The HP8709A is a synchronizer for the HP8690 series of sweepers

HP8709A use with HP8690 by Rien Eradus PA0JME, [email protected]

An old acquaintance, Bert Bruntink living in DL for some decades, obtained a HP8709A which he wants to use to phase-lock his HP8690 sweepers. This lead to some thoughts on this subject, hence this story.

The HP8709A is a synchronizer for the HP8690 series of sweepers.

The HP8690 series sweepers can accept plug-in units with a BWO as the basic oscillator.

The helix voltage of these BWOs is rather high voltage and is swept by the mainframe using HV tubes in it’s output stage. The helix voltage can be up to 2200V.

The helix voltage is shaped with a diode-resistor network within each individual plug-in, and fed back in to the sweeper mainframe to have a linear voltage to frequency relation.

Some time ago I made an external input for the HP8690 to enable external sweeping with a low-voltage from -4 to +4 Volt from my PC DAC.

The divided down helix voltage is normally output to the mainframe’s front via BNC labeled with “SWEEP OUT”.

The low to high-voltage is done by shunting the manual sweep pot-meter with a HV-transistor current source, such that the voltage across this pot-meter changes from about 0 to 280 V.

The circuit is checked using a HP8690B with a 8696A plug-in (18-26.5 GHz)

It showed that not the entire range from 0 to 280 V is effectively used a sweep voltage.

The input voltage of -4.096 V equals a DAC value of -2048 and 4.096V equals a DAC value of 2048 and the figure shows an approximate linear behaviour from -1700 to 1300, i.e. (1700+1300)*8.192/4096 = 6V span. The effective EMK voltage is from -3.4 up to 2.6 V

Using the 18 to 265 GHz plug-in, the input sensitivity is (26.5-18)/6=-1.3 GHz/V

The minus sign is due to the negative slope of the sweep-voltage output w.r.t. the input voltage.

HP8709 interfacing

Referring to the HP8709A manual, there are two outputs of the phase-detector. The 1^st one, labeled HELIX OUTPUT, has an output impedance of 200 Ω and -12 up to +12 V output-swing. The 2^nd one labeled SERVO OUTPUT has and impedance of 70 kΩ, with -.8 up to .8 V output-swing. Obviously the SERVO O/P is .8/12 x HELIX = 2/30.

The HP8709A’s modulation sensitivity has 3 positions which refer to the HELIX output

Position	Modulation sensitivity HELIX o/p	Modulation sensitivity SERVO o/p
1-4 GHz	0.5 to 1 MHz/V	7.5 to 15 MHz/V
4-12.4 GHz	1 to 2.5 MHz/V	15 to 37.5 MHz/V
12.4-40 GHz	2.5 to 6.0 MHz/V	37.5 to 90 MHz/V

First, neglecting phase, the servo output and tune-voltage need to be added in proper proportion to each other. The mean output in the higher range is √37.5 x 90 = 58 MHz/V or 17 mV/MHz and the tuning slope is 1300 MHz/V, i.e. 770 µV/MHz. This boils down to an additional attenuation of 1/22 of the servo voltage.

Figuur 1 suggestion for a circuit to add phase detector voltage to a locally generated tune-voltage

The circuit operates as follows. The detector voltage is buffered by a voltage follower built around A2. The 8709 SERVO output is fed into “in”. The 10 turn pot-meter creates the tuning voltage and the ratio of R9 and R10 determine the approximate required 1/22 attenuation. If the circuit is to feed the proposed low to high voltage converter, a nominal output on A4 should be between -4.096 to 4.096V for the highest and lowest frequency respectively. The phase inversion of A4 together with the phase inversion of the low to high voltage converter cancel as far as control theory is concerned.

A1 and A2 are voltage followers to adjust the required top and bottom voltage for the tune pot-meter. The output voltage of A1 is positive and has to translate to -3.4 V on A4.

The gain of A4 is -1 for the tune-voltage, so A1 shall be adjusted for +3.4 V using R7.

Evidently A3 output shall be adjusted for -2.6 V using R3.

R12 feeds the low to high-voltage converter.

Low to high-voltage converter

Back in the late eighties or early nineties I built a PC ISA card published by the Dutch Elektuur magazine in May 1988 to control my swept measurements. This is an eight channel 13 bit ICL7109 ADC and a single channel PM7548 12 bit DA output with an output resistance of 10 kΩ and an open voltage from -4.096 to 4.096 Volt

The ADC is pretty slow though, about 120 ms, so you should not be in a hurry. The board also hosts a 8255 parallel IO and a 8253 counter, but I never used the latter one.The sweep-speed of the circuit below operates OK at least up to 1 kHz. At 10 kHz the op-amps appear to be too slow. You can parallel the input of the circuit with the J1 BNC connector on the instrument’s front. The trigger circuit is AC coupled and of no need if PC sweeping. The circuit at the right is the manual sweep pot-meter, which is paralleled with the transistor

I built my circuit on a piece of experimenting board as a piggyback on power-supply card A14, the input is routed to free pin 8 on the board, a shield wire is routed from connector A2-pin 44 and shield to A2-43, those are connected to the input trigger BNC. Another separate shielded wire is routed from pin 12 of A2 to the collector of the BF859. Don’t use any of the other presumed free pin’s of A14, since they are not free on the mother-board !!!Note that you cannot use the manual sweep anymore unless the input is negative biased such that the transistor is in cut-off i.e. the new sweep-input should be held below -2 Volt.At open input the emitter pot should be tweaked for 2mA collector current. The gain of the first amplifier is 0.5 x since the input resistance is 20 kΩ, i.e the sum of the PC-card output’s resistance and this amplifier input resistance, while the feedback is 10 kΩ.The lower op-amp creates a -2V bias from the -6.3 and 20 Volt supply, to enable a suitable sink for the current source formed by the right-hand upper op-amp, the transistor and it’s emitter resistor. At 4 Volt from the PC card the output of the 1^st op-amp will be 2Volt so forcing 4 Volt/1 kΩ = 4mA current. This result in about zero volts at the top of R1 pot-meter, which under all circumstances should have it’s wiper contact in the “top position”, i.e. at A2-R5’s end. At -4 Volt from the PC-card the transistor will be cut-off and the voltage at junction R5 and R1 will be max. The result is measured in the figure below. The assumption that the complete range of the manual pot-meter is used proved to be wrong, part of the range is saturated. This can be altered by changing some of the resistors in the circuit, i.e. offset and gain.

Figuur 2

Possibility of using the HP8690 EXT FM input

The HP8690 has a BNC labelled EXT FM which was considered to be used as a phase detector input too. Since the -3dB bandwidth of this input is however just a few kHz, I expect it to be too low to obtain a stable control. You can experiment with it though, but the 8709’s SERVO O/P should be used and the voltage should be additionally attenuated by 550/58 = +/- 10x using a HP8696A 18.0 to 26.5 GHz plug in. The input impedance of the EXT FM input is 100 kΩ.

The table shows some suggested attenuation values for different types. An advantage is that just simple hardware is required and there’s no need to open the HP8690.

TYPE	Ko on EXT FM MHz/V	RANGE GHz	SERVO O/P MHz/V	Attenuation
HP8691A/B	63	1.0-2.0	7-15 ≈ 10	63/10 ≈ 6
HP8692A/B	126	2.0-4.0	7-15 ≈ 10	126/10 ≈ 13
HP8693A/B	252	4.0-8.0	15-37.5 ≈ 24	252/24 ≈ 10
HP8694A/B	277	8.0-12.4	15-37.5 ≈ 24	277/24 ≈ 12
HP8695A/B	352	12.4-18.0	37.5 – 90 ≈ 58	352/58 ≈ 6
HP8696A/B	550	18.0-26.5	37.5 – 90 ≈ 58	550/58 ≈ 10
HP8697A/B	850	26.5-40.0	37.5 – 90 ≈ 58	850/58 ≈ 15