cti ocxo low in freq

[shabaz]

Hi,
Ordinarily it wouldn't be advised to swap for a different frequency for various reasons. One reason is that the oscillator circuits are intended for a particular ballpark, so if you use a crystal of significantly different frequency, it may not oscillate.

Another reason is that things are designed around using a particular reference and then constructing any other precise frequency from that (with a frequency synthesizer).

If your requirement is to simply have a reasonably precise frequency signal source at an arbitrary frequency, you may wish to consider one of various PLL or DDS frequency synth boards (often very low-cost from AliExpress), or ICs (if you wish to build from scratch). A few examples are AD9954 (very easy-to-use for generating signals up to 160 MHz or so), MAX2870 (various boards on AliExpress, usually with a touch-screen fitted, for approx. 28.5 MHz upward into microwave), or even a very low cost Si5351 (useful up to a couple of hundred MHz approximately, a little awkward to program from scratch, but there are dozens of libraries, including for Arduino and so on). Some of these boards may use a TCXO for their internal reference; others may allow connection of an external 10MHz reference.

[ME]

I may cut mine open and change the xtal and see what happens, It seems to be dead anyway so nothing to loose. It was low in freq anyway so guess the xtal  was in its death throws, think i will cut the xtals legs and tack another xtal to the remains and see if it restores output on the output pin just for a test.

[ME]

Well i cut this open, What looks like a vreg has a bleb and a burn hole, Its dead jim dead!.

[floobydust]

Which IC is blown? I thought your module worked but freq was low? WS06 flipped 90SM LDO w/ref cap, or 1ACL Tinylogic, or ASIC 501?

We have to figure out the VCOCXO ASIC IC, it's either "501" or "303" maybe NDK makes it. I can't find this mystery IC. Could conquer it and draw a schematic if we have that info.
One dud module I have is centered high in freq. (32Hz 3.2ppm) and the control voltage does nothing over a wide window. It doesn't seem to pull the crystal unless the voltage is far out there like 0.4V or 4.5V so I suspected a varactor diode got old, it must be in the ASIC.
I'm not sure what happens with old crystals, how they behave. These modules have a selected crystal in them I think, QR code tracked.
Precision 2ppm Taitiencrystals are $45 at DK.
I'll just order more modules for the fun.

edit: had SMT marking code upside down.

[ME]

The blown ic is top left in pic, As far as i know xtals go down in frequency as they age.

[ME]

the pic rotated so the dead ic is now lower left,  Whatever it is has 6 legs.

[floobydust]

That burnt IC is the output buffer SOT-23-6 "1ACL" or "1ACW" likely a tiny logic part 74LVC 5V inverter etc. but I could not find the marking code.
The IC on the other side is the voltage regulator SOT-23-5 "90SM" and I couldn't find it either. "90S" is Ricoh RP104N311B LDO but it has no low noise ref pin for the cap.
The ASIC could be Seiko NPC or NDK, I suspect these are Japanese manufacturer.

It's even possible to be using a 20MHz xtal and a divide by two.
Module Pin 1 OUTPUT is the square corner, not the rounded corners.

Did see a patent US 6628175 Pericom "Voltage-controlled crystal oscillator (VCXO) using MOS varactors" for the VCXO function. I think the ASIC is using these.

[ME]

I received 5 new tcxo bricks today, I removed  the resistors and pot from the frequency ref chinese pcb ,I then glued the pot on the rear of the pcb with one outer leg to ground the center leg to the vref leg on the tcxo can and the other outside leg of the pot going to the 2-5v output leg of the lt1009, I hooked this upto my scope on one channel and my 10 Mhz gpsdo onto ch2,after turning the pot i get the traces locked with eachother,see pic, Does that mean its working ok ?.

[floobydust]

I think that's one way to see if the module under test is ballpark for frequency.
You would trigger the scope off the GPSDO and look at the module under test to see if the trace is rolling. But any rolling could be super slow to observe and jitter from the scope's trigger section I think could make things fuzzy? You should see moving until both are stable.
What module control voltage do you measure when it's set to match?

[Uunoctium]

"Accuracy = (for a one-cycle drift)
1 second drift/cycle = 10^-7; 10 second drift/cycle = 10^-8; 100 second drift/cycle = 10^-9"

Source: xl microwave 3400 counter manual, Calibration: see P.54

https://www.ko4bb.com/manuals/88.73.135.189/3600-3400_40-60GHz_Counter_w_Powermeter_Service_Manual.pdf

And according to their drawing, they also use an TEK475

[ME]

I think that's one way to see if the module under test is ballpark for frequency.
You would trigger the scope off the GPSDO and look at the module under test to see if the trace is rolling. But any rolling could be super slow to observe and jitter from the scope's trigger section I think could make things fuzzy? You should see moving until both are stable.
What module control voltage do you measure when it's set to match?

So do i just make sure to say have the gpsdo on ch1 the chinese jobby on ch2 but set the scopes trigger to ch1?.

[Uunoctium]

calibration against a reference standard is precisely described in the mentioned xl-microwave manual

[ME]

Well i dont have any microwave manuals apart fro the one for heating my food!, I triggered from boh chanells and triggering  from either gave a rock steady trace in respect of each other so guess its ok then?, the only thing of note is it draws 260ma after a few mins warmup and not 250ma as per the datasheet.

[floobydust]

So do i just make sure to say have the gpsdo on ch1 the chinese jobby on ch2 but set the scopes trigger to ch1?.

Yes, that's pretty much it. I'm assuming that you are only looking for ballpark frequency, in order to determine if a module is in spec. and not way off in frequency.  Just like tuning into WWV and using a beat frequency I think.  Timenuts would not like this method for any great sub-ppm accuracy.


From the XL Microwave manual:

5-7. REFERENCE OSCILLATOR ADJUSTMENT [stuff deleted...]
5-7.2. Allow the instrument to warm for three (3) hours for the TCXO, 24 hours for the OCXO...
5-7.3. Two ways are given to calibrate the counter’s internal reference oscillator. The first way (analog) is simpler, requiring only a scope and a traceable frequency standard. It also offers instant visual feedback of the result of the tuning adjustment. The second way (digital) requires more equipment but is faster and yields a more accurate number.
5-7.4. Analog. You will need the following test equipment from Table 5-1: a100 MHz Oscilloscope and a traceable 10 MHz Frequency Standard. Connect the equipment as illustrated below in Figure 5-1.
Figure 5-1. Reference Oscillator Test Setup, analog version.
5-7.5. Set the scope to show 1 to 5 cycles of a 10 MHz signal. Connect the 10 MHz signal from the frequency Standard to Channel ‘A/1’ of the scope and trigger on Channel ‘A/1’. Connect channel ‘B/2’ of the scope to the ‘REF OSC OUT’ BNC connector on the back of the counter. Watch the scope’s display and note the time it takes for one complete cycle of channel ‘B/2’ to drift one cycle compared to channel ‘A/1’. On the Oscillator CCA (5999-45561, figure 6-6) adjust the ‘Oscillator Trimpot’ (RA1) for minimum drift of the signal on the oscilloscope display using a nonconductive tuning tool (remove the tool when noting drift as the tool will affect the observation). Accuracy = (for a one-cycle drift)
1 second drift/cycle = 10-7; 10 second drift/cycle = 10-8; 100 second drift/cycle = 10-9.

[ME]

I would like to build a pcb around my new ocxo cans using eagle, does anyone know where i can get an eagle footprint for the CTIosc5a2b02 ocxo ?, thanks.

[Solder_Junkie]

I would like to build a pcb around my new ocxo cans using eagle, does anyone know where i can get an eagle footprint for the CTIosc5a2b02 ocxo ?, thanks.

In many (most/all) PCB design packages you can make your own footprint, otherwise an oddball part would leave you stuck. My own favourite software is Sprint Layout by Abacom, it is about 49 Euros.

SJ

[Gyro]

I would like to build a pcb around my new ocxo cans using eagle, does anyone know where i can get an eagle footprint for the CTIosc5a2b02 ocxo ?, thanks.

In many (most/all) PCB design packages you can make your own footprint, otherwise an oddball part would leave you stuck. My own favourite software is Sprint Layout by Abacom, it is about 49 Euros.

SJ

The OP double-posted a separate thread on the subject... https://www.eevblog.com/forum/eagle/cti-osc5a2b02-eagle-footprint/

[alex843]

Is the xtal  in the teardown pics a 10Mhz unit, can it be changed for another freq?.

You might be able to replace the crystal, but it's good to remember that the common AT-cut crystal resonators that you can purchase for most radio and microprocessor applications are optimised for stability at room temperature, while the ones used in the OCXO are likely to be SC-cut crystal resonators designed for the higher temperature oven controlled oscillators.

There are other OCXO frequencies that you find for sale if you want them, including:

13 MHz used in GSM base stations
- 13 MHz/48 = 270.83 kHz which is the GSM Air Interface bit rate.
- 13 MHz/65 = 200 kHz, the GSM radio channel spacing.

38.88 MHz used in SDH/SONET/ATM optical systems
- 38.88x4 = 155.52 Mhz, the STM-1 line rate
- 38.88x16 = 622.08 Mhz, the STM-4 line rate.

I've also seen some 20 MHz and 40 MHz OCXO for sale.

[ftg]

Other two cheap OCXO frequencies are 26MHz (also used in GSM base stations, that's where all of mine came from) and 30.72MHz, used in LTE base stations.

[Uunoctium]

Only for the protocoll.
I now tested or installed all 4 that I purchased last year. Or repaired them.

Result: 2 DoA! 
One without heating, a second with distorted output signal (rather sawtooth). I replaced the heating transistor with 2SC5706, but draws a slightly higher current (600..700mA in the first heating phase).
The distorted output was caused by the Inverter/Schmitt ( bottom left corner). I threw in what was available in the workshop, 74SU04 -two pins reversed -, square yes, but round corners and wrong duty cycle – too slow, no HCT. Doesn't matter, because in the target device anyway only sine works. That comes for free, because at input of inverter is a pure sinewave, but only 150mVpp. Wired it to the free (NC) contact.

There seem to be very different versions, because in contrast to the pictures here in the thread, mine all had a TSSOP-20 chip instead of -16. And the supply of the output buffer is either direct wired to U(unreg) (the one with the defective output buffer) or behind the 3V3 regulator and so protected against overvoltage.

Differences in the tuning voltage are also huge: 2 x 1.8 V, 1x 1.4 V and 1x 2.02 V.

[bingo600]

Just attached the "interesting" Cal. page from the counter manual above

[BillyO]

Just attached the "interesting" Cal. page from the counter manual above

Yup, that's how it's done.

[Uunoctium]

By the way, the tech nerds at HP also count seconds