cti ocxo low in freq

[ME]

I have a chinese 10Mhz cti ocxo ,It is running at 99999861Hz not 10MHz as per spec, this is after doing the mods as per pic and after running for an hour, any ideas what to try next to raise the frequency to 10Mhz?

[807]

My reverse engineered circuit is different to the one provided by Transradio. I had to reduce R1 & R3 in my case, as my oscillator was running high, rather than low. In your case you may need to increase R1 or R3 (or both), or decrease R2 to provide more voltage on the Vref pin.

[SteveThackery]

Out of interest,  what is the accuracy spec for that oscillator?

Another question: what are you measuring its frequency with, and are you sure of its accuracy?

[ME]

I am measuring with an advantest tr5823,the specs are:
Description
Report Item
OCXO-10MHz rate reference source:
Current consumption: preheating stage <750mA, working stage <300mA
Preheating time :≈5 minutes
Calibration accuracy :±001Hz
Frequency stability :5e-12
Output level :dBm: 10dBmt2dBmas09;
VRMS:07071V@5022:
VPeak: I.0V@502:
VP-P:20Va502.
Output impedance :50Ω
Interface type :SMA outer screw inner hole
Product Usage: constant temperature 0MHz ultra-low phase noise, frequency reference module, frequency reference source,OCXO rate reference source, used for sound system, decoder, shortwave radio, frequency meter, signal source and other instrument reference.
as its so small with tiny resistors, I am thinking of doing a rebuild on a new custom pcb using a tl431.

[SteveThackery]

I have a question about OXCOs in general, if anyone can help.

What is the purpose of the Vref input?  If it is to adjust the frequency, then doesn't that mean that the device doesn't meet the published specs until the user has tweaked it?

If so, I don't understand. I thought an OXCO was supposed to be usable as a high accuracy, high stability frequency reference. But if the customer needs another frequency reference to set it up... well, isn't that defeating the purpose?

And how much is the frequency affected per volt or millivolt on the Vref pin?  Does the Vref input need connecting to a precision voltage reference with a guaranteed stability in order to ensure the requisite frequency stability of the oscillator?

In my ignorance I always thought that when you bought an OXCO precision frequency reference it would meet the stated accuracy and stability specs straight out of the packet.

[ME]

any ideas of a ballpark value of resistors to try?, The max i can get at the vref pin is 7.8v. Thats with the pot at max travel,its cw as i turner the pot around thru 108 deg on the pcb so cw increases the freq and ccw decreases it.

[floobydust]

My reverse engineered circuit is different to the one provided by Transradio. I had to reduce R1 & R3 in my case, as my oscillator was running high, rather than low. In your case you may need to increase R1 or R3 (or both), or decrease R2 to provide more voltage on the Vref pin.

Newer boards (REV.0.0.2) come with a 200k trimpot instead of a 100k. You can read the "204" on the top marking. The other resistors are the same.
It gave a V_CTRL of 1.95-2.15V span, which did not do much of anything towards changing frequency, although my resolution was only 8 digits at the time.
OSC5A2B02 module spec. is nominal 2.0V +/-2.0V for -1.0ppm to +2.0ppm.
The LDO LM2940CS-5.0 is shit quality, line regulation is poor 4.75-4.85V finally up to 5V with 10V in but then it runs warm. It has terrible line regulation, not a great clone.

any ideas of a ballpark value of resistors to try?, The max i can get at the vref pin is 7.8v. Thats with the pot at max travel,its cw as i turner the pot around thru 108 deg on the pcb so cw increases the freq and ccw decreases it.

Uh this is a 5V part and what are you doing measuring 7.8V there? Something wrong there.

[hgl]

Frequency stability :5e-12
ultra-low phase noise

Alternative facts or misinterpretation of ppb ? ;-)

[edpalmer42]

I am measuring with an advantest tr5823,the specs are:
Description
Report Item
OCXO-10MHz rate reference source:
Current consumption: preheating stage <750mA, working stage <300mA
Preheating time :≈5 minutes
Calibration accuracy :±001Hz
Frequency stability :5e-12
Output level :dBm: 10dBmt2dBmas09;
VRMS:07071V@5022:
VPeak: I.0V@502:
VP-P:20Va502.
Output impedance :50Ω
Interface type :SMA outer screw inner hole
Product Usage: constant temperature 0MHz ultra-low phase noise, frequency reference module, frequency reference source,OCXO rate reference source, used for sound system, decoder, shortwave radio, frequency meter, signal source and other instrument reference.
as its so small with tiny resistors, I am thinking of doing a rebuild on a new custom pcb using a tl431.

Have you checked the accuracy lately by comparing it to a known standard like a GPSDO, off-air standard broadcast, etc.?  Even the best equipment will give bad results if it's out of adjustment.

Do you remember when they launched the Hubble Space Telescope?  All the pictures were blurry.  It turned out that they'd checked the mirror by two methods.  The low precision equipment said that the mirror was bad.  The high precision equipment said the mirror was good.  It turned out that the high precision equipment was completely wrong because it was out of calibration.

[edpalmer42]

I have a question about OXCOs in general, if anyone can help.

What is the purpose of the Vref input?  If it is to adjust the frequency, then doesn't that mean that the device doesn't meet the published specs until the user has tweaked it?

If so, I don't understand. I thought an OXCO was supposed to be usable as a high accuracy, high stability frequency reference. But if the customer needs another frequency reference to set it up... well, isn't that defeating the purpose?

And how much is the frequency affected per volt or millivolt on the Vref pin?  Does the Vref input need connecting to a precision voltage reference with a guaranteed stability in order to ensure the requisite frequency stability of the oscillator?

In my ignorance I always thought that when you bought an OXCO precision frequency reference it would meet the stated accuracy and stability specs straight out of the packet.

Crystals drift.  Always.  Forever.  Can't predict the direction or magnitude, no matter how hard you try, although you can say that it will probably stay within a range.  Today, the frequency might drift up, tomorrow down.    

The higher quality oscillators are the hardest ones to deal with.  They can be absolute divas!  Don't shake it.  Don't breathe on it.  Don't let a sunbeam fall on it.  Pretend it's full of nitroglycerine!     Some oscillators have to run for 90 days before they meet their published specs.  If the power even glitches during this period, the aging can completely change.

Of course this CTI oscillator isn't in the same class as the divas.  So the most likely possibilities are that it's just broken or that the frequency counter is out of adjustment. 

[SteveThackery]

So you should set Vref to pull the oscillator to the correct frequency?  How sensitive are they, typically, in terms of Hz per volt?  That doesn't appear in the quoted specs

[ME]

my post of the max vref voltage was a typo,it should of said 2.376v,thats with the 100k pot at max travel ok.

[ME]

if i make a new pcb for this,would i be better of using a 7805 to power it with its tab soldered to the ocxo can, provided the can is only floating or common to gnd?, just thinking in terms of thermal stability?.

[PA0PBZ]

A 7805 is a regulator, not a voltage reference so that will not be stable enough I think.
The 'official'(?) pdf says the following about the tuning range:

0V: -2.0 ~ -1.0ppm
2V: -0.2 ~ +0.2ppm
4V: +1.0 ~ +2.0ppm

1ppm on 10Mhz is 10Hz (if I had enough coffee already)

https://dl6gl.de/media/files/ocxo-cti-osc5a2b02-datasheet.pdf

[ME]

What i meant was keeping the lt1009 but using the 7805 in place of its original vreg (lm2940s-5) but have the tab of the 7805 soldered to the osc can.

[SteveThackery]

A 7805 is a regulator, not a voltage reference so that will not be stable enough I think.
The 'official'(?) pdf says the following about the tuning range:

0V: -2.0 ~ -1.0ppm
2V: -0.2 ~ +0.2ppm
4V: +1.0 ~ +2.0ppm

1ppm on 10Mhz is 10Hz (if I had enough coffee already)

https://dl6gl.de/media/files/ocxo-cti-osc5a2b02-datasheet.pdf

So it looks like there is no chance of the OP's OCXO reaching 10MHz using Vref.  9999861Hz* is 139Hz low, which is 13.9ppm.  Is that right?

The spec sheet says Vref can pull the oscillator over the range +2ppm to -2ppm.  So -13.9ppm is way too much to be fixed with Vref.  Looks like it's faulty, or maybe the OP's Advantest TR5823 has gone out of spec.


*Did anyone else notice that the OP seems to have added an extra 9?

I have a chinese 10Mhz cti ocxo ,It is running at 99999861Hz not 10MHz as per spec, this is after doing the mods as per pic and after running for an hour, any ideas what to try next to raise the frequency to 10Mhz?

[PA0PBZ]

So it looks like there is no chance of the OP's OCXO reaching 10MHz using Vref.  9999861Hz* is 139Hz low, which is 13.9ppm.  Is that right?

The spec sheet says Vref can pull the oscillator over the range +2ppm to -2ppm.  So -13.9ppm is way too much to be fixed with Vref.  Looks like it's faulty, or maybe the OP's Advantest TR5823 has gone out of spec.


*Did anyone else notice that the OP seems to have added an extra 9?

I have a chinese 10Mhz cti ocxo ,It is running at 99999861Hz not 10MHz as per spec, this is after doing the mods as per pic and after running for an hour, any ideas what to try next to raise the frequency to 10Mhz?

There's either an extra 9 or OP forgot a dot and it's really 9999986.1Hz which is a lot closer to 10MHz, the TR5823 has 8 digits afaik.

[CaptDon]

Do I recall correctly from other threads on EEVBLOG that these units are 'fall outs' from the O.E.M. that didn't meet spec in the first place? Did someone raid a dumpster full of these non-conforming OCXO's. Anyone old enough to recall the days of 'Budget' vacuum tube suppliers buying fall outs from the major brands GE,RCA, Sylvania and Raytheon and rebranding them as names like TelRad, RadTel, Universal, KenRad and a hundred others? Usually sold at drug stores where there was a big 80 socket tube tester that mostly measured emission and possibly shorts. Fun times. The new tubes could be worse than your old ones!!

[CaptDon]

I think we also want to know when the O.P. last verified the accuracy of his counter? I have a Tek 500 series counter module that is very accurate but only after a 4 hour warm up. It has an OCXO and it will actually overshoot two times during warm up and then settle down. The nice thing is after warm up it always falls within +/- 1Hz with a 15MHz reference. I have a 15MHz PLL locked to WWV as my bench reference.

[edpalmer42]

So you should set Vref to pull the oscillator to the correct frequency?  How sensitive are they, typically, in terms of Hz per volt?  That doesn't appear in the quoted specs

Datasheets rarely, if ever, state Hz per Volt.  The one posted by hgl states that the tuning slope is positive (higher voltage = higher frequency), lists the tuning voltage range, and lists the overall tuning range in ppm.  This is a fairly typical format for the information.

So, you power up the oscillator, wait a day or two for the initial warmup & stabilization and do an initial adjustment of the frequency.  Wait a day or a week and repeat.  Is the aging dropping?  Good!  Eventually, the aging will stabilize at some value and you can, if necessary set up a monthly, quarterly, or whatever task to tweak the oscillator.  In the case of a GPSDO, the feedback loop will keep the oscillator on frequency but you still might want to tweak the tuning voltage to ensure that the loop doesn't run out of range.  Depending on the unit, you might have both a mechanical and electrical adjustment.  Mechanical adjustments typically have more range, but less resolution than an electrical adjustment.  i.e. the mechanical adjustment can be thought of as 'coarse tuning' while the electrical adjustment is 'fine tuning'.

[807]

So you should set Vref to pull the oscillator to the correct frequency?  How sensitive are they, typically, in terms of Hz per volt?  That doesn't appear in the quoted specs

In my particular case, I could adjust the Vref voltage between 1.9400v and 2.1085v. This corresponded to a frequency of 1.80Hz and 3.49Hz above 10MHz. So a 1.69Hz change for a 0.1685v change. That's 10Hz/v. I didn't take any measurements between those min/max readings, so don't know if it was a linear change. Assuming it was linear, I calculated that I needed around 1.76v for 10MHz.

[807]

I have a question about OXCOs in general, if anyone can help.

What is the purpose of the Vref input?  If it is to adjust the frequency, then doesn't that mean that the device doesn't meet the published specs until the user has tweaked it?

If so, I don't understand. I thought an OXCO was supposed to be usable as a high accuracy, high stability frequency reference. But if the customer needs another frequency reference to set it up... well, isn't that defeating the purpose?

And how much is the frequency affected per volt or millivolt on the Vref pin?  Does the Vref input need connecting to a precision voltage reference with a guaranteed stability in order to ensure the requisite frequency stability of the oscillator?

In my ignorance I always thought that when you bought an OXCO precision frequency reference it would meet the stated accuracy and stability specs straight out of the packet.

Any type of Xtal oscillator is a secondary standard & will need tweaking to get it close to 10MHz. The OCXO is the better of the Xtal oscillators, but it needs to be compared to a higher order standard. The link shows comparisons between different oscillators. The primary standard being Caesium (or is it Hydrogen MASER these days?). There's also GPSDO to consider but it's not included in the table.

Many items of test gear such as frequency counters have OCXO timebases, but you still see a hole in the back of the case to trim it (or in some cases it's trimmed in software).

https://www.sciencedirect.com/topics/physics-and-astronomy/frequency-standard

[ME]

So is changing the original vreg to a lm-7805 and mounting its tab to the ocxo a good idea?.

[edpalmer42]

So is changing the original vreg to a lm-7805 and mounting its tab to the ocxo a good idea?.

Why?  It won't help you get the oscillator on-frequency, if that's what you're thinking.  If it really is 139 Hz (14 ppm) low, the OCXO is broken and nothing will fix it.  The datasheet shows that it only has a tuning range of 2 ppm.

In general, you don't want to add or remove heat from an OCXO.  It's designed to have a certain amount of heat transfer from the oscillator to the environment.  If you change that, the oven temperature control circuit will be disrupted.

Has it been powered up for the past few days?  Has the frequency drifted towards 10 MHz?

Have you checked your TR5823?

[ME]

The only way i have to check my tr5823 is a very old thandar tf-200,that gives nearon the same readings,guess i need to buy a gpsdo!,thanks all.