Sunday 31 January 2021

GPS with PPS output

Note: all files and sketches on M0IFA.me

I have been using a  GPS board with 4 pins VCC-GND-TXD-RXD in a clock project and it works very well using the TinyGSP Arduino library.

QCX+ on top of GPS CLOCK

The sketch is  GPS_CLOCK_TFT. To get out the PPS signal from the GPS module which did not have a PPS pin output, I soldered a tiny wire to the 1K resistor connected to the GPS receiver pin 3.

Wire on PPS output of receiver


On test with the QCX+ inputting the NMEA GPS TX signal and the PPS signal via the "Paddle" input to the QCX+ I found that it did not operate reliably. So I checked, and found the output voltage levels from the GPS module were 0-3.2V. In the QCX+ manual it says the GPS inputs should be 0-5V levels and at least < 1.5V to > 3.5V. So it seems I cannot directly interface the GPS and QCX+...

So I wired up this 3v3 to 5V converter:


And low and behold the QCX interface worked just fine. i was able to easily calibrate both the Reference (menu 8.11) and the System (menu 8.12) oscillators from the GPS PPS input. While calibrating I set the Practice mode (menu 4.7) "ON" to prevent any possible transmissions, I don't know if the is mandatory or if the calibration happens without the GPS triggering the TX.

Here's the inside of the GPS CLOCK with the converter installed

Inside the GPS clock
The GPS clock, Time, Locator and Date


Saturday 16 January 2021

QCX+ SSB DSP decoder

Note: all files and sketches on M0IFA.me

Wow its works, boy does it work!

QCX+ and Teensy DSP SSB decoder

Teensy DSP Hilbert filters for QCX+ SSB reception

One possibility, to have the QCX+ able to receive LSB or USB SSB signals, is to wire out the I & Q signals from the Tayloe detector via a couple of 100nF capacitors... (Teensy audio input impedance is 30k so 100nf is enough for good LF response)

Tayloe Detector, I & Q outputs, wire up from QCX+ PCB

...to an external DSP SSB decoder. For example using a Teensy processor. This DSP can be programmed with a couple of Hilbert filters to give a narrow BW and +/-45deg phase shifts.

DSP processing

The code can be generated using the PRJC Audio library and their on-line designer. The full Teensy code for this application is TEENSY_QCX_100. The Teensy used in v3.2 with its Audio processor board, from PRJC,

Teensy 3.2

Audio board/shield

Simple wiring to the QCX, the Audio board output goes to an audio amp & LS

Teensy DSP SSB decoder, USB or LSB switchable


The circuitry could be built into the QCX case. The audio output can also be programmed to be by the Teensy USB connector if you want to feed it to a WSJT program running on your PC for FT8 reception.

Boxed up

This is it working with the QCX+

Switch down LSB, up USB


Tuesday 12 January 2021

QCX and IQ SSB reception

QCX analysis

SSB? Well it was a bright idea. 

The QCX+ has pins on the mother board for picking off the Tayloe  detector I & Q audio signals. So how about bring these out via a couple of capacitors to a 3.5mm jack socket ? To which I could connect an audio cable to my PC audio input (Mic). Actually my Mac Mini does not have a mike input (!), so I used a USB external soundcard plug in dongle.

Grey wire is GND

Doing this I had the hope that I could run an SDR program like HDSDR and get 96kHz bandwidth display of frequencies around the centre RX tuning point (700Hz lower than the displayed TX frequency remember).

But it was not what I expected, what I got was this


FT8 signals on 40m band.

And the bandwidth is not +/- 48kHz but more like +/-3kHz!! I guess this is due in part to the low frequency response of the op amp stages after the CMOS Tayloe detector switch and the integration capacitors on the switch. Any way it was a big disappointment. Of course it could still be used for SSB reception by tuning the RX across the band to keep the wanted signal in the centre of the display... but it's not SDR.

Bandwidth

So where is this small bandwidth controlled? In two places:

1. By the input impedance, nominally 50R antenna plus the 10R on resistance of the FST3253 switch = 60R total, on for 1/4 of the time, so actually 240R*, and the 470nF shunt capacitors. 

* As pointed out by W6OQ this makes a filter of 4.8kHz, thanks

2. By the feedback across the op amp of 10k and 10nF, which together make a lowpass filter of 1.6kHz!

Putting this in  LTSpice change components as you wish to simulate... we have

Schematic

and a response of

Response,  Green overall, Blue first filter

So there we are, we cannot get a wide bandwidth SDR display from the I & Q signals from the QCX+.  But it should work OK for a station tuned to the centre frequency where the overall BW is around 2.3kHz at -6dB down.

How to improve for SDR?

We cab change the component values, perhaps like this

New component values for wider BW SDR
and get this

Wide bandwidth possibility

And feed the I&Q signals to a PC based SDR program.

Right now I am reluctant to start tearing out 6 caps and replacing them. We will see later.

How about the QCX's own phase shifters

This is quite a clever piece of circuitry, but with limited performance - OK for the narrow band SDR use, and USB only. Essentially it two of these networks, one with fixed R2, C1 R5, C2 values to give a phase shift across 100-5000Hz. like this


Giving a phase shift (dotted line, ignore the solid) like this

Phase = Dotted line

This gives a reference phase shift of -125 to -320 deg. 100-5000Hz. Then a second network like this


Giving a phase shift as close as possible to 90 deg smaller, like this

Phase = Dotted line

If you take the difference between these you get the grey line in the chart below

Grey = phase difference, Green = reference, Blue = 2nd network

The second network is adjusted during the radio alignment to optimise the differential phase shift to as near as 90deg as possible to detect only USB signals.

But to note is that the 90deg shift is not too bad from 100-2000Hz and so would be OK for USB SSB signals. See the SSB reception below.

Conclusion

QCX+ is an excellent design for its purpose, a good Tayloe detector with limited audio bandwidth, a satisfactory phase network for USB reception and an innovative narrow CW tone filter. All good design.

As many have looked into it can receive SSB in a restricted audio bandwidth, picking off the audio from just before the CW filter, but is not a wide band SDR.

Post Script

I just connected the IQ outputs to my MacBook running CubicSDR program. The QCX is tuned to 7050kHz, this is the IQ spectrum across +/- 48 kHz audio

IQ Spectrum, very narrow. SDR at 7,050,000Hz 
On the left my own strong local CW transmission at 7,020,000Hz

And here's the same CW signal at 7,020,000Hz with the QCX+ tuned to  TX 7,020,000Hz, thus receiving at RX centre of 7019300Hz. You can see how much more gain there is at lower audio frequencies.

CW received on USB in the narrow audio passband, 
TX at 7,020,000Hz
RX & SDR at 7,019, 300Hz USB gives 700Hz tone

Actual SSB reception
This is a test transmission on 7.1MHz LSB made by my ELAD FDM-DUO and received by the QCX+ and decoded using CubicSDR program

Testing with a transmission from my ELAD FDM-DUO
7100kHz LSB

As you can see the QCX+ is capable of narrow band SSB reception when signals are decoded from the  I & Q signals on the main board. These were fed to a StarTech Analog Digital convertor running at 96kHz and then to the CubicSDR program running on my MacBook. They were not readable on the QCX+ itself, due to the CW filter bandwidth muffling them of course.



Saturday 9 January 2021

A morse practice generator

 Note: all files and sketches on M0IFA.me

Many moons ago I made, as part of the Banbury Amateur Radio Society BARSICLE project  (making a digital VFO, RF POWER METER AND DC RECEIVER as a practical home brew projects) I designed a Si5351 digital VFO (See 2018 August BARSICLE 6 project). This design has also a real time clock (DS3231) included, and so can be used for many things, a straight VFO up to 150MHz with 3 outputs, WSPR generation, QRSS and other beacon codings... 



It can also be programmed to generate random CW. And that is what I have now done. The Arduino code is called CW_LEARN.

This a a simple program which could be added to for example to change the WPM of the transmission and the spacing between characters. It currently transmits on only 7020kHz.

I am using it locally to be received on my new QCX+ and it works well. It sends 0-9 numbers or A-Z letters but skips punctuation such as "> @ etc" (not even sure if these have morse codes?)

Progress is slow, keep forgetting.

Thursday 7 January 2021

Built a paddle key!

 Looking on Thingiverse 3D printing web site where people post hundreds of STL files for download and 3D printing I found a morse code paddle key. My son has a couple of 3~D printers so I persuaded him to download it and print it.

So here is my new set up, on the left a very low cost Chinese Switch Mode Power Supply (how can you get 5A/12V from such a small device? Incredible these SM PSU! But very very RF noisy!), centre is my just completed QCX+ transceiver (40m version) and to the right my 3D printed paddle. Don't like the colour much but that's the "inks" my son had threaded in his printers...

PSU, QCX, Paddle

Now the real challenge - to learn morse code and learn to use this paddle.

Friday 1 January 2021

QCX+ has arrived, building experience

 I DREAM

I need a new project. So how about... a QCX+ 5W CW transceiver?  And learning CW??? Or just doing WSPR TX to monitor my propagation???


A PACKAGE ARRIVES

So the package arrived, ahead of expected delivery time. 

Package in post
IN THE BOX?

I  opened it up. Wow. Impressive. Excellent quality of components. Beautiful panels. I will drill an extra hole in the back panel eventually to feed out the raw IQ Tayloe audio signals to an SDR program like HDSDR. (which I have running on WIndows 10 under Parallels on my Mac Mini). Connection will be via a USB soundcard plug-in dongle MIC input.

Contents!

What a pile of bits and pieces. I had not expected so many, nor such a complex PCB. This will take some time to assemble, even though I am a quite experienced builder...

GETTING READY TO WORK

My build station, with ELAD FDM-DUE SDR for FT8 contacts 
and Macbook for instructions

And now let's read the manual - twice!

THE DREADED COIL

The coil 39 + 5 + 5 + 5 turns

The coils took me hours. I tried the method in the instructions. But in the end I did it my way. First the 39 turns, marking off with a tally every 5 on a piece of paper. Then 1, 2 & 3 coils of 5 turns. Each set had its start and end wires twisted together. And after I  puzzled out which holes to thread them though. Finally, check, check again and solder. Hold the iron on for the full 10 seconds recommended to burn off the wire coating. Test for continuity, all OK.

Phew, have a long rest. No need to rush it. This will take days...

THE  PCB

Time to have a look round the PCB and mount some intitial pieces, start with the ICs

Coil and some ICs fitted
Onwards an upwards...

16 x "104" CAPACITORS

Scattered all over the board are 16 x "104" decoupling capacitors. These are next to install. 

A note: the capacitors come fix along a cardboard strip, just cut them off, there will be plenty of wire left for mounting, don't try to tear them off the mounting strip this will just bend the leads and leave glue on the leads.

16 x "104" capacitors installed

5 x "474" CAPACITORS

Now I mounted the 5 x "474" capacitors. The group close together at the top are the Tayloe detector filter capacitors to detect the IQ audio signals before the amplifiers.

5 x "474" capacitors

TAKE A BREAK - that's enough for now let's see if I can work some European FT8 non 40m... then carry on... 

Was it a mistake to empty out all the components? Lots of searching for the right ones!!!

Tipped everything in a box of chocolates lid!!!

RESISTORS

Most capacitors done, starting on the resistors. On these Chinese blue background resistors it is quite impossible to read the colour code - need to check every one with a meter. 

PROBLEM

Got to section 3.30, "Install 2 x 36K resistors R32 & R33". There were two "orange-blue- black-red-brown" in the kit but when checked with a meter they were actually 39k! LTSpice re-simulation of the CW filter does not change much with the 39k resistors.

LTSpice with 39k resistors on third filter stage

I have emailed QRP-Labs (Hans) with a query about this. Awaiting a reply. This is where I am at right now

Up to section 3.29 this is my build

Oh dear. I have found the batteries in my test meter were low, I have replaced them and now the 36k resistors read 36k. Have written to Hans to apologise.

Nearly made another resistor measurement goofy when checking the 750k resistor R35, it measured 560k until I realised I was measuring the resistance of my finger!!

WOW, that's all the resistors done. Here it is



Finished the main board resistors

OTHER COMPONENTS

Trimmer, electrolytics, transistors. Neat way the output power transistors are mounted and held down by a screw and washer, Very neat idea.

Output power transistors

Now for those wretched toroids. Two things will help. The first is the web site toroids.info. Here you will find the listed toroid core types T37-2 and T37-6 and for each the dimensions and more impportantly a way to calculate the inductance from the number of turns (or vice versa). And to the right length of the wire needed. This is great as you can cut a length of wire and not have to pull the whole length through each time, but just the shorter length you need. For example for L2 in the Low Pass Filter - 24 turns

T37-6 24 turns needs 35.6 cm of wire, cut 38 or so to be safe

And the second trick is to photo the coil when it is done, then zoom in to count the number of turns is correct...

Yep, its 24

So let's do all the coils 21, 24 and 21 turns

All the coils mounted

Next is the Antenna connector, a BNC type - though I would have referred an SMA to match all my other bits of gear.

FINISHED

Completed mainboard

NOW THE FRONT PANEL

Components mount on both sides of the front panel. First the front side resistors

Five front side resistors

Now the rest, back and front. Three wire have to be used to connect the volume control.

Front
Back

I used red, yellow and black wires for the volume control, thought it would look prettier.

The LCD is installed with mounting screw and nuts. One problem is that the righthand nut on the back cannot be fitted as it collides with the volume control, I temporarily loosened the volume control to get the nut on to have the correct spacing before soldering in the 16pin header. This remains a problem as it cannot be fitted again when the QCX+ is fitted in its enclosure.

Finished!!!

ALIGNMENT  & ENCLOSURE

I have aligned the radio, there are two things to adjust, the input tuned circuit to peak on the 40m CW end of the band and the detector IQ phasing & balance to reject the wrong sideband - CW RX reception is on USB at 700Hz below the displayed TX frequency.

I have also purchased the  excellent enclosure  for the radio, and here it is before I put the top on. I did have a queer problem. When I tightened the nut on the BNC connector it pulled the main board backwards it disconnected the front panel, and the screen went dark - bit of a shock, but all was well when I removed the nut and pushed the two boards back together.

Finished radio
Second problem was crackling when the boxed unit was "tapped". This turned out to be poor solder joint on one of the LPF coils. It IS important to solder then on for the full 10sec to burn off the wire coating!!! 

MORSE

Now the challenge begins, to learn morse code!!! First a paddle morse key? I am building this one.

3D printed paddle morse key