cti ocxo low in freq

[ME]

I was thinking having the vreg tab thermaly fixed to the ocxo would make the whole caboosh more stable as in theory should the oven then not control the temp of the vreg as well sort of, Or am i missing something?.

[edpalmer42]

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.

Do you have or can you borrow a synthesized signal generator that will accept an external 10 MHz reference?  If so, you can use your OCXO as the generator's reference and then use the generator's output to interfere with an off-air radio signal.  By zero-beating the signals you can determine if your oscillator is on-frequency.

For example, if you have a local station at 1 MHz, tune the generator to 1 MHz and with a short piece of wire as an antenna, you should hear some noise over top of the local station.  If your oscillator is lower than 10 MHz, you'll have to adjust the generator up in frequency to get the same effect.  It's a very cool effect!   

Here in North America we're spoiled by having WWV at 10 MHz.  We don't have to use a synthesized generator, we just need a radio.  A piece of wire attached to the OCXO's output would broadcast enough of a signal to hear the interference as a loud squeal in the radio.  By tuning the OCXO to minimize the pitch of the squeal it's easy to adjust the OCXO.  If the radio has a signal strength meter, you'll see the meter rise and fall when your OCXO is within a fraction of a Hertz of WWV's frequency.  Ham operators have used this trick for decades.

[edpalmer42]

I was thinking having the vreg tab thermaly fixed to the ocxo would make the whole caboosh more stable as in theory should the oven then not control the temp of the vreg as well sort of, Or am i missing something?.

No, the surface of the OCXO will be much cooler than the oscillator's core.  Some oscillators have insulation around the oven, some just use an air gap.  Regardless, they won't appreciate it if you ask them to try and keep something else at a constant temperature.  The heating circuit is only designed to keep the crystal and sometimes the core oscillator circuitry at a constant temperature.

[SteveThackery]

I was thinking having the vreg tab thermaly fixed to the ocxo would make the whole caboosh more stable as in theory should the oven then not control the temp of the vreg as well sort of, Or am i missing something?.

Your OXCO is off by 139Hz (14ppm). That means it is faulty. There is little more to be said, and especially it is pointless messing about with bodging in a 7805.

The Vref input can only affect the frequency by -2 to +2ppm.  That is less than the 14ppm error, so you can forget about fiddling with Vref - it's flogging a dead horse. 

Either the OXCO is broken, or your Advantest TR5823 is way out of calibration. Assuming the former, the only way forward is to buy another.

[807]

@ME...you never clarified the actual reading you were getting on your frequency meter. You stated it was 99999861Hz, which is nearly 100MHz. Did you add an extra 9 at the beginning, meaning that the oscillator was 139Hz out, or did you mean 9999986.1Hz, meaning the oscillator was only 13.9Hz out?

[shabaz]

Either the OXCO is broken, or your Advantest TR5823 is way out of calibration. Assuming the former, the only way forward is to buy another.

It's also expected that some may be faulty, since it's likely not much care is taken extracting them from their old equipment. It makes sense to buy (say) at least half a dozen or more, since they are (relatively) low-cost, and try each one, fully expecting some to need to be thrown away. I think I purchased 20 (still costs less than a couple of $ per oscillator) and will be happy if just one or two work out great. I have not got around to doing anything with them yet, since I don't have any decent strategy thought out how to set up the test for trying several of these at a go.

[BillyO]

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?

Not all of them, if any.  The ones I got a couple of years ago were from upgraded cell tower equipment.  China did a huge upgrade of their cell infrastructure a few years back which caused a flood of these on the market.  That they are all "fall outs" strikes me as the result of the popular sport of Sino Phobia.  The 10 I got all met their specifications without any issue.

[ME]

Either the OXCO is broken, or your Advantest TR5823 is way out of calibration. Assuming the former, the only way forward is to buy another.

It's also expected that some may be faulty, since it's likely not much care is taken extracting them from their old equipment. It makes sense to buy (say) at least half a dozen or more, since they are (relatively) low-cost, and try each one, fully expecting some to need to be thrown away. I think I purchased 20 (still costs less than a couple of $ per oscillator) and will be happy if just one or two work out great. I have not got around to doing anything with them yet, since I don't have any decent strategy thought out how to set up the test for trying several of these at a go.

fwiw heres a pic of it on the counter.

[BillyO]

Hmm, Advantest state the accuracy of the TR5823 as: +/- 1 count +/- timebase accuracy.  However, they give no specification on base timebase accuracy for either of the available timebase  option.  However, they quote the aging at 5x10^-7 / month.  Depending on how far off it was when new, it could very well be considerably out of whack.  I think the "error" you are seeing is a combination of both the counter being off and the OCXO being off.

I would just try putting a 100K pot between 5V and 0V with the wiper going to the vref pin and see if you can adjust it to your liking

[floobydust]

Oh... I see a varactor diode is used in OCXO's. My experience with varactor diodes is they age and drift badly, I kind of hate them as an unreliable beast.
This would explain the used modules (pulls) that are grossly out of spec or that do not respond much to their tuning voltage.

edit: these teardown pics of the OSC5A2B02 don't seem to show a varactor diode.  Mainly an ASIC with "303" or "501" markings, SOIC-16W (wide body).

Schematic of larger CTS 10MHz OCXO 1960017 (taken from The Sync Channel Blog) uses SMV1237 varactor diode SOT-23.

[ME]

Well heres an update on this turd, I decided to have another look at this, I tried two different counters and none was counting just reading zeros!, I thought i would have a poke around with my scope, Both sma outputs on the pcb have no outputs!, The only signal i get is on the nc pin on the ocxo block, its a clean sine wave at just over 1mv peak to peak, I guess this ocxo was on its last legs when i got it and has finaly died,I am waiting for a refund from aliexpress after i asked for a refund and they found in my favor, Does anyone know a decent seller for these so i can order a replacement, I even tried hooking up a pot like mentioned a few posts back, That allowed me to adjust the vref pin from 0v to 2.4volts but no joy!.?

[Uunoctium]

Bought 4 pcs bare OXCO's from this seller. Came without the pcb fragments.

https://www.ebay.de/itm/144769640308

Project is upgrading an very old fashined Yaesu YC500s (late '70) which is totally out of spec with the build in Toyocom TCXO.
That's mine solution – reading is stable over a dozen of hrs
Wiper of pot is aprox at 2/3 upper position

[SteveThackery]

I even tried hooking up a pot like mentioned a few posts back, That allowed me to adjust the vref pin from 0v to 2.4volts but no joy!.?

Why the question mark? We've explained this already. That Vref pin can only adjust the frequency by +/- 2ppm; your OXCO is off by 14ppm.  Hence the no joy.  Is there something unclear about this?

[shabaz]

They are (practically) a black box, if it doesn't work and you're convinced you're using it right, then you may need to try a different one.

Someone I trust tested about four or five of them, and although they all generated output, the best of that bunch had a ten- to twenty-fold difference in stability over a period of time, compared to the remainder. If that matters to you, then you'll likely need to try more than one.

I don't have a full strategy nor the complete test equipment for it (for instance, a GPS disciplined oscillator to be able to trim the end frequency), hence why mine are waiting as a back-burner project for when I do have the knowledge/capability.

But even from a five-second visual inspection, you can see that the best care was not taken in extracting and storing them. I'm not saying that doesn't make it worthwhile, but you have to be pragmatic and simply accept that not all will function (or function well).

[floobydust]

I think each of these modules has been tested and ranked after being pulled.
The oddball mini QR code on the underbelly- it's the wrong place for them, is unique for each module, and as a thermal label none would survive reflow - although I'm sure these modules are not compatible with reflow processes.
So what is this QR about then?

I saw one module taken apart and with no glue, leaving an air gap between the thermistor and thermal plate inside. That can't work very well lol. Other's have glue. So there might be a quality problem with some.

[shabaz]

You're right, they may have done a go/no-go test and measured frequency maybe at low/high control voltages, but a stability test would take a long time, so it's probably economically unviable for that to have been done and anything thorough at the low cost they were charging.

I'm willing to take apart one or two of these units to see what's inside (it might take me a few days to find some time, unfortunately). It will be interesting, anyway, just to see how similar this lot looks inside compared to other examples on the 'net.

[CaptDon]

As mentioned before and in different threads on EEV, these are pullouts being sold at two cents on the dollar of their original purchase price. The best of the best pullouts are probably being bought up and used by the Chinese military and the rest being dumped to eWaste marketers. You get what you pay for and from whom you are buying. Adjustable by +/- 2ppm but off by better than 10 ppm after warmup? You bought yourself some expensive target practice. Add this thread to the many others concerning TCXO / OCXO devices with very poor or totally unacceptable performance.

[807]

As I mentioned in one of the other threads, if you want to see what's inside, IMSAI Guy took one apart...

[floobydust]

Do we know who makes the ASIC? That video it's a "303" but recent teardowns it's a "501".
I note these VCOXO's modules sell for $100-$200 typically. So an entire board for $20 suddenly looks attractive, even if I have to cut open the thing and repair it lol.

[ME]

Well i have ordered 5 bare osc cans,the lot were only £5 ish, Also ordered a gpsdo , We shall see what the next lot are like, aliexpress for some reason aggreed with my refund then had an email saying final solution=no refund!, Not sure why, does anyone know if the xtal can be changed in these for different freqs?.

[shabaz]

I decided to take one apart. I appreciate there are already teardowns out there as someone mentioned, but I'd rather take my own hi-res photos.

It turned out to be just a three-minute task, using a Dremel cutting wheel.

I have not had a chance to look in detail at the photos yet, but they are attached.

This OCXO was purchased in March 2024.

EDIT: The transistor on the underside is 2SD1803, and you can see what looks like it may be an RTD or thermistor, between the transistor and the crystal. The crystal, transistor and that temperature sensor are all bonded together and to the metal plate, on top of some insulator that looks finely corrugated.

[floobydust]

I would guess the zener "B9", or crystal or ASIC have drifted somehow causing the frequency to be off. How the freq. is skewed by the control voltage, varactor diodes live somewhere. The tantalum looks OK.
I'm not sure of the oven target temperature for the crystal - but the outside QR code includes the crystal marking label as well. Hmmm.
Did you notice the oddball PCB layout for pin 2? It floats in space lol.

[Doctorandus_P]

Best I know, the things from Ebay / Ali / China are second hand OCXO (are you sure it has an oven? I don't trust what the cinese print on stuff) and salvaged from old equipment. The thing may be 10 or 20 years old and "well aged". If it's out of spec now I guess you're just out of luck and you'd better try another one. Also look at the scratches and dents on the photograph posted by shabaz. Does that look like new stuff to you?

You can of course always try to increase the adjustment range on the trim pin and see how far you can get. But even if you manage to get it within spec, how long will it stay that way? Once it starts drifting it's probably simply reached the end of it's life.

[shabaz]

I would guess the zener "B9", or crystal or ASIC have drifted somehow causing the frequency to be off. How the freq. is skewed by the control voltage, varactor diodes live somewhere. The tantalum looks OK.
I'm not sure of the oven target temperature for the crystal - but the outside QR code includes the crystal marking label as well. Hmmm.
Did you notice the oddball PCB layout for pin 2? It floats in space lol.

Hi, yes just noticed that pin 2 now! It's unsoldered as you say.
Just out of general curiosity I tried buzzing out parts of the circuit just now, and I believe the black boxes are small chokes (doesn't feel like ceramic, but might be plastic potted), the top SOT23-5 part (90SM) is likely a voltage regulator (that tantalum positive end is connected to it too), and possibly the oscillator output is pin 1 on the main IC (I'll power it up at some point and confirm), and DC is going into pin 1 too (via one of those chokes).

Presumably the lower-left SOT23-6 is an amplifier or RF switch maybe.

The power NPN on the other side is simply directly across the power rails.

[ME]

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