Getting to grips with DMR

I’ve been monitoring developments in the amateur radio world for quite a while now, eager to keep abreast of the latest trends as the world dashes headlong down the slippery road of technological advancement. Most of the action seemed to me to be happening in the digital domain.

Digital voice is where most of the commercial and military world is headed, so I zeroed in on DMR as an emerging amateur radio mode.

D Star has been operating in this space for quite a while now but the proprietary nature of this venture has many sitting on the sidelines. One of the reasons is probably because of the high price of the hardware needed.

Being an open source computer fan for many years now, I became aware of a new mode that is gathering pace world-wide: DMR-MARC.

DMR-MARC is an all-digital group of over 500 DMR-MARC repeaters in 48 countries with 33036 registered users. And both lists are growing all the time. In Australia there are repeaters in NSW, VIC, WA, QLD, with the latest in the ACT currently being commissioned as I write. These repeaters operate on the 70cm band.

My friend Wallace, VK4CBW, also became interested in this fast-growing mode. He decided to take the plunge and imported two DMR hand-held transceivers from a factory in China. The brand is Vitai, which neither of us had heard of before.

When they arrived, we were very pleased with the quality of construction.


The Chinese make excellent rigs these days.

The first thing we had to do was to register as a DMR user. This we did through the DMR-MARC web site and it wasn’t long before we received our new ID numbers.

2016-07-01 (1)

Now it was time to sample the marvels of digital voice communications. We began by working simplex between us; what quality! And our conversations were private too; no other user could listen to our conversations.

It was now time for me to try for some DX. For this, I discovered that most operators monitor a very useful web page, which acts as a control centre for VK operators: it allows us to see who is working on the various channels or talk groups.


All I had to do now was to select the talk group of interest and hit the PTT button on my hand-held and I would be up and away.

So far I have worked stations in Malta, South Africa and Finland, all with 4 Watts from a hand-held transceiver!

Softrock RXTX v 6.3: RX Switching and RX Muting

This stage handles the muting of the RX section when I PTT goes high. After soldering in all the component, the testing went according to plan, until taking resistance readings of the power rail.

The initial current readings were as expected, but when I switched the DMM over to read resistance levels, nothing registered. I was expecting to see around 7 Meg ohms at the 12V test point, 950 Ohms at the 5V test point and 10 K Ohms at the 3.3V test point. Resistance readings of the band pass filter’s secondary windings were fine.

That’s where I left things for the night. My philosophy is to sleep on it when things get tough.

It was at around 4am the next morning that I woke suddenly with the answer as to why the resistance readings for the power rail were non-existent: I was taking the readings with the power to the PCB turned on! So I took readings again, this time with the board dead and all was just as it should be.

Still flushed with success, I decided to push on and continue with the rest of the testing. All readings on my DMM were as they should be, so it was time for the best part, to solder in a temporary antenna, connect up a lead to the input of the computer’s sound card, start up the software (Rocky in my case) and see if the rig could detect a test signal on 7.046 MHz (the centre frequency of the 40m band that the rig is tuned to).

This is the spectrum before sending a test signal from my FT-817.

This is the spectrum before sending a test signal from my FT-817.

Now came the moment of truth.

The test signal is clearly visible now. The spike at 7046.7 is the signal. It has a mirror image at roughly the same distance to the left of the centre frequency. This is due to ground loops on the PCB.

The test signal is clearly visible now. The spike at 7046.7 is the signal. It has a mirror image at roughly the same distance to the left of the centre frequency. This is due to ground loops on the PCB.

Once the rig has been completed and installed into an enclosure I will try to filter out any signal images that mayexist.

This completes the build of the main board. Next is the difficult part: the PA filters.



Softrock RXTX v6.3: PTT circuitry

The PTT circuit is all about connecting the PTT and Keyer inputs up to the SDR software via a serial interface. However, the DB9 connector will be installed in a later part of the build. I will be using instead, a USB I2C interface as both my Surface Pro 4 and my Compaq laptop, which runs Arch Linux, don’t have serial ports.

This stage involved installing four caps, 12 resisters, a diode, an RF choke and four transistors.

All appeared to go well until it was time to carry out some tests. Current tests proved spot on, and so did the initial voltage tests. Until it came time to prove that the transistors were turning on when voltage was applied to the PTT_IN connection.

What is supposed to happen is this: when 12V is supplied to PTT_IN, Q1 turns on, pulling R21 and PTT_IN to a low level. Q2 then turns on and I should be able to measure about 12V at S12V. This means the rig is transmitting.

I did not see this on my DMM. My reading was of the order of 0.02V.

So it was out with the schematic once more (I had become quite familiar with this piece of paper). I started by tracing the power supply to the transistors to see what was wrong and why Q2 wasn’t turning on. After much thought, I noticed that two vital resisters were missing on my PCB.

These resisters were missing on my PCB

These resisters were missing on my PCB

It was then that I decided to check if any other components were also missing. What I discovered was that I had omitted to solder in all the capacitors and all the resisters for this stage! No wonder Q2 wasn’t turning on.

Once this slight oversight had been corrected I ran through the test once more, with perfect results.

Next will be the RX Switching stage.


Softrock RXTX v6.3: TX Mixer

The next stage in the project was to construct the TX Mixer stage. The job of this stage is to provide the modulation of the Dividers’ output signals by the four I and Q signals from the Op Amps. The result is a double sideband RF waveform that will be coupled into the PA stage.

This stage centres around U3: FST3253 which is a SOIC-16 Dual 4:1 Mux/Demux Bus Switch. There were also four resisters, a capacitor and two connector sockets that completed the build.

When that was done it was time to test if all was as it should be.

This was when I hit a snag.

First, I had to jumper the hairpin bend of R26 (which hadn’t yet been installed) to ground and then jumper pins two, three and four of socket J1.

R26 needed to be jumpered to ground.

R26 needed to be jumpered to ground.

Current readings were fine, and so were the initial voltage readings. But when it came to measuring the voltage on pins 7 and 9 of U3, instead of getting around 2v I was reading 0.01v.

I tried again but this time noticed that when I turned on my power supply (12V DC) the current surged to around 2A before settling down to more normal levels. I cut the power and began scratching my head.

All the solder joints looked fine as did the components, which were all in the right places. That’s when I decided to take a break and sleep on it.

After the dust had settled, I decided to consult the schematic. I started by tracing the 5V power route, through U4 and into U3. I could see its path to ground was through pins 1 and 15, then on to the as yet uninstalled R26 to ground via C43. So that’s why I needed to insert a jumper.

I then connected up my DMM and swung the switch to the continuity setting. Probing the jumper connection I had inserted into R26 produced nothing. So I probed R26’s other hole and bingo. I had jumpered the wrong whole!

My mistake was immediately apparent.

I had presumed the jumper needed to be in the whole marked with a white circle around it (to indicate that’s where the body of the resister fits). The instructions called for jumpering the hairpin of R26, which I suddenly realised was not the hole indicated by the circle, but the other one.

Resoldering the jumper took only a few seconds, but the satisfaction I received from a full set of good readings lasted quite a lot longer.

This little exercise highlighted to me the importance of being able to read a schematic diagram.


Softrock RXTX v6.3: Building the TX Op Amps

This stage has a fairly high component count, so patience was the order of the day. I decided to take it nice and slowly so as to enjoy the process. It would also make sure I didn’t make any mistakes.

The stage consists of four unitary gain op-amps, arranged in pairs. The left channel’s input resolves to two signals: 0° and 180°. The right channel’s input resolves to two signals: 90° and 270°.

Each of the 14 resisters was checked with my DMM to ensure I had the correct ones for insertion into the PCB; it’s easy to mistake brown for red in the colour coding on the tiny resisters.

The Op Amps themselves (IC SOIC-8 dual Op-Amps) were also tiny beasts each with eight pins that required careful soldering so as to ensure no solder bridges or spashover on any of the adjacent empty holes. For this task I used a very handy suction tool that Wallace, VK4CBW, gave me some time ago.


Positioning U1 on the underside of the board with the suction tool.

Once all the components had been soldered into place and I was certain there were no cold solder joints or solder bridges, it was time to conduct the usual current and voltage tests according to the instructions.

Thankfully my patience paid off with all reading being as expected.

Next is to tackle the TX mixer.

Softrock V6.3 build: The RX Mixer Stage

The mixer stage acts like two traditional direct conversion mixers operating in tandem. It centres around U10, a SOIC-16 Dual 4:1 Mux/Demux Bus Switch. A 2-pin and 3-pin socket also needed to be installed along with three resisters. And for testing the stage, an 80/40m Band Pass Filter board also needed to be built.

Care had to be taken once more when soldering the tiny pins of U10 into position.


This is the underside of the PCB with U10 top middle.

Building the BPF was relatively simple, with the exception of winding the two coils; these take time and care but aren’t particularly difficult to do.


The 80/40m BPF board in place.

Testing went well. Current and voltage readings were as expected. The only test I was unable to carry out at this stage was to test it with Rocky, a SDR software package that shows a chunk of spectrum. This was due to my Linux laptop’s sound card not showing up in Rocky.

Next will be to build the TX Op Amps.


Softrock RXTX Dividers – Stage 3

The dividers stage takes the local oscillator’s signal and divides it by four, producing two output signals that are said to be ‘in quadrature’. This means they are out of phase with each other by 90 degrees.

The trickiest part of this stage’s construction was soldering in U9, a 74AC74 SOIC-14 SMT that has fourteen small pins (or legs). And even though I took great care soldering them onto their respective pads using flux and a very fine soldering iron, when it came time to test if all was okay, the readings I obtained suggested otherwise. So it was out with my more powerful iron (with a larger tip) and with care, the re-soldering exercise produced near perfect readings on my DMM.

Next on the agenda was something I had been waiting for with anticipation: the frequency output test.

To accomplish this I would need my Hantek 6022BE USB DSO, and both probes.

This is what I observed.


As can be seen, the two signals are 90 degrees out of phase.

Next to do is to build the RX Op Amp.