Marconi TF2431 Frequency Counter

[brian_mk]

I have almost completed my repair of a vintage Marconi TF2431 frequency counter.
This is the second example of this instrument I have aquired over the last couple of years.

See the original thread from 2020 here:-
https://www.vintage-radio.net/forum/...d.php?t=169327

In both cases MD102 Schottky diodes in the AGC detector had failed.
The original diodes are no longer available so I replaced them with FH100.
The second counter is now basically working but I still have a few questions and issues:-

First of all I question the specified input impedance.

The specification states 1Meg in parallel with 25pF.

If you look at the schematic for the input stage, that cannot be right.
The schematic is for the TF2430 but I believe the input stage is the same as the TF2431.

The input has a 1 Meg series resistor in parallel with 15pF feeding into one of the gates of a dual gate MOSFET. The gate has a 10Meg resistor to ground.
To me that suggests the input impedance is more like 11Meg in parallel with a small capacitance.

I don't have the necessary equipment to measure the actual input impedance, so I created an LTspice model. You feed the input from a voltage source with a series resistor in parallel with a capacitor and create a bode plot. You can then adjust the series resistor and capacitor to give a 6dB drop that appears flat over the frequency range of interest. When I do this I get 11 Meg in parallel with 5pF.
The actual capacitance is probably slightly higher by a few pF due to the protection diodes, strays in cables, connectors, PCB etc.

I have some other issues connected with the frequency response and stability of the wideband input amplifier. I will raise those in a follow up.

[brian_mk]

Now we come to the low frequency response of the wideband amplifier...

Both the TF2430 and TF2431 are specified as working down to 10Hz.

The schematic for the TF2430 shows a 100uF decoupling capacitor (C5) in parallel with a 47 Ohm resistor (R11) connected to the source of the dual gate FET.

In the TF2431, this has been changed to 10uF in parallel with 330 Ohms.

I created an LTspice model of the complete front end for the TF2431.
To get a bode plot of the input amplifier, I disconnected the AGC from the dual gate MOSFET in the model and replaced it with a fixed voltage source. That way the transconductance for the dual gate FET remains fixed.

The bode plot shows the low frequency 3dB point is somewhere between 200Hz and 300Hz. To me this seems far too high for something that is expected to work down to 10Hz.

Changing C5 to 100uF gives a 3dB point between 20Hz and 30Hz which seems more sensible.

There is another issue with an unwanted peak at the high frequency end of the bode plot that I will come to later. Watch this space.

[brian_mk]

Now to the problem with the 10dB peak in the frequency response at around 100 MHz.

This causes problems with stability in the amplifier, made worse by external ground loops. I have found that the amplifier can actually oscillate at around 100 MHz giving noisy and erroneous frequency measurements.

The peak in response is mainly due to the 69pF capacitor (C9) in parallel with the 39 Ohm resistor (R19) on TR7 emitter. It is also affected by the 68pF capacitor (C20) in parallel with the 100 Ohm resistor (R51) that feeds the AGC detector.

I assume C9 was added to increase the bandwidth at 200 MHz. With the capacitor included, the gain at 200MHz drops by about 1.5dB.

Simply removing C9 reduces the peak to around 2.5dB. However the gain at 200 MHz then falls by 9dB.

If C20 is also removed, C9 can be reduced to 15pF. This results in a drop of 2dB at 200 MHz with a slight peak of 1dB.

At this stage I am not sure what other unwanted effects removing C20 may have.