No luck on my first attempt at acquiring a cesium frequency standard

[edpalmer42]

It does seem like Datum units aren't common.  I'm surprised and a little disappointed that we haven't heard from anyone else who has one.

The bezel in the picture is for a 3.2" LCD, but I'm wondering if a better way to go is with a 7" or 8" Windows tablet.  A few things to think about are:

- Cost can be < $100 (ebay).
- Use USB/serial adapter to connect to the 4040A's serial port.
- Drill one or two small holes in the front panel of the 4040A for USB and power cables.
     -Right angle cables would be sweet if you can find them!
- Use velcro or magnetic sheet to hold the tablet to the front panel.
- Can run Monitor3.
- Touchscreen means you could create your own program with whatever buttons you need.
- Forget about GPS to get the time.  Windows/Wifi can use NTP.
- Log data to internal flash.

A few issues/concerns

- Battery will die while the unit is powered down (assumes a day or two of operation every few months).
     - Cllock will be wrong since tablets usually don't have a CMOS battery - Windows doesn't like that.
     - Could keep it plugged in (storage mode).
- Monitor3 won't be very easy to read on a small screen.  A custom app tailored to the screen size would be better.
     - Could use VNC or remote desktop to view the screen from a different computer with a larger monitor.
         - You could even use a tablet with a broken screen!

Ed

[hydrogen maser]

Yes, I am surprised also that you and I are the only FTS/Datum owners here that want to discuss these units. I know how rare it is for a hobbyist to own a working Cesium standard so perhaps it is not that surprising after all. I get the impression that the 5045A is a pretty robust piece of hardware in general (other than the 1173 failing) and there have to be more of them out there in hackers hands somewhere (Yes, I have seen a few out there on the web like http://www.prc68.com but not many - and yes there has been discussion of them on the time-nuts list) . The other side of that is these were expensive pieces of equipment (don't know what the cost of a 4040A/RS new was but I have been curious to know) and I would assume most of them stayed in service until they ran out of cesium and were then disposed of because they were never going to work again. Another reason they might not end up on the surplus market that often is because cesium is a hazardous material and I assume many companies have policies about how to dispose of these kind of things due to various laws and regulations and such. Yes, there are plenty of different cesium standards for sale on ebay at any particular time but when you consider the world wide population vs how many of these units are for sale on the surplus market at any particular time you are really only talking about a tiny handful of devices. The vast majority of the cesium standards I have seen for sale are HP and the Datum/FTS ones are very much in the minority.

Still thinking about what to do with the 4040, I keep going back and forth on it. I want to do something fun with it but not quite sure what yet.

[edpalmer42]

The other side of that is these were expensive pieces of equipment (don't know what the cost of a 4040A/RS new was but I have been curious to know)

Check out this auction - particularly the last picture.  Oct. 2002 vintage, $24102. 

http://www.ebay.com/itm/FTS-4050-CESIUM-FREQUENCY-STANDARD-/331765084794

Ed

[hydrogen maser]

OK, so I think I have a general sort of plan now. Display to be mounted on the 4040: Mostly I know what I don't want for that at this point: something last looks chintzy. I want a proper industrial display and I am willing to pay more to get it. I think the best way to go is with a display that is as large as possible and is just that - a display only. There is not a lot of extra space inside the 4040 and I don't want to try to put a DC to DC converter in it and tap off the 4040 PS - I don't know how much margin there is in the 4040 PS so my plan is that everything is really external to the 4040. The only connection from the "4040" is the rs-232 and that is not really coming from the 4040, it is coming from the 5045A. Display on the front of 4040 with a hole drilled behind it - all wiring goes out the back and into another rackmount unit that has the computer for the display. That rackmount computer can be as sophisticated as necessary and have multiple video outputs, a larger one for detailed info and another one for the 4040 display. Not sure what to display on the 4040 yet, it likely will need a custom program for that. I may wish to mirror the rs-232 output from the 5045A into another system with multiple timebases on whatever bus I decide for that but I think the best way to go about this that has the largest chance for success is a modular system. I really need to rearrange my lab because I have outgrown it and have too many things stacked up so the rack mount system is something I have been thinking about for a while and it is just time to do that and get the counters and function gens and spectrum analyzer (TBD - 70000 HP looking like a nice option) mounted up and out of the way. So, a general plan is formed now.

[edpalmer42]

Sounds like an ambitious plan.  It'll be interesting to see how it develops.

Ed

[guido]

Got my hands on three (...) Datum PRS-45A units (2x Rev C and one J). Just powered up the first one (the J). Locked in some 20 minutes 
I already prepared a PC with Monitor3, that worked in one go. Lots of interesting numbers.

The J unit is from 2002, run time 33540 hours. That is only 3.8 years.  So it seems they did not run 24/7, or where switched off long ago. I know these units are not from a telco operator. I don't know how long they have been decommisioned. A piece of paper suggests 2007, but it is hard to read. Vacuum is good, the pump is now at 0uA after running for 1.5 hours. Got to 0 quickly.

Tube is a 7613A/077.

I see a Dallas RAM+battery. I'm wondering what data is present in there. Could be the log and the running hours. Need to look into replacing those (including copying of the data). I'll also backup the ROMs.

So which numbers are important? After one hour running:

• Clock servo Ramsey/Rabi errors are at some mVolts
• Zeeman Ramsey/Rabi are at -6 and -12. C field is 29mV
• Ion Pump is now at 0uA.

Empty alarm list (apart from one power supply that i did not connect).

Time to look at the other two...

[uncle_bob]

Hi

After a short run, vacuum and beam current are the two big things. Next on my list would be the EFC on the OCXO to see if it's drifted badly.

Bob

[guido]

Thanks.

Vacuum i presume is the Ion Pump current; stays at 0 uA.
Beam current I presume is C-Field (?), which is 18.90 mA, no idea at the moment if that is good or bad 
OCXO CV is stable at 3292mV

All of these seem stable for the 1.5 hours after warm up/lock.

[uncle_bob]

Hi

Vacuum is indeed the ion pump. OCXO at 3.3V is a bit high, but not worth doing anything about.

The C field is the magnetic field that the beam runs in. You "tune" the Cs by varying the C field. It's not worth digging into right now.

You should have a reading in there somewhere for the beam current. The beam is generated by "boiling" Cs off of a source inside the tube. As the source depletes, the current drops. Monitoring the current is one way to guess how much life a tube has left.

I'm an "HP Guy ( 5061's) " rather than a Datum owner. I can't help with the "where is the beam current in the menu?" question that obviously comes next. Of course if somebody sent me a couple of Datum's I *could* become a Datum guy 

Have Fun

Bob

[guido]

Looks like C-Field current is the beam current. The manual says that it gives an alarm >22 and <16V. Current in volts.. I assume it is a typo in the manual and it should say mA like the monitor software. So 19uA would be "half way", which somewhat matches with the run time.

Number two just locked.      I'll leave it running for 1.5 hours now.
OCXO at 2804mV, Ion Pump went up to 9uA (edit: probably much higher, but did not notice), but is now at 0. C-Field on this one is 19.1mA. Which would match the running hours, it's some 3.5 years on this one.

Also from 2002, but a Rev C. But same tube and same firmware.

[edpalmer42]

Congratulations on your score!  Don't bother buying any lottery tickets for the next lifetime or two - you've used up all your good luck! 

The only info I've got on the PRS-45A is some pictures, so take this info with a grain of salt.  The main signal strength is often called something like 'Ramsey Level', but Monitor3 doesn't use that term.  I think 'Clock Level' is the term it uses.  'Signal Variance' shows the noise level.  For a standard tube it should be < 200 or 225.  A high performance tube should be < 100.  As the tube ages, the value will slowly creep up and the Clock Level will drift down.  Eventually, the signal will become really noisy and the unit won't lock.

As Bob said, C-Field is a magnetic field used to fine-tune the frequency.  It has nothing to do with the signal level.

Take a look at this document.  http://www.ampere.com.mx/pdf/Manual_CsIII.pdf .  It gives some details about Monitor3 and the meaning of the terms and info.  Be careful about using any of the acceptable values since it's a different model.

I have no info on your tube or what an acceptable OCXO tuning voltage is, but 3V3 doesn't sound bad.

Ed

[guido]

And number three also locks without problem. This one is from 2001 and has some 3.9 years of run time.

Thanks for the document, i've got a lot of reading to do. When i search for "life" in the pdf, there are a number of parameters mentioned:

OCXO Control Voltage. Typical 2-4 VDC during lifetime. My values are 3.2, 2.8 and 3.0 after running for ~2 hours. So well within spec.

DAC Gain: 1.4 to 256, where 256 is end of useful life. My values are 3.88, 4.27 and 3.08. It's 256/N, where N varies from 180 to 0. So an N of 66, 60 and 83. "On" hours: 33540, 31071 and 34406. That does not add up, the one with the longest hours has the highest value. It would add up if N goes up, instead of down to 0. But you are right, this is for a different instrument. Maybe the value goes up here? I will see where it goes in the future.

Think the parameter for "Signal Variance" is Signal Deviation, or Deviation. 120-150 for a normal tube and 70-90 for a high performance tube it says in the doc.
I've got: 145, 141 and 147. But 120-150 is for the CsIII. You are mentioning 200/225, that's better And also the values went down during the hours they ran. E.g. from 200 after lock to 147. So it could well be going down further after more "on" time.

I found the loggin function in the monitor program. For number three i've made a logging for the whole time. You see the Ion Pump go up to 70uA, going down in the ~25 minutes it takes time to lock. When it locks the OCXO jumps to the control voltage and then slowly creeps down again (further warm up of the oven).

Unfortunately there are no such details in the PRS45 manual. And i did not find a manual for the PRS45A at all.

Some pictures:


The alarm is because there is only one power supply connection used (Fuse A open). No other alarms.









Yes it was a score. I payed peanuts for all three...

[notfaded1]

Dragging up this necro'd thread.  I hope some of you guys are still around?  I picked up an FTS Datum 4040A that's on it's way to me now.  I'm not sure what to expect being 20 years old though.  The pics make it look like it's been taken care of.  It was bought in a "freight auction" I'm not sure what that means?  When I get it what should I check?  Does anyone have the manual for this?  Many of the links in this thread were dead.  I did find the manual for the MODEL 4065C and it refers to the 4040A.  Here are some of the original pics.  I've always wanted one of these so I pulled the trigger... I figure it's worth a try right?  I did manage to download monitor.zip which is good.

Bill

[notfaded1]

I found a document that explains the difference between the 4040A and 4040A/RS... thankfully mines an RS which (it makes sense) is a reference to the serial RS-232 monitoring functionality.  This is a relief because reading here I was afraid I might not be able to interface directly to the clock at first.  I also found the original specs page for the clock.  I couldn't find a manual specific to 4040A/RS but the MODEL 4065C manual references it and I'm hoping the interface and commands are mostly similar.  I attached it here as well.  It seems to continue this previous thread made sense so all the information is here in one place.  I'm hoping after I start working on this that others might comment and potentially assist.


Bill

[edpalmer42]

Did you find this thread?  https://www.eevblog.com/forum/metrology/datum-fts-4040a/

When you get your unit, you'll be able to find out what's inside.  If it turns out to be the 5045A version you'll be able to talk to it via RS-232 using Monitor, or preferably, Monitor3 http://www.w2hx.com/x/Datum/PRS45A/Monitor3.zip .  Otherwise, it looks like there's not much in the way of monitoring/diagnostic.

Obvious first steps are to check for bad caps, power supply voltages, OCXO frequency & EFC, obvious burned/broken parts, loose ground connections, etc.  If you've got a high voltage probe and you're feeling brave you can check the high voltage supplies - carefully.

Be sure to read this entire thread.  There are many nuggets of info sprinkled through it.  For example, the manual for the 4040B at http://www.mediafire.com/download/6icdc7v35c68t16/4040B_mnl.pdf , failure modes, internal power supply voltages, etc.

What other measuring equipment and reference sources do you have?

Ed

[notfaded1]

Thanks so much for the response Ed.  I've been reading all of your posts on the threads about this.  Ironically my only other references are Symmetricom Syncservers S200's.  One has the x72 rubidium oscillator in it.  I have a scope, MM's and collection of old counters.  I really need another separate reference.  I've been reading about the software (thanks for posting the monitor3 btw... I've been looking for it all day and just sent a PM to EMUD asking him about it since all I could find was the original monitor.exe DOS program.  I'm just hoping that there's cesium left in the tube.  The serial number is 02013.  From what I can see it does say 4040A/RS on the front so I'm really hoping it's able to talk through the 25 pin female RS-232 port on the back.  I noticed in many other pictures it's a male connector but I suppose that doesn't mean much... just something I noticed.  It looks like it may require a null modem as well since the manuals say it's DTE to DTE in the diagram when hooking to a PC serial port.  I'll open it up and take pictures as soon as it gets here to AZ from UT.  It's being sent by UPS ground.  It must not be light... I could see in the manual for your unit it lists around 75lbs. if I'm not mistaken.  I've got a good known working USB to 9 PIN serial cable I use for the symmetricom units that should work well with an added 9pin to 25 pin serial converter and maybe null modem.  Oh and Happy almost New Year 2020!  History wise the seller said he acquired the unit in a freight auction and didn't have a use for it so was selling it.  The anticipation is killing me... I know it powers up and he said he only plugged it in just to see if it would turn on.  Fingers crossed!

Bill

[hydrogen maser]

Mine is also the 4040A/RS - it is silk screened as just "4040A" on the front but the label on the back says 4040A/RS and you do need a null modem cable to talk to it over serial. I got this cable from Amazon (I have db9 serial ports on the computer) and it works on mine: https://www.amazon.com/gp/product/B000I1EXT0/  monitor3 is the way to go if you can because monitor.exe only works in dos box on windows 7 (Win8/10 ) and you have to fiddle with a serial configuration file to get it to work but monitor3 just works - much easier. If yours is like mine monitor3 will misidentify your unit as being a different model and the log function will not work but otherwise it is much better than monitor.exe. Kind of on the fence as to - just fire it up and see what happens vs tear it apart and see if you have bad caps in the dc to dc converter in the 5045a like mine did - your call. BTW mine still works, I turn it on every 6 months and let it run until the pump current goes to zero. 

[edpalmer42]

Thanks so much for the response Ed.  I've been reading all of your posts on the threads about this.  Ironically my only other references are Symmetricom Syncservers S200's.  One has the x72 rubidium oscillator in it.

I'm not familiar with the Syncserver.  A quick browse through the datasheet didn't show any 10 MHz or 1 PPS outputs.  Are either available either externally or internally?

I have a scope, MM's and collection of old counters.  I really need another separate reference.

The scope will get you started.  If you can pull data out of one of the counters, that will be the next step.  You may have to improvise by using either the 4040A or the Syncserver as the reference for your counter while you measure the frequency of the other.  Not ideal, but you use what you've got.  Ideally, you use a Time Interval Counter to measure the time between the rising edges of the 1 PPS output from your reference and the DUT and collect the data on a computer.  The outputs won't (and *can't*) be in phase, but if the difference is constant, the frequencies are identical.

I'll open it up and take pictures as soon as it gets here to AZ from UT.  It's being sent by UPS ground.  It must not be light... I could see in the manual for your unit it lists around 75lbs. if I'm not mistaken.

Yes, it's a beast.  The UPS guy who delivered mine looked like Arnold Schwarzenegger!  Be sure to check it carefully for damage, both internal and external.  There's a good chance it will have been dropped - at least once! 

Remember that when you first power it up it'll have to pump down the tube before much else happens.  I'd recommend that you hook up the RS-232 right away, if possible, so that you can monitor the process.  The most important number initially is the ion pump current.

[CJay]

Threadjacking but I'm still in need of some info about the DC-DC converter in my Datum Cs standard, does anyone have one of these tand would be willing to check some things for me please?

I've a few pads/tracks damaged by the capacitor leakage and a diode that's got the cathode floating, it looks to have burned up in the past and been replaces but whoever did it seems to have pulled the thru hole plating out so there's no connection to the cathode.

[notfaded1]

I have a tracking number and the shipping details says:  Weight 50.40 LBS.  It says it'll be delivered by Friday so fingers crossed.  It's one of those label created tracking messages though so who knows if it's really moving yet?  Thanks for the tip on the null modem cable HM...  weird that mine has female DB25... not that it really matters but most pics of units I've seen it's male.

I'm guessing that some of my counters that have A, B, and C inputs that looking at an A/B ratio might tell something to compare a one source to another but I know that generally frequency counters and interval counters aren't the same but are similar and use similar components.  For used the HP5370A and B models look pretty good?  Wow some counters specifically called time interval counters can get expensive too!  The SRS SR620 look like the more recent cadillac of TIC's!  I've always been somewhat of a voltnut but now I've really got bitten by the timenut bug... in my day job we do a lot to support the brand new Block III GPS satellite constellation so it just makes sense to have an interest in this and time in general.

I'm thinking it'll be normal to have a bunch of alarms on unit until the tube pumps down?  I've read (by you guys) it can take even days to pump down if it's been sitting for a long time?  Geesh I still can't believe it's 2020 already o.O!  Here's to hoping I can track 2020 with precision I didn't have in 2019!

Best Regards,

Bill

[edpalmer42]

I'm guessing that some of my counters that have A, B, and C inputs that looking at an A/B ratio might tell something to compare a one source to another but I know that generally frequency counters and interval counters aren't the same but are similar and use similar components.  For used the HP5370A and B models look pretty good?  Wow some counters specifically called time interval counters can get expensive too!  The SRS SR620 look like the more recent cadillac of TIC's!  I've always been somewhat of a voltnut but now I've really got bitten by the timenut bug... in my day job we do a lot to support the brand new Block III GPS satellite constellation so it just makes sense to have an interest in this and time in general.

If your counter has A and B channels, it probably has a Time Interval mode.  Good counters can be pricey, but you can use almost any counter at this point so don't rush out and buy one yet.  A good start would be a counter that has a minimum resolution of 1 ns. like the Racal Dana 1991/1992 or Agilent 53131.  Many of the Racal units have bad buttons so move carefully.  An economical alternative is the TAPR TICC.  The SR620 is one of the best available.  The 5370B is close behind.  I have a Fluke PM6681 which is an OEM version of a Pendulum counter.  Tektronix also makes some high end counters.  If you want something big and heavy enough to compete with your 4040A, you could even get a Wavecrest DTS unit.  Warning:  it's weird and it's LOUD, but it does have potential to be the best of the bunch.  Mine has an ADEV noise floor of about 5e-12@1 sec.  You can even sort of use it as a digital scope with 100 Gsps equivalent sampling!  Don't buy one without talking with me.  All of these units are available on the auction site we love to hate.

What counters do you have access to?

I'm thinking it'll be normal to have a bunch of alarms on unit until the tube pumps down?  I've read (by you guys) it can take even days to pump down if it's been sitting for a long time?  Geesh I still can't believe it's 2020 already o.O!  Here's to hoping I can track 2020 with precision I didn't have in 2019!

My 4065A always starts with an alarm that just tells you that it restarted (in case there was a power failure).  Depending on the state of your tube, it could take many, many days to pump down the tube.  I just dug mine out.  I didn't realize that it had been so long since I ran it.  After sitting idle for 30 months, the Ion Pump current peaked at only 101 ua and by the time the unit locked (about 10 min.) the current was zero.

I was watching the startup behaviour very carefully and learned something new.  The first thing that happens is the OCXO warms up and changes from a COLD state to a WARM state.  That takes about 10 minutes.  Once that happens, the Ion Pump and Cesium oven start up.  About ten minutes after that, the unit locks.  If you're monitoring the output frequency, you can see it rise slowly to within a few Hz of 10 MHz, then stall for a while, and finally swing above and below 10 MHz before it locks.  So total time from power-on to lock was about 20 minutes.

Always remember the first rule of Cs standards.  Don't run it unless you need to.  It has a limited amount of Cesium and whenever it's running, you're using up that supply.  Once that's gone, the unit is junk.  There's no way to find out how much Cesium is left in your tube.  Some (most?) newer units have runtime counters, but that still doesn't tell you how much is left.

[notfaded1]

I have many counters (I especially collect nixie tube counters) but since you mentioned it I have a Racal Dana 1991... it's kinda strange on it's own because it has a 1992 body with 3 inputs on it but inside it actually only has 2... it does have the OCXO inside the back corner which I think was an option?  When it boots up it says 1991 not 1992 like the outside markings say.  I thought about giving the guy that sold it to me bad feedback but I don't think he really knew and I got it cheap.  I'm pretty sure it'll work like a 1991.  Oh and yeah I know about the buttons on the Racal's... mine are ok.  I actually did get some replacement buttons I was going to install on it but I think mine are good enough that it works at least now.  I actually bought a HAKKO FR-301 desoldering gun just for the job because its a LOT of buttons and a LOT of connectors!

Regards,

Bill

[edpalmer42]

Threadjacking but I'm still in need of some info about the DC-DC converter in my Datum Cs standard, does anyone have one of these tand would be willing to check some things for me please?

I've a few pads/tracks damaged by the capacitor leakage and a diode that's got the cathode floating, it looks to have burned up in the past and been replaces but whoever did it seems to have pulled the thru hole plating out so there's no connection to the cathode.

A model number and a picture or two would be helpful.

[notfaded1]

Just got the 4040A out of huge box and cleared enough space to put this monster.  I haven't opened it up yet maybe tomorrow but I wanted to plug it in at least and see what happened.  I did hook up the RS-232 at 2400 7 2 Odd with null modem cable and didn't see anything on putty at all.  I'm wondering if I should wait a while and see what happens or?  I had to hook up the Racal Dana 1991 though and at least look at the 10MHz output.  Attached is the picture.  When I plug it in I hear a brief click then that's really it and you see the lights on right now.  I kinda expected it wasn't going to just lock up in under 30 minutes.  The align/initialization and operation alarm/fatal alarm indicators are LEDs and buttons.  I'm guessing it's been sitting a while.  Weird thing on the back the storage/operation switch isn't there it has a blank in the space.  I'll take more pics tomorrow.  I guess leaving it on for now won't hurt anything.

Bill

[notfaded1]

Here's that normal storage where switch is on some units.

Bill