OCXO & TCXO

[Rene Tschaggelar]

I was looking for an ovenized oscillator to find none on stock.
The usually expected series order is 100 pieces up.

Then I had a look at temperature compensated oscillators.
They had a tempco of 1e-6 and ageing of 1e-6 and
initial accuracy of 1e-6. Somehow they appear not be
self-upgradeable to an OCXO with a peltier.

What kind of crystals do they have in OCXOs to achieve
these unbelievable low values of drift ?

BTW, beside the low drift, a low phase noise is also of concern.

Rene

 

[Fred Bloggs]

I was looking for an ovenized oscillator to find none on stock.
The usually expected series order is 100 pieces up.

Then I had a look at temperature compensated oscillators.
They had a tempco of 1e-6 and ageing of 1e-6 and
initial accuracy of 1e-6. Somehow they appear not be
self-upgradeable to an OCXO with a peltier.

What kind of crystals do they have in OCXOs to achieve
these unbelievable low values of drift ?

BTW, beside the low drift, a low phase noise is also of concern.

Rene

See Valpey-Fisher and the units with thin-film heaters deposited on the
crystal- ultra-low phase noise achieved by using the new fundamental
mode monolithic crystal technology.
http://www.valpeyfisher.com/Main/VFTProductSpecifications.asp

 

[Fred Bloggs]

We used to use AT cut x-tals. for both TCXO and OCXO oscillators. With the
proper choice of temperature, the ovenized x-tals. characteristic freq. vs.
temp. curve can be set to zero (or nearly zero) slope. Additionally, use of
cold weld x-tals. will usually exhibit more stable characteristics over
time. Use a heat sink when soldering to avoid thermal shock, and use a
capacitor with stable time/temperature characterists to drive the x-tal.
X'tals. in the 1 - 25 MHz. range seemed to exhibit greatest stability, and
keep the drive power within specified limits.

Some day, remind me to tell you about the bad batch of x-tals. we received
right after Three Mile Island. The freq. vs. temp. characteristic showed
extraordinary discontinuities, making temperature compensation virtually
impossible excepting over a very limited temperature range. Very scary
stuff!

Lemme see- which manu is located in the Catskills?

 

[Padraig FitzGerald]

Connor Winfield in Aurora Il, do good SC cuts OCXO's excellent phase noise
and drift over temp(as good as 10ppb)

Stratum 3E - http://osc.conwin.com/ocxo/cx/cx007.pdf

Stratum 3 - http://osc.conwin.com/ocxo/cx/cx011.pdf

in the uk www.rfx.co.uk are very good also.

 

[Ken Smith]

What kind of crystals do they have in OCXOs to achieve
these unbelievable low values of drift ?

In general, the SC cut crystal gives the best drift values.

In an OCXO, the oven temperature and the crystal's tempco curve can be
matched to each other. AT and SC cut crystals have 3rd order curves for
freq VS. temperature. As a result, there can be two points where the
curve is horiz. The oven temperature is usually set right on the flat
point. This way the drift in the oven circuit doesn't show up as
frequency drift.

BTW, beside the low drift, a low phase noise is also of concern.

I suggest you look at Greenray and Corning. (no I don't work for them)

Beware that many makers of oscillators will say "Low phase noise" but I
have found a 30dB range in the specifications. At one point I even went
back to the maker with the question "Are you sure that isn't an error?"

Depending on your application, you may also want to check the AVAR
specifications for the oscillator.

 

[Baphomet]

Then I had a look at temperature compensated oscillators.
They had a tempco of 1e-6 and ageing of 1e-6 and
initial accuracy of 1e-6. Somehow they appear not be
self-upgradeable to an OCXO with a peltier.

What kind of crystals do they have in OCXOs to achieve
these unbelievable low values of drift ?

BTW, beside the low drift, a low phase noise is also of concern.

We used to use AT cut x-tals. for both TCXO and OCXO oscillators. With the
proper choice of temperature, the ovenized x-tals. characteristic freq. vs.
temp. curve can be set to zero (or nearly zero) slope. Additionally, use of
cold weld x-tals. will usually exhibit more stable characteristics over
time. Use a heat sink when soldering to avoid thermal shock, and use a
capacitor with stable time/temperature characterists to drive the x-tal.
X'tals. in the 1 - 25 MHz. range seemed to exhibit greatest stability, and
keep the drive power within specified limits.

Some day, remind me to tell you about the bad batch of x-tals. we received
right after Three Mile Island. The freq. vs. temp. characteristic showed
extraordinary discontinuities, making temperature compensation virtually
impossible excepting over a very limited temperature range. Very scary
stuff!

 

[Rene Tschaggelar]

Thanks,
I'm having a look at Connor Winfield. My past experience with them
is somewhat not the best. I once required a 250MHz Osc. After a lof
of mail, in the final ordering stage they required a university be
the customer, I was not sufficient. Then, they increased the minimum
order to 3 pieces at 200$ or so each.

Rene

 

[Baphomet]

Some day, remind me to tell you about the bad batch of x-tals. we
received

right after Three Mile Island. The freq. vs. temp. characteristic showed
extraordinary discontinuities, making temperature compensation virtually
impossible excepting over a very limited temperature range. Very scary
stuff!

Lemme see- which manu is located in the Catskills?

I worked and lived in Manhattan in those days

 

[Rene Tschaggelar]

I was looking for an ovenized oscillator to find none on stock.
[ snip ]

Thanks to all who responded.
Oddly none of the mentioned manufacturers had a list of
stocked parts. I sent them inquiring mails.

Rene

 

[Tweetldee]

I was looking for an ovenized oscillator to find none on stock.
[ snip ]

Thanks to all who responded.
Oddly none of the mentioned manufacturers had a list of
stocked parts. I sent them inquiring mails.

Rene

How many of these OXCOs do you need? What frequency? If you only need one
or two, Ebay may be a good resource, since there are several sellers that
regularly list them. Of course, the frequencies are limited only to what
the sellers have.
If a standard frequency like 10 Mhz is acceptable, you might look for an HP
5328 counter with the high stability time base. The time base in those
counters is the HP 10544A or the newer 10811A. Those oscillators are very
highly respected, with an aging rate of 5x10e-10/day. The oscillators are
frequently sold alone.
--
Tweetldee
Tweetldee at att dot net (Just subsitute the appropriate characters in the
address)

Time is what keeps everything from happening all at once.

 

[Padraig FitzGerald]

Was this back during the busy period, 3 yrs ago? They've become a lot more
accomodating now.

Most of their products are custom designs so they didn't usually like small
qtys, but now I'd say they'd be willing to do anything for you.

 

[Rene Tschaggelar]

Was this back during the busy period, 3 yrs ago? They've become a lot more
accomodating now.

Most of their products are custom designs so they didn't usually like small
qtys, but now I'd say they'd be willing to do anything for you.

Yes, that was about 3 years ago.
I have to admit, other were not even interestesd or had
nothing.
And now no manufacturer appears to publish stocked parts.
I wrote a few, to get more questions back.

Rene

 

[Baphomet]

Oddly none of the mentioned manufacturers had a list of
stocked parts. I sent them inquiring mails.

Rene -

Try http://www.bliley.com . We used them many years ago and had excellent
results. They make superior cold weld x-tals.

 

[Tom Bruhns]

I was looking for an ovenized oscillator to find none on stock.
The usually expected series order is 100 pieces up.

....
What frequency do you need? Must it be a new one? Surely you can get
5MHz, 10MHz and a few others as stock items. I have some I'd be
willing to part with, mostly 10MHz, most somewhat power-hungry.

Cheers,
Tom

 

[Rene Tschaggelar]

...
What frequency do you need? Must it be a new one? Surely you can get
5MHz, 10MHz and a few others as stock items. I have some I'd be
willing to part with, mostly 10MHz, most somewhat power-hungry.

Thanks for the offer.
No, it doesn't have to be new. Especially for the prototype.
It preferably is 15MHz or above. A 10MHz one would require me
to PLL multiply it before use. I sent a few mails to the manufacturers
and some of them have specials for one-of pieces. I'll try to get such
a one first.

Rene

 

[ddwyer]

Thanks for the offer.
No, it doesn't have to be new. Especially for the prototype.
It preferably is 15MHz or above. A 10MHz one would require me
to PLL multiply it before use. I sent a few mails to the manufacturers
and some of them have specials for one-of pieces. I'll try to get such
a one first.

Rene

I bought a small quantity of the highest grade OCXOs on the surplus
market some years ago and can post 1 to 10 to you..
The 16.384 MHz (binary divides to 1kHz) 12V OCXOs can be adjusted
electrically 1ppm to take out ageing units employ SC cut crystals to
give 1ppE-8 / year and 1ppE9 or better over the temp band.
Note SC cut are better than AT cut for temperature control and minimum
ageing; used to sell at $500 perhaps less now $US from me or equivalent
in any wholesome currency!

 

[ddwyer]

now $85 US from me or equivalent