000

1000.000

MHz

Output Signal Harmonic Content 2nd

3rd

0 -10

dB

-20 -30 -40 -50 -60 0.10

1.00

10.00

MHz

100.00

8

6. The rf input, 8 ways output combiner, protection, monitoring and control circuitry Coupling together eight 625 W units will require, of course, input signal distribution and an output combiner. These two items have not yet been developed but should not present too big problems. The total gain of the amplifier has been set at 50 dB. Considering that the input and output impedance are different (50 Ω and 6.25 Ω), this means that for the full power we need at the input 0.8 V peak (6.3 mW or 8 dBm on 50 Ω). For full power the 625 W modules need 0.5 V peak so that a simple 8 channels current buffer could do the job. Since the output load impedance is 1/8 of 50 Ω, no transformer is required at the output. Using a hybrid coupler each winding will handle a maximum voltage of 250 V peak. This corresponds to twice the voltage handled by the equivalent winding of the 4-ways combiner described above. Therefore using twice the number of turns will limit the losses in the ferrite to the same level, which proved to be acceptable. The cable length will be obviously longer and thus the frequency response limited but still higher than the desired 50 MHz. Testing of the amplifier on different load conditions will show whether an active protection against load mismatch is required. Since the final stage mosfets are well overdimensioned, this will probably not be the case and protections will then be limited to thermal survey of critical parts and power supply voltage. Monitoring will include rf output voltage and, for each 160 W module, thermal protection status, dc supply voltage and current. Control circuitry will be limited to local and remote ON/OFF control plus remote indication of failing elements. Reduced power operation with a broken 160 W module per 625 W unit might be possible. 7. The power supplies Eight 40 V, 60 A sources, each supplying one 625 W module, plus a 20 V, 60 A source common to all modules installed in a 5 kW unit are required for CW operation. This solution can be implemented with industry standard 3 kW power supplies and has the advantage of de-coupling the DC sources. For operation with low duty-cycle a cheaper solution could be implemented using a single source and a capacitor bank. To make this alternative attracting not only the duty-cycle should be low but also the rest current should be kept as low as possible Using a single 50 V, 60 A power supply, a 800 mF capacitor bank and setting the mosfets rest current to 0.5 A, the full power can be delivered for 10 ms every second. One can see from the data plotted in chapter 3 that the third harmonic distortion increases for low bias current values and this parameter should also be considered. 8. Mechanical layout The 625 W modules are independent, shielded units 265 mm high, 104 mm wide and 500 mm long. Four of them can be installed side by side in a standard 19” rack. The input signal distribution, eight ways power combiner, monitoring, protections and control circuitry can be housed in a 3 units Europa chassis. Considering that a 3 kW switching power supply is typically two units high, the 5 kW amplifier will need 33 to 35 standard 19” units. A possible layout is shown below.

9

Conclusion Although a certain amount of work is still needed for the development of missing parts and finalisation of the existing prototypes, the production of a 5 kW, 0.1-50 MHz amplifier seems to be feasible. Decisions should be made on what kind of operation is required, CW or pulsed, so as to match the cooling hardware and power supply to the needs. In case the proposed amplifier would be retained, pre-production of two final version 625 W modules could be done within 1999, and a first 5 kW test unit in year 2000.

Proposed Mechanical Layout