Datum 4040A Problems

[edpalmer42]

I might be wrong, but believe it's less frequencies and not twenty-four.

I didn't believe it either but .....

Fr = Ramsey (main) peak frequency
Fz = Zeeman peak frequency

Fr +- 5 KHz - Initial Ramsey frequency lock
Fr +- unstated, but much less than 5 KHz - fine Ramsey frequency lock
11 distinct frequencies - (values unstated) to measure / adjust
   - C - field
   - Ramsey peak to background
   - Zeeman peak to background
   - Central peak symmetry
   - Clock resonance pedestal symmetry (whatever that is!)
   - Microwave power level

Fz +- 7 KHz - initial Zeeman frequency lock
Fz +- 350 Hz - fine Zeeman frequency lock

Fr +- 386 Hz - test to ensure that you're actually locked to the Ramsey peak

So, 2 + 2 +11 + 2 + 2 + 2 = 21

My mistake.  I double-counted the Ramsey and Zeeman peak / background ratio frequencies.  They compare the Ramsey peak to the background level and the Zeeman peak to the same background level .  Those three frequencies were included among the '11 distinct frequencies'.

   

[ITDH]

The process is described in the 4065C manual.

WAIT - WHAT there is a Manual? I had no idea - I've just downloaded it and have only briefly skimmed though it - however its a GOLD mine of information, explaining SO much!!!

If I had known earlier it would have made my job of annotating the PCB schematic SO much easier 

Thank you for pointing it out to me - I'm a very happy Bunny!

I'll update the schematics with clearer annotations

[bostonman]

Be aware, while I haven't read it or proof read it, at some point manuals were combined with pages from other manuals.

One quick example is page A-17 is a screen capture from really old monitoring software. Still equally educational, but you may find errors that point to areas that aren't in the manual or wrong references.

You're lucky to have found that because as time went on, the manuals shrunk.

[edpalmer42]

The process is described in the 4065C manual.

WAIT - WHAT there is a Manual? I had no idea - I've just downloaded it and have only briefly skimmed though it - however its a GOLD mine of information, explaining SO much!!!

In a situation like this, I make a point of spending days sniffing around online to find every shred of information I can - every document, every mention of the product (or related products).  I look here on eevblog and of course, for Time-Nuts stuff I check the archive of the Time-Nuts mailing list.  Just asking for help isn't enough.  You have to do the digging to find the gems.  Weeks later, I do it again to see if I can stumble across anything I didn't find the first time.  When I run into broken links, I get on a 'first name' basis with the Wayback Machine.   

In this case, I downloaded the manuals of every FTS/Datum Cs standard I could find.  There's a family resemblance between them so lots of the info is transferable.  I found a manual for the CsIII that describes some of the parameters that are poorly covered elsewhere.  The 4060 is a totally different beast from ours, but the service manual has schematics!  Interesting!   

[ITDH]

Your making me feel old This new fangled "Internet" - I just took it as fact that there was little to no information on the 5045 Module apart from what was published in the marketing and sales material...

The user's manual is very well written - obviously from a time when engineer's where still allowed "In the Loop" - I wish I was aware of the manual before redrawing the schematic as it would have eased with the task of annotating the signal lines... but at the end of the day it made me smile as it confirms my understanding of the unit's operation..

When you said that there where 24 different frequency's I erroneously interpreted this as meaning a continuous "Multitone" Spread Spectrum type signal - but in reality "Frames" are periodically stolen on a ~650mS basis to squeeze in House keeping / System performance check "frequency hops" for up to a couple of Frames (each frame is 17.92mS wide - so the Frame cycle rate is 55.8036Hz)  - The repetition rate of 55.8036 Hz is filtered out by the longer time constant VCOCXO PLL loop (Apparently 5 seconds is the "default" loop filter TC setting).

I'm Looking forward to getting my unit operational (I wonder what the DAC Gain setting will be - as this is indicative of my units CBT health) and then being able to make some measurements to confirm "our" understanding of the units operation. Yesterday, I finally had time to order the 74HC series logic from Mouser - the shipment should arrive Monday or Tuesday - however here in Monaco, France handles the customs processing which is never good and is much worst at this time of the year as everyone is on Holidays

[bostonman]

I downloaded the manuals of every FTS/Datum Cs standard I could find.

You'll find most manuals have the same information, and, as previously stated, probably errors in newer ones due to combining pages from old to new. The errors I think were just page numbering and general stuff; and maybe you'll see the font is different.

The old manuals were much better and larger with more emphasis on the theory of operation. When CSIII (i.e. 4310, PRS4500, etc..) came out, the manuals became very thin with much less information.

For many years customer support was handled locally via phone, email, and 24/7 cell, with both sides having technical discussions trying to walk customers through issues in hopes of solving the problem without returning the unit.

In later years they decided to let new hires support the product. They got "trained" for about two-days on what a Cesium unit is and how to support every make/model (and not just Cesium - everything, including units that had custom settings/functions per individual customer needs); then moved support to CA. This timeline is basically when manuals became much lighter and phone support was more or less taking customer information so they can immediately issue an RMA.

Also, in CSIII, the 9.192GHz is generated differently. Old style uses a multiplier and the new style is technically a divider. I could be wrong, but you won't find much information regarding this design; and the oscillators switched to 5MHz.

On a side note, it was stated at one time not to bang the unit when the tube is hot because the ionizer could break. I forgot the time frame to allow the unit to cool, but it's good practice to leave it to cool. I think this was more regarding shipping the unit and the banging that occurs such as boxing and stacking it onto a delivery truck.

[ITDH]

Components arrived and I rebuilt the PCB last night - replacing all the CMOS logic IC, UART, PIA, the failed SRAM and added PLCC / DIL sockets etc.

The logic IC tested OK when removed from the PCB - but after being stressed by overvoltage I'd not trust they will still meet there original performance or trust in there long-term reliability.

Failed parts so far found on the Servo board:-

74HC74

65764 (SRAM)

1uF 35V Tant Cap.

Finger's crossed when I test later this evening - I wonder how its going to go!!! will I be in luck? - I'd not bet on it!!! 

[ITDH]

Progress update on my FTS4040 repairs - this CS clock is now just playing with me!

So after rebuilding the PCB and replacing the SRAM I have for the first time a initialization LED and after say 10 minute's the unit goes to lock) for a couple of seconds then Major fault

Communication with the Monitor3 is very hit and miss as on each powerup or Reset the Baud rate is different - I've even seen a instance when the RX / TX Baud rate differed, and the Clock select pins to the HD-4702 is constantly changing during TX / RX (or this could have been happenstance - but for sure they where alternating at a constant cyclic rate).

The EEPROM is also starting to full with what appears as nonsensical data - I'm presuming some memory field is related to the serial Baud rate - is it possible to have different Baud rate for TX / RX modes?

When the unit briefly locked, I noticed a large frequency error in the 10's of Hz... which then had me doubted my GPS locked lab standard. I opened Lady Heather and YES my GPSDO was indicating PLL unlocked (plenty of active satellites) - sure enough the VCXO DAC was at 97% meaning the Samsung STP2945 LF OCXO in the GPSDO had drifted too far off frequency for the GPSDO to compensate... I did a quick search and others had reported the same issue with this OCXO. There is no accessible trimmer to re-center the OCXO frequency - so the OCXO is dead. I'll have to order a replacement from eBay.

I'm lucky that I had Leo Bodnar's Mini Precision GPS Reference Clock on hand to use as a temporary replacement - these are great little units and "just work", Highly recommended as a "simple and affordable" GPSDO.

So - one dead OCXO... I don't know why, but I then removed the Datum 1111B (HP 10811 "type" OCXO) from the FTS4040 and decided to test it on the bench.

On first power up with 24Vdc the OCXO heated with 520mA current that quickly dropped to 28mA - but 10's of Hz negative Frequency offset - and each time I powered the OCXO from cold the time required for initial heating appeared to be getting shorter and shorter.

I could find no information on the Datum 1111B - so having nothing to work with and no clue, I opened the Oscillator and started from basics...

The HP 10811 should heat at a current limited 500mA which then drops to around 120mA when reaching operating temp - however the 1111B is a different design so I was not sure how relevant this is.

SC cut crystals are very temperature sensitive and have turn-over point around 85Deg - my 1111B was internally marked with as 82.1Deg. For sure my "oven" was nowhere near 82 DegC and even with a negative 10V on the VC pin I could not pull the Frequency to 10MHz... (the 5045 servo controller can pull the OCXO VC pin about +/- 4500mV).

Having little to work with, I started by removing the Semi-conductors and Tantalum Capacitors to measure them out of circuit - when I noticed this:-



As I was desoldering the Tant Cap. (C11) I noticed that the Positive band was connected to the Ground plane!! hold on there's no negative supply in the 1111B? Looking closer to the silkscreen markings and YES the Tantalum had been fitted in the wrong orientation from the factory 

I quickly traced out the circuit and sure enough the Tant Cap. formed an RC filter on the Gate of MOSFET in line with one arm of the Temperature compensation Bridge.. I'm guessing as part of the current limiting circuit.

I replaced the Tant cap - reassembled the unit and and on testing I measured 520mA on power up - which then after a few minutes reduced to about 110mA just as the frequency was nearing -0.5Hz offset. I trimmed the OCXO to 10MHz with -410mV on the VC pin, and the oscillator was nice and stable

So in one day I've had 2 bad OCXO's - Surely I'm being Trolled by OCXO's here!!

So does the unit work now? - Off Course Not!!! - with no apparently difference - looses lock after 1-2 seconds after coming out if Initialization (and with a large frequency offset)...

However, I now believe that the original problem (where the unit would randomly start to loose lock) was the failing OCXO - and not the PSU. Sure the capacitor's in the PSU where in a bad way, and sure the unit has been subsequently further damaged due to the failed PSU repair - but I now believe that the OCXO was the initial culprit.

I'm at a bit of a loss ATM - I'll give it a break for awhile to gather some thoughts and ponder my life choices - I also need to get to the root cause of the misbehaving MCU section.

[edpalmer42]

Oh, the drama! 

A month ago, a Datum 4040A/RS sold on ebay for $1300 with a guarantee of full functionality.  Today, I see the same seller is selling the same model.  The pictures show the same serial number, but the new description doesn't mention a guarantee.  But both auctions are listed as 'used' rather than 'parts'.

I wonder if the unit is working......

Hmmmmmmm.....