Understanding and Repairing Clock Generator on Adret 742A UHF Generator

[timeandfrequency]

Hi George,

Two rows IDC connectors always have an even number of pins (usually between 6 and 50).

On both CPU boards pictured in this thread (here and here), I guess that the 'CPU card to motherboard connector' you mention is the light grey one at the bottom of the board, a bit on the right side, which has 31 pins, as printed on it (1-3-5-7-9-11-13-15-17-19-21-23-25-27-29-31).
Concerning the 'TRELEC TM31' model part # written in the SM, '31' stands for...31 pins.
The extender and the connectors I pictured also have exactly 31 pins.
All of the replacement parts I suggested have 31 pins (see page 206 & 207 here).

Well, I might be wrong, but is there another uncommon connector on the CPU board ? In the affirmative, can you provide a picture ?

[George Edmonds]

Again someone who does not read diode specifications.

Part of the specification for the 1N 540X series says “For capacitive loads derate maximum current by 20%”  This had not been done hence the high diode failure rate.

I said I have never seen that connector, it was never used in UK and US electronics, even Adret stopped using it in the 7000 series, they are also prone to bent pins.

G Edmonds

[PerArdua]

Evening all.

With the successful test of the power supply (unloaded), the instrument checked for visible signs of damage, etc I decided to attempt to boot the instrument. Upon apply AC mains, the standby LED came on and everything appeared stable. I the flicked the standby switch to ON and the instrument appeared to try and start (a few numbers flashed on amplitude display) and a flashing LED above the execute button. However, at that point I began to smell some acrid scent, so I turned the supply at mains. Opening the VHF module I discovered a tant. capacitor for +15V in the output amplifier section (C716 on schematic, C16 on the PCB view) was charred. Furthermore, opening the UHF section (the section containing the two frequency doublers, with no schematic) revealed a charred capacitor in the section containing the attenuator EPROM for the -15V supply. There is a possibility I missed the first capacitor (though I believe small), but the second capacitor was definitely damaged on this attempt (photographic evidence from the first repair to the +15V capacitor haha).

I am running out of time for further tests this evening, but do we think these caps are failing due to a past catastrophic event, and it is worth chasing down and repairing each cap now, or is it more likely there's a PSU issue that needs to be addressed first? I will replace the two caps above, and will then measure the startup behaviour of the PSU outputs under no load (disconnected). Depending on those results, I may reattach the PSU to the instrument and again measure the rails.

Would it be beneficial to measure the resistance across each supply rail on the instrument motherboard with the PSU disconnected? Whilst I get this might change during operation, would it at least be a good idea to verify the 'cold' state?

I am both a bit bit encouraged and disheartened by these results - on one hand the flashing LED gives me hope that the CPU is doing _something_, though chasing this PSU/blown caps is a bit repetitive. ah well. 

Attached are combined schematics for the VHF PCB. I've combined the four pages onto a single sheet - the manual shows this as two different schematics.

[George Edmonds]

Hi Time and Frequency

Thank you for the information on the PCB connectors, I now understand what they are and why I have never seen them, nor had anyone else in the UK that I asked.  They appear to have been used solely in the telecommunications industry, until recently the telecommunications equipment industry in the UK has been dominated by two UK companies and an ex nationalised industry.  Shall we say that it was not an open market place.

The shrouded DIN multi pole connectors I am very familiar with as I have a number of Rohde & Schwartz instruments which make extensive use of them, I even have PCB extenders for them.  A quick search in the UK for the type used by Adret reveals no stock held with the only stockist being in Germany at high prices and expensive delivery costs.

When I used to get a 700 series instrument in for repair I automatically shotgun replaced all of the power supply diodes with BY550-400 and never had a diode fail again, these are 5A 400V PIV and dimensionally almost identical to the 1N5404.

Hi Per Ardua

Please do not give up, you have tackled a complex instrument and are making real progress.

Before you power the instrument up again please replace ALL the bead tantalum capacitors before you have one go on fire and cause considerable damage.  When they were first used they were sold as a magic solution having low ESR and being able to work very close to the specified working voltage.  Over time and as they age this has certainly not proved to be the case, they need to be specified to be some 50% over their operating voltage, for example on a 15V rail use 25V ones not the 16V ones used.

This is NOT an Adret only problem, it is industry wide.  Racal had massive problems with failing bead tantalums which sometimes caused extensive damage.

It would also be a wise move to  replace all of the aluminium electrolytics, over time they go high ESR and lose capacitance, after all they must by be thirty years old.  They can also leak electrolyte which is also corrosive causing PCB damage and is conductive.

Please keep us informed of your progress.

G Edmonds

[timeandfrequency]

Hi everybody,

@George
About the PCB connectors : you are right when saying that those open models were never used in UK and USA TE. But that's only half of the world, and there might have been some regional preferences/choices. I could also spot these connectors on Siemens automation systems designed in the mid/late 70' (see attached pictures), which is about the time the 74x was designed.
Furthermore, Adret management teams first target was the telecommunications industry.
You may buy these connectors here and also here, for immediate delivery.

The more common shrouded version (which only exists with even number of contacts : 16, 32, 48, 64 & 96) was also much used in VME racks/boards.


@PerArdua
George's advice to replace all of the tantalium capacitors is wise. It's a nightmare for every vintage TE in which they were used. That some of them prevent your Adret from working properly is not so surprising.
You may replace them by low ESR 105°C Electrolytic capacitors from first tier manufacturers like Panasonic, Cornell-Dubillier, Nichicon or Rubycon. Sometimes it is wise to also add an 100 nF MKT capacitor in //, in case not other plastic capacitor prowls in the surrounding.

For low voltage rails, a direct replacement for tantalium capacitor are also those ones. With no liquid electrolyte, they last forever, but are rather expensive, exist only in SMT form factor and are 10VDC maximum. The fact that they are SMT should not refrain you from trying : sometimes, you can solder them on the...solder side of the PCB.

Thanks to his experience on this stuff, George could spot the usual failures of the PS inside the Adret, so replacing the diodes and put new electrolytic capacitors is also desirable.

Would it be beneficial to measure the resistance across each supply rail on the instrument motherboard with the PSU disconnected? Whilst I get this might change during operation, would it at least be a good idea to verify the 'cold' state?

Yes you can measure the resistance of each supply rail but the test is not always valid (you will get some false negative). You may detect dead shorts, but tantalum capacitors sometimes have a voltage threshold where they begin to become a short... And the voltage provided by your Ohmmeter (usually <= 2VDC) might not be sufficient to trigger the short.
Do not attempt to use your lab PS to make this test : only supplying one of the ±15VDC rails could kill all of the opamps. To be on the safe side, you would need a dual lab PS with tracking mode and common overcurrent trip mode, which is a rather rare piece of equipement.

And thanks for the rebuilt of the VHF board schematic, which is now much easier to read. A new item just appeard on my wish list : an 8K PC monitor.

You have already made good progress in the repair of your TE and going further would be really interesting. It is true that reparing the RF and analog boards is more demanding than the PS & CPU assemblies. It usually requires specific TE for probing inside RF sections. But maybe you get lucky and the sole issues you'll find are charred/shorted capacitors.

[PerArdua]

Thanks for both of your replies - I am certainly going to keep working at this.

I'll replace all tantalums I see, and will get to work on aluminium electrolytics too. Thus far I have been replacing the tants with 25V rated components, they are slightly larger but fit easily. Does anyone know if the attenuator assembly has any capacitors in it? I tried removing it last night but could figure how to remove it from the instrument. I removed the nut retaining the N connector from the front of the instrument, the SMA input, and four hex bolts sitting below the VHF module - are there any extra screws somewhere? Aside from that I'm comfortable with accessing the rest of the instrument (though I might try and keep repairs to the VHF module on the component side so I don't have to deal with all the feedthroughs etc).

[George Edmonds]

Hi Per Ardua

I cannot immediately find a schematic for the 740 attenuator, but I have looked at the one for the 730, there is only one possible tant and that is a 4.7uf 35V one, it is in the RPP circuit directly on the 15v supply rail, so probably no problem

G Edmonds

[timeandfrequency]

Hi Per Ardua,

I had an in-deep look in other Adret technical paper documentation (in that case for the 740A; 0. to 1120 MHz RF Generator) and could not find much infomation about the step-attenuator assembly.
It is only mentionned on the bloc diagram (see attached picture) ans seems to be controlled via the approach board. But globally the ouput scheme of the 740A seems to be different to the 742A you own, so unfortunately, not much infomation can be gained this way.


I'll resume searching and update if I can find something useful.

[George Edmonds]

Hi

Look at the 730A information it may be the same

G Edmonds

[PerArdua]

Thanks all - in that case I shall leave the attenuator assembly as is - I can't figure how to remove it and if there's unlikely to be a dodgy capacitor in there then why risk damage.

I spent this evening taking pictures of the PCBs, which I will resize and upload here in short order.

I've also started creating a shopping list of tantalum capacitors, minimum voltage rating of 25V.

For rectifier diodes - would the 6A4-TP be good candidate? Key specs: Vf - 0.95V, If(av) - 6.0A, Ifsm - 400A (https://media.digikey.com/pdf/Data%20Sheets/Micro%20Commercial%20PDFs/6A05-6A10.pdf)

Best wishes, 

[PerArdua]

Interestingly I think I have found that there are at least two quite different revisions of the analogue PCB in the 742A - the right hand side of the PCB in the manual provided by K4OBB (first attachment) and the position of potentiometers does not correspond with what I have in my unit (see second picture).

[timeandfrequency]

Interestingly I think I have found that there are at least two quite different revisions of the analogue PCB in the 742A - the right hand side of the PCB in the manual provided by K4OBB (first attachment) and the position of potentiometers does not correspond with what I have in my unit (see second picture).

Hi Per Ardua,

Indeed interesting : the PCB version of the analog board inside your Adret seems not easy to find in the currently available documentation. Any serial numer on the PCB beginning with D97027' ? Did you spot an issue on this board ? Does it need recalibration ?

About the attenuator, as @George underscored, the 730A SM includes its schematics, completed with the reverse power protection circuit.
C8 (4.7µF/35V) might be a tantalium cap on the +15VDC rail.

However, I don't know if 730A and 742A attenuators are identical.

[PerArdua]

Hi timeandfrequency,

the only part number I can find for the PCB seems to call it H12774902 (unsure whether the first character is H). I've attached a high resolution image of the reverse of the PCB which shows this in the lower right quadrant.

Regards,

[George Edmonds]

Hi Per Ardua

Looking through my Adret archive I have found a zip file of the Adret service manual for the 740A.

This zip file also contains images of a version of the firmware and EEPROM along with some photographs.

Most importantly it does contain full details of the attenuator assembly, including a schematic.  Regrettably it does not contain any information on the final frequency doubler stage, you need a 742A manual for that if it was ever provided.  I suspect that the final frequency doubler stage uses devices containing Beryllium, in which case it would have been a factory serviced only module with no schematic provided.  The original UK sales and service representatives for Adret were Racal Dana and they operated a policy of factory service only for anything containing Beryllia.

There appears to be details of yet another different Analogue board in this zip file.

I was always under the . probably mistaken, impression that the only major difference between the A and B models was the case, the A case has four plastic rear corner moulding whereas the B case does not.

This zip file is some 24mb so emailing it to you may not be possible.  I possibly can upload it to the Marconi Test Instruments group on groups.io for a few days only as I am a joint owner of the group, but it cannot remain there as the group has a very limited file storage allocation.

Let me know what you want to do.

G Edmonds

[PerArdua]

Hi George, if you could send me that via email I would be very grateful thank you. I will send you a PM with email address.

I am currently putting together a digikey order for replacement capacitors, and a couple of values of resistors I did not have to hand when repairing the CPU PCB (ie completely fresh, completely corrosion free etc). When that arrives I will add them to the instrument. In the mean time I have a couple of EPROMs I am using to build and test my reader with, before using the devices in the 742A. It is unfortunate that it does not use a Z80 CPU - I have already about half a dozen of those to hand. 

[George Edmonds]

Hi

Email failed.

I have uploaded the 740A Zip file to the files area on Marconi-Test-Instruments@groups.io.

It is in a folder called Adret 740A and is some 25MB

G Edmonds

[PerArdua]

Thank you George - I've downloaded the Zip. 

[PerArdua]

Apologies for the lack of update for a while - work as been pretty intense and will likely be so for the next few weeks as I change posting.

Good news though, in that all tantalum capacitors have been replaced with at least 25V or more rated components, and I have a fresh set of aluminium electrolytics on order from Digikey this week for the power supply! I will make some time this weekend to give a proper update and share some more images.

[PerArdua]

Update - some good some bad,

I've replaced all tantalum capacitors, as well as the four big aluminium caps on the power supply rails. When I switch on the instrument, it appears to be stable - no fires, or burning just yet. All tantalums were replaced with at 25V rated tantalums, the 10000uF, 25V capacitors replaced with 10000uF 35V capacitors, and the 6800uF 40V capacitors replaced with 6800uF 50V capacitors - all from Digikey. I couldn't find any that had the right pitch and sized screws, so had to resort to soldering in replacements.

However, the power supply seems to be starting itself. When applying AC mains in standby, the fan turns on and the following voltages are measure:

+15V: 1.33V
-15V: -11.10V
+5V: -0.47V
+15V pilot: 6.22V
+5Vfloat: -0.77V

When I turn the instrument on from the front panel, a 'random' assortment of LEDs illuminate, as well as maybe 1 or 2 characters on the seven segment displays. The voltages become:

+15V: 4.2V
-15V: -15.0V
+5V: 4.2V
+15V pilot: 5.6V
+5Vfloat: 5.2V

It is a bit concerning to see these voltages and the PSU behaving incorrectly. However, I take some solice in the seven segment display showing a correct character - perhaps this indicates that at least part of the ROMs are be ok assuming higher probably for a legible character to appear, rather than a random assortment of segments.

I plan to disconnect the PSU from the backplane tomorrow, and see if it produces sensible voltages that match those which I measured in an earlier post. In the meantime, please enjoy a few snaps from the current configuration:

[George Edmonds]

Hi Per Ardua

It is some four years since I last had to re capacitor an Adret power supply, even then I had to use snap in capacitors and solder them in.  Screw fixing ones were available at a stupid price but were of a much lower spec than the snap in ones so the use of snap in capacitors became a no brainer.

With what you are seeing with the power supply voltages my first reaction would be to check the continuity of all of the 0V/ground connections between the PSU board and the rest of the instrument given the crap connectors that Ardet used.


G Edmonds

[PerArdua]

Hi George, thanks for your message. It's quite interesting to see the difference material science has made in shrinking the volume of capacitors.

The connections between boards appeared good. I had soldered the ribbon cables between boards and bent the cables so as to not strain the soldered joints. All the same, I desoldered these and disconnected the connector to the OCXO PCB - essential leaving the PSU on its own. I powered it up and measured the following:

5Vfloating: 0.0V (measured from bottom)
+15Vpilot:  11.2V (measured from bottom)
+5V:          0.0V (measured from PSU Control PCB)
+15V:        1.3V (measured from PSU Control PCB)
-15V:        -3.4V (measured from PSU Control PCB)

I've attached a picture of my test setup - just to ensure I've not done anything completely stupid.

In this disconnected state the fan still turns on - but not sure what to make of that given the floating state of the PSU enable now that is separated from the CPU PCB. However, the lack of 5Vfloat and 15Vpilot leave me concerned that something isn't quite right. The AC input voltages are all correct, though I did notice that my meter (Fluke 15B+ captured some high startup transients for one measurement period - around 20VAC on the 10VAC rails, and 35VAC on the 20VAC), however I'm not sure whether to trust these readings - could they be some measurement artifact or something that could've caused damage?

The capacitors I used were:

6800uF: https://www.digikey.co.uk/en/products/detail/nichicon/UVR1H682MRD6/588859
10000uF: https://www.digikey.co.uk/en/products/detail/nichicon/UVR1V103MRD6/588834

I also noticed a repair on the PSU control PCB, It appears D22 has been resoldered, and C28 (next to it) resoldered or replaced. Circled in the attached image.

I'm thinking the next step is to get the 5Vfloat supply working again. I'm unsure the best way to do this given the difficulty in accessing the power transistor and the control PCB, all I can think to do is to solder wires to useful points and measure with everything in the instrument. Any times/suggestions would be very welcome.