RACAL 9081 - synthesized signal generator - REPAIR

[tozitom]

Hi EEVBlog forum,
although it is my first post I have received lots of help in this forum thanks to the contribution of all members, so I thought I could give some contribution back and here goes my ongoing Racal 9081 repair... (btw sorry for any english grammar error, I can not blame google for it only myself )
Just to contextualize, this instrument belongs to me for many years but I've not been using it for about 2 years now, and the day I needed it... well.. it just don't start.
I'm aware of the tantalum fault of those instruments, and after opening it up and checking the different power supplies voltages there was no 12V - a quick measurement on the output shows a short to ground and after desoldering the 7812 output I get 11,95V, so regulator is OK.
I've started to look for visual fault and found 3 smocked resistors, R24+R81 on "FM LOOP ASSEMBLY 19-0844" board and R63 on "LOGIC ASSEMBLY 10-0842" board. All connected to a power supply line and with associated tantalum cap shorted (C16+C46+C30) - so the tantalum cap story begun!


After replacing all those components I continue to get 12v shorted. I've started to follow the 12v line (trought wires) - not an easy task I can say! - and found another shorted tantalum!!! new cap on and finally get something on display counter!  but frequency is always jumping up and down 

I then thought and thought and... started my tantalum journey: replacing all the tantalum caps and a full renewing of this instrument! I'm I crazy? did anyone tried this before? I hope so, can not be alone on this .

I make a list, ordered caps and started to replace them board by board. So far so good - after replacing caps on one side o the 9081, frequency is not jumping any more, but cannot ajust it with spin wheel yet , however switch range is apparently working fine.

An example of result before going to the other side of the 9081 

and after - no more red drops!


stay tuned!

[tozitom]

Time for an update...

After working for long hours on this 9081 I finally get all new tantalum to its place; look of all those old caps! I've got all "styles": shorted ones, ESR >100R ones, bad values ones (1uF instead of 22uF, etc).


After switching it on, and some debugging I found a new fault on the Standard - not working! seems to be a common fault on those 9442/12 STANDARD = ceramic caps (I have a cracked one, see picture).


As I could not find the schematic (yet! but does anyone have it?), I've tried to replace a fez ceramic caps around the crystal and it's working now  got a clean wave on the output, but as I do not have correct ceramic cap values I do not know if I got the exact frequency out.

So I've a few question for you:
1) STANDARD is supposed to be 5MHz (at least it's what it's written on it), so why is my output only 1MHz? my service manual also speaks of 5MHz on block diagram + schematic! do I have the correct manual?? I got a clean wave of 1000027Hz on the standard output (as shown on my counter, but cannot say it is calibrated, so I might have an measuring error).
anyway I suppose the 1MHz is correct, as the manual says to feed external standard with 1MHz too, and 9081 seems to be working fine with internal standard...
2) what's the working temperature of this Standard? It's very hot when touching it, cannot stand my finger on it for a long time? it that normal?

Thanks for watching!

[wn1fju]

I have the very close Racal 9082 and I feel your pain!  Luckily, mine only had two shorted tantalum caps.  But one of them was one the main tuning encoder
which required removal of the front panel.  The service manual gives no clue how to disassemble this beast and it was actually quite difficult.  But that could
have been because I didn't do it correctly.

My oscillator puts out 1 MHz.  It looks different than the picture you showed.  Perhaps I have the standard oscillator and you have one of the options.  Mine
runs ice cold (measured after 30 minutes).  I can't tell from your picture whether the copper shield is a heatsink.  If it is, I suppose it would run naturally warm.

[Chris56000]

Hi!

Will this help?

https://elektrotanya.com/racal_9081_9082_h_1.5_5..520mhz_signal_generator_sm.pdf/download.html

It clearly says it's applicable to both the 9081 & 9082 on it's cover!

Chris Willams

[tozitom]

thanks both for the reply.
the board of the standard is inside an insulated "box"

and the copper serves to get the transistor close to the crystal not quite a heatsink, just to conduct heat better I think. The temperature is not warm, it's hot (cannot measure it right now, can't find my thermocouple...); I can not leave my finger on it for more than 4/5 sec... I would say it's >50ºC - is that normal?

For the manual, I already have this version and also another one with more schematics of 9081 version, but could not find reference of standard circuit on both (or reference to the oven working temperature).

[Chris56000]

Hi!

If Racal doesn't give the accuracy/stability specification for the oscillator inside the box, I would certainly be looking at about 25–50 ppm over about 90 days – this can be checked with any reasonably good 8 digit frequency counter, and anything within ±50 Hz of the frequency marked or specified should be fine – it's more important that the frequency o/p of this module is stable rather than exactly spot–on, as long as it meets the published specs!

If the module turns out to be defective, any type of crystal oven or stabilised module of the same basic frequency o/p that meets thd p.p.m. tolerance and operates off the generator's supply should do, but a lot of the modern TCXOs/OCXOs are +3.3V rather than +5V!

I would NOT attempt to disect the original module unless you can obtain another at a reasonable price – it might have been built on a ceramic substrate using specialist parts and not amenable to easy repair!

PS!

https://www.google.com/url?sa=t&source=web&rct=j&url=http://bee.mif.pg.gda.pl/ciasteczkowypotwor/Racal/Racal_9400_series_OCXOs_annotated.pdf&ved=2ahUKEwi9jMi-t7nhAhU3VBUIHRjfDoEQFjALegQIARAB&usg=AOvVaw0laokytIawoo31_Njt8bsa

Pinout:–

1 5 MHz Output relative to pin 7
2 Internally stabilized +5.~ V varactor
diode control voltage
3 +7.5 V stabilized monitor output
4 +12 V supply
5 Internally stablized +2.8 Voutput
6 Varactor diode connection
7 0 V (also connected to can).

PPS!

Several posts on the "Time–Nuts" forum suggest these things can be taken to bits and repaired – they suggest the internal wiring crumbles to bits, very likely considering the internal temperature ( >80°C!) these devices warm up to, and it also apparently causes deleterious effects on the transistors & ceramic capacitors within!

If you do have a look inside, take the utmost care with substituting any caps, etc.,they will have a specified temp. Coefficent or be the "NP0" type!

Several chaps on "Time–Nuts" have said they've drawn the insides of these things out, but I've not found a post where anyone's willing to share it!


Chris Williams

[wn1fju]

Assuming it is an ovenized oscillator, high temperatures are expected.  The HP 10811 series ovenized oscillators, used by HP for decades, use a pair of Darlington transistors to produce the heat.  It keeps the inside of the can at 80C to 84C.  You'd get a nice burn if you touched those for any length of time.  The whole
point of an ovenized oscillator is to keep the crystal at an artificially high temperature, much larger than ambient, so that room temperature changes do not
affect the frequency.

The fact that yours is labelled 5 MHz is a bit of a mystery.  As I said in my earlier post, my oscillator produces 1 MHz.  But it is the standard, non-ovenized
version, not the one you have.

[tozitom]

thanks Chris for link, the info is consistent with what I have, mine is 9442/12 (except the output freq  )
wn1fju: in fact it is an ovenized oscillator, and your description of the HP10811 series corresponds to mine too, the oscillator is inside the box from this picture above with one transistor to produce heat, perhaps not as high as the 80/84ºC but maybe close to 60/70ºC.
Anyway mystery remains with regard to its output frequency; the link posted by Chris clearly states 5MHz output, also the sticker on my standard shows the same, but my oscilloscope and counter only shows 1MHz!
It must be correct, as 9081 is working fine, with correct frequencies output, I was just curious knowing more about this 9081, that's all. It may have been "tweaked" somewhere in the past...

To close this thread I will soon post the final result after disassembling all the front  , clean it up + sandpaper on lot of dents + some polish, aluminium paint, etc...

[bd139]

That doesn't surprise me with the OCXO. Racal are a bit crazy in this area. I'm surprised it even works at all TBH. I have two Racal 9915 counters with a 9442/5 in, one of which drops out occasionally that sort of works and one that the OCXO crystal had aged past the adjustment range. Total nightmare the things are. I replaced the drifted one with a surplus Bliley OCXO as described here in the following link. It turns out that counter ASIC needed a 10MHz reference and they'd multiplied the 5MHz OCXO up to 10MHz so I just bypassed half the counter:

Incidentally I've got a PCB design in progress which allows a Bliley OCXO (aliexpress £12) to be dropped in on the same mounting holes. I may add an optional /10 stage (74hc390) to the design before sending off PCBs as that would satisfy the 1MHz reference requirement for people as well.

[wn1fju]

I looked a little bit at the service manual.  Perhaps this will address the 1MHz/5MHz problem.  Apparently, the reference oscillator is supposed to
be 5 MHz.  When operated on "internal," there is a divide-by-5 integrated circuit that chops it down to 1MHz.  That is also fed to the output jack
on the back panel.  This is why, at least on my unit, I read 1MHz on the counter when I hooked it up to that jack.  When operated on "external," the
divide-by-5 integrated circuit is inhibited and bypassed, thus feeding the externally applied 1MHz signal to the subsequent circuitry.

So it looks as though if you could monitor directly the output of the ovenized oscillator, you would see 5MHz.  And similarly if I had somehow clipped
onto my oscillator.  But if you look at the rear panel jack, you will see 1MHz as the 9081/2 really is always operating with a 1MHz reference.

At least, that's my reading of the manual...

[tozitom]

wn1fju: you're completely right! I did not yet take a closer look to this part of the schem, and was assuming the output of the oven was direct to rear panel BNC  cause as all wires are tied together, a quick follow-up of the oven output misdirected me to the rear panel BNC. When measuring directly on the oven's output I get the correct 5MHz!
Also I had a 100Hz (+/-) deviation on this standard as I did not had the correct ceramic caps value at the time I repaired it, but now with all new caps I could adjust the correct 5MHz value  (tied to my counter calibration, or the lack of it )

[bd139]

Worth sticking a thermocouple inside the oven if you can and making sure temperature is stable. The iffy one I have jumps up and down nearly 20 oC which is probably why it glitches.

[tozitom]

bd139: thanks for the tip, will do as soon as I found my thermocouple  and have time for it.

Right now I focused on getting it work, and a "quick" face renewal. I do not have time now for the all case renewal (unfortunately), even though this 9081 deserves it and also needs it...

Just a quick before/after:
 

 

[tozitom]

some more details:
   
 
   

[tozitom]

and final result:



will have to do same job on all other panels, but not in next times to come...

[bd139]

Excellent work. Nice to see some quality restoration 

[coromonadalix]

Excellent in deed