I have held an amateur radio licence since 2007, and this website shows a couple of electronics projects done since then.
I hope this will inspire you to do your own experiments and try out something new or something old. If you have any question about one of the projects mentioned here, please get in touch!
2023
Since first of january amateurs in Switzerland are authorised to use the 4m band. This is a great excuse to build upon the Picardy 2020 and design a new transceiver.
The design uses a main board with vertical daughterboards for XTAL filter, RX frontend, TX frontend and PA. Front elements (display, buttons and connectors) are connected by cables.
The project is ongoing in the picardy repository. See README-4m.md and the kicad-dart-70 folder.
Current state: power, microcontroller, user interface, Si5351 clock generator, RX frontend and TX frontend are assembled. The TX chain didn't have enough gain, reworked it. On the RX chain, the AGC still misbehaves, and has too low gain. But strong signals come through. First PA assembled, aiming for about 2W output. Output filter not yet done, and unclear if spurious emissions are going to be an issue.
2023
Sending APRS information over something else than AX.25 AFSK is possible, and the lora-aprs tracker project is a nice development. The APRS434 project aims to shorten the message lengths to improve successful packet reception probability.
As a small experiment, I used an STM32H7 NUCLEO board, RFM98 LoRa module and u-blox GNSS receiver to make a small proof-of-concept, and a raspberry pi receiver, also using an RFM98 module.
Implementation of the compressed message formats proposed by aprs434, both in the tracker firmware and in the i-gate code, is ongoing. Plus I am designing a tracker with a power amplifier, and now have an I-Gate written in python on raspi running at home: Code repository.
Update september 2023: I've assembled the first PCB for my tracker with power amplifier, and with the current matching circuitry it doesn't have much gain remaining at 433MHz.
December 2022
Last time we activated a castle with HB9HI we had two HF stations a bit too close together, and it was impossible to use both simultaneously. I built two filters that should handle 100W PEP.
References: "Receiver Band-Pass Filters Having Maximum Attenuation in Adjacent Bands" by W3NQN and its implementation by DL2NBU.
I measured about 55W out for 60W input for both. We will see how well they work in the field next time we activate a castle.
January 2023
The Fresco Logic FL2000 integrated circuit used in some VGA dongles supports continous output of samples. This was discovered in 2016 by Steve Markgraf and resulted in osmo-fl2k.
A friend of mine wants to use this device as an exciter for an ampliphase amplifier, a scheme described in the 1930s, and I prepared a tool written in Rust for him.
2022
My son enjoys pushing buttons, so I made him a box with buttons. Contains an astable vibrator, sports different kinds of switches and knobs, and runs from a small LiPo battery.
Summer 2022
Previous experiments in february 2019 receiving QO-100 with a PLL LNB showed me that some form of frequency locking was necessary. I built an ADF4002 based 27MHz PLL around a 10MHz GPSDO I already had, and replaced the internal LNB crystal.
PCB design in the ADF4002_27MHz_PLL repository.
In addition to that, I build a helical feed for the uplink, which was difficult to tune and probably still is off frequency. The station consisted of SDRAngel, LimeSDR mini for both receive and transmit, and a Wifi power amplifier. On 2022-06-18, I did 13 QSOs in total with HB9HI.
References: OE2KHM and DC8PAT for the helix feed.
March 2022
The FT-100D does not have an IF or RF output in order to connect an SDR receiver at the same time. I built and inserted an HF probe circuit presented by DL1DWG, and connected it to the FT-100 base board on J1015 pin 8, and to an SMA connector at the back.
More or less at the same time, I replaced the front panel encoder that started to get really flaky with a PEC11R-4020K-S0024. The detent clicks are missing, but at least it's usable again.
December 2021
A friend of mine is very much into Teletext, and just to see if it was possible we tried to transmit teletext pages over VHF in a totally non-standard way. Teletext pages converted to vbit2 format and streamed in a carousel, which was received by a python script that created a fake MPEG-TS stream (just header and padding to correct length), which was sent to leandvb at 20ksym/s, 800ksamps/s. Leandvb output went into a GNURadio flowgraph driving a USRP B200 at 144.700MHz, into an MRF101AN power amplifier driven to about 20W output.
Decoding it didn't work, probably because of intersymbol interference due to the non-line of sight propagation path between our locations. But it was a fun experiment.
August 2021
The site hosting the HB9G Repeater suffered lightning damage, and I replaced faulty components. Less bad than last time, it still required replacement of many components, both digital and analog. It is somehow a miracle that most circuits are still intact.
In october we put the repeater back in service.
July 2021
Using thick coaxial cable and a soviet vacuum capacitor KP1-4 10kV, 10 to 500pF. With these dimensions, the loop was tunable from 4MHz to about 13MHz, which misses both the 80m and 20m bands. After two FT8 QSOs on 30m and some listening on 40m it was disassembled again.
First I manually tuned it, and then I added a geared DC motor for remote tuning. But it's difficult to reach the right spot, and a stepper motor would probably have been a better choice.
This magloop got revived in march 2022 for a 30m WSPR transmitter that uses the Si5351 eval board.
Spring 2021
I had designed an eval board around the Si5351 before I did the Picardy project, and I reused that eval board to make a CW and FELDHELL transmitter. Of course, the Si5351 needs some filtering and amplification, and I built my first totem-pole amplifier as an experiment. This used MMRF1021NT1 transistors, which are anything but ideal for the application, but adequate enough to learn, and I had a bunch of them.
On the left, the user interface, eval clock board and a 2N3019 driver hidden underneath. The driver amplifier is inspired from SV1AFN's Low-noise broadband HF preamplifier. The totem-pole PA is on the right.
The clock eval board KiCAD project is here, the amplifier here.
The transmitter was used combined with an AirSpy HF+ SDR receiver.
March 2021
Built a PA0RDT mini-whip antenna with 2N3819 and BSV52LT transistors.
December 2020
Assembled the MRF101AN starter kit for 2m and a low pass filter.
This PA was of course tested with the Picardy 2020 transceiver, but the secondary emissions were too high, and the whole setup was very prone to self-oscillation. The relay used for antenna switching didn't have enough isolation, and the receive path inside the Picardy 2020 let too much energy flow back into the transmit path.
The low-pass filter used 22p 200V C315CC220K2G5TA7301 capacitors first, but only for about two seconds before one burst into smoke. The second iteration used two 33pF ATC 100B capacitor in series.
The MRF101AN PA was however later useful for the teletext transmission project.
Second half of 2020
I have always been frustrated by the high current drain of the Yaesu FT-100D when used as an IF rig for the 10GHz transverter. When I stumbled upon F6FEO's Picardy project I set myself the goal of reimplementing it in KiCAD and try to use it as rig for the transverter.
However, I also wanted to use the Si5351 as clock source, and was curious to integrate some ideas from G4DDK's Anglian 3L transverter, and so Picardy 2020 ended up being a bit of a mashup of these existing projects.
Before being used with the 3cm transverter, the Picardy 2020 also got some standalone usage, for which transmit gain and antenna switching was needed. This led to several experiments around MMRF1021NT1 transistors I had received from a friend, and taught me about RF amplification.
The first 2m SSB QSOs were on 7.12.2020, before the rig was put into an enclosure.
In march 2021 I participated in the 3cm contest from JN36DM and made 7 QSOs.
The project files are located in the picardy repository.
Summer 2020
Every now and then I try out new HF antenna ideas for my location, on the fifth floor, without access to the roof, and without good grounding. This time, I wound cheap garden wire around some PVC tubing, extended with telescopic whips.
The antenna is centre-fed with something that looks like a delta match, and a 1:4 Guanella balun. Resonance and a reasonably good match were easy to obtain. I could move the turns towards centre or the edge of the tube to change resonance frequency, and tap the coil in different places to find match.
During a PSK31 QSO where I increased the power to 30W, the PVC started to melt, which put this experiment to a conclusion. Either the garden wire wasn't good enough, or the PVC dielectric is too lossy. Probably a bit of both, but I made a few QSOs in total, which counts as a success.
Summer 2020
During the may VHF-and-up contest I received some feedback about the bad modulation quality in SSB, and I found out the SRF7043 final had a visible crack on its ceramic package. I followed KA7OEI's guide and VK3KBC's guides about substituting a MRF1570.
Capacitors also needed replacement, and I ordered some ATC 100B caps.
The first MRF1570 didn't work, it was either faulty or counterfeit, so I ordered another one elsewhere. Now the FT-100D is usable again on VHF and UHF, albeit with intentionnally reduced power to avoid having to do the same operation again in the future.
May 2020
I followed DH2PA's explanations on how to build a HB9CV for 23cm, and connected my USRP B200 to a SPF8189Z board followed by an MC7862 amplifier hybrid module for transmit, and a commercial connectorised LNA. A manual switch powered amplifiers and the coax relay.
On software side, I used F4EXB's SDRAngel to modulate and demodulate. With this decidedly QRP setup I managed one QSO with F5AYE across the lake.
Spring 2020
Up to now I only operated on 3cm using a small horn. The RSGB Microwave Handbook contains information about a penny feed for deep centre-focus dishes, and since I had received such a dish I wanted to put it to use. The local hardware store happens to stock rectangular aluminium extrusion that has the right size for 10GHz (23.4mm x 12.5mm inside dimensions, giving a wavelength of 36mm at 10386MHz)
However, I had never soldered aluminium before, and I was told it wasn't easy. I can confirm this now.
Turning the "penny" of the feed
I was lucky to be able to do my first rain-scatter QSO with DK3SE in april with this setup:
Operating portable on the site of the HB9G VHF Repeater, made 5 QSOs
Spring 2020
In april 2020 we installed the Glutte Coulomb-Counter project on the repeater site. New batteries were installed in summer 2019 and we found out that measuring battery voltage alone wasn't a good indicator for state of charge. The principle of the charge estimation is to measure (dis)charge current ten times per second and integrate over time, which gives Ampere × seconds = Coulombs. Current is measured through a shunt, amplified by an AD8210 followed by an LTC2400 ADC, and the integration is done using an ATmega328.
An excerpt of the schematic:
Board installed in april 2020:
The resulting estimation in Ah is sent by the repeater every even hour in the CW beacon, and below a threshold value the repeater is put in low-power mode. Values are also graphed using munin. Later, we want to use the three relays to disconnect other consumers, e.g. the 70cm and 23cm beacons.
The above repository contains KiCAD design and C firmware. PDF Schematics
Summer 2019
The HB9G VHF repeater logic sends out a CW identifier with some stats every even hour. We thought completing this with a longer report every evening at 22:00 local time would be interesting, and chose to transmit PSK125 over FM. This is quite easy to receive, placing a handheld transceiver next to a laptop microphone actually works, and the datarate is high enough to send some interesting statistics.
Example report:
HB9G www.glutte.ch HB9G www.glutte.ch Statistiques du 2022-07-18 Version= e1cb67c Uptime= 57j7h49m U min,max= 11V6,14V6 Temps QRP= 0% U heures pleines= 12V2 12V4 12V1 12V9 12V3 12V4 12V0 13V0 13V1 13V9 14V0 14V2 13V0 13V0 13V0 13V2 12V8 13V1 12V9 13V3 12V7 12V7 11V6 12V2 Capa heures pleines= 1648 1649 1646 1649 1648 1649 1646 1649 1649 1650 1650 1650 1649 1650 1650 1650 1650 1650 1649 1649 1649 1649 1647 1647 Nbre de commutations eolienne= 22 Temp min,max= 15C7,26C1 Nbre de balises= 60 Nbre de TX ON/OFF= 220 Nbre anti-bavard= 0 QSO le plus long= 0h17m32s Sat GPS= 12 Disjoncteur eolienne= On
The logic controller was replaced in 2015 when the repeater got struck by lightning. It is responsible for transmit/receive switching through a state machine, temperature measurement, 1750 Hz tone detection, DTMF tone decoding, CW identification and information, reacting to high SWR. The controller is based on the STM32F4 Discovery Board carried on a protoboard.
For more details, please see the repository, especially the doc folder for schematics.
2017
I had a cheap transistor amp cabinet, and when a friend of mine asked me if I could repair his tube amplifier, that triggered a new interest. I ended up replacing all the electronics of my amp by a two-tube circuit inspired by the AX84 project
More pictures here
Before 2018
I started taking lab notes in 2018, and everything that was before is a bit too incomplete to document after the fact. Maybe that's alright, because I feel these older projecets are also less relevant.
Among those projects were: The assembly of the Elecraft K1 kit, which served as an excuse to "finally" learn CW, a goal still unfullfilled; assembly of the Downeast Microwave 10 GHz Transverter, mentioned above already; repair of a HP8563E spectrum analyser, which only worked for spans lower than 2 MHz; and probably many other projects that were forgotten since.
