Any tips on opening a XTAL oven?

[PM3295]

I have a HP 8920A comms monitor with a faulty oven oscillator. There is a 10 MHz output coming from it but with a lot of noise in the 10 M sine wave. It was made by "Oak Frequency Control" which is no longer in existence. I found a OFC unit on eBay (picture in link below), but the operating voltage and pin-outs are different. This xtal oven fit inside the A19 reference module so there is a height restriction of around 15mm for any suitable replacement.

http://tinyurl.com/lapdfa8

Before going into modifications, I want to open it up and see if it can be fixed.

My question: What is the best way to open this without causing further heat damage to the inside?   

[CocaCola]

My guess is that it's potted inside the can, so when you open the can up you will likely be faced with a block of epoxy that will need to be ever so painfully scraped away...  IMO no worth the trouble, there has to be a suitable replacement out there...

[pickit2]

most xtal ovens of the type shown in the link are first a tight fit, and have a thick solder filling of any gap.
not tried to open one but if you have a faulty one, you have nothing to loose,

Have you tried a standard 10 Mhz xtal module as a replacement, some have a better spec than the old xtal ovens.

[PM3295]

Any replacement must have an electronic frequency adjustment pin. In this instrument, the frequency can be calibrated by setting a coarse and fine DAC value via the service menu. It also contains circuitry so the internal 10 MHz reference can be phase-locked against an external 10 MHz reference. This HP unit is fitted with the "High Stab" option, so I will have to be careful not to degrade the performance with the replacement I choose.

If i can find a smaller suitable replacement and mount it in the same housing of the old osc, that would be great. I will have to go and research the stability figures of this old osc and try to find something smaller with equal or better specifications.

I suppose double-oven means double oven cavities? 

[pickit2]

http://www.quartzcom.com/en/through-hole-type-_content---1--1052.html

[solutions]

Bliley OVCXOs are comms grade, high stability, stuff. About the size of a soft drink can if you were to give it a box shape. These were used in verrrrry low bandwidth, extremely low phase noise, PLLs one level down from atomic clock masters in the phone network and could be disconnected for up to three days and not drift out of frequency tolerance. Vectron are another good source.

All OVCXOs are not created equal and an instrument needs to be right at the bleeding spec edge. The little 20mm cubes are toys.

But, you knew that.

[fpgaguy]

http://www.abracon.com/
A0CTQ5

I use them for something but, a bit less $$ than others

[PM3295]

I spoke to a person well familiar with repairing these instruments and got the confirmation that the ovens are not potted. The inside is packed with a heat-resistant sponge type material. He suggested to apply rapid heat to the top of the enclosure (like dipping it upside down into a solder pot), and the expanding air pressure inside the oven will pop it open while melting the solder seams at the same time.

Now I will have to go find a solder pot to borrow that the oven will fit into.

[solutions]

Use a propane or acetylene torch to melt the solder quickly.

Seriously.

We've used them to salvage soldered memory chips

[titi]

Hi PM3295,
In this web page, it is explain how to repair an Morion MV89A (I think this one is bigger yours):
http://www.rbarrios.com/projects/MV89A/
He is using an small propane torch to open the cover.
I had the same problem in the past with the same OCXO and I could open it with a heat gun.
It was the same problem as on the web site the output capacitor has lost it's value from 10nF to hundred pico-Farads.
Best regards.

[PM3295]

My one friend has a Hakko solder pot which should be ideal. It has a square pot, and the oven will fit into it nicely. I am picking it up over the weekend. I will post an update here soon.

[solutions]

Gradual heat will kill it.

Were you planning on immersing it in solder or just sticking the oven in an empty pot?

[PM3295]

I am going to get the pot with solder up to temperature and immerse the top into the molten solder to have it heat up the enclosure quickly.
I have the solder pot now and will do it later today. Hopefully, it works out.

This is the type of solder pot I borrowed. http://tinyurl.com/mflkmx2

[PM3295]

Wow, did this opened with a bang! It took about 10 sec in the solder. It sounded like an electrolytic cap exploding. Well after this excitement, I managed to do some measurements. I have a good clean 10 MHz sine-wave up to an SMD device marked "E1." After this device, the signal is low and noisy. The package looks like an SOT-236. The closest I get from my code book for E1 is a BFS17, but the package is SOT-23. Under my microscope, it looks like only three pins are used. I wonder what this could be?

[pickit2]

http://www.s-manuals.com/smd/e1
there is a LDO regulator that is SOT-236

[Ichan]

I managed to do some measurements. I have a good clean 10 MHz sine-wave up to an SMD device marked "E1." After this device, the signal is low and noisy. The package looks like an SOT-236. The closest I get from my code book for E1 is a BFS17, but the package is SOT-23. Under my microscope, it looks like only three pins are used. I wonder what this could be?

ESD protection diode (array) ?

-ichan

[h0nk]

If the OCXO is specified for CMOS/TTL-output, it could be a single/double gate inverter.


Best Regards

[PM3295]

This package is near the output pin, so I suspect it is some amplifier or buffer. The output on this reference is a sine-wave.

One of the pins of the SOT-236 is tied to the internal 5V regulated bus. This oven runs from 12V. It has two separate 12V supply pins, one for the heater control circuit and other for the rest of the electronics.

I found another posting on opening an XTAL oven here: http://www.eevblog.com/forum/repair/how-to-open-an-ocxo/msg480836/#msg480836  

[PM3295]

Repaired! Well, I took a chance on the device acting like an emitter follower* since I could trace a resistor between pin 6 and 1 as well as a resistor between 1 and GND. Pin 4 went through some LC's (maybe a LPF) to the output pin. So I soldered an NPN RF transistor (I think it is a BFR106 or something) like shown. I was rewarded with a nice strong clean 10 MHz output! It all worked out well. The oven is back in the instrument and seems to be working perfect after I calibrated it against my GPS frequency standard.

I want to thank everyone for your helpful comments and suggestions.

*(Pin 6 was tied to the 5V rail, so I figured that could be the collector of the emitter follower)

[solutions]

It would have been interesting to see a spectrum analyzer noise plot of your "RF transistor" follower to see if the OCXO still meets specs.

[PM3295]

Do you mean measuring the phase-noise spec? I don't have a phase noise test set, but I did check the harmonics on a spectrum analyzer. The worst was about -40 dBc.