|
Vintage Amateur and Military Radio Amateur/military receivers and transmitters, morse, and any other related vintage comms equipment. |
|
Thread Tools |
24th May 2020, 1:22 pm | #21 |
Nonode
Join Date: Oct 2010
Location: Tintinara, South Australia, Australia
Posts: 2,339
|
Re: LO for genuine 10m FM10D Pye Cambridge
You need to load the output with 50 Ohm if you have selected the 8mA drive option.
Use an attenuator on the output if you need a lower level rather than dropping the drive level. I forget what the output impedance's are for the other drive currents and they are not listed the datasheet. And be aware the datasheets contain errors as well. Last edited by Terry_VK5TM; 24th May 2020 at 1:29 pm. |
24th May 2020, 2:21 pm | #22 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: LO for genuine 10m FM10D Pye Cambridge
Thanks again. The output of my PCB is not dc blocked and so far I've not experimented with different drive currents as I didn't know this was possible. I could try changing the drive current whilst adding a dc block and a termination to see if that helps.
On a scope it currently shows something like a 3V pkpk squarewave for signals in the HF band. I really am still in short trousers with this part and there's a lot for me to learn I think.
__________________
Regards, Jeremy G0HZU |
24th May 2020, 4:29 pm | #23 |
Nonode
Join Date: Nov 2015
Location: Welwyn Garden City, Hertfordshire, UK.
Posts: 2,015
|
Re: LO for genuine 10m FM10D Pye Cambridge
Registers 16,17 and 18 of the Si5351 control output drive strength and On/Off for CLK0,1&2
Sending 0x4F will set them for 8mA + Enabled, 0xCF for 8mA Disabled. (This assumes the source for them is from PLLA and Multisynth 0,1,2 respectively - See pages 19,20 & 21 of AN619.pdf) HTH Last edited by Jon_G4MDC; 24th May 2020 at 4:35 pm. |
25th May 2020, 3:38 pm | #24 |
Tetrode
Join Date: Mar 2018
Location: Nottingham, Nottinghamshire, UK.
Posts: 88
|
Re: LO for genuine 10m FM10D Pye Cambridge
I once built an Si5351 based local oscillator for a Pye F27AM using a PIC18F4520, a rotary encoder and two line LDC. If anyone wants to use it as a starting point I'd be happy to supply the code. It's written in C and works reasonably well. I also started a six channel conversion for a Cambridge but it got shelved due to other projects. I may well pick it up again soon.
|
25th May 2020, 11:20 pm | #25 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,556
|
Re: LO for genuine 10m FM10D Pye Cambridge
Ian, I think the problem is that Jon and Jeremy are both dismayed by how exotic the output spectrum from these modules is. I await their findings with interest.
|
26th May 2020, 12:39 am | #26 |
Nonode
Join Date: Oct 2010
Location: Tintinara, South Australia, Australia
Posts: 2,339
|
Re: LO for genuine 10m FM10D Pye Cambridge
Being a square wave output, without additional filtering, exotic is about all you can expect out of these chips.
|
26th May 2020, 9:02 am | #27 |
Nonode
Join Date: Nov 2015
Location: Welwyn Garden City, Hertfordshire, UK.
Posts: 2,015
|
Re: LO for genuine 10m FM10D Pye Cambridge
It it were only harmonics we could deal with them!
I'm still wondering about a clean up PLL actually. I wonder how much mileage there is in that...? We like the 5351 for it's programmablity down to Hz resolution but we don't like the grass that comes with it. I could live with the RX LO with a bit of bandpass filtering at 35MHz - any spurs will only affect me anyway as RX spurious responses. On TX it needs 2.5MHz output but it only needs to cover 50kHz or so. A 4046 or such could handle a PLL for that. Last edited by Jon_G4MDC; 26th May 2020 at 9:08 am. |
26th May 2020, 9:47 am | #28 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,556
|
Re: LO for genuine 10m FM10D Pye Cambridge
In one of the Pye / Philips selcall / trunking control PCBs there is an interesting scheme for generating a crude sine wave output from a digital clock input.
The ten outputs of a 4017 are connected to a network of resistors, so that on each successive step the output from the resistor network is the next step in a crude ten-step sinewave. This is followed by an analogue filter stage which smooths it into an actual sinewave. To generate a sinewave of frequency 'f', you feed the 4017 with a squarewave of 10 X f. If there is some variant of the 4017 which runs comfortably at 25MHz you could use the same principle to generate a 2.5MHz sinewave by feeding it with 25Mhz. If you can find a device which produces more output steps (16, as per 74154 or equivalent?) then you could produce a more finely graded initial sinewave but of course then you'd have to feed it with (f X 16). |
26th May 2020, 10:00 am | #29 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: LO for genuine 10m FM10D Pye Cambridge
Quote:
__________________
Regards, Jeremy G0HZU |
|
26th May 2020, 10:39 am | #30 |
Nonode
Join Date: Nov 2015
Location: Welwyn Garden City, Hertfordshire, UK.
Posts: 2,015
|
Re: LO for genuine 10m FM10D Pye Cambridge
Thanks for the idea about 4017. Pye also did a similar thing with a 4017 and capacitors to make a commutating notch filter.
The 5351 already includes an output divider which is capable of 2,4,8,16...to 128 division. My first attempt at the 2.5MHz problem was to first generate 1/3 of the 10m final frequency with an output divider of 4 for Fc/12 output overall. This looked like a field of grass. So far I had no time try another scheme. If I can find the time I will try to get a screenshot of it today. Have to make a hole in the junk on the shed bench first. Glad that the doc was useful & enjoy the floor and ceiling functions in there. (I had forgotten all about them) I wonder what is thought of the clean up PLL idea - not necessarily with a 4046 but just generally as an approach? |
26th May 2020, 12:19 pm | #31 |
Nonode
Join Date: Oct 2010
Location: Tintinara, South Australia, Australia
Posts: 2,339
|
Re: LO for genuine 10m FM10D Pye Cambridge
Another option to use two Si5351's at reasonably high frequencies compared to what you want as the product and feeding them into a DBM (SBL1 or similar).
Make one fixed frequency and the other variable to shift channels. That should take out a lot of the hash and give you a sinewave out. |
29th May 2020, 6:32 pm | #32 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,556
|
Re: LO for genuine 10m FM10D Pye Cambridge
Quick heads up, 10m is very lively at the moment (early evening, Friday 29th May).
|
30th May 2020, 1:51 pm | #33 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: LO for genuine 10m FM10D Pye Cambridge
Sadly, I haven't had a chance to look at the Si5351 board again and I really need to buy another one to make comparisons.
Quote:
It might work best with the existing modulator and a cleanup PLL with a wide loop bandwidth? I'm really just guessing.
__________________
Regards, Jeremy G0HZU |
|
30th May 2020, 2:49 pm | #34 |
Nonode
Join Date: Nov 2015
Location: Welwyn Garden City, Hertfordshire, UK.
Posts: 2,015
|
Re: LO for genuine 10m FM10D Pye Cambridge
Thanks for the input. Fortunately no need for direct FM. The reason 2.5MHz is necessary in the first place is a phase modulator in the original line-up. I would like to keep that original.
Of course there is nothing to prevent the clean up PLL being run at some higher frequency and divided down to 2.5MHz. Just so long as all the grass gets cleaned up. I still haven't got the lashup board on the spectrum analyser either!! I did clear up a bit of bench but it stalled after that. |
31st May 2020, 6:50 pm | #35 |
Nonode
Join Date: Nov 2015
Location: Welwyn Garden City, Hertfordshire, UK.
Posts: 2,015
|
Re: LO for genuine 10m FM10D Pye Cambridge
Today the lash up board met the spectrum analyser. Actually the results were much better than expected and I was going to post some plots which I thought showed a minor victory. Spur levels were much better than measured before.
Then more messing around started to debunk those plots until I'm so confused I don't know what is right and wrong. It all concerns the load presented to the output of the Si5351. If I connect direct to my spectrum analyser it is DC blocked by the capacitor in the input cable of 100nF. If I introduce a resistive 10dB pad directly at the Si5351 output (without DC block) the quite nice spur levels degrade considerably. I have yet to try DC block > 10dB Pad >SA. I hope that will produce consistent results with Si5351directly connected to SA without the pad. It doesn't seem to be SA mixer related since SA input attenuation setting has no effect. We program these for an output current strength so to DC block them or not? |
31st May 2020, 7:31 pm | #36 |
Nonode
Join Date: Nov 2015
Location: Welwyn Garden City, Hertfordshire, UK.
Posts: 2,015
|
Re: LO for genuine 10m FM10D Pye Cambridge
Added the needed DC block. Now the spur level remains the same with and without an in line attenuator. If you don't need DC coupling, I suggest avoid it.
Later edit - these observations are related to the 35MHz RX LO so far. Last edited by Jon_G4MDC; 31st May 2020 at 7:49 pm. |
2nd Jun 2020, 10:31 am | #37 |
Nonode
Join Date: Nov 2015
Location: Welwyn Garden City, Hertfordshire, UK.
Posts: 2,015
|
Re: LO for genuine 10m FM10D Pye Cambridge
It's been an interesting couple of days. Adding the DC blocks stopped the whole thing working at all.
The PIC was still running as the tuning pot ADC was being read and the display was updating. No I2C between PIC and Si5351.... It took a while to find the invisible fracture on the SDA line link wire. Now fixed and I'm back in business. I'm happy to say that things don't seem nearly as bad as my first measurements led me to believe. It's cleaner than I thought. It has some 150kHz spurs on both the 35MHz and the 2.5 MHz outputs. These appear to be at a fixed spacing from carrier and unaffected by tuning but more work is required to confirm. The 35MHz output has a collection of -65dBc spurs which tune forwards and backwards at various rates. These might be real or partly due to spurious mixes in the Spectrum Analyser front end. I'm working wideband so far. It would be interesting to look again through a low pass filter. I might knock one up but I will probably have to fix my HP8754 Network Analyser first! If they are real then I think I could put up with them on the RX LO. It remains to be seen what the TX LO spurs will look like after multiplication but at least the field of grass I saw before must have been an environmental effect. It wasn't done in a screened room! Last edited by Jon_G4MDC; 2nd Jun 2020 at 10:53 am. |
2nd Jun 2020, 10:45 am | #38 |
Nonode
Join Date: Nov 2015
Location: Welwyn Garden City, Hertfordshire, UK.
Posts: 2,015
|
Re: LO for genuine 10m FM10D Pye Cambridge
I got to thinking about those 150kHz spurs and why hadn't I noticed them before...?
Rookie error! I was running the board from the Pickit 3 programmer because I had been troubleshooting it. Hey fool! It has a DC/DC in it. Run from the onboard analogue 3V3 regulator instead. Last edited by Jon_G4MDC; 2nd Jun 2020 at 11:05 am. |
2nd Jun 2020, 10:47 pm | #39 |
Triode
Join Date: Jun 2005
Location: nr. Melrose in the Scottish Borders
Posts: 11
|
Re: LO for genuine 10m FM10D Pye Cambridge
Could the AD9833 set for sine wave output be a possible candidate?
Very easy to get up and running in the Arduino environment. Bob GM4CID |
3rd Jun 2020, 3:18 pm | #40 |
Nonode
Join Date: Nov 2015
Location: Welwyn Garden City, Hertfordshire, UK.
Posts: 2,015
|
Re: LO for genuine 10m FM10D Pye Cambridge
Thanks for the suggestion Bob. AD9833 has 12.5MHz max Fout which is too low for the RX LO of 35.2 to 35.7MHz.
Of course it could be multiplied up. |