OCXO from ebay

[grumpydoc]

What it also does is to attract moronic or even fraudulent buyers who take advantage of the buyer protection

This is about the only upside to managed returns that I can see - if buyers are repeatedly returning stuff then that should flag up to ebay. Whether they do much about it we'll see.

[Flump]

Thanks very much babysitter fir the explanation 

I am sending my 2 OCXO to grumpydoc next week
to check against his rubidium reference,
I will have 2 accurate ocxo + the FLUMPILLATOR  O0
so I will be all set then

[grumpydoc]

I am sending my 2 OCXO to grumpydoc next week to check against his rubidium reference....

...which arrived a few days ago, I've just had chance to start to have a look at them.

The Oscilloquartz unit which is definitely the better of the two is in good condition. With no connection to the adj pin it floated at about 3.3V and the output frequency was 9.99999886MHz or about 1.114Hz slow. Pulling adj to 4.745V gave an almost spot on 9.999999997MHz - 0.3ppb out and about as close as I can measure anyway.

Given that the output is at 10MHz with the control voltage very close to the centre of its range (0-10V) means there's plenty of adjustment left and drift will have been well within spec. The current draw shows the oven control to be OK as well - starting about 0.5A, rising slightly to peak at 0.53A and settling at 0.22A with the unit warmed up (just slightly over spec as it's supposed to be < 2.5W warmed up but it's not too far off).

I think Flump was most interested in the micro crystal unit - I'll have a look at that and some more checks on the Oscilloquartz unit next week.

[dannyf]

0.22A with the unit warmed up

1w when warmed up is fairly typical.

I have found that most of the times, a TCXO (2.5ppm for example) is fairly good, as they typically come in at around <1ppm unadjusted. The ones I am testing now consumes < 2ma and "warm" in seconds.

[grumpydoc]

0.22A with the unit warmed up

1w when warmed up is fairly typical.

The 8663 is a 12V supply so this is 2.64W which is a bit over spec (< 2.5W) and it's warm in the workshop tonight - about 28^oC so it shouldn't really need to be working that hard.

However it's quite possible any insulating foam has deteriorated through use, explaining the higher draw. Certainly the outside of the unit is fairly warm.

[dannyf]

They are usually warm to the touch so that doesn't surprise me.

Maybe yours does run hotter. Or your room temperature is lower or has better ventilation, etc., causing the oven to consume more energy to maintain its desired temperature.

[Flump]

I made a mistake buying the Micro Crystal unit, I thought it was going to be near enough spot on frequency like the Oscilloquartz one but better because it was new,
Still it will be usefull to have another to use as a reference.

I look for to see what the MC one is like and if it is stable.

Thanks so far grumpydoc 

[grumpydoc]

I made a mistake buying the Micro Crystal unit, I thought it was going to be near enough spot on frequency like the Oscilloquartz one but better because it was new,
Still it will be usefull to have another to use as a reference.

I look for to see what the MC one is like and if it is stable.

I don't think that it was a mistake.

However it is clear from the specs that the Oscilloquartz is the better unit.

Turning to testing the Micro Crystal one....

Hooking it up initial current draw is 190mA, falling to 60mA as it warms up so that looks good and all in spec.

Leaving V_adj floating it settles at around 2.97V and, at that, the output is 21.27Hz high - so Flump's original measurement of 25Hz high was pretty close to mine. Certainly close enough for most of the difference to be explained by supply voltage differences (following measurements at V_DD = 5.002V), or ambient temp.

Pulling the V_adj pin to 1.974V gets it pretty close - 10.000000114MHz just now, but it's a pig to adjust by hand and is drifting somewhat making it almost impossible to pin down better than about 1 in 10⁷.

Part of the reason it's harder to adjust is that the pull range of the MC unit is 10x the 8663 - ±3ppm (30Hz) rather than ±0.3ppm. This, in turn, is because the 10 year stability of the 8663 is 10x better than the MC one.

The 8663 also has a better rejection of supply variations as well: 3x10^-10 over the 12V ±5% supply voltage range vs 1x10^-7 for the MC oscillator. Running on the breadboard I'm finding it hard to pin down the last mV of stability on V_DD and V_adj and it definitely affects the MC unit more than the 8663.

the 8663 is also massively better on variation vs ambient temp. The spec is 4x10^-9 over the operating temperature range of -20^oC to +70^oC. The corresponding figure for the MC unit is 150x10^-9 even for the "high stability" variant.

At the end of the day this oscillator is probably in spec but playing with it mirrors the experience I had with another unit that I (coincidentally) bought off the same vendor - I just couldn't get it to settle down to much better than 10^-7 stability. Conversely another one I had (again from the same source) is much better behaved. I suspect the MC unit was designed assuming fairly tightly controlled supply voltage and probably the frequency control voltage coming from a DAC rather than a multi-turn trimpot.

I think that if Flump wants a reference oscillator the 8663 is by far his best choice - if he drives the frequency control pin at 4.75V it will almost certainly be within 0.1Hz of 10MHz. To reliably get any more accurate I think will take a Rubidium standard or GPSDO - sadly both now seem very rare on ebay. In fact even cheap OCXO's are not as cheap or plentiful as 12-18 months ago.

[Flump]

Grumpy thanks sooo much for checking them both out.

On one hand I bought the MC XO which was a bit of a let down
but on the other the 8663 is really good
so I got what I needed in the end which was a decent 10Mhz reference.

How long would the MC XO stay accurate for do you think ?
10.000000114 is still pretty good.

would the MC make a good reference for my dana counter ?
as it just uses a standard 1Mhz crystal but it has an external
socket for a reference, i could use a 10:1 divider on it.

I'm loving the 8663 right now though lol
 

[grumpydoc]

Grumpy thanks sooo much for checking them both out.

On one hand I bought the MC XO which was a bit of a let down but on the other the 8663 is really good so I got what I needed in the end which was a decent 10Mhz reference.

How long would the MC XO stay accurate for do you think ?
10.000000114 is still pretty good.

would the MC make a good reference for my dana counter ? as it just uses a standard 1Mhz crystal but it has an external socket for a reference, i could use a 10:1 divider on it.

I'm loving the 8663 right now though lol

I've turned them off but over the space of this evening it drifted up about 0.1Hz to 10.000000208MHz. You can probably assume it's going to settle to within 1x10^-7 but I'm not sure it's going to be easy to get it better than this.

If your Racal frequency meter has the 04C option it would be a better oscillator, if it has the 04T then maybe - if it's 04A or 04B then those are ovened oscillators and much closer to the 8663.

I'd probably make a little PCB to go inside the Racal with a trimpot accessible from the rear of the counter. If you do this you need to divide by 2 as it's actually a 5MHz  standard internally and watch for the fact that the OCXO supply is 12V, not 5. I actually bought mine to go in some Racal 19xx series counters as I acquired one with no oscillator and one with a broken one. As I said one is fine, the other I can't quite get to settle - rather like yours.

Yes, the 8663 is one of the nicer OCXO's kicking around ebay - I used one to fix my GPSDO see https://www.eevblog.com/forum/repair/rapco-1804m-repair-10mhz-conversionbodge-(-teardown)/msg137237/#msg137237

[Flump]

heres my osc board
and the pdf mine is the 9914

I was going to use the xo as an external reference to leave the counter original and i think its 1mhz for that



http://www.drop22.com/images/dana/Racal%20Counters1984-5%20%281%29.pdf

[grumpydoc]

heres my osc board

Yes, that's the 04C - un-compensated simple crystal osc. The idea was that would get the unit working but you'd use an external ref for any measurements.

I was going to use the xo as an external reference to leave the counter original and i think its 1mhz for that

It is. Doing it by replacing the PCB eliminates a box and some wiring but it will work just as well as an external standard.

[wyphy]

I'm a bit late to this party, but hey, I just joined the forum.

I'm actually working on my own OCXO reference so the title caught my eye. (Shameless self promotion : https://www.eevblog.com/forum/projects/10mhz-ocxo-frequency-reference-project/)
I also noticed that you mention the Yaesu 897D, so I presume you're an amateur radio op, which doubly made me think of something that W2AEW posted a couple years ago, and will get you to better than the +-20Hz you mentioned with the CW tuning indicator.
Isn't quite so easy with a digital S-meter, but borrow something with a real analog S-meter and you'll be fine.  My Kenwood TS-530 will do just fine for me.

Cheers.