Racal 9839 UHF frequency meter

[factory]

Well I've bought another faulty Racal counter for my collection, this time a model 9839 that has a pre-scaler for up to 560MHz.
It appears a previous owner has done quite a bit of work before giving up (possibly why an IC is missing).


I like how they thoughtfully sleeved the fuse holder & mains socket, but then fitted a filter board with exposed connections above these, the transformer connections are right above the +5V adjustment pot too.


Does anyone know why the track from the +200V fuseholder has been cut (dotted green line) and a link added between R60 & the electrolytic capacitor? The 60mA fuse is OK.


The display only has a few digits working, I should be able to work through the faulty sections using the 9835/7 manual as the display boards are similar, some IC's have been replaced & a number of tracks repaired by a previous owner, I can see one broken track that got missed.
I've already adjusted the +5V rail as it was around +5.2V.

David

[factory]

A quick update, I used the HP 8443A to check the frequency standard of the Racal 9839, after an hour there is just under 10Hz difference between the two standards, the HP 8443A reads either 1.00000MHz or 1.00001MHz.
And using the 1MHz output from the 8443A I found both the low & high frequency ranges of the Racal were working fine, even if I only have 2.5 working digits.


Next task, find the spare boards I have & replace the first confirmed dead number readout bulb, to find out if the pre-scaler high speed divider is working, this is a separate circuit to the rest of the counter section.

David

[Vintage_RC]

I have had a Racal 9839 for many years and it was and still is my main frequency counter. It has given very little trouble over the years. One of the Nixie tubes developed an internal short that caused two digits to display together, a replacement tube sorted that out. A more recent problem documented on this forum was when all digits of all tubes started displaying together. It was a strange fault that just disappeared during investigation, I had re-seated a number of connectors which might have cleared the fault. The ovened reference oscillator in these counters has, in my experience, excellent long term stability. Most of the components are easily obtained but I think the pre-scaler chip, a Plessey SP630B, would be a real problem if defective.

[factory]

I'm hoping the Plessey IC will be OK as both inputs are working.

I have a larger Racal 9025 with a failed crystal standard (much larger unit to the one here). Another forum member who is on eevblog suggested checking the standard as they have had a couple of failures too.

The smaller number tubes (or Indicator Tubes as Rodan call them, as fitted to this Racal) do seem more prone to getting shorts between digits, had that with some in my HP counters. The two tubes next to the socketed IC's in my 9839 seem completely dead, I suspect the impact that bent the case up has damaged them, I do have some used spares that I've already tested to swap in if needed.

David

[PETERg0rsq]

Originally Posted by factory

]

Does anyone know why the track from the +200V fuseholder has been cut (dotted green line) and a link added between R60 & the electrolytic capacitor? The 60mA fuse is OK.

Looking at the schematic and your pictures, as far as I can tell, the pictures now reflect the schematic.

So my guess is the board was fabricated wrong (in the days of draughtsmen/women and no ratsnest software packages) so was likely a correction in the factory.

I have come across this a few times in Racal equipment, so always resist the urge to re-instate cut traces without checking why they are cut.

Also in my 9087, updates and modifications are done this way as the products mature, and several components "spliced" into a circuit this way.

What manual do you have, is it one I uploaded to KO4BB many years ago? I now have two original paper manuals for the 9835/9837 (1973 and 1975) which I could scan for a better resolution.

[factory]

The modification could indeed be something to correct a layout error, as pulling the 60mA fuse does stop the display from lighting as well as the gate lamp. Certainly not done by whoever attempted to repair the display board as the link uses proper sleeving, just the flux residue had me questioning it. R85 in the power supply area has also been removed, but the flux residue has been cleaned away, as has the unlabelled IC from the front on the main board (this makes me think the same board was maybe used in another counter).

Here is what I have to deal with this afternoon, some proper insulated links (kynar wire) from the driver IC's to number tubes will be fitted and I'll clean up the excess solder that could short adjacent traces. At least one broken track got missed by a previous owner too.


I also managed to find a paper copy of the 9835/9837 manual (I also have the pdf from KO4BB), it's certainly helping with the display board, the power supply section differs a bit as there is a second bridge rectifier (next to 60mA fuse) and that 33uF capacitor for the 200V rail. I noticed the 9835/7 uses a half wave rectifier & no capacitor.

David

[buggies]

That last picture in post #1 (solder side near fuse) seemed to show a lot of through-hole "griplets" which kept us service engineers in a job during the late 60s... another test-gear manufacturer BTW.

[factory]

There are quite a few of those griplets, most are used for soldering wires to, the four rectifier diodes have one on each of them to space them off the board too.

Originally Posted by Vintage_RC

A more recent problem documented on this forum was when all digits of all tubes started displaying together. It was a strange fault that just disappeared during investigation, I had re-seated a number of connectors which might have cleared the fault.

This fault can be recreated by disconnecting the +5V supply to the display board, did exactly that earlier when re-checking some spare boards, no harm done.

If anyone needs the pre-scaler board (19-0604) diagram it can be found in the 9057/9059 manual scanned by PETERg0rsq, it's available on KO4BB here (pages 51 to 53 of the pdf); http://ftb.ko4bb.com/getsimple/index...3.1.74-125.pdf not much else the same as the 9057/9 counters use those newfangled LED displays.

David

[factory]

An update on the repair of the Racal 9839, all the IC's are OK, it turned out six of the number tubes were faulty, most of those required more than 250V on the Megger to get a faint glow, one required more than 500V, a few of the leads have some rust where they meet the glass seal, making me think the neon inside is contaminated with air.

Luckily I had purchased a few spare five digit boards years ago, when they were still reasonably priced, I re-tested those boards yesterday using the connector I had made up, only one tube was faulty.
Six of the good tubes were fitted to the 9839 board, as well as the missing IC's, I removed the IC sockets & repaired the damaged traces at the same time.


After cleaning the board & refitting it, the counter is now fully working on the 40MHz to 560MHz range. And mostly working on the 10Hz to 60MHz range, but the sensitivity seems poor, I'll clean the connectors & pot tomorrow to see if it can be improved or if there is a fault.


David

[factory]

Originally Posted by PETERg0rsq

Looking at the schematic and your pictures, as far as I can tell, the pictures now reflect the schematic.

So my guess is the board was fabricated wrong (in the days of draughtsmen/women and no ratsnest software packages) so was likely a correction in the factory.

I have come across this a few times in Racal equipment, so always resist the urge to re-instate cut traces without checking why they are cut.

Also in my 9087, updates and modifications are done this way as the products mature, and several components "spliced" into a circuit this way.

What manual do you have, is it one I uploaded to KO4BB many years ago? I now have two original paper manuals for the 9835/9837 (1973 and 1975) which I could scan for a better resolution.

The Racal 9835/9837 manual I have is from issue 6 from Oct 1974, it has an amendment sheet with an added 1uF 250V capacitor (C38) from the lower end of FS1 and the 0V rail.

As no manuals exist for the Racal 9839, I've traced out the +200V supply circuit in mine and the 33uF 350V capacitor would also have been connected as the 1uF in the 9835/7 if the track hadn't been cut.
Instead it is across the second 220K resistor R61 (in parallel with a 0.22uF capacitor), quite frankly I can't see it doing anything useful where it is and the supply to the tubes is unsmoothed as a result.


The counter after cleaning up the case and straightening out the dents as best I could.


David

[Jon_G4MDC]

You have done a nice job there. I like it.

[PETERg0rsq]

Just checked all three manuals I have, and all show the same circuit, which is completely different to what you have. (half wave rectifier with no smoothing capacitor.

So the logic behind the capacitor, and moving it is anyones guess.

Only thing I would say is it is not necessary to smooth the 200V line, as the effect of 50Hz ripple would be the same as 50 Hz ripple on an incandecent bulb!

[factory]

Thanks for having a look at your manuals, I thought I would trace that part out as it differed a bit, it was a little tricky with the main board having traces on both sides.

I guess the reason for the change will remain a mystery and I've left it as I found it, I certainly can't see any flicker, nor can the camera I used, unlike the Philips signal generator used for testing (very noticeable flicker in the camera from the LED display).

Some of my HP counters also use an un-smoothed supply for the tube display, but not all of them do.

David

[Cobaltblue]

Re-opened at OP's request

Cheers

Mike T

[factory]

Thank you.

Originally Posted by factory

After cleaning the board & refitting it, the counter is now fully working on the 40MHz to 560MHz range. And mostly working on the 10Hz to 60MHz range, but the sensitivity seems poor, I'll clean the connectors & pot tomorrow to see if it can be improved or if there is a fault.

Well I never got round to looking at the sensitivity problem with the lower frequency range, until yesterday.

I've now had a look at the problem, using the 9735/7 manual as a reference as the input circuitry is similar, but probably not identical, the PSU circuitry certainly isn't as previously mentioned.

Yesterday I went through the input circuit with a scope and the signal was getting through all the stages, albeit a bit lower than I expected.
It then occurred to me that I had forgotten to check the separate -6.8V supply rail, that is used for the main input circuit, well it turned out to be stuck at zero volts. Checking with the ohms range revealed the -6.8V rail was shorted to the ground, this could have been the tantalum cap C2 , or disc cap C4 (unlikely), or the Zener D8, nothing was burnt.


The tant fell to bits on removal, but wasn't the cause of the short, next I checked disc cap C4, again it was fine, this left the 6.8V Zener D8 and yes it was now a nice 1Ω resistor.
The original was a Z5B 6.8, from a once common 400mW Zener range that BR used a lot, but now obsolete, my spares had a draw with some modern 500mW replacements, as well as some dodgy looking parts the same size that had been relabelled.


In went the new 500mW Zener and a replacement tantalum capacitor of slightly higher voltage (all I had), I cut some of the coating away on the original and yes it tested fine.


Now re-assembled & fully working at 60MHz with a 10mV input, it ran out of steam at approx 83MHz.


David