order of magnitude bias for the diode mixer

I want to replace a 1N78C diode with one of BAT15-03W or SMS7630-079LF in a mixer. This mixer has bias settings for more frequencies. I would like to know what bias current I need for these Schottky diodes. Since the original diode was melted I have no idea about the magnitude of this bias. Is it the order of 0.5mA or should it be chosen so that the diode enters a bit in conduction? The device's manual says the setting is for maximum sensitivity, but I do not know if the bias circuit has failed.

Your post is confusing:

If the original diode 'melted' there is something seriously wrong with the bias circuit and the current is magnitudes higher than it should be.
If 0.5mA is flowing, the diode IS in conduction.

0.5mA seems a bit high to me but it depends on the level of LO drive. You should adjust it as recommended for maximum sensitivity. The current requirement should be quite low and the voltage somewhat less than 1V.

Brian.

Meanwhile, I replaced the mixer diode with a B-E junction from a Ge transistor. Measured on the DVM diode indicates 150mV. So it means 1mA per junction is 150mV drop voltage . This junction mounted in place of the original 1N76C diode receives bias as follows:
10Mhz-1,8Ghz ........... 175mV / 0.9mA
2,5-5Ghz ..................... 112mV / 0,108mA
5-9Ghz ........................ 174mV / 0.880mA
9-12Ghz ....................... 249mV / 4,3mA
12-18Ghz ...................... 242mA / 3,8mA
Obviously, during this time OL is in operation.
The bias source is a UA741 in repeater configuration and there is no limitation resistance on the output. Maybe at most an RF shock inside the mixer block but does not have measurable resistance.
This UA741 which has a differential supply of -15V and -18V according to the diagram and in fact flows out with no load output between 0.12V and 4.2V depending on the frequency range. Each bias can thus be adjusted between 1V-4.2V with no load . Calculations according to values from the schematic diagram check this range. I can not calculate what is happening with the diode connected since the UA741 resistance is several tens of Ohm (under 75) and the diode is nonlinear I / U.

Mixer diodes usually are not DC biased. The diodes are opened by the LO signal. The diodes should behave like ideal switches.
Meanwhile, the RF detector diodes should be DC biased, to increase detector sensitivity and linearity performances.

Not right. 75 ohm is DC open loop output impedance, closed loop output impedance of buffer amplifier is 75ohm /loop gain, below 1 mohm. Do you prevent RF from loading the OP, otherwise the effects or actually hard to predict.

Apart from this point, I completely agree with vfone that mixers should be operated with high level LO and no bias.

test systron-donner762_320 2.rar
Indeed, the output impedance is very small. If I do not get wrong input impedance increases. I've reviewed the mixer and it seems like I was wrong about the bias injection. Between the bias point of mixer block and the cathode of the diode can measure 1045 Ohm. I mounted a BAT15-03WE diode in the 1N76 diode cartridge and it looks like the mixer works as before. For this operation I mechanically processed the cartridge. In a few days I will receive a new 1N76 diode and I will compare it especially to the noise. You are right with the bias in the idea that as it rises it decreases the level at the output of the mixer. It works best without bias or anyway up to a maximum of 200mV per diode. I do not know what was in the engineers' head 40 years ago. I attach the schematic diagram of this bias for mixer. I have tested with a generator (8-10dBm & 26dB fixed attenuator & 20-50dB variable attenuator) , up to 4Ghz and the harmonics appear strong and clean beyond 12 Ghz with the BAT15 diode. Maybe I will try to dismantle this mixer, although I'm afraid of being disturbed.