EIP 545A 18GHz counter questions. Signature analysers?

[G0HZU_JMR]

You should be fine to use the HP8568B if you use a x10 scope probe on it and an external limiter and/or 20dB external attenuator as has already been suggested.

I'm still a bit bewildered as to why it isn't tripping the comparator inside the TP4 window on range 2A with 158MHz or 20MHz.

With 158MHz it trips it AFTER the TP4 window but this is due to the range 2B tests that follow immediately after.

Your latest 158MHz test shows the comparator is healthy because you can see the TP12 output fly up to 4V once the mixer is given a 370MHz LO.

You could try reducing the 4k7 resistor R57 by putting a 3k3 in its place so the mixer gets biased harder during the range 2A sniff window?

I'd be amazed if that worked but it might be worth a try. Don't go much lower because you only want to pass a few mA into the mixer diodes.

[Chris Wilson]

Very nearly right with that Jeremy.

I still thought the issue maybe around Q12 and it had two resistors of 2%, one being "SAT" (select at test?), so I persevered a bit more in that region. I knew putting a 1k resistor across E and C of where Q12 was removed made Band 2 work, so with the new Q12 reinstated I shorted out R53, grounding the base of Q12. The counter showed zeros with a 50 MHz input at -10 dBm. So going back to basics of biasing I though R55 (750 Ohm, 2%) could have gone lower, or R54 (11k 5%) could have gone higher. Rather than pull more stuff I put a 10k across R54 and it started counting correctly. I pulled R54 and checked it out. It was quite a bit high and temperature sensitive, to boot. I fitted a 10k for now and it works perfectly across all bands

http://www.gatesgarth.com/scope/resistor-R54.jpg

I am pretty confident we have a fix. Thanks a million Jeremy and those who have put forward help and ideas.

[G0HZU_JMR]

That is good news

That Q12 switching circuit is a very odd design because it always bleeds current through the 11k R54 and the 750R (with TP4 at either logic state of 0V or 4V) and the difference in bleed current is less than a factor of about 2 between the 'off' state and the 'on' state. Most odd for a design of a switch...

If you pull the predicted 0.82mA through the 750R you get nearly 0.62V across Vbe of Q12. This ought to be enough to get sufficient bias current (eg 100uA?) to get a few mA of collector current to drive the diodes.

I'd expect it to work reasonably well even with a 13k resistor in there?

So maybe the other resistors have slightly changed value as well or maybe R54 is drifting when in circuit and this throws the basic analysis above out the window?
The only other thing that would aggravate this issue would be if the 12V supply at R56 was drifting quite low as it would affect the Vbe calculation above.


Anyway the good news it all works!

[Chris Wilson]

Still working, I may change the other two resistors (R55 and R57) and put an actual 11k in for R54, as my 10k is only a 1/8 Watt. Or maybe I should screw the lid down and leave it alone now?

Seriously, thanks a bunch for your patience, which my lack of knowledge, typos and fumbling with scope set ups must have sorely tried. You have been an absolute star and I greatly appreciate
you and everyone else sharing your wealth of knowledge and experience, and of course, your valuable time.

[flyingtech55]

I think you deserve credit for persevering with this unit and tracking down the fault to a single component using proper fault diagnosis methods. I wonder how may good bits of kit have been thrown away because of the failure of one component.

Well done sir,

TimR

[Chris Wilson]

Tim, it's Jeremy G0HZU_JMR who deserves the credit, he's an absolute star, and I could never have done it without him. I have the correct value, correct tolerance resistors for the area of the circuit around Q12 now, so to finalise things will replace the three critical components with exactly the right components.

Tek 7633 `scope next......