The 83592A manual page 3-16 shows the power meter recorder output (0V at zero input, 1V full scale according to the HP 432A manual), directly connected with a BNC.

Page 8-28 under the heading of A4 ALC assembly mentions an inverter circuit for the power meter j out. This makes sense when considering the internal ALC has a negative detector diode.

Page 8-94 mentions crystal detector voltages of -10 to -200 mV and power meter inputs of 0-1V range.

It mentions signal routing when [MTR] power meter is selected.

Plug in a BNC cable to the recorder output. Hit the [MTR] button to select power meter ALC.

I'm using the 3325B service manual on Keysight's website from March 1990. Specifically the schematic right after page 8-F-4 which is Figure 8-F-2 Fractional N Analog A21 on PDF page 130. CR3 and CR4 are facinging each other at the end of the Phase Comparator and Integrator. I did get Anode and Cathode reversed. The cathodes connect while the anode of CR3 connects to Q3 of the Phase Comparator while CR4 anode connects to several things including R71 from the Integrator.

For CR3, the anode connects through Q3 and Q4 as you say. The most direct route to power is the collector-emitter junction of Q3 to?+5V (I called it?+15V as my copy is very hard to read).? Q3ce does have a resistance but in the worst case (failed transistor) it will be a direct short to +5V. The most direct path to power for the cathode of CR3 is the source-drain junction of Q18 (I think) but that is only?+V which CR3 will block. The next best route is through the source-drain junction of Q10 (I think) and then through the dual collector-emitters of U2A then U30A (another guess) which is a 1150 ohm resistor to -15V. So the worst case looks like?+5V -|>|---/\/\/\/\--- -15V. That is 20V across 1150 ohms or 17.4mA. To calculate the power dissipation across CR3 (not the power flowing through CR3 which is 20V * 0.0174A or 348mW) we need to know the voltage drop from 17.4mA at the operating temperature of the diode in the worst case scenario. This is different for each diode so we need to pick one and see. Let's use the 1N4448. It is not the best choice but has the most detailed datasheet. We will use a reputable version from Vishey. The rated power dissipation is 440mW @ 45C which is the maximum amount of heat the diode can eject continuously in normal circumstances. We only care about power that is converted to heat which comes from some kind of internal resistance. The resistance comes in the form of a voltage drop across the diode. You often hear that silicon diodes have a 0.6V drop and Schottky diodes are somewhere around 0.35V but that number changes with current and temperature. Page 2 Figure 2 of the 1N4448 datasheet is a forward current vs forward voltage graph. Forward voltage is the voltage drop at a given forward current. We can approximate 17.4mA with the 1st line above 10mA. We find the forward voltage drop to be about 0.8V. Now 0.8V at 17.4mA is 14mW which is the worst case power dissipation from CR3.

Doing the same for CR4 we will use the same cathode connection of -15V through a 1150 ohm resistor. For the anode side we follow R72 (1k) through W2 (assuming jumpered in the test position which is a valid?non-fault condition) past TP10 through the collector-emitter junction of Q9 to +15V. If Q1 FET were to fail we could also see a direct short to +15V. So max operational current is 30V (+15 to -15) through 1150 ohms (fault) or 2150 ohms (non-fault) for 26/14mA. The forward voltage drop of 1N4448 at 26mA is maybe 0.85V. 0.85V at 26mA = 22mW power dissipation.

Modern silicon diodes exceed the specs of many of these old Schotkky diodes when they were used mainly for their low forward voltage to achieve a lower power dissipation. I hope that helps.

The 1N5711 is a matched part for 5082-2xxx series. The HP datasheet can be found here

It is probably not the same part as the 5082-5509. It seems Broadcom deleted any reference to the part when they took over ownership. So the most we know is:

HP 3325B A21CR3/4

HP-PN 1901-0518 manufacures by HP Div 5 MSD (specification PN)

MFG-PN 5082-5509 (real part PN)

NSN 5961-00-430-6819 (US military universal PN)

70V reverse breakdown voltage

Schottky small signal diode hot carrier

In this circuit CR3 and CR4 are used to sum the inputs from the phase comparator and the integrator. The anode of both connect to -15V through a 1k15 resistor. The cathode of CR3 (phase comparator) connects to +15V (26mA). CR4 (integrator) connected to +15V through 1k ohms (14mA). At 26mA the 1N5711 drops 0.7V for a total power dissipation of 18mW.

Let's figure out what is needed based on the circuit.

The phase comparator and integrator operate at 100kHz. The speed or CR3/4 equate directly to the maximum resolution of the phase comparison and integration. So faster is better.

100KHz or 10μs = 360° or ±1Hz (very bad)

1MHz or 1μs = 36° or ±100mHz or 1 decimal place

10MHz or 100ns = 3.6° or ±10mHz or 2 decimal place

100MHz or 10ns = 0.36° or ±1mHz or 3 decimal place

1GHz or 1ns = 0.036° or ± 100μHz or 4 decimal place

10GHz or 100ps = 0.0036° or ±10μHz or 5 decimal place (excellent)

This instrument has a resolution of 1μHz and accuracy of ±5mHz. 10ns might work but 1ns is better.

All together you are looking for:

Vrrm > 30V

If > 30mA

Pd > 800mW

Trr < 5nS

Irl = lowest available for best noise performance?

The HP 1N5711 is 70V 15mA 430mW 100ps diode which wouldn't survive the worst case scenario.

1N5711-1/6857-1/6858-1 are 33mA 200mW diodes but are expensive obsolete parts. Oddly Microsemi doesn't specify reverse recovery time or V/s. However 2pF at 1MHz normally equates to 1-5ns which is good enough.

1N4448 100V 200mA 4ns 4pF 500mW 780mV@28mA(22mW)

BAW76 50V 300mA 2ns 2pF 500mW? 1V@100mA(<100mW)

1N4151 same as BAW76 with slightly higher forward voltage drop so slightly higher power dissipation but still well under limits

1N916 same as 1N4448 but with worse power dissipation

1N6638/42/43 150/100/75V 300mA 4.5/5.0/6.0ns 2.5/2.8/2.8pF ~900mV@28mA(25mW)

All of these should work. With BAW76 being the best. If you look harder you can probably find some 1ns or even 100ps diodes?