OCXO from ebay

[Flump]

I'm mad and frustrated and trying to keep a cool head about this.

After the replies in this thread
https://www.eevblog.com/forum/projects/frequency-standard-ideas/

I decided I wanted a very accurate 10Mhz oscillator so I found this one on ebay and messaged the seller
http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=251355497721&ssPageName=ADME:X:RTQ:GB:1123
I asked him if he checked each of the ocxo's before sending them out he said no there was no need and it would be
expensive to do that etc.
So I decided to purchase one and waited a few weeks for it to arrive and it came this morning, I was so excited
like it was christmass and then I hooked it up on my bench to find it approxomately 25Hz high which is massivly out of spec.

my refernces are
1) tcxo a that babysitter on here made for me then calibrated and adjusted it to be 9.999990 which you can see that in the above link at the end of the thread.
2) ocxo which I bought secondhand from ebay http://www.ebay.co.uk/itm/300802349781?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649

WWV the atomic clock on 10Mhz has a transmitter accuracy of 1 part per billion but the signal may degrade over air
to 1 part in 100 million, thats +/- 0.1 Hz

now my Yaesu 897d ham radio has a CW Tune indicator, when I tune in to WWV at exactly 10Mhz on the radio the Tune indicator is lit and stays lit if you go +/- 20 Hz , if you go +/- 30Hz the Tune indicator goes out, so that is how I know my
radio is revieving accurately on frequency (and the fact it was £800 and is a few months old)

So if I hook up the secondhand OCXO both of my counters report that it is exactly on frequency like +/- 1Hz
also the signal comming from the Oscilator is exactly the same as from WWV, so I am happy to belive this Oscillator is
reasonably accurate.

The tcxo send to me from babysitter on this forum he tuned to be 10Hz below 10Mhz and on both my counters
it reads 9.999990 +/- 2Hz , so again thats good enough for me to belive my counters are accurate.

but as I said the new ocxo from ebay is 25Hz high, the guy returned my email saying I can return it to him and once recieved
i can get a refund or replacement but thats not so easy for me because I am dissabled and house bound, also sending it
from the UK to to HK is going to be a lot more expensive I think than it cost him to send it to me, he did say he will pay the return post fees, but I need to get a qoute first, so two trips to a post office.

The thing about this I am not sure about is at the end of his message he said the following
"But I have to remind you that this is an ocxo, so the voltage control pin on this oscillator can't be left at floating state which may affects the frequency. And a normal oscillator with the control voltage would get you adjustment of 40Hz off 10MHz on up and down side. This may override the initial tolerance although this is not the main purpose. It is the temp stability that put the ocxo a top rating."

My second hand ocxo's VC pin is disconnected and it's still accurate

I think this is a mute point because even disconnected it should never be 25Hz out ?
or am I doing something wrong ?

[rolycat]

You do know you can adjust the frequency by applying a voltage to pin 1?

Range is +/- 40Hz, which should let you set it exactly.

[Flump]

rolycat I messed up when i made the post, there is a lot more info there now, sorry about that

yeah you can adjust it but the reason i bout it is because it is meand to be accurate already
the only way i could adjust it is using my counter and that is calibrated to my other ocxo
so I am not gaining anything by doing that, the reason i got the new ocxo is it should have been
more precise and accurate than what i already have.

[grumpydoc]

yeah you can adjust it but the reason i bout it is because it is meand to be accurate already

I can't see the ebay auction (ebay blocked at work) but a lot of modern OCXO's only have a voltage adjust pin for trimming.

They drift a small amount - typically 0.5ppm/year which is 5Hz for a 10MHz osc - they have enough adjustment to give a 10 year lifespan.

To be honest the ones which have a trimmer to adjust the frequency could be adjusted by the seller but it's likely that the trip through the mail with its attendent jostling and mechanical shocks would move the frequency more than 25Hz

[johansen]

no, you can't leave it floating. seems the data sheet for those devices specifies the calibration frequency with the voltage adjust pin grounded.

note that the calibration frequency may not be 10Mhz because the voltage adjust limits are .5 volts to 5 volts. so, with it grounded, it may not be centered at 10mhz, and given the age, that's probably expected anyways

[grumpydoc]

OK, I can see the auction now - I've bought a couple of those OCXO's from that seller.

seems the data sheet for those devices specifies the calibration frequency with the voltage adjust pin grounded.

No, that's a specific variant which would have YA or YB in the part number in place of the V5 in the part number being sold.

You have to drive pin1 with a voltage - I've always assumed that the "zero point" is approx V_DD/2 and it would be sensible for the manufacturer to bias the pin so it sits there if unconnected but I don't know whether they do.

Remember that OCXO's aren't about accuracy - that comes from an external source - but about stability.

That said if yours is 25Hz out either pin 1 floats towards one end of the adjustment range (+/- 3ppm or 30Hz) or you have a marginal unit.

I'd start by getting a couple of 1k resistors connected in series. Tie one end to V_DD, one end to ground and  connect the centre point to pin1, then see if the frequency looks any better.

[Flump]

here are some pictures

babysitters 9.999990 TCXO


second hand OCXO



new ocxo from ebay

[grumpydoc]

here are some pictures

Have you had a go at biasing pin 1 to 2.5V and checking?

As you're in the UK if you want me to check it against my Rubidium standard and let you know what control voltage it needs for 10.000000MHz drop me a PM.

[Flump]

Hi grumpydoc
thanks for the advice

here is the voltage divider made from 1k resistors

I would love it checked but I am trying to work out if it is faulty at the moment.
I thought just connect the supply and gnd and you would get 10.000000 out.

[babysitter]

Aging might have prevented it from being at exactly 00000... at 2.5V - its a control voltage, you might need to go a bit up or down to hit exactly, you might use a 10K pot with its ends between 0V and 5V (or a VREF out pin of the OCXO?) and see if you can trim it above and below 10 MHz. If it is not symetrically, dont panic. In case you cant get it from one side of 10.000 MHz to the other - only then consider shipping back !

[grumpydoc]

OK, that's better and well within the 3ppm pull range but still not great.

I don't know where the seller gets his "frequency tolerance 0.15ppm" from - that figure isn't in the datasheet, nor would I really expect it there.

I'd sort-of expect a new module to be fairly close to 10MHz - common sense would suggest they'd set it that way after some initial ageing at the factory.

As I said, I have bought a couple of these. At least one would/will not settle down very well at all. The seller says they are "new" but they could still be units that have failed QC, with sales like this one hopes that they are just part of a batch bought and no longer needed for whatever reason but it's always hard to tell.

As I said, as long as you pay postage both ways I'll run it against my Rubidium unit (or the GPSDO if I'm feeling keen) and let you know what control voltage to to feed it. It will likely still drift though. Drop me  a PM.

[grumpydoc]

Aging might have prevented it from being at exactly 00000... at 2.5V - its a control voltage, you might need to go a bit up or down to hit exactly, you might use a 10K pot with its ends between 0V and 5V (or a VREF out pin of the OCXO?) and see if you can trim it above and below 10 MHz. If it is not symetrically, dont panic. In case you cant get it from one side of 10.000 MHz to the other - only then consider shipping back !

I'd have thought it would trim fine - it's only 1.8ppm out and the trim range is supposed to be 3ppm.

[Flump]

Update on this I replaced the 2 1K resistors with a 10K multiturn VR
and i can adjust the OCXO from 9.999939 to 10.000060

Vin = 4.999

1.82v applied to voltage control (pin 1) bring the frequency to 10.000000 +/- the accuracy of my counter.

So would we say that the OCXO is working correctly as intended ?

[retrolefty]

Update on this I replaced the 2 1K resistors with a 10K multiturn VR
and i can adjust the OCXO from 9.999939 to 10.000060

Vin = 4.999

1.82v applied to voltage control (pin 1) bring the frequency to 10.000000 +/- the accuracy of my counter.

So would we say that the OCXO is working correctly as intended ?

 I would. As others have implied you didn't purchase a 'primary standard' but rather a 'secondary standard' that when trimmed to agree with a primary standard will stay within specified accuracy.

[grumpydoc]

So would we say that the OCXO is working correctly as intended

Well, the control voltage range is OK - +/- 6ppm - nicely symmetrical which is actually fairly reassuring.

We don't know anything about phase noise, temperature coefficients or age related drift but you're probably no too interested in those to be honest

As others have implied you didn't purchase a 'primary standard' but rather a 'secondary standard' that when trimmed to agree with a primary standard will stay within specified accuracy.

Primary standards aren't exactly household items. About the best you can get on ebay would be a rubidium unit or a GPSDO but these rather seem to have dried up at the moment.

[Stray Electron]

OP said:

also sending it
from the UK to to HK is going to be a lot more expensive I think than it cost him to send it to me, he did say he will pay the return post fees, but I need to get a qoute first, so two trips to a post office.

   If you're still intent on sending it back OR just for future reference, there's no need to make two trips to the post office. If the seller wants it back then take take to the PO and send it. You don't need to get an estimate ahead of time. Seller can do that online if he wants one! Per E-bay's rules (in the US at least and I think world wide), if the seller sold you a "new" or "used" item that proves defective then they must reimburse you for the return postage plus your costs including shipping and any fees that they charged.  It's somewhat buried in Ebay's rules but you can look it up under returning an item.  I just went through this with an automotive part that I bought. The seller described it as "used" which per E-bay's rules means that it has to be 100% functional. But when I got it it was a piece of crap and had obviously never been checked.  I immediately opened a case against him and after that he resolved the case promptly.

   Having said that,  if I'd gotten a reply like you did to your inquiry, I'd have never bought the item! It is NOT expensive to check an OCXO!

[dannyf]

Maybe your expectation is unrealistic or unreasonable.

there is a multitude of reasons why stuff on eBay is cheap, terrible service being one of them. If you want more service, maybe you should buy from someone offering such services, like a local shop, etc. And be expected to pay for such services.

in this case, the seller told you upfront that he isn't offering such services and you decided to buy from him anyway.

he delivered a product that performed in-spec per your measurement. And told you correctly that you can trim the frequency to your own spec.

to me, he held his side of the bargain and earned the money. It is great that he offered a refund and offered to pay for the return postage.

quite frankly I don't know why anyone would sell over eBay, unless he is making a killing. As much buyer protection as eBay offers, it sucks sellers dry.

[grumpydoc]

quite frankly I don't know why anyone would sell over eBay, unless he is making a killing. As much buyer protection as eBay offers, it sucks sellers dry.

It's getting to be more work to sell on ebay, that's true.

It's also true that this is no bad thing, ebay is trying hard to clean up its act - I've felt for a long time that it wants to compete with the Amazon Marketplace and for that it needs reliable sellers so it has slowly been forcing sellers to improve (or jump ship).

The latest wheeze is to manage the returns process - of course charging the seller for the return delivery.

This does annoy me a bit. Not because I don't want to accept returns - far from it. I try to be as fair as possible with anyone who buys stuff from me and will happily take back stuff that doesn't work (rare, I hope) or doesn't work out. BUT I want to control that myself and use the most efficient service I can - not one which costs twice as much and co-incidentally happens to give ebay a cut.

For that I have to thank not only ebay but all the sellers who refuse to take stuff back and generate a poor reputation for the rest of us.

However if you want to sell 2nd hand test equipment it's essentially the only marketplace.

[Flump]

Hello danny

being new to OCXO's in fact the only previous experience I have of them
is from the 2nd hand unit i recently bought from ebay which did not need any
form of intervention to be +/- 2Hz, so I assumed this new OCXO would be the
same but more accurate, after all there is nothing in the advert that says
it had to be adjusted before use.

Micro Crystal OCXOVT-BV5
NEW OCXO Oscillator
10 MHz / 0.15ppm (-20C ~ +70C) / DIP/ +5V Single Supply
Frequency Tolerance : +/- 0.15 ppm max

I agree with what you said mostly,
he said after the sale that the frequency control pin could not
be left loose, if i had known that before the sale then I would not have
bought it.

So yeah you are right he delivered a working unit and held up his end
of the deal but I feel it should have been made apparent that the OCXO
needed to be trimmed before use if you expected any kind of accuracy.

Anyways all is well that ends well and I have given him good feedback
and i have learned a few more things about these
and thats never a bad thing.

[dannyf]

it needs reliable sellers so it has slowly been forcing sellers to improve (or jump ship).

I can understand their attempt at disciplining their sellers: to attract more buyers, which in turn makes ebay more valuable for sellers.

What it also does is to attract moronic or even fraudulent buyers who take advantage of the buyer protection. This will drive up the prices for every good buyer. So there is a natural balance.

In the end, I think a more efficient market place is beneficial to all. That means rating sellers AS WELL AS buyers, and to potentially offer different grades of services.

Hopefully Alibaba's entrance into the market will help eBay find that compromise.

[dannyf]

so I assumed this new OCXO would be the
same but more accurate,

I wouldn't assume that.

The datasheet says 3ppm frequency stability. Your measurement, if it is done after sufficient time for the oscillator to warm up, is 2.5ppm.

after all there is nothing in the advert that says
it had to be adjusted before use.

I would rely on the datasheet instead.

Actually, as an OCXO, yours is fairly inaccurate at 3ppm: most of OCXOs are <1ppm. Performance-wise, yours sounds more like a TCXO.

[Flump]

Actually, as an OCXO, yours is fairly inaccurate at 3ppm: most of OCXOs are <1ppm. Performance-wise, yours sounds more like a TCXO.

At the moment I am not very impressed by it, I'm going to shoot grumpydoc a PM to see if his offer
is still on the table.

[dannyf]

You should measure the current into the ocxo. Or put a finger on it when it is powered. It should feel warm / hot.

[Jay_Diddy_B]

Hi,

To the original poster.

They are well insulated, I doubt you will feel the case warm up.

Since an OCXO is an Oven controlled oscillator you should see the oven warm up.

The way to test for test for this:

• Leave your frequency plugged in an on for several hours. This is to make that the oscillator in the counter is warmed up and not drifting.
  Take the OCXO that is cold, not been operated for several hours.
   
  Monitor the current consumption of the oscillator. The current consumption will drop when the OCXO reaches the operating temperature.
  
  The oscillator should drift in frequency until the operating temperature is reached. The temperature and frequency typically stabilize after 10 minutes.
  
  If the oven in the OCXO is broken you will see not see the frequency or current consumption change.
  

I have fixed some HP 10811-60111 OCXO by replacing the thermal fuse. I have no experience with the unit you have.

Regards,

Jay_Diddy_B

[babysitter]

Let me summarize a bit about care and feeding of unknown OCXOs:

Try to get the data sheet of your OCXO, it will tell you almost everything you need.
The pin connected to the case is always GND. Try to find a hint at the internet which pin is actually supply voltage. Dont be surprised when discovering 2 supply voltages, one for the oven, one for the oscillator.

Certainly you will find a Pin called control, Vtune, Utune,EFC, UEFC or such, this is a voltage that will pull the oscillator frequency around by means of a varicap inside. Sometimes you might find this pin internally biased to the midpoint, either VCC/2 or for example VREF/2 or something different which could be telling you "I am half the maximum voltage!"

You can control this voltage with a potentiometer between ground and either oscillator supply or a additional UREF, VREF pin if available.
You can reduce the adjustment range and subsequently have easier fine control by adding a series resistor between the wiper of your pot and the control input.

The power consumption of the oscillator should be rather low but quite constant, I would consider 3 to 50 mA reasonable. The oven will start out drawing a lot of current in the beginning but decreasing after a few minutes. Above 1.5A initial, the oven current would look suspicious to me.

Now you can watch the oscillator dance for a while - while the oven current starts up high but will settle to around 1/4th of inital current, the current consumption shall be quite stable after 5-60 Minutes. Oscillating current consumption points to a bad regulation, needs repair.

The oscillator will start out far from its target frequency necessarily, most probably hundreds of Hz lower. It will home in on its target frequency with a similar behaviour as the oven but I saw that it will take longer usually. (Sample size 2.)

If both don not stabilize, consider that you need more supply voltage.

Setting the control voltage: Make sure that varying the control voltage varies the frequency in a consistent manner. lower voltage=higher frequency or vice versa, without kinks and bends. If you have a trusted 10 MHz Source, check if you can tune above and below 10 MHz.
Half-range control voltage should be close to 10 MHz. It isnt necessarily symmetrical as aging might have moved the center f. Only if you can only tune above OR below 10 MHz, youre basically screwed - this damns you to memorizing or noting the difference you settled for and do some calculations if you use it.

Coarse tuning can be done with any garden variety frequency counter. "Copying" the same frequency of a trusted source can be done by connecting both oscillators to a o-scope, triggering on the trusted one. you will see that your target oscillator signal runs relative to your trusted signal. When adjusting the tune voltage you will realise that it will either run faster (wrong direction!) or slows down (right direction) - and getting up to speed as soon as you are pulling behind the right frequency (getting wronger again). You want your target oscillators trace moving as slow as possible relative to the sources trace. Still standing over minutes or hours is perfect :-)

@flump especially:

You are not lost. The OCXO will provide a good stability. If you set it right now when both your fixed oscillaltor and also my -10 MHz source are new, they are probably quite accurate. Time will play less well with them. Setting your well-heated (over 24h)OCXO soon according to them will be like a backup of their performance of nowadays!