Demo station for AMSAT-UK colloquium

  • Nice progress. I could have helped you with the 40mm adapter. I did the same as you but in aluminium. However, the recommended POTY lens is one taken from a rocket LNB and they usually come with a 40-20mm adapter as well. Unfortunately, people are reluctant to keep the lens and throw out the LNB, even though it cost under €10.


    Mike

  • Hi Mike.

    Didn't the rocket LNB become 'unobtainium' several months ago. Even the LNB suggested by Heiner seems to be 18 Euros ish on e-bay.co.uk. If the performance is 'poor' in comparison to the LNB in front of a small Sky dish, I'll chase this update after the upconverter has made some progress. Not sure Nylon 60 is totally suitable for 10G.


    David

  • Hi David


    I don't know as I have not looked but I doubt it. A PLL Ventron Red Rocket might be hard to get hold of and the one I ordered turned out to be a DRO anyway, but other DRO ones are still available. E.g the "Venton Rocket Twin LNB EXL-T 0.1dB Full HD 3D Ready" is £11.89 on Amazon this morning.


    Mike

  • Its funny how demand and supply resulted in higher prices for these things. I suppose it is a reflection of the tremendous success of QO-100 and the receiver designs based on these cheap PLL LNBs. There is someone asking 68Euro for one, it might just be a decimal place error but I suspect not. Glad I got my ordering in early.


    Mike

  • The upconverter board is 90% complete and only one minor issue has been found so far, - a missing coupling capacitor.

    The board uses 2 bandpass filters, I thought I would use different types to evaluate performance.

    Filter A in the image below is the popular SF2124E. It has 6 connections: Input, output and 4 ground pads. Importantly, there is no DC connection from input to output, or from input or output to ground. This means you do not need coupling capacitors and I have a direct connection from the filter to the input of the ERA2 that follows it.

    The second filter (B) is a TDK DEA252450BT. It has 4 connections including grounds that take up the entire length along the sides of the device, making it easy to fit to the board with a regular soldering iron. Unfortunately this device has a DC connection between all of the pins. This property means I need to add a coupling capacitor between the output of the ERA2 and the filter input. I think I have some 100pF 0402 caps somewhere. There is just about enough room to squeeze one in if I remove about 1mm of copper track.


    Next task. Programme the PIC12F629 and attach the Local Oscillator module. Then some measurements and fitting the the final ERA2.


    73 David


  • More steps forward and some steps back. - Here is the 'good'.

    I managed to programme the 12F629 with the code from G7DOE adapted for 1967.500MHz using interger N as suggested by Remco.

    The Chinese pickit-3 put up a small struggle, but I finally got the 'build successful' and 'verify' responses from the free MPLab-X software telling me that the software had been loaded into the processor.
    It took some time before I managed to get the synth board to lock as, for some reason, the data is not transferred successfully when power is applied. What is needed is that after power has been applied the MCLR pin near R11 on my board needs to be momentarily connected to ground. After the voltage comes back up, all is well and the 'lock' LED on the synth board is illuminated.

    Frequency is good (see later) power output is also good at 0dBm. The programmed level was +2 dBm but I've assumed this is measured as the total from both outputs rather than just the one that I'm using. Finally I replaced the drifting 25MHz crystal with a 10MHz TXCO. So what could go wrong?..........


    73 David


     


     


  • And the not so good:
    The local oscillator is operating, but looking at the signal on a FUNcube dongle shows a couple of issues which need to be improved.

    Firstly, the TCXO is manufactured by ECS and is available from Mouser. While it holds the frequency stable over a period of time it has internal correction that uses a D to A converter. This produces frequency corrections as 130Hz jumps in frequency every time it drifts outside a preset limit. Not too bad when warmed up, but not pleasent when starting from cold or if mounted in the same box as a power amplifier. If anyone can suggest a 7 x 5mm 10MHz TCXO that is not 'digital', please let me know. The spectrum of the LO is shown below on an image from SDR#. The signal on the left is the ADF4351 with the TCXO 'jump'. The signal on the right is a reference signal from a signal generator on 1967.5kHz.

    The second, lesser problem, is spectral purity. I have used integer N operation to minimise spurs and hopefully improve noise performance.

    Unfortunately, the carrier appears to have odd bursts of 'phase noise' or interference that extends +/- 90Hz from the carrier. Beyond 90Hz the signal is very clean, so I'm not sure what this can be. I have copied the register settings from the Analog Devices software below. If anyone has a better performing set, please let me know. This is not visible on this SDR sharp image, so it may be an intermittent issue that will go away with proper shielding and supply line decoupling.


    73 David (Yes...I know. I wouldn't get frequency jumps with GPS locking ! )



  • David G0MRF try PFD of 2.5 MHz (or 500 kHz) as PFD and use INT-N.

    You now do not have INT-N (N = 393.5)


    2.5 MHz is GCD of 1967.5 and 10 MHz.


    1967.5 / 2.5 = 787 (which is an integer ; -)


    (you can also see it's not an INT because FRAC = not zero)


    Use R-counter = 4 for PFD 2.5 MHz, or R-counter = 20 for 500 kHz PFD

    (set 'channel spacing' accordingly, or omit the errors ; -)


    (and I am personally a fan of 4/5 prescaler)

  • In the words of Homer Simpson.....Dooh !
    Yes, of course, how can 393.5 be an integer. There are so many settings I just selected INT-N and didn't check the rest.

    OK, I will find the PICKit and try again.

    I've already found a rather ancient ovened 10MHz oscillator module, But at 40mm cubed I just hope it does not draw more power than my PA. (hi)


    Thanks, adjusted for +2dBm and done...... David


  • @PA3FYM Hi, I have compared different register settings to generate a 1968 MHz LO signal for a mixer. One with FRAC and R=1 counter versus INT and R=25 counter. Reference is 25MHz in both cases.

    The FRAC spectrum has heavy signals every 25MHz from the carrier, while the INT spectrum looks much cleaner wideband.
    However close to the carrier (1...2 KHz) the FRAC gives about 6-8 dB lower noise compared to the INT.
    I think for narrowband operation (SSB) the FRAC is the better choice, because the 25 MHz spurs will be filtered out after the mixer anyway.

    What is your opinion and why do you suggest to work with INT rather than FRAC?

  • DH1DA This ADF4351 is actually a complex and versatile device. For generating signals there are several routes to Rome.


    From what I learned from the 'black science' of PLL's and (especially!) loop filters, is that a 'normal' INT PLL with lowest N as possible (and therefore highest PFD possible) delivers the cleanest signal. As a rule of thumb phase noise increases with N^2 (N-squared, or 20logN).


    The FRAC mode isn't bad, but depending on the loop filter on these board. So .. for this given loop filter it needs a little experimenting with different settings (PFD and BW) to find the most optimal settings or you have to replace the loop filter components on the ADF PCB, which is a tedious job.


    I built several applications with these PLL's (also passive, i.e. external oscillator and 'seperate' ADF PLL (like ADF4001, 4118, 4113, 4153 etc) and found that for every case I needed to fiddle and burnt hours struggling with ADISimPLL ;; -)

  • Any interest in this? Should I make it?



    So perhaps I should say what it is for. It is a board to take a 25 MHz TCXO or VCTCXO reference, filter and amplify the reference and triplex it with DC and the IF of the LNB which it is designed to lock to said reference. The IF from the LNB is sent to the SMA connector.


    The LNB itself has to be a PLL type and modified to use the 25MHz reference rather than it's own crystal.


    I made one of these last year but I mistakenly thought using Connor Windfield D75F was a good idea. That device was useless due to the poor frequency stability of those TCXOs, possibly a bad batch but my trust was lost. However there are many other TCXOs so the board now has footprints for two common types. The prototype worked OK but I have added an extra buffer who's gain can be selected as appropriate depending on the TCXOs output level.


    Mike

  • Water bottles these days have a new meaning, but this 'old school' version provided an easy method of assessing the performance of my "high stability 10MHz source" bought at a radio rally a couple of years ago.

    For a 25 to 60 degree temeprature change this Murata oscillator drifted 6 kHz at 1967.5 MHz. Probably no more than a big crystal and a TTL inverter housed in a large metal can. Buyer beware !!

    Oh well, the ECS TCXO may have digital jumps, but it does compensate very well and stays within 200Hz .......Another couple of bits in its A to D converter and it would be much better.


    So with 10 days to go before setting up the demo station, it's time to get these parts into a box and wired up to the Spectrian PA.

    Perhaps a test by this weekend?