Eagle PCB for 100MHz OCXO PLL

First draft of the design for the 100MHz OCXO PLL to lock the Valpey-Fisher 100MHz reference oscillator to my 10MHz Rubidium atomic clock reference.  First time I’ve used Eagle in anger.

The completed board is 43 x 63mm.  It is not at all optimised, I could probably make it half the size with practice, but this was a learning exercise.

I’ve submitted the .brd file to http://www.eurocircuits.com/ and five proto boards on 7 day delivery with no masks, legends or gold finish work out at just over 8 euro each plus VAT and 2 euro postage.

If this works out OK, I’ll be using this outfit again, the website made it amazingly simple.  Just hope the boards arrive and are up to the quality of the ones in their Youtube video


Thoughts on my new frequency standard

I have an Efratom 10MHz Rubidium standard I found on ebay, it runs 24/7 on a Meanwell switchmode PSU, which feeds a pair of 12V 4AH gel-cells and a low-drop-out 24V regulator. The Meanwell is set to 28.4V to keep the gel-cells in top condition. The output of the Efratom is fed to 8 port and 4 port distribution amplifiers from http://huprf.com.

In the same rackmount chassis, I have a 10MHz Morion MV89A double-ovened OCXO, which I can switch to if the Rubidium source ever dies. That is encased in two inches of polystyrene foam, along with a 22-turn trimpot and series resistors, which are all glued to the case of the Morion to keep them at a constant temp.

I also have a G3RUH 10MHz GPSDO, which uses an external GPS powered antenna up on the roof.

I have an AD8302 monitoring the output levels and phase difference between the GPSDO output and the Efratom, and a second AD8302 looking at the difference between the GPSDO and the Morion. Those drive simple analogue meters at present.

I feed the sinewave 10MHz from the Efratom/DA to both of my Elad FDM-DUOs through SMA 6dB attenuators, and to both spectrum analysers, the 20GHz Racal frequency counter, my 3GHz HP signal genny, the 2m, 23cm and 13cm transverters and to a set of ADF4531 synth LOs for the pair of 144MHz G4DDK Anglian transverters I use for SHF IFs into the DUOs. The 4351 LOs run at 2556MHz for 3cm and at the fundamental for 9cm, and I have others for 70cm and 4m.

Also in the frequency standard chassis is a Valpey-Fisher 100MHz OCXO. I am just working on a PLL board to keep that locked to the 10MHz Rb source, using an ECL /20 prescaler, and a single-gate SOT23 XOR chip. The 10MHz Rb ref feeds a comparator and a /2 divider into the other input of the XOR. The output is filtered and has a time-constant of around 5 seconds. I will need to build another 4-port distribution amp for that, it will be used to lock some SHF local oscillators based on a set of YIGs for 3GHz up to 26.5GHz fundamental, to give me stable low-noise sources for 24GHz, 47GHz, 76GHz and higher, but that’s another story.

The key elements for locking the DUO are just the Efratom and the DA. I have been measuring the stability of this free-running MV89A against the long-term ref of the G3RUH GPSDO and it is stunningly good.

I used the MV89A to generate a test signal from an ADF4351 pll chip on 2,300,000,000.0Hz, after the Morion had been running for a week. Receiving that on the DUO and the locked transverters, after a careful check of the Efratom calibration, there was less than 3Hz difference. That is about 0.02Hz off on 20m, which I think is good enough, especially as the DUO only does 1Hz steps….

For the fine calibration of the Efraton and MV89As, I am looking at a beat frequency of less than 1mHz, or a phase shift of less than 18 degrees per minute, so the 3802s and a high-stability voltmeter are needed, or a dual-beam scope and a lot of patience.

Once I have the 100MHz OCXO locking circuit finished, I am going to make a new GPSDO based on a UBlox LEA-M8F with Galileo support and proper direct integration with an MV89A. I’ll also run the AD8302 boards into a PIC with A/D and Flash to monitor the long-term stability and variance, with USB monitoring. Next will be another full frequency standard for my motorhome, with a Ublox, MV89A and this PTI 100MHz OCXO

If I wasn’t so totally obsessive about frequency precision, I would just use the MV89A in a block of foam and set that to zero beat with a decent standard frequency transmission. It is never more than about 12Hz out at 10GHz, so really the Efratom is a bit redundant. I guess I just liked the idea of having that Rubidium thing running here in my lab racks.

160m receive loop experiments

Back in the late 1990s, I was a fanatical Topbander.  Here is a piece I wrote about my experiences with receive loops.

On 26/12/2016 21:07, alan@g3xaq.net [rsgbtech] wrote:

Please can you explain in more detail about your paired loops? On first reading it sounds more like you had two loops in broadside than in endfire as you explicitly note, so I am clearly not getting it. Can you upload a diagram perhaps?

Alan, the paired loops were in the same plane.   The outcome I was looking for was to be able to put a null on the loud EU stations arriving at high-ish angles and it worked fairly well.  I used a 7 metre 2 inch fibreglass pole at the centre, turned by the rotator on a short stand made from welded angle iron and scaffpole.  I used four fishing poles fixed at right angles to the mast, fixed to two support plates.  The wires were just taped in place.  The control wires and feeds were taped to thin bamboo canes with fishing line supports.

The null is off the direction opposite to the feedpoint.  I can’t remember the technical detail , but I think it worked as two pairs of close-spaced phased verticals.  I tried a number of tricks including relays to switch in different lengths of coax between the two feedpoints and the combiner, but I didn’t try any reactive elements.  The amps use two MMICs with a tuned bandpass filter between them.

It was certainly a lot better than an EWE, but I didn’t have  switched K9AYs with which to compare it.  It probably helped me to work 20 or so of the 250+ DXCCs I worked on 160 during the trough between cycles 22 and 23. The antenna was wrecked in a storm in the early 2000s and I didn’t bother rebuilding it, it was a bit of a beast.  Best bit was being able null out some of the endless callers and policemen who were co-channel with the DX.

It was a fun experiment, but it would need much better engineering to be practical.  Keeping noise out of the preamps on single loops was a major challenge, at one point I tried a balanced-input amplifier, but I found that if I combined the two loops, although the signal was reduced by 10-15dB, the noise dropped rather more.  I tried using 1k pots in series with 470 ohm carbon resistors for the loads, but then I remembered reading something about Vactrols being used.  The ones I used were not ideal, I had to use resistors in addition to get the roughly 800 ohms that gave best results.  I had a controller with three switches to connect in different coax lengths, and two pots to adjust the resistance of the Vactrols for best nulls/noise in conjunction with the rotator.  Happy days before I was infected with microwaves.

Bit of Top Band nostalgia

On 27/12/2016 13:09, alan@g3xaq.net [rsgbtech] asked about my experiences with phased receive loops.  I rather over-answered his query with a raft of nostalgic reminiscence about when I was a proper Topbander.

With an output around -35dBi I’m wondering if one could get away without outdoor preamps? I wouldn’t have thought you really need the buffering of the preamps if a decent hybrid combiner was used. Or would that just exchange the noise pickup in the preamp that you suffered for noise pickup on the less-than-perfect coax back to the shack?

I certainly found that in-shack and near-house noise sources were a problem, so adding a fair slab of gain helped to raise the wanted signal over the local noise floor.  With the loops alone, I couldn’t detect any increase of band noise compared with a screened dummy load at the far end (or near end) of the coax.

I also tried a pair of loops a quarter-wave apart to see if I could steer the nulls. Results were a bit inconclusive, and by then the sunspot numbers were on the way up and I rather lost the fanatical urge for 160.

One thing for certain is that you must detune the transmitting antenna or you get a lot of induced noise and pattern distortion from that.

My tx antenna was a 70ft tiltover pole, pivoting on top of a vertical 20ft piece of 4 x 6 inch steel box section using a 1 inch steel pin for the bearing.  The base section of the mast pole is made from schedule-80 steel water pipe with 2 inch ID, then steel scaff pole, then thickwall and then thinwall aluminium tube.  On top of the thinwall section is a 30mm fibreglass tube about 2ft long.

The base of the pole is in a 5ft hole with a concrete collar top and bottom, and brick fill between.  The pivot pole is earthed at the base to 6600ft of buried radials and a 20ft square mesh mat, plus about 30 long ground stakes and was also bonded to every sheet of the 30x30ft corrugated iron roof of a low outbuilding.  In the configuration I was using for 160, I ran a 2.5mm copper wire from the top of the ally mast out about 70ft, to one end of a 2ft spreader, which was tensioned by a braided Dacron lanyard round a pulley fixed to a tree at about 60ft, with bungees to keep a steady tension.  The wire then went along the spreader and from there back to the top of the 2ft fibreglass pole.  From there the wire went to another piece of fibreglass tube at right-angles to the mast, then down to about 2ft above ground, where it was fixed with a barrel tensioner to a section of alkathene water pipe bolted fixed to a steel anchor plate set into the concrete. I connected the tilting section to the main pole with a section of 1 inch wide braided copper strap.

I fed it with a huge and heavy 4 inch diameter ferrite ring autotransformer to match the approx 110 ohm impedance of the folded hairpin to 50 ohms.  To detune it I used a vacuum relay to switch in a motorised 500pF vacuum capacitor (left over from an earlier version where I just tuned the thing as a simple inv-L using an omega match to a cage of wires from the 45-ft level to a ring just about ground)

To balance the tension of the horizontal section, I ran a long lanyard to haul up a 160m dipole fixed to a 35ft pole back on the house.  I fed DC up the dipole feeder to a vacuum relay to switch in a 330pF mica cap in one leg to detune the dipole on receive.  I used a simple 2 inch diameter ferrite transformer at the feedpoint to match the approx 20 ohm feedpoint impedance to the coax, with a 10k carbon resistor from each leg to the coax outer to bleed off any static.  That dipole was WAY better for ZL and VK6 and 9M2 greyline, also for those rare QSOs with KL7 from inside the auroral oval.  The vertical was best for west coast US, JA and just about everything else.

I welded up a conduit made of thirty or so scrap 6ft sections of 1/4 inch wall 4 inch steel tube and ran the feeders and control cables through it, and buried it about a foot deep.  That is also bonded to the earth system.

I dropped the big pole in summer 2000 with the idea of doing a bit of maintenance, but never got round to it.  Now lot of trees have grown around it.  It is currently stuck at about 30ft agl, so without major tree climbing and chainsaw work, it can’t go up or down.  Also I replaced the corrugated iron roof with a painted version last year and I couldn’t raise enough enthusiasm to sand a corner of every sheet, drill it and do all the stainless bolts and crimp tags all over again.

I had a huge amount of fun messing about with antenna systems on 160.  Looking back, I must have had an awful lot of spare energy then.  I sound more and more like an OT with a bad case of nostalgia.  To cure that malady, I have gone right to the bottom of the learning curve to have a proper bash at all this Four Metres and Down nonsense.

Simple tracking generator for HP8562A Spectrum Analyser

I’ve been trying out a simple tracking generator for 0-2700MHz to use with my HP8562A spectrum analyser.  The plan is to use a 3910.7MHz signal from an ADF4351 locked to the same 10MHz ref as the 8562B, then mix it using a Minicircuits SIM-83+ to get the required 0-2700MHz baseband signal at about -10dBm.

This is the measured response of my 23cm LPF:

There are a few issues.  Main one is that there is a fair bit of leakage of the signal as I don’t have it all boxed up, but it is at about -50dB, so not bad.  There is also a bit of ripple every 180MHz or so, not sure what that is, could be the result of using only a 10kHz bandwidth, I need to get everything boxed up and screened and retested.

I took a spectrum of the fairly-flat output with no DUT using trace B, then took a trace A and did a subtraction.  It was pretty much completely flat at 0dBm. Next, I inserted the filter and that resulted in the trace above.

Apart from the ripple, the curve of the LPF looks a reasonable facsimile of the manual curve I took previously http://www.g4dbn.uk/?p=323. OK, the scales are both different, but the bottom one took me 20 minutes, this is doing 20 scans per second…


I am fairly sure that it needs a decent 3-7GHz isolator (or isolation amplifier) and on the IF output from the 8562A to prevent the fundamental signal getting back in that way, and maybe a 3GHz HPF as well, but the principle looks good for baseband.

No time to do the testing at the next band up, other projects on the bench to finish first.


CppSim modelling of 100MHz OCXO PLL

Following intricate discussions on the RSGB Tech Yahoo group, I have been working on the detailed design for a phase-locked loop to lock a Valpey-Fisher VFT-22H 100MHz OCXO to my Efratom Rubidium reference clock running at 10MHz.  While I am waiting for bits and boards to be delivered, I decided to take the opportunity to learn how to use some circuit modelling software.

This model was built using the Sue2 schematic capture element of the CppSim package from Michael Perrott.

The schematic uses standard building blocks which are then compiled as C++ objects to maximise the speed of the models.  So I could try out CppSim, I have created a model of the XOR-based version of the PLL.

Step 1: Schematic capture using Sue2

Next step is to compile and run the simulation in CppSim after selecting the time steps and which elements should be probed in a test.par parameter file.  In this case I am interested in the ref, vin, out and div signals.

Now to run the simulation.  This is a huge computation with 5 billion steps at 1ns intervals, but that is too much for the graphical display tools, so I have to select shorter time periods of a few tens of million steps.  The simulation took 30 mins on my desktop PC.

Next step is to display the captured signals in the probe list which are saved in the .FST file by CppSim, using GTKWave.  The initial phase offset between the ref and div signals was about 20 degrees.  After 5 seconds, the phase has pulled to 90 degrees and the vin has settled to a sawtooth with a mean value of -0.19 mV and a peak to peak value of 4.5 microvolts. Top trace shows the 100MHz out, next is the 10MHz ref, then the divider output and finally the vin at the vctrl pin of the OCXO. Signals are referred to 0V, in the real system the ref will be 2.5V.

This is using a simple RC filter with a time constant of 1/30 s.

At the beginning, the vin starts at 0.0 V and pulls negative to try to get to 90 degree phase difference:

Running a sample once per microsecond for 5s shows an overdamped response as expected from a simple RC filter:

Initially, the voltage drops to -0.5 but pulls in so that at 5s, the offset is only 0.2mV, better than 1mHz, which is better than the precision of the reference.  In reality there will be an offset caused by the non-infinite input resistance of the OCXO.

Next step will be to model the response using a lead-lag filter and a resistive input with a small DC offset error.

Verdict on CppSim software (which is freeware) is very positive so far.


Fluke 8845A

At last I have a serious multimeter. OK, I do have a venerable Avo 8 in a leather suitcase, but this thing is lab-quality and comes with a valid calibration certificate from Fluke Park Labs in Everett, WA.