Racal Dana 9913 counter. Poor sensitivity.

[Station X]

My Racal 9913 Frequency Counter has become very insensitive and requires a large input voltage to ensure reliable counting. This type of fault is normally caused by high power RF from a transmitter being applied to the input, which destroys the FET in the input stage, but that wasn't the case here.

All the components preceding the FET in the input stage have been checked and test OK.

I've taken voltage readings from the transistors in the input stages and have posted them below. However no voltage readings are given in the service manual for me to compare my readings with. Semi-conductors are not my strong point and the circuit is complicated, to me anyway, in having both +5V and -5V supply lines. Can anyone advise me as to whether the readings look to be correct?

I've attached an extract from the circuit diagram and part of the circuit description taken from the manual.

The next step will be to track the signal through the input stages using a 'scope, but I don't know what to expect.

Q1.This should be a W300A FET, but as these are unobtainable it was replaced with a 2N3819 as part of a previous repair.

Drain 4.95V. Gate -0,02V Source 0.32V

The remaining transistors are all BFX89's and I've given the readings in the order Collector, Base, Emitter.

Q2. 2.473V 0.788V 0V.
Q3. 2.45V -0.137V -0.940V
Q4. 2.790V -0.176V -0.980V
Q5. 4.95V -0.529V -1.316V
Q6. 2.119V -0.532V -1.337V
Q7. 2.118V 0.798V 0V

Many Thanks.

[ronbryan]

The DC conditions you have measured seem fine to me. All of the single ended and differential stages are AC coupled, so your scope check with input signal applied should show whether any of the 220uF and 47uF electrolytics are in trouble as you work your way through the stages searching for a sudden reduction of signal.

Ron

[Station X]

Ron.

Thanks for confirming that the voltage readings look OK. I'll 'scope the circuit and report back.

The coupling caps are tants, so not above suspicion.

[orbanp1]

Hi Graham,

Did a quick simulation of the input circuit.

The source voltage of the input FET looks suspiciously low in your case, though FETs could have high variation.
I would start by replacing Q1 with another 2N3819 or similar.

If you replace tantalum caps with regular electrolytic caps I would add a 10nf - 100nF disk capacitor parallel with them.

The overall gain of the input stage is about 21dB.
The high frequency response is shaped by the capacitors (C7, C13 on the factory drawing) between the diff amp emitters.
For me the best results were given by 15pF caps, though I used MPSH10 BJT transistors, and those have an Ft of 600MHz versus the 1.1GHz of the BFX89.

I enclosed the drawing of the transfer function at the various stages of the circuit, that should help you to find the failure in your case.

Regards, Peter

[Wellertcp]

Yes, the transistor readings look ok. If you haven't already, its worth checking the power supply rails.
I had an old Data Precision voltmeter; which worked fine when set to (any) manual range; however when left to its own devices to auto-range; it would continuously ramp up and down thru' the ranges without settling.
All voltage readings checked-out ok, however, when digging deeper, I found that the bulk caps on the analogue rails (+/-24VDC iirc) had turned into 100k resistors!
Changing these caps restored normal operation.

[Station X]

Thanks to everyone who's responded so far, especially Peter for doing the simulation.

I removed the input FET and it tested good out of circuit. Nevertheless I replaced with another 2N3819 which also tested good out of circuit. I checked the voltages on the FET and they were very similar to what they were last time.

There's quite a difference between the 0.32V I measured on the source and the 1.58V shown in the simulation. This makes me think there may be a problem with one of the components around the FET?

Tests with a scope have shown that there's a considerable voltage loss between the input socket and TP1.

[Radio Wrangler]

The 2N3819 is driving into a shunt-feedback stage, consequently the output impedance of the FET forms part of the gain-setting impedance ratio of the two stages seen together.

2N3819 has quite wide bounds in its DC parameters which can make biassing uncertain. Two devices out of the same batch might be similar, but the low-ish Gm may be the cause of low gain without necessarily upsetting the DC conditions from that with the original FET the circuit was designed around.

It might be worth tying a higher Gm FET.

Marconi used the 2N4391, 2N4392, 2N4393 family as RF amplifiers in some of their earlier high dynamic range receivers like the H2900. This family has high Idss and high gm. At HP we used them pretty much exclusively as analogue switches, but maybe the less extreme one might be a possibility in this circuit.

That counter doesn't happen to have two channels giving a good one for comparison? It's be helpful to compare signal levels down the two. With an input via a 1 meg pot, I don't think this input is particularly wide bandwidth.

David

[Station X]

A bit of an anti-climax I'm afraid gentlemen.

The problem is down to the sensitivity control, designation 1R1, a 1Meg log pot.

It tests out perfectly DC resistance wise, but wouldn't pass a 10MHz signal. I proved this by temporarily bridging the "top" tag to the wiper tag using a croc clip. This restored full operation and the counter is now very sensitive with response up to the specified 200MHz limit.

I dismantled the pot, cleaned the track and re-tensioned the contacts, but it's made no difference. I'll have another go at it tomorrow. If no joy I'll solder in a wire bridge for the time being so I can get on with the job I set out to use the counter for.

Replacing the pot will be difficult. It's a special unit with a ganged change over switch which operates when the control is rotated fully anti-clockwise. Its function is to switch in a divide by 4 pre-scaler on the VHF range and change the gate time by a factor of 4.

I'm thinking I could perhaps use the make contacts on a normal pot to release a reed relay with changeover contacts. Or I could fit a separate NORMAL/VHF switch, but that would spoil the unit's appearance.

[Radio Wrangler]

Got a variety of small relays if that'll help. There may even be some Teledyne TO-5 can ones!

David

[Station X]

The sensitivity pot is a lost cause with portions of the track worn away. It has two wipers at different radial distances from the axis of the shaft, but this makes no difference.

I've ordered up some 1M log pots with 4mm shafts, as per the original, but they lack any kind of switch. I'll see what I can do to marry one up to the original switch which will mean extending the shaft by some means.