Upgrading Counter 53220A to Option 010 Ultra High-stability OCXO Timebase?

[carl_lab]

Does anybody know, if you can buy the oscillator module as a spare part (P/N, cost?) and is it sufficient to swap the modules?
Or is additional configuration change requided?

[plesa]

It require new motherboard.
You needs to replace 53220-66111 ($1808) with 53220-66112 ($1929).
So external reference ( like use rubidium will be much cheaper ).
Or open instrument and check what is missing.

[carl_lab]

It require new motherboard.
You needs to replace 53220-66111 ($1808) with 53220-66112 ($1929).

Thanks for your reply.

That's not, what I expected. On old 53131A replacing the oscillator module did the job, nothing else required.

On 53220A mainboard a standard 4-pin-quartz-oscillator is implemented (DIL footprint).
But there is an additional empty footprint for a bigger oscillator module (19x19mm=3/4x3/4", 5 pins).
So changing the module is a possible way to go?

Any pictures of an Opt. 010 Ultra High-stability OCXO Timebase available?

So external reference ( like use rubidium will be much cheaper ).

Do you know a "cheap" rubidium standard with calibration certificate?

The counters will be installed in production ATE - not in the calibration lab, where an existing external frequency reference could be used.

[plesa]

Try to contact Agilent, in manual they are no so strict and maybe they will populate the OCXO instead of board swap.
Manual says :  order 53200U-010.
If you needs certificate I cannot help you.

[HighVoltage]

I have a 53220A without OCXO and the drift was horrible but an upgrade was expensive.
Actually, I got a 53230A with installed OCXO for less money than then upgrade of the 53220A.

[carl_lab]

I have a 53220A without OCXO and the drift was horrible but an upgrade was expensive.
Actually, I got a 53230A with installed OCXO for less money than then upgrade of the 53220A.

Can you post a picture of the OCXO and its footprint on PCB, please?

[dr.diesel]

You can get Trimble GPSDOs on eBay for < $150 with antenna, at least in the US.

[carl_lab]

We cannot install separate GPS antennas for each production ATE using a 53220A.

So rubidium maybe an alternative...

But:
I looked at the 53200A mainboard a little bit closer.

There are 3 jumpers (0-Ohm chip resistors) with instruction nearby:
"Remove R827 and R828 when installing Y800".
Looked at the other side of PCB, and the bigger oscillator footprint is named "Y800" - tada!

So I try to find a parts list to identify P/N of Y800...

[carl_lab]

Try to contact Agilent, in manual they are no so strict and maybe they will populate the OCXO instead of board swap.
Manual says :  order 53200U-010.
If you needs certificate I cannot help you.

OK, found it in the component level manual:

"Retrofitting to Option 010 (Ultra-High Stability Timebase) must
be performed by technicians at a Keysight Technologies Service
Center. Refer to the list of Keysight Sales and Service offices in
the Appendix for the office closest to you.
NOTE To order the Ultra-High Stability Option 010, you will need to
place an order with Keysight for 53200U-010 and identify
which counter you have (53210A, 53220A, or 53230A)."

Those cowards don't want to sell the OCXO as a spare part? 

[plesa]

Try to contact Agilent, in manual they are no so strict and maybe they will populate the OCXO instead of board swap.
Manual says :  order 53200U-010.
If you needs certificate I cannot help you.

OK, found it in the component level manual:

"Retrofitting to Option 010 (Ultra-High Stability Timebase) must
be performed by technicians at a Keysight Technologies Service
Center. Refer to the list of Keysight Sales and Service offices in
the Appendix for the office closest to you.
NOTE To order the Ultra-High Stability Option 010, you will need to
place an order with Keysight for 53200U-010 and identify
which counter you have (53210A, 53220A, or 53230A)."

Those cowards don't want to sell the OXCO as a spare part? 

BTW Rubidium has poor short term stability and better longterm stability and OCXO has better short term and worse long term. So check what you needs.
Question is if you really needs certificate I do not think KS as default providing any certificate and calibration will be additional fee.
Do you know Timelab software?

[HighVoltage]

I have a 53220A without OCXO and the drift was horrible but an upgrade was expensive.
Actually, I got a 53230A with installed OCXO for less money than then upgrade of the 53220A.

Can you post a picture of the OCXO and its footprint on PCB, please?

Both of  my counters are still under warranty and have Keysight calibration stickers.
I am hesitant to remove them.
May be I will try Dave's way of removing them in a safe way, when I am back in my lab tomorrow.

[carl_lab]

Question is if you really needs certificate I do not think KS as default providing any certificate and calibration will be additional fee.

I'm not quite sure about.
Maybe changing the OCXO to a 3rd-party part and calibrating the counter with tolerances of the original Opt. 010 is sufficiant. 2000 bucks for upgrading is "just a little bit" too expensive for my understanding...

Main purpose of the 53220A counters is adjustment of clocks.

Do you know Timelab software?

No. It's not a laboratory application, so how could this software help anyway?

[carl_lab]

Both of  my counters are still under warranty and have Keysight calibration stickers.
I am hesitant to remove them.
May be I will try Dave's way of removing them in a safe way, when I am back in my lab tomorrow.

Oh, please don't do it!
I don't want to be responsible, if you lose your warranty.

[nfmax]

It looks like these counters have a reference frequency input. Can you not use an external reference source, with suitable calibration certificate? This could be almost anything with a sufficiently stable oscillator and a reference output. You might need to use a distribution amplifier, depending on the cable lengths, but for just two inputs a power splitter should be OK.

[lukier]

We cannot install separate GPS antennas for each production ATE using a 53220A.

Sounds extravagant, I think typically you would have one GPSDO for the facility and then a distribution amplifier to spread the 10 MHz around to the counters.

[carl_lab]

I think typically you would have one GPSDO for the facility and then a distribution amplifier to spread the 10 MHz around to the counters.

You're right, that's the better way of course, but I think installation of an antenna on roof of the production hall will not be allowed whilst other alternatives are possible (rubidium?).

[nfmax]

We use a lot of GPS aerials here, mounted inside a typical industrial estate metal-clad building with plastic roof light panels. Presumably enough signal makes it through the polycarbonate (?) panels to make it work. We use GPS to control the time-of-day clocks on the data acquisition systems we make, so we need the 'real thing' for testing, not just a GPSDO.

[Dr. Frank]

,...
Main purpose of the 53220A counters is adjustment of vehicle clocks (dashboard instruments).
....

Hi, for which company are you working?

Frank

[Dr. Frank]

I think typically you would have one GPSDO for the facility and then a distribution amplifier to spread the 10 MHz around to the counters.

You're right, that's the better way of course, but I think installation of an antenna on roof of the production hall will not be allowed whilst other alternatives are possible (rubidium?).

I think, for that purpose, an internal or external OCXO is fully sufficient, Rb or GPSDO is overkill.
Requirement for the uncertainty of this clock adjustment is on the order of 0.1ppm. Maybe a frequently calibrated TCXO will also do the job.
This can easily be done by means of a GPSDO, but in the cal lab, where it belongs to.
Does your company have a lack of budget, as such an instrument w/o proper OCXO had been ordered? 

Frank

[plesa]

I just tried to disassemble unit (stickers are not problem, they are not brittle) but the OCXO is below RF module So sorry I will not remove it.
Like other mentioned only R854 0R resistor is placed on bottom. Module footprint is really small.

[dr.diesel]

In this case I'd suggest no personal modifications, in some obscure innocent you could be liable if something went wrong.  If this is something you need for production, order the correct instrument or use a proper external timebase.

[Vgkid]

I will agree wit Dr. Diesel, though you could always make a small enclosure and mount a small ocxo inside of that. Though with the dip 14 sized ocxo's you could easily breadboard one inside of the unit...

[HighVoltage]

OK, Plesa was right, these stickers come off very easily, so, what is life without a risk.

Here are some pictures of my 53230A with all options installed
It seems the UOCXO is hiding well.

[SoundTech-LG]

Nice photos.
Use my TCXO equipped 53230A with EXT Ref Rubidium, or GPSDO...

[carl_lab]

Thanks for the photos.

As I supposed bigger footprint is used for (U)OCXO, inner one for standard quartz oscillator.



Could you be so kind trying to look at the (U)OCXO for identifying maker and type?
(I don't know if removing  removing RF module is required)

[plesa]

For type of U OCXO you must remove RF module and after that plastic cap. You are fight that standard OCXO use the inner holes and it is much smaller.
Maybe try to look into to other service manual for Keysight gear for type.
I suppose they are using same UOCXO across platforms, like hey did it with 10811.

BTW why you are trying to modify some production equipment which is probably covered by warranty?
Did you asked repair center for quote?
Why is not external UOCXO an option?

[HighVoltage]

OK, since I was curious as well, I took it apart again and removed the GPIB and 6 GHz module.
The UOCXO is nicely covered with a plastic hood.
Here are the pictures

The UOCXO is of following type
- VECTRON
- C4550A2-0281
- 10.000000 MHz
- 22158-0097
- Made in 1240

[HighVoltage]

More pictures

[HighVoltage]

A few pictures of the GPIB interface board.

[HighVoltage]

Interestingly, there is a CR2032 on the Front Panel

Ok, now back together ...

[Dr. Frank]

HighVoltage, that's very interesting.. could you please also provide the dimensions of this OCXO, the pinning, and the output waveform (square wave or sine)?

I think, VECTRON offers OCXOs with standard pinning, which might fit, but with customized specifications for KS.
Maybe something like their OX-208: https://www.vectron.com/products/ocxo/OX-208.pdf

Frank

[carl_lab]

Thank you very much!
Now I can try to get those (U)OCXOs as spare parts.
1900+ euro (as offered by Keysight distributor) for changing one of these seem to be a bit over the top...
 
I don't know which jerk ordered counters without OCXO for clock adjustment... 
What I know, those counters replaced out-dated 53131 opt. 010.
The specs of 53220A (without option) is inbetween 53131 without option and 54131 opt. 010.

I'm not sure about warranty. The counters maybe out in the meantime...
I also don't know, if warranty stickers are attached.  Most gear I see here in the lab is without stickers...

[carl_lab]

HighVoltage, that's very interesting.. could you please also provide the dimensions of this OCXO, the pinning, and the output waveform (square wave or sine)?

Pinning is 19x19mm, case dimensions 1"x1" (Vectron C4550A2).
-> www.vectron.com/products/ocxo/C4550.pdf



The grey plasic cap seems to be the same, that is used with standard oscillator.

[HighVoltage]

I don't know which jerk ordered counters without OCXO for clock adjustment... 
What I know, those counters replaced out-dated 54131 opt. 010.
The specs of 53220A (without option) is inbetween 54131 without option and 54131 opt. 010.

You mean 53131?

The 53131 without opt 010 was as bad as the 53220A without option 010

After a few month of being off, I turned my HP GPS Time Domain on again today.
Here are my three counters after a few hours of warmup and GPS-Lock on the HP Z3805A.

The 53132A was upgraded by me here:
https://www.eevblog.com/forum/repair/agilent-53132a-repair-(and-possible-upgrade)/msg843138/#msg843138
This one is right on target in the shortest possible time.
Every single time I turn it on, it is fluctuating only the last 2 digits

The 53230A with all options was the one I opened in this thread
This one takes much longer to warmup than the 53132A to be stable after a cold start
But after one hour it has a stability about the same as the 53132A

The 53220A is without Option 010 and is horrible in comparison
After every cold start it finds a different spot where it settles
Interestingly, this one was just repaired and calibrated at Keysight Germany in (warranty) but
is far off. Why having 15 digits, when the last 6 have no meaning?

[carl_lab]

The 53220A is without Option 010 and is horrible in comparison
After every cold start it finds a different spot where it settles
Interestingly, this one was just repaired and calibrated at Keysight Germany in (warranty) but
is far off. Why having 15 digits, when the last 6 have no meaning?

7 digits, not 6.   

You mean 53131?

Yes, exactly: 53131A.

Bad combination:
Laptop keyboard and my "wurstfingers"...

[carl_lab]

could you please also provide...output waveform (square wave or sine)?

Waveform is square 5Vpp, supply voltage 5V.

[andyturk]

My project for the evening was opening up a 53220A with the standard timebase and dropping in an "aftermarket" OCXO. The specs on the Abracon AOCJY3A-10.000MHz-E seem pretty close to the original Vectron part used by Keysight, so I figured I'd give it a try. At worst, I'll have to open it back up and reverse the surgery.

After replacing the oscillator and removing the two 0-ohm resistors called out on the silkscreen, the unit booted right up and passed its self test. It also registers the 010 option on the "About" screen. Surprisingly, it didn't complain about being out of calibration, but it was clearly off by ~2 Hz.

I let it warm up for a while and calibrated it with 10Mhz from a rubidium source (PRS10). It seems to work well, although I'm not a time nut and don't have anything better than PRS10 to compare it to.

[TheSteve]

Always good to have more options. I'd give this a try if I was looking for one - https://www.ebay.com/itm/181776010665

[HalFET]

Rubidium reference modules run <300 bucks from the surplus telecom stockpile these days. So it'd most like be cheaper to just strap one of those to the top of the unit and get that the rubidium reference certified. Added advantage is that marketing can then claim things were calibrated atomic clocks.

[HighVoltage]

Great to know, that the replacement worked
Thanks for this info, andyturk

One of these days I will get a Rubidium standard as well.
I just have no idea, which one to get, there are too many options.

[swixo]

Going to try this upgrade today. 
Got my vacuum desolder, RU GPSDO reference for calibration and the part on its way. 
Will update how it goes.

The stock oscillator is absolute crap.  Couple that with a sloppy at best factory calibration,
I am a little disappointed in Keysight here.

[swixo]

Well it works!

Swapped out the small TCXO and installed the OCXO.  Cleaned up the solder flux
mess -- it looks super clean.

Self tests pass.  About box says OCXO installed!
System is MUCH More stable after warm-up, which takes about 45 minutes.  No issue there.
Calibrated with RU standard and I am good to go.  Total cost about 320$ which included
Saturday delivery of the part -- AND the Vacuum desolder!

[HighVoltage]

What model and type of OCXO did you install?
Pictures?

[swixo]

I used this:

https://www.mouser.com/productdetail/815-aocjy3a-10m-e

Didnt take 'after' pix, sorry.  But the pix earlier in this thread are spot-on.

[HighVoltage]

Good to know, that this mouser supplied part works well for the upgrade.
I assume you did not have to change any settings and the instrument detected the new OCXO being installed.

[swixo]

The procedure is changing the TCXO for the OCXO, then removing the two
zero ohm resistors near the clock zone. 

The firmware detects that change all by itself.  Whether it is the presence of the
new part alone - or the removal of the resistors is not known.

s

[TheSteve]

FYI the Vectron C4550A1-0213 I linked to above works just fine, and was cheap. I got mine from a different seller for 17 bucks though.

[swixo]

Steve - Did you install one of those EBay OXCOs in an instrument?  How's the drift?

[TheSteve]

Steve - Did you install one of those EBay OXCOs in an instrument?  How's the drift?

Yes I have one in a 53230A. I haven't checked drift lately, I'll fire up the unit and see how it looks. When I also had a factory 010 here it was every bit as good. Could be a crap shoot with a used ebay sourced one I suppose.

edit:

Just fired it up and compared it to my GPSDO. I do have the standby mode selected that keeps the OCXO warm. It has been more then 6 months since the last cal from the same GPSDO.

[swixo]

Seems like a lot of drift, doesn't it?

[TheSteve]

Seems like a lot of drift, doesn't it?

Not at all from what I can tell. The specs of the factory unit gives +- 50 ppb per year and another +-50 ppb from cal. That would be +- 1 Hz. Now this is only half a year but the error is only 0.05 Hz - seems pretty darn good to me.

[swixo]

Fair.

But when your unit can show 15 digits, it seems like a lot.  I am looking at this OCXO:

http://www.morion-us.com/catalog_pdf/mv291.pdf

The ULN version is 8E-13 short term.  Imagine THAT in your counter.  This
unit appears to be able to fit in the spot for the OCXO, and will fit under the
cover.  I have another Morion OXCO, and it clocks in the E-12 all the time.

s

[TheSteve]

In my case it is all just for fun, my entire lab is locked to the same reference source(GPSDO or Rb.)

[swixo]

All of this is just for fun!  I lock to an SRS FS740, super stable. 

Also want the built in oscillator to be good!

[EE-digger]

Good thread.  I have a 53132A with a Ebay - Polish reference osc. and Rb lab reference for now.

I'm disturbed by Agilent / KS no clean flux residue.  I hope our scopes aren't full of that stuff.  I don't trust it.  Yes, it should be benign (low SIR, good ROSE numbers) but it also acts as a "gatherer" for dust, dirt, stray solder bits (ok, this can be good  ).

[Andrew]

I'm considering a similar upgrade to a 53220A using the Abracon AOCJY3 10ppb part.

A couple of questions in relation to ocxo's:

1) Abracon offer a 5ppb option in the AOCJY3 series however it's temperature range is restricted to 0 to 50C. The 10ppb part is -20 to 70C. What would be the maximum ambient temperature that the 50C part reliably operate?

2) The lock range to an external reference is narrower for the oem ocxo part, 0.2ppm, versus 2ppm for the tcxo part. Does this mean there is some performance gain using the ocxo even with an external reference? Perhaps this would be noticable on the faster 53230 model?

Thanks
Andrew

[jpb]

If you're using the counter indoors I'd think that the 0 to 50C part should be fine (temperature wise - I don't know about its suitability generally), it is very unlikely that the internals of a counter would get significantly hotter than the ambient - they are not high power devices and the main source of heat would probably be the ocxo but it would be in quite a lot of empty space. If you're planning on using the counter in on expeditions to the Sahara or the South Pole it might be better to use the wider temperature range or if you keep the counter in a rack with hot, high power gear on either side.

The narrow lock range is because ocxos generally have a much narrower tuning range (frequency versus voltage gain). They are better quality crystals (SC cut) but for the purposes of a counter there would be no advantage when it is locked to an external source as the external source will determine the timing over longer periods such as that of the counter gate. For RF generators the phase noise is more important but for a counter it is of less importance.

Here is a quote from the Stanford Research Systems SR620 counter manual:

EXTERNAL TIMEBASES
The SR620 has a rear panel input that will accept either a 5 or 10Mhz external timebase.  The SR620 phase-
locks its internal timebase to this reference.  The phase-locked loop has a bandwidth of about 20Hz and thus
the  characteristics  the  the  SR620's  clock,  for  measurement  times  longer  than  50ms,  become  that  of  the 
external  source.    For  shorter  measurement  times  the  clock  characteristics  are  unimportant  compared  to  the 
internal jitter (25ps rms) of the SR620.  Thus, if the signal from a Cesium clock is input into a SR620 with a 
standard TCXO oscillator the short-term and long-term stability of the SR620 will become that of the Cesium clock. 

[Andrew]

Thanks jpb that answers many questions.

Interesting that measurement times shorter than the PLL bandwidth are not considered important compared to the internal clock jitter. I would have thought this would be an issue for counters like the SRS620 25ps and Keysight 35ps 53230A but apparently not.

[pamphonica]

I just put in a Vectron C4550A1-0213 from Ebay (from China) into my 53220A. It oscillates fine but the internal reference voltage output seems to be missing (looks like open-circuit), so the calibration routine failed.

My fix (for now) was to re-install R828 (0R) which re-connects the internal +5v reference to the adjustment circuitry. This strap was needed for the original smaller oscillator module as it has no reference voltage output.  Cal worked perfectly once this was done.

Has anyone put in one of these specific new/used devices from Ebay and tried the "Cal Step" routine? If this is a feature of this secondhand module, so be it, but if yours worked fine, than please let me know so I can get a replacement from the vendor.

The datasheet shows that ref voltage output can be removed as an option.

-Jeremy

[TheSteve]

I just put in a Vectron C4550A1-0213 from Ebay (from China) into my 53220A. It oscillates fine but the internal reference voltage output seems to be missing (looks like open-circuit), so the calibration routine failed.

My fix (for now) was to re-install R828 (0R) which re-connects the internal +5v reference to the adjustment circuitry. This strap was needed for the original smaller oscillator module as it has no reference voltage output.  Cal worked perfectly once this was done.

Has anyone put in one of these specific new/used devices from Ebay and tried the "Cal Step" routine? If this is a feature of this secondhand module, so be it, but if yours worked fine, than please let me know so I can get a replacement from the vendor.

The datasheet shows that ref voltage output can be removed as an option.

-Jeremy

I've used that exact OCXO - you should not need to install R827 or R828 for it to work perfectly.

[pamphonica]

Thanks Steve, I've asked the vendor for a replacement.  I know that these modules can be ordered with no reference voltage output.  It's re-assuring to know your identical one worked fine with the 0R resistors removed as indicated.
-Jeremy

Update: the vendor says that these modules have no connection to pin 4 (voltage ref out). There are a number of vendors on Ebay but all at higher prices.  Maybe I'll try another vendor and see if these ones are cheap because of faults.  Weird.

[rkeller]

Is there a consensus on the OCXO DIY upgrade.
Hesitant to take apart and start soldering in Chinese sourced Vectron C4550A1-0213's that have been removed from other equipment.
Tried to find these new but no Joy.
Has anyone got a concise DIY upgrade path.  I checked Keysight https://www.keysight.com/main/editorial.jspx?cc=US&lc=eng&ckey=2190178&nid=-33609.959902.02&id=2190178
It states the 010 upgrade is "not available" for those of us with non 010 units.
Seems simple enough, but for now the OCXO with the GPS timebase at 10 Mhz going in the back works but isn't the best solution especially if I need to carry the unit to another work station.
Thank you,
Russ

[tkamiya]

High-end counters often don't come with OCXO standard and built-in oscillators are usually awful.  This is because it is expected that customer will either use shop standard or pick an OCXO.  Shop standard being more common, later customers don't have to pay for extra.  This will be a significant amount of money when buying ten or more counters!  Standard oscillators are there just to check that the equipment works.  It was never meant to be used for measuring anything.

I typically don't like internal OCXO on counters.  It means separate calibrations and they will never be exactly the same.  This makes it impossible to check correlations.  Also, counters will keep OCXO turned on while equipment is turned off.  That means heat generation and power consumption.  Having few of these, it adds up. 

If this is for business, I would really NOT mess with it myself.  I would have certified hands do certified work and come back with certificate of the job performed and the result.  Can't imagine what would happen if unthinkable happens and thousands of miscalibrated products going out the door.  Compared to that, what HP charges is cheap.

[F64098]

I just put in a Vectron C4550A1-0213 from Ebay (from China) into my 53220A. It oscillates fine but the internal reference voltage output seems to be missing (looks like open-circuit), so the calibration routine failed.

My fix (for now) was to re-install R828 (0R) which re-connects the internal +5v reference to the adjustment circuitry. This strap was needed for the original smaller oscillator module as it has no reference voltage output.  Cal worked perfectly once this was done.

Has anyone put in one of these specific new/used devices from Ebay and tried the "Cal Step" routine? If this is a feature of this secondhand module, so be it, but if yours worked fine, than please let me know so I can get a replacement from the vendor.

I bought a bundle of 5 used OCXOs and 3 of them have the 4V Reference output enabled. Looking at the markings there are no differencies in type or mfg date.
Try your luck: https://de.aliexpress.com/item/4000760950500.html

Put one in my 53220A and i'm very happy 

Best regards

Frank

[SoundTech-LG]

No Luck...

Leider, ist dieser Artikel nicht mehr verfügbar!

[F64098]

Oh i'm very sorry,

got my items just a few days ago and c&p the link without
an actual click on it.
But there are other sellers with similar items.
From my view it is now shure, that the suffix -0213 don't say
anything about the optional reference voltage on pin 4.

Another user made the same experience: https://www.eevblog.com/forum/metrology/how-to-unsolderremove-ocxo-from-pcb/msg4030306/#msg4030306

HTH

Frank

[hartvige]

Just got a couple of C4550A1-0213 devices off Ebay, and both of these do not drive the reference voltage (4.0V nominally) out on pin 4.
It seems really odd that devices with the same marking and P/N (-0213) behave differently in this regard.

Someone mentioned that the pin 4 output was a selectable option in the P/N.
However, looking at:

https://www.vectron.com/products/ocxo/C4550.pdf

This does not seem to be the case. As I read the DS, all devices should drive this pin. Which begs the question: the -0213 devices that don't do this, are they simply broken?