Test Equipment Anonymous

This came from a rally in the summer. It was claimed to be working. It was working on the B range (10Hz-60MHz) but the A range (40MHz -520MHz) was dead.

Opened it up and checked the supply line voltages. The 5V and -5v lines were about as specified in the manual. When the A range is set, part of IC14 is used to divide by two 1MHz derived from the main frequency reference. This 500KHz signal is used to drive Q7 on and off and energise the primary of T1. The voltage in the secondary of T1 is rectified and added to the +5V line to create a 24V supply for IC 26, a wide band hybrid amplifier. It is smoothed by a filter comprising C30 and C31 (4.7uF and 1.5uF tants) and L4, a 3.3uH inductance.

Test point 2 is provided to check the output of this PSU arrangement for IC26, the hybrid amplifier. It was 3.7V with the B range selected and 4.3V with the A range selected, not 24V. Q7 (ZTX 450) was hot to the touch. A scope showed a 500KHz square wave from IC14 driving the base of Q7.

I suspected one of the tants, (C30 and C31) was short circuit, or IC26 an OM335 hybrid circuit had failed, and was drawing heavy current. The replacement parts in that part of the circuit are mostly quite cheap, so I ordered the lot from CPC and had to wait a time for them all to arrive. The OM335 is still available on ebay for about £8, but I could see no point in jumping to conclusions about that.

I desoldered C30 and C31. For some reason it was very hard, even with a desoldering station. I then had the idea of lifting one end of L4 to isolate the OM335. There is no L4, it was replaced by R166 a 68R resistor. This took some time to work out. I had the idea of putting in a dummy load for IC26 which draws 35mA according to its datasheet. 680R was about right. A 3uF 50V electrolytic was put in place across it. This arrangement was connected between the cathodes of D6 and D7 and earth. The instrument was powered on and the voltage across the load was monitored with a scope. 3.7V on range B, 23V and little ripple on range A.

Next I soldered in the replacements for C30 and C31. I soldered the cut lead of R166 back together. I'd rather replace a part than do that, but it was such a pain to remove the two caps that I couldn't face removing and replacing it. It seems to be a sound joint. Tweaking R166 with long nosed pliers couldn't break the joint.

It was still possible that IC26 was faulty.

With everything replaced and checked. the instrument was powered on. 22.9V at TP2 with A range selected, which is within spec, and it read about what was expected when fed from an RF signal generator.

After it had been allowed to warm up, it was set up against a 10MHz GPSDO. It seems to settle to less than 1 part in 10^7 and return there after it's been powered on after being powered off for a time. It has the lesser of the two OCXO options Racal offered. I had thought of replacing the OCXO with something based on one of the Aliexpress cell tower pulls, but if it continues like this I won't.

Piccies

A pdf of the relevant bit of the circuit for anyone inclined to follow. A picture of the circuit board around the 24V PSU area. The two orange blobs in the middle are the new C30 and C31. R166/L4 is above them. The orange/brown long thing above R166 is the OM355. TP2 is between the OM355 and R166. The grey, square thing to the right with an orange dot is T1. Q7 is the closest transistor beneath T1. Diodes D6, D7 and D8 are beneath and to the right of T1. The polarity of a lot of the diodes is indicated incorrectly on the silk screen, which caused some head scratching.

A picture of the 9916 showing as close to 10MHz as I could adjust it.
Final thoughts. A fairly simple repair with no nasty and expensive surprises. The manual is a lot better than nothing, but with a few omissions. Not a thing I found pleasant to work on. This thing is packed full of tants. At least one has died. I suppose most of them could be replaced with electrolytics with a ceramic in parallel, but removing them is a PITA.

Izzat one of those that came in a nice chonky cast aluminum enclosure?

mnem

No, this one's cast aluminium.

Yes, it's got really solid cast sides and front and back pieces, and even the sheet metal on the top and bottom doesn't flex easily. The earlier ones had extruded sides but there was a cover which went over them which seemed a bit flimsy.

They have the odd special IC which is unobtainable, but mostly they don't give any trouble, apart from the OCXOs which are starting to suffer with age. I was slightly surprised that this one had a problem.

Yup... those are the ones I was thinking of. Nice score.

mnem
*toddles off to nuke a PS3*

I had exactly the same failure on the last 9917 I had. Same caps. They managed to charcoal the series inductor entirely. Had the same difficulty desoldering them as well. I suspect this is due to the hefty copper used on the boards in these. IC26 was fine. It just sprang into action afterwards. I did consider replacing all the tants but I ran out of arsed by then. Can't find the link now on EEVblog. It was in the TEA thread somewhere!

Inevitable reference in case anyone else trips over this. Replacing a duff Racal OCXO with a 10MHz aliexpress lump: https://www.eevblog.com/forum/testgear/ ... sg2167945/

It was difficult to desolder anything on that board because everything is pretty tightly packed, they bent the solder side of the leads over at right angles and the holes are a tight fit for the leads. Usually, removing capacitors is easy, just melt the solder at one end and tweak the cap out and do the same on the other end until it's free - not with these. Even with the caps and bits of lead removed, the holes had to be resoldered and sucked out with a desoldering station a few times, before they were free. I wasn't about to go desoldering more on that board than I had to. The other thing is that it's not easy to find the solder side corresponding to most of the components.

Those tants must have a hard life, smoothing 500KHz with harmonics. I was pleased it was just the tants and not the hybrid amplifier or the transformer. This was an absolutely typical rally acquisition. Cheap enough, and mostly working, but with some fault that the seller knew about and either couldn't fix or couldn't be bothered to fix, which is why they wanted to get rid of it. It usually turns out that the fault isn't that hard to fix.

I found your EEVBlog post on using the Aliexpress OCXOs which was useful. Fortunately, the OCXO in this 9916 is fairly OK. After being left off for a few days it takes two or three hours to settle down to about 1 part in 10^7.

I also have a 9906 where the OCXO is definitely old and tired. I can't find a 9906 cct diagram, but the 9905 is reckoned to be nearly the same. Those Racal counters have a quirky scheme where 5MHz from the OCXO is doubled to 10MHz. I suppose there were good reasons for not using a 10MHz OCXO. Anyway, I've decided to replace the OCXO in the 9906 with an Aliexpress surplus unit. It's been too long since I used KiCad and etched a board. I decided not to use your approach, because I'd rather not mess with the PCB in the counter, but to divide down the 10MHz from the OCXO with a 7474. This will be a really quirky scheme where 10MHz is divided to 5MHz for the counter to double back to 10MHz.

Makes sense to me. Sounds good. The only reason I went for the modification is I didn't have anything to divide it down with lying around and this was a Sunday afternoon project and I didn't really give a shit about the counter enough to keep it original.

I have a big bag of about 200 pieces of 7400 series TTL I bought at a rally some years back for 50p. It's produced a few 7474s of various shades, but no 74141s. When I bought it ,a friend sort of knowledgeable in electronics, asked me whether I'd use all those chips. I said it was just possible that one or two might remain unassigned when I shuffled off this mortal coil. Since they cost at least 25p a pop and I've used half a dozen, I think I'm ahead of the game on that one.

Sentimentality doesn't really apply to TE, and especially commonplace things like Racal counters. It applies in spades to expensive domestic radios and tellies and to some extent to boatanchors. I didn't want to mess with the circuit because changing two caps on the 9916 had been an ordeal. I haven't found an authentic cct diagram anyway, and shits are in short supply when it comes to the question of tracing out the circuit. There were no thoughts of preserving originality or I might have fitted the new unit into the original Racal OCXO case. Come to think of it. that could actually be a good mounting solution.

I was also seduced by the matter of getting back into KiCad and making a circuit board. I don't like Ferric Chloride and found I have Hydrochloric acid but no Hydrogen Peroxide. I thought I had a bottle of Hydrogen Peroxide but no. Apparently it's in short supply. After checking the local pharmacy and a couple of places in the local town with no luck, I checked the pharmacy in the next village and they had several bottles in several strengths. They also had acetone, so I bought a small bottle of that too. There were no questions. The last time I did that I had to explain that I was not buying it to make TATP and that if my thoughts were running along those lines, RDX would be my preference.

Anyway, I've not got it working yet. I'll keep you posted.

I have several boards full of TTL here as well waiting for a rainy day. I might bounce them because quite frankly I'm feeling a little attacked by the amount of crap I've collected.

As for the ferric chloride, I disposed of that when I moved. I don't want to bother making PCBs by hand any more unless it's dead bug. It's not worth it. China is too easy and the results are good if you want something production ready.

bd139 wrote: ↑Tue Dec 13, 2022 1:55 pm I have several boards full of TTL here as well waiting for a rainy day. I might bounce them because quite frankly I'm feeling a little attacked by the amount of crap I've collected.

As for the ferric chloride, I disposed of that when I moved. I don't want to bother making PCBs by hand any more unless it's dead bug. It's not worth it. China is too easy and the results are good if you want something production ready.

"It might come in" is a phrase that will be coming back into fashion and relevance

Manhattan is good

bd139 wrote: ↑Tue Dec 13, 2022 1:55 pm As for the ferric chloride, I disposed of that when I moved. I don't want to bother making PCBs by hand any more unless it's dead bug. It's not worth it. China is too easy and the results are good if you want something production ready.

Ditto here... I got rid of all that when I left San Damntonio... over a decade ago. OshPark was my "It's just too easy" plugged into that equation... but now of course we have oodles of micro-fabs to choose from in the land of oolong and 5-clawed dwagons. They're like this generation's FotoHut.


mnem