I bought the very Last Symmetricom SA.22C in the shop!

[Smile2016]

Z04 are likely these: nc7sz04 "TinyLogic UHS Inverter"

that's the one + datecode

[Smile2016]

Is it possible to change the factory reference crystal on the unit? 

I am planning to buy a unit that doesn't output 10Mhz it only ouput 9.8304MHz from factory.

How to check reference crystal frequency? my unit shows 60.00MHz at startup and 39.38MHz if I use "i" command .

SA22C by Symmetricom, Inc., Copyright 2006
SA22 Version 6.01C of 7/2006; Loader Version 3
Mode CN01 Flag 0000 [D04D]ok
Unit serial code is xxxxx-h, current tuning state is 6
Crystal: 60000000hz, ACMOS: 10000000.0hz, Sine: 10000000.0hz
...
r>
and with command i:

r>i
SA22C by Symmetricom, Inc., Copyright 2006
SA22 Version 6.01C of 7/2006; Loader Version 3
Mode CN01 Flag 0004
Unit serial code is xxxx-h, current tuning state is 6
Crystal: 3938700hz, ACMOS: 989680.00000000hz, Sine:989680.00000000hz
...

opened one unit , cannot directly access a physical crystal that could be exchanged anyways...

[Johnny B Good]

Is it possible to change the factory reference crystal on the unit? 

I am planning to buy a unit that doesn't output 10Mhz it only ouput 9.8304MHz from factory.

How to check reference crystal frequency? my unit shows 60.00MHz at startup and 39.38MHz if I use "i" command .

SA22C by Symmetricom, Inc., Copyright 2006
SA22 Version 6.01C of 7/2006; Loader Version 3
Mode CN01 Flag 0000 [D04D]ok
Unit serial code is xxxxx-h, current tuning state is 6
Crystal: 60000000hz, ACMOS: 10000000.0hz, Sine: 10000000.0hz
...
r>
and with command i:

r>i
SA22C by Symmetricom, Inc., Copyright 2006
SA22 Version 6.01C of 7/2006; Loader Version 3
Mode CN01 Flag 0004
Unit serial code is xxxx-h, current tuning state is 6
Crystal: 3938700hz, ACMOS: 989680.00000000hz, Sine:989680.00000000hz
...

opened one unit , cannot directly access a physical crystal that could be exchanged anyways...

 Seeing this going unanswered for the past three and a half weeks has bugged me enough to answer your post.

 Did you ever figure out that the crystal is in a TO5 case (see attached image in reply #40)?

 As far as the i command reply is concerned, I'd be inclined to treat it as a total nonsense (see reply #1 where I posted my own i command results as per the quote below)

The i command displays this:

r>i


SA22C by Symmetricom, Inc., Copyright 2006
   SA22 Version 6.02C of 7/2006; Loader Version 3
   Mode CN41  Flag 0004

Unit serial code is 1211EC3234-h, current tuning state is 6
Crystal: 3938700hz, ACMOS: E4E1C0.00000000hz, Sine: 989680.00000000hz
Ctl Reg: 004C, Res temp off: BFC53F7D., Lamp temp off: BFF9418E.
FC: enabled, Srvc: low

 The ACMOS frequency in my case, as you can observe, appears to be a hexadecimal equivalent of 989680 but it is not. Using a scientific calculator to convert it to decimal gave me a figure of 15,000,000 Hz (15Mhz - the initial frequency setting that had been in force at that time - no doubt, having since used the o command to set it to 10MHz, I'd expect to see it display a new hexadecimal value).

 Well, bugger me sideways! The hexadecimal representation for 10,000,000Hz just happens to be 0x989680Hz!!!!

Hmm, I wonder... Yep! the crystal frequency is also represented in hexadecimal. 0x3938700 and translates to 60,000,000.

IMHO, whoever decided to use a hexadecimal representation in the i command report without that all important 0x prefix is a cnut (damn my dyslexia )

Now that I've solved the mystery of those weird (nonsensical) frequencies shown by the i command versus the power on startup message, that does rather beg the question as to whether you measured the actual output frequency or compared it against a 10MHz gpsdo reference or simply assumed it was meant to output that particular CDMA reference frequency (the 9.8304MHz frequency appears in Wikipedia's list of crystal oscillator frequencies)
 https://en.wikipedia.org/wiki/Crystal_oscillator_frequencies

 However, there is one remaining mystery in my own i command results, namely the out of place hex representation of 10MHz for the Sine frequency when the ACMOS is correctly showing the hex representation of the 15MHz output frequency it had been set to as it was originally received.

 Now, if anyone has any pearls of wisdom to offer, please feel free to chip in.

[Smile2016]

Is it possible to change the factory reference crystal on the unit? 

I am planning to buy a unit that doesn't output 10Mhz it only ouput 9.8304MHz from factory.

How to check reference crystal frequency? my unit shows 60.00MHz at startup and 39.38MHz if I use "i" command .

SA22C by Symmetricom, Inc., Copyright 2006
SA22 Version 6.01C of 7/2006; Loader Version 3
Mode CN01 Flag 0000 [D04D]ok
Unit serial code is xxxxx-h, current tuning state is 6
Crystal: 60000000hz, ACMOS: 10000000.0hz, Sine: 10000000.0hz
...
r>
and with command i:

r>i
SA22C by Symmetricom, Inc., Copyright 2006
SA22 Version 6.01C of 7/2006; Loader Version 3
Mode CN01 Flag 0004
Unit serial code is xxxx-h, current tuning state is 6
Crystal: 3938700hz, ACMOS: 989680.00000000hz, Sine:989680.00000000hz
...

opened one unit , cannot directly access a physical crystal that could be exchanged anyways...

 Seeing this going unanswered for the past three and a half weeks has bugged me enough to answer your post.

 Did you ever figure out that the crystal is in a TO5 case (see attached image in reply #40)?

 As far as the i command reply is concerned, I'd be inclined to treat it as a total nonsense (see reply #1 where I posted my own i command results as per the quote below)

The i command displays this:

r>i


SA22C by Symmetricom, Inc., Copyright 2006
   SA22 Version 6.02C of 7/2006; Loader Version 3
   Mode CN41  Flag 0004

Unit serial code is 1211EC3234-h, current tuning state is 6
Crystal: 3938700hz, ACMOS: E4E1C0.00000000hz, Sine: 989680.00000000hz
Ctl Reg: 004C, Res temp off: BFC53F7D., Lamp temp off: BFF9418E.
FC: enabled, Srvc: low

 The ACMOS frequency in my case, as you can observe, appears to be a hexadecimal equivalent of 989680 but it is not. Using a scientific calculator to convert it to decimal gave me a figure of 15,000,000 Hz (15Mhz - the initial frequency setting that had been in force at that time - no doubt, having since used the o command to set it to 10MHz, I'd expect to see it display a new hexadecimal value).

 Well, bugger me sideways! The hexadecimal representation for 10,000,000Hz just happens to be 0x989680Hz!!!!

Hmm, I wonder... Yep! the crystal frequency is also represented in hexadecimal. 0x3938700 and translates to 60,000,000.

IMHO, whoever decided to use a hexadecimal representation in the i command report without that all important 0x prefix is a cnut (damn my dyslexia )

Now that I've solved the mystery of those weird (nonsensical) frequencies shown by the i command versus the power on startup message, that does rather beg the question as to whether you measured the actual output frequency or compared it against a 10MHz gpsdo reference or simply assumed it was meant to output that particular CDMA reference frequency (the 9.8304MHz frequency appears in Wikipedia's list of crystal oscillator frequencies)
 https://en.wikipedia.org/wiki/Crystal_oscillator_frequencies

 However, there is one remaining mystery in my own i command results, namely the out of place hex representation of 10MHz for the Sine frequency when the ACMOS is correctly showing the hex representation of the 15MHz output frequency it had been set to as it was originally received.

 Now, if anyone has any pearls of wisdom to offer, please feel free to chip in.

Yes, I identified the TO5, though I wasn’t expecting this package/case. Looking at the components around the crystal, it seems like a simple VCO where designer didn’t put much effort into making it very stable or performant. I’ll probably sacrifice one of my units for the sake of science and remove the crystal to check if it’s one of those SC cut types, which are a bit more thermally stable. It’s most probably possible to change the crystal without a software change, as long as we don’t care about the ‘i’ command response or boot-up. All possible frequencies stated in the datasheet can be synthesized from crystals in a relative small range of 58.9824 to 65.0000 MHz using integer division factors. Extracting the image from flash and ROM would be interesting too. It seems the way numbers were represented was not the primary concern.
My units were initially set to 15.000MHz. I adjusted them to 10.000MHz and ran them for approximately 10 days with a PPS from GPSDO. Now, they are reasonably well-adjusted. More updates to follow…

[Johnny B Good]

 The apparent lack of stability with that vcxo circuit is by design. It has to respond very quickly to the tuning control signal from the
6.8GHz microwave frequency produced by the ground-state hyperfine transistion of the 87Rb atom in the vapor cell inside of the microwave cavity in order to remain in frequency lock.

 The crystal isn't ovenised but merely has to warm up sufficiently to the internal ambient temperature to stabilise its frequency sufficiently enough for the FLL to achieve and maintain lock. I too had the impression of insufficient stability when I was trying to trace a mysterious and intermittent 1 or 2 ns phase modulation that had suddenly made its appearance earlier in the year in my LPRO-101.

 I thought it might have been a dry joint or fractured trace around the crystal so started poking it with a plastic pen, discovering just how touchy it was. I could force it to transiently shift phase by 20 to 30ns by applying steady pressure to the top of the crystal during which time it would drift back onto its original phase until I slowly removed the pressure whereupon it then shifted the other way before settling back. If I kept a steady pressure and then suddenly released it. I could get it to fall off its perch and lose lock for the time time it would take to regain a fresh lock (I could also obtain the same result by a sudden application of pressure).

 Convinced there was a dry joint, I desoldered it from the board and cleaned the old solder from the pins and the holes before resoldering it back in place. Testing revealed that hadn't made any difference to this weird issue so I reassembled it and ordered a pack of ten HCU49 20MHz crystals from eBay. However, I didn't get round to replacing the crystal on account the mysterious phase modulation effect had quietly slunk off to another hidden dimension just a few days later.

 Considering this experience over the next few weeks, I finally concluded that this 'touchyness' was an inevitable phenomena created by design and the abnormal physical interference of probing fingers. Yes, you can excite modest transient phase shifts by gentle to hard thumping of the LPRO-101's case. In this regard, this characteristic hadn't altered in all the time it had been in my care.

 This conclusion is essentially a hypothesis on my part but if, as I suspect, you'd had a similar experience when probing around the crystal, that would help validate this hypothesis.