Well, it's been three weeks since that radio ham rally event in Blackpool that I mentioned in my last marathon posting about the voltage tests on my PS board modifications. I'm afraid to report that it failed to live up to my hopes regarding my desire to accumulate OCXOs and the other bits 'n' bobs I was hoping to acquire in face to face deals with the various traders who attend such events.
It wasn't a complete wash out since I did manage to get my hands on a 13MHz CQE branded OCXO for just 4 quid that I'd spotted amongst a rather random collection of stuff laid out on one of the trader's tables. It wasn't the desired 10MHz but, for a mere 4 quid, I was confident I'd be able to use it as a 13MHz sourced 10MHz clock supply (it
was an OCXO - the
only OCXO I'd managed to spot for sale- and
was only 4 quid! ).
Typically, whenever I enquired about OCXOs (and other items) that any trader looked like they might normally hold stock of, yet unconscionably failed to have on display, the answer was often, "Yes, I have them back at the shop but I didn't bother to bring any due to limited space in the van." or similar such lame excuse.
I do appreciate the limited carrying capacity of their transport (and the strain on their backs in carting such stock from the van to the table(s) and then back again) but, for an event like this, especially these days, I'd have thought such items as high precision frequency components like OCXOs and RRIO opamps and PLL and divider chips and so on would have been high priority stock items on such outings.
I can only presume that this is a sad reflection of the state of the hobby now being one of cheque book solutions to the advancement in building up the contents of a modern day radio shack rather than via technical expertise in home brewing or DIY modification of existing commercial amateur radio kit as was once the case several decades past.
Anyway, all that sadness aside, I did manage to grab hold of a pitifully few other bits 'n' bobs before having to admit defeat and call it a day. Needless to say, I failed to find even a suitable PSU enclosure to re-home the FY6600's PS board into with most of my purchases being more along the lines of speculative "might prove useful" items such as a couple of 10MHz crystals which did prove surprisingly useful in creating a crude filter to clean up the PPS 10MHz output from my u-blox M8N based GPS module I'd bought a few months earlier.
That 13MHz OCXO that I'd managed to snag so cheaply didn't have any information regarding its pin out or any hint of whether it was a 5 or 12 volt part (the trader professed a total ignorance as to any such information - probably why he was happy to accept my offer of just 4 quid for its purchase). The best I could do in my on line searches for any sort of a data sheet was a single page image of a 13MHz 12 volt Vectron clone. I got the pin out details ok but I had my doubts as to whether my CQE version was, likewise, a 12 volt part. Discretion proved the best part of valour in my subsequent voltage probing tests from which I was eventually able to conclude that, rather unusually, it was a 5 volt rather than the more common 12 volt part.
I lashed up a breadboarded circuit to generate a 10MHz signal via a doubler to feed a divide by 13 with 26 MHz which then fed the resulting 2MHz into a second clock multiplier chip to generate the required 10MHz (the 3N502's minimum input clock frequency being 2MHz, hence the multiply by 2, divide by 13, with a multiply by 5 sequence to convert 13MHz into a perfect 10MHz square wave signal).
Since this was all powered from a single 5 volt rail, I was tempted to use this as a replacement to my original TCXO upgrade but although the peak current demand by the OCXO alone of 280mA (reducing to around the 170mA mark once up to temperature) looked likely to be just within the limits of the FY6600's 5v PSU rail, I rather thought the additional 70 to 90mA from the multiplier, divider, multiplier chips would break this particular camel's back.
Further searches for a more specific data sheet for my 5 volt 13MHz CQE OCXO eventually disclosed a cheap source of 10MHz CQE clones at just £4.99 each from the following supplier:
<https://www.ebay.co.uk/itm/CQE-CRYSTAL-OSCILLATOR-10MHz-REDUCED-TO-CLEAR/253081992039?hash=item3aecdcbb67:g:x1gAAOSw~qNZh2rl>
<https://tinyurl.com/y6pl2xvk>
I initially bought three of them just over a week ago before committing to another four to provide myself with a "Lifetime's Supply" at a bargain basement price. I hadn't wanted to disperse the news of my bargain of the century find for fear of creating a stock depleting demand on the supply. Having now secured my supply, I'm only too happy to now spread the good news as a "Public Service Announcement" to all and sundry here.
Be aware though, the stock is slightly mis-described as "
unused NOS" when in fact on close inspection (of the last four examples in my case) they're obviously used stock - two of them clearly had remnants of a through plated hole clinging onto a pin or two. All seven have proved to be in excellent working condition and showed no sign of the more usual cosmetics typical of some Chinese recovered offerings which often hint that they'd been dislodged from their PCBs with a hammer before being slung into a skip.
Quite frankly, why the trader in this case felt it worth the risk of damaging his credibility by claiming "Unused NOS" in the description on an item he could easily have sold for twice the asking price even when correctly described as "Used stock", is a mystery to me. Force of habit, maybe? Anyhow, they seem to be well worth taking a chance on as a dirt cheap bargain basement 10MHz 12v OCXO clone of the Vectron units described in that photocopied data sheet page I posted in my u-blox M8N GPS module thread. The only discrepancy being the 4Vpp ac coupled sine wave output versus the 4Vpp squarewave output described in the Vectron data sheet.
In spite of the complication of adding an extra 12v 300mA smpsu board (recovered from a small wallwart), I decided to utilise one of these 10MHz OCXOs to upgrade my FY6600 instead of the original 13MHz 5 volt only unit. The extra power demand by the 10MHz OCXO and its smpsu once at temperature is just a mere 1.3W which additional thermal loading is no longer an issue once the signal generator has been blessed with a small cooling fan as it has been for the past 5 months with my own generator (for anyone contemplating modifications to their FY6600/6800s, a cooling fan must take the highest priority - I used a 50mm square by 10mm deep 12v fan powered off the 5v rail in my case, very quiet yet so very effective at dispersing the heat build up).
Going the "12v OCXO plus additional 12v smpsu add-on" route proves a very effective way to apply such an upgrade. Unlike Arthur Dent, I chose to connect the little 12v 300mA smpsu board directly to the mains inlet socket connections so I could keep the OCXO powered up whilst plugged into a mains supply even when the whole generator is switched off by its back panel on/off switch rather than burn the additional 5 watts of standby power it consumes when placed into standby mode on the front panel button.
Also, unlike Arthur's use of a change over switch arrangement with the external 10MHz reference input socket he'd fitted to his unit, I plan on using the "injection locking" technique to synchronise the internal 10MHz OCXO to an external GPSDO 10MHz reference sans the complication of a change-over switch and the risk of disruptive glitches which could confuse the FPGA logic into a locked up state.
The only problem with this 'neat idea'(tm) despite my on line researches, being that I need to experimentally figure out an injection locking circuit to achieve this goal. It's a well document technique (often discussed in the context of avoiding unwanted injection locking of XOs of any type to external interference from close in frequency clock sources - so, "How hard can it be?").
Unfortunately, I haven't been able to find any example circuits with real world OCXOs so I'm on my own with regard to solving this particular mystery. I started a new thread a couple of days ago entitled:- " Injection locking the 10Mhz OCXO to external reference (upgrading a FY6600)" but so far, haven't had any 'takers' despite the attached images having been viewed 14 and 13 times so far. Don't worry, I do plan on eventually presenting a report on this new tangential approach to adding an external 10MHz reference to the FY6600/6800 AWGs we all seem to have a love/hate relationship with. I don't know how long it will take to develop a working solution so don't wait with bated breath on this.
In the meantime (in the tradition of my recent posts into other threads), here are a couple of attached photos for those who may not have seen them already. The first shows my collection of OCXOs, sans the one fitted into the FY6600, and the second shows the effect of my crude two crystal and 4K7R 'T' 10MHz filtering of the GPS's PPS signal on the 'scope traces when comparing the 10MHz output from the signal generator.
There's, hopefully, just enough resolution to reveal sufficient detail in what's going on. The picture is a fair representation of the reduction of the non-integer division induced jitter of a 48MHz clock down to a 10MHz sine wave although it can't show its mitigation of the sawtooth jitter to something less jittery and more wavery in character which makes attempts at calibrating a 10MHz OCXO to within 100ppt a little less fraught.
Unfortunately, such filtering is no substitute for a disciplined OCXO since the filtered raw PPS signal still contains the +/- 10 to 12ns variations reflected in the Deviation Map window in the u-centre app which typically shows excursions of 3 to 4 metres from a median position (which itself wanders around). It's this effect which makes the matter of getting the OCXO calibrated to better than 50ppt a matter of pure dumb luck. Clearly, there's no substitute for a GPSDO when it comes to sub ppb precision in calibrating OCXOs. Anyway, enjoy the pictures!
JBG