Another mini teardown - HCD-66-SC 5MHz ocxo

[grumpydoc]

OK, not quite as small as a single IC but still not large...

It looks as though I'm not going to be able to easily find a replacement 5MHz ocxo so I guess repair is the only option. Well, I gould get a new one for USD700 but that's a bit beyone what's reasonable :-/

The first problem is that there is a solder seam the whole way around the base as can be seen in these photos (click on picture for larger version)





Some exploratory attempts proved my initial suspicion that none of my irons were large enough - the biggest is about 60W or so, and the hot air gun didn't touch it even set to 480deg C

So, I reached for the next larger source of heat which is a small blowtorch I have, set a small flame and ran it round the whole seam then pulled the top off with a gloved hand. Things did get a bit warm but only for a short while.

The first thing I encountered was some insulating foam - this was rather sticky. Not sure whether it had gone like that over time or due to my heating the whole thing a bit too much. On three sides the foam was separated from the case by what looks like some thin SRBP. This had blistered a bit so obviously things were a bit hot just there but the foam was much the same whether "protected" by the SRBP or not so I'm wondering whether this has just disintegrated over time.

Pulling/washing the foam off reveals three PCBs and a further enclosed metal case with the crystal. A TIP21 bolted to this acts as the heating element. Originally there had been some cable ties anchoring the crystal enclosure but these were brittle - presumably from the heat of the oven. You can see them in one of the photos.

Hooking the unit up to 12V and an oscilloscope shows that it has survived the encounter with the blow torch and produces a nice 5MHz sine wave at 2.5V p-p (into 10Meg ohm). Irritatingly it now doesn't really want to misbehave - presumably because whatever didn't like the heat is now cooler with the insulating foam removed. The inner metal case gets too hot to touch but the thermostat seems to be working because the current drawn drops from about 380mA to 240 or so. It should probably be lower but, again, without the usual amount of insulation it's likely to draw
more keeping the oven up to temperature.

So, in terms of fixing it I'm not totally clear which way to go. The output does drop to about 1.8V p-p when the oven is fully warm which might be related to the original fault. I suppose the fact that the fault has "gone away" eliminates the oven assembly or crystal itself as the source of the problem. I suspect a thorough clean followed by careful examination of all the joints and components is the way to go - certainly that seems to be the feeling on Time-Nuts


Whoops - things might have gotten a little hot!


Three PCBs inside - the black stuff is remains of insulating foam.
This looks like the oven control PCB


And this is presumably the 7V reference.


TIP21 as heat source at the top of the photo - the crystal itself is enclosed in the box - not tearing that down as it comes up to design temp so if the fault were here it should still show itself.


Finally just another view of the internal construction.

[SeanB]

Clean the boards and look under a magnifier for dry joints and cracked ceramic caps. The toasting probably fixed a dry joint, but you can try a little freezer spray on the caps on the boards to see if it faults.

[grumpydoc]

Clean the boards and look under a magnifier for dry joints and cracked ceramic caps. The toasting probably fixed a dry joint, but you can try a little freezer spray on the caps on the boards to see if it faults.

Yes, that's the plan.

[muvideo]

Nice, given the few components, I would pass the joints
with the iron to refresh them. In the past a pair of times
I found broken smd resistors that split in two once touched
by the iron.

For coincidence I read your thread on time nuts.
Dont know if can help, and if the size and charachteristics
are ok, but also the Racal 1992 opt04E ( and probably others)
contains a 5MHz ocxo, that is doubled by a small pcb that's
mounted on the outside of the oven.
Given the known keyboard problems it's easy to find cheap
units around.

Fabio.

[grumpydoc]

Don't know if can help, and if the size and characteristics are ok, but also the Racal 1992 opt04E ( and probably others) contains a 5MHz ocxo, that is doubled by a small pcb that's mounted on the outside of the oven.

I have a 1998 with the 04E oscillator, it's quite a nice unit, just bought randomly from ebay it's still within 5x10^-8. In fact that's one reason I want the Rapco running as there's no real point trying to adjust the timebase on the 1998 unless I can come up with something that's closer than about 1 in 10^-10. When running fully disciplined the Rapco gets to about 3 in 10^-11.

The 04E oscillator isn't suitable as it is to large to go in the Rapco case (which is 1U). Also I don't think it has a voltage control input.

The one item I have which has an ocxo which is PBC mounted and probably the correct pin-out and dimensions is a Marconi 2022 sig gen which is faulty. However I haven't really sat down with that one to see if it's fixable so I don't want to condemn it as "parts only" just yet. Also I'm pretty certain that it has a 10MHz oscillator.

Given the known keyboard problems it's easy to find cheap units around.

I haven't seen one on ebay or at any of the local radio rallies recently. I did pick up a "working" 1991 which needed one button replacing which I will be selling at some point. They do well with small tactile switches in place of the originals and the keytops can easily be drilled out to fit the new switch - there's a teardown of a 1992 somewhere on the forum.

[muvideo]

button replacing which I will be selling at some point. They do well with small tactile switches in place of the originals and the keytops can easily be drilled out to fit the new switch - there's a teardown of a 1992 somewhere on the forum.

I know 

Good luck with your repair.
Fabio.

[grumpydoc]

I know 

Ah, yes, sorry - that was yours wasn't it!

Clean the boards

Talking of which does anyone have any suggestions as to with what? I've tried IPA, ethanol, acetone and even some of the tiny amount of 1,1,1 trichloroethane I have left (yes, I know it's banned). Nothing so far has had much impact on the remains of the foam or the remains of the sticky tape (the yellow residue on some of the components in the photos).

I'm running out of ideas as to what might clean it off without damaging the board.

[Fraser]

When I read that you were having difficulties finding a new 5MHz OCXO I was a little surprised. I have bought many via eBay over the years.

A quick search on ebay.co.uk found this good quality unit for a very reasonable price:

http://www.ebay.co.uk/itm/Symmetricom-106824-5Mhz-oscillator-12V-OCXO-Oscillator-Digital-EFC-/300663293785?pt=Vintage_Electronics_R2&hash=item4600edc759

I have dealt with 'fluke.l' in the past and he is a very honest seller with great quality 'second user' items.

By all means repair the excellent quality HCD unit but you may wish to consider buying a spare unit. You would not normally need an exact drop in replacement as small mods can be made to the control line and supply to match the 'new' unit.

Regards

Aurora

[grumpydoc]

Yep, spotted that one.

I have nothing against the vendor, I've bought a few items from China and apart from one which took truly ages to arrive haven't had any problems (yet anyway).

The reason that I didn't buy the Symmetricom is that I could find no data on it anywhere.

I did later turn up some specs in another auction. However if the pin-out given there is for the ocxo itself (which has 7 pins, so it might be) then it's clear it won't work in the Rapco which expects an oxco with a voltage controlled frequency adjust, not a digital one. I wouldn't be able to get that working without a lot of effort.

There are a couple of further 5MHz units which are too big for the 1U case of the Rapco. Finally there is this one but, again, I can't find a datasheet - even on the Vectron web site (which lists a 351 series in the archive but not 251). Also I'm a bit wary of that particular vendor - they have good feedback but auction a lot of stuff that looks like it has been through a war zone (and back) - there are several previous threads on the forums here. I don't mind military surplus but preferably not stuff which has seen the wrong end of live ordnance.

So the difficulty is not finding a 5MHz ocxo per-se but I can't see one I can use in the Rapco.

There are plenty of 10MHz units around which look to be a good fit and 10MHz would be a more useful output frequency so I'm still considering buying one of those and putting a D-type flip flop in the signal chain to the processor to divide it by 2. In fact the Rapco does come in 10MHz output versions so it's always possible the firmware is set up so that a 10MHz ocxo "just works"

[Fraser]

Ah I see your problem now.

I have the same Rapco unit, bought on ebay a little while ago. I missed the application of your OCXO when posting my original response. I was considering modifying it to 10MHz but was advised against it. A good friend who knows more about GPSDO design advised that some have firmware that is very carefully 'tuned' to the OCXO and its characteristics. You would be well to repair your HCD in in this cae. I am considering using  a frequency doubler and buffer amplifier on the output of my Rapco so as not to degrade the units performance (I need a 10MHz reference frequency). I bought my Rapco out of curiosity as I already use a couple of Trimble  Thunderbolts, which, if I am honest, I prefer. I am also using a Quartzlock 2A 'Off-Air' radio standard (Radio 4) that is great for quick set up when working away from the lab. They seem quite rare these days but are a useful bit of kit to own.


As a side note..... personal experience has stopped me trying to desolder the cans on units such as yours.....too much heat and you end up with a mess inside including unsoldered components. I now use a very fine modelling saw to trepan the upper secion of the case from the lower. Re-securing afterwards may be done with conductive adhesive copper tape or small solder bridges around the cut line. I suppose a Dremel disc cutter would also work but is a little less controllable.

Very Best of luck with the HCD OCXO

Aurora.

[DRT]

I too bought one of the Rapco units from eBay some time ago. Can't help with you OXCO problem, but I thought I would mention that I found the unit to work well apart from the very deaf GPS receiver. I had to add quite a bit of amplification to the GPS signal to get it to work.

I also compared the 5MHz output with a Quartzlock 2A. I found the Quartzlock to have quite a bit of short term jitter (comparing phase on a scope) . Longer term I'm sure they would be up to spec.

Just as an exercise I built a simple UI for my Rapco, based on an Arduino. It decodes the serial output to display GPS signal strengths and time on a 20x2 LCD. Also, I upload some of the stats to a web page - see here: https://cosm.com/feeds/43878

David