The Southwest Cluster

The Southwest Cluster History

Linking of repeaters.

The concept of linking repeaters is not new in fact it can be traced back to the introduction of amateur repeater stations in the 1970’s. However,  with the recent and rapid growth in Digital Mobile Radio and its associated repeaters here in the UK, linking has really moved to the forefront and has advanced up the list of requirements for a repeater.  It now appears to supersede the RF planning and predicted coverage area of a repeater at the very beginning of a new repeater application.

Looking specifically at DMR technology, as opposed to the modes of D*Star and Fusion which confuse things by also being referred to as DMR.  This specific arm of the hobby uses repeater linking. In fact  DMR linking has been pushed by the amateur community way past its original design.

It is worth noting that much of the DMR equipment available up to around 2014 was purely commercial sourced. Being commercial it was designed for the commercial environment, by this we mean it was designed to work on a specific private network, providing links to other repeaters in that privately owned network. For example, a large government roads department where you want to provide comms to all your workforce across the country, but keeping communications private and secure from say the local cab drivers in a number of cities where your working.  On top of this you want to be able to communicate with, say team leaders or supervisors throughout the network, but keeping that communication private from the normal work force whilst allowing the standard worker to communicate to the rest of the team.

(C) Hytera

Commercial DMR equipment allow this limited separation of managers and workers by using a single RF Channel (commercial RF licensing is a lot of money so keeping spectrum usage to a minimum is a priority) and digitally chopping it into two, you can then switch between these two sub channels so fast the human ear cannot notice the gaps in communication this causes. This is basically what DMR is doing. A single channel is chopped up into things called Timeslots, each timeslot is given a tag, if you have a tag for that timeslot you can communicate, if you do not have the right tag, you cannot hear or use that timeslot. This is a VERY high level look at commercial DMR systems.

Of course ‘us amateurs’ see these commercial world as restricted and as a technological challenge that we must push beyond this original design specification both in terms of RF and its network design and function. Welcome to (in the case of DMR) the DMRmarc group in the USA. There is a lot of information on this group out on the internet, so I will not waste page space repeating this. 

They created a network, linking a number of American counties together. But they saw the need to create a lot of ‘Talkgroups’ (Tags) on each timeslot, so that for example: A person wanting to know what  amateur events were going on nationally could select that talkgroup to listen to and interact with communications about national events, yet another amateur that was not interested could monitor the repeater channel and Timeslot, but not hear the events chatter as they had not dialed into that talkgroup. However if someone called a talkgroup they WERE monitoring they then could interact on the radio without the need need to re-tune the radio. It just worked.

The DMRmarc network spread quickly world wide, so did the number of talkgroups. These work further broken up into, Interests, county level and even country level. So in the matter of a few months they had gone from the simple commercial design of 2 timeslots each with a couple of talk groups, to two timeslots and hundreds of talkgroups!

There was one BIG restriction with DMRmarc.  This was the ‘marc’ bit or Motorola Amateur Radio Club. They used Motorola equipment but by design Motorola repeaters when connected to a network, such as the internet, can only speak to other Motorola repeaters, RF side not a problem, any vendors radio worked just fine. So not only was the DMRmarc network restricted to using Motorola equipment, but you could only connect 15 repeaters together.

To get over the restrictions of 15 repeaters maximum you had to invest in another commercial bit of equipment called a ‘Bridge’ which allowed a larger number of repeaters to be linked. Of course being amateur radio, other groups started using other commercial radio’s such as Hytera and Tait.  Each designed to link to other repeaters as long as they were manufactured by the same company.

This lead to rivalry and fallings out between a number of amateurs (The wrath of politics again!). I will again spare the commentary as there is an a mass of information on the web for your viewing pleasure.

By around 2013/14 there were at least 3 big networks, each supporting a large number of talkgroups and each isolated from each other.


Back in 2013 the Bristol 70cms Repeater Group were looking at the growth in DMR and decided to look at the feasibility in running such a repeater at our Lansdown site which already had the well established Analogue repeater GB3BS.

The initial look at the frequency assignments being made for Digital repeaters made for some interesting RF calculations as the Digital Tx to Rx splits were now at 9MHz and with the repeater transmit channel being at the high end of the 70cms band.

This placed the DMR transmitter very close to the existing GB3BS receiver input frequency and vice versa with the Digital receive channel. In fact there was, at best, only some 700 KHz of separation!  This arrangement was nuts, how the band planners came to this decision is, at best, interesting! Never the less we took to the challenge.

To make things even more interesting we decided that our Digital DMR Repeater would also share the same antenna as our Analogue counterpart, GB3BS, see what I mean about wanting to push things beyond their original designs?

The Orginal network layout

In late 2013 we submitted an application for a NoV to run GB7BS. We had expected that such a NoV would take some 6 months to come through, giving us time to gather hardware together, decide on a design and find a way of raising funds to purchase the Motorola DR3000 repeater which at the time ran into the £1000s.

January/February of 2014 we were taken aback when the NoV to run our new digital repeater was granted. Traditionally UHF repeaters had to go to the MOD for vetting which added months to the wait time for the NOV. Added to our shock was the number of other UHF repeaters that were granted their NOV’s at the same time. It was like a plug being pulled out of the bath.

Forming the Cluster.

The early part of 2014 we had, by chance, discussions with other repeater groups within the South West and soon found that we had similar ideas that linking of several repeaters would be a good project to undertake while keeping it divorced from the world wide DMR-Marc and the other over specified networks. The name South West Cluster was born!

As time marched on the various repeater groups involved, proceeded to prepare their own digital DMR repeaters. First to come on air was GB7AA and operated as a stand alone repeater until others were placed on air. After few delays, in October GB7BS came on air. During the following months the other repeaters GB7SD, GB7JB and GB7DR also came on air.  Of course at this point we were still working on getting the internet to our own site.

Each repeater was owned and operated by its own individual repeater group, bought and paid for by them. I know this flies against a common Idea that repeaters are paid for by the RSGB, this is not the case. In fact I approached the RSGB to investigate if they would like to support out concept. I was told “No,  This technology will never catch on, and therefore we do not want to support such limited equipment”.

Linking (known as the Back Haul).

Most repeaters are located on remote hill tops to help increase coverage. This is no different for a number of the repeaters that are members of the South West Cluster. Although it brings a massive coverage area improvement, a remote hill top does introduce internet provision problems and cost.

A few of the repeater groups either already had broad band on site or managed to share a connection with an existing site user. For the rest, we had to find a back haul link to the nearest broad band service.

GB7BS was not immune to this issue. We did not want to go down the route of requesting a commercial broadband service to install a connection on the Lansdown site, as this would not only cost a lot of money to provision, and more money to maintain the connection. We did not feel it right to approach our land lords with a request that may disrupt their operations while the installation took place.

We invested instead in a 5GHz link from the Lansdown site back to my home QTH, around 6.8Km away.

The 5GHz link description is elsewhere on our website. Once installed, we could then start linking the repeaters together.

It was amazing how fast the cluster grew. However, we did have a few repeater groups leave as well, but other than the lack of politeness in pre-warning us of their disconnection, leaving the SWC was again down to the individual repeater groups.

In 2019 we published a high level view of the network, it was amazing how big it had become, equally it also started alarms ringing as we grew, we got ever closer to the maximum number of 15 repeaters that we could support.

We investigated into buying an off the shelf Bridge unit, but this was a LOT of money, and that and on going maintenance support costs and a helpdesk in the USA, the purchase and running costs were eye watering this stopped us in our tracks.

In Mid 2021 we started looking at the FreeDMR network. And with the help of the authors of the server software we built a Virtual server on a converted high end spec PC.  The Authors were running a network that was primarily made up of MMDVM repeaters with one or two Motorola repeaters. This was one of the things that interested us, not only had they broken the 15 repeater restriction, but also allowed different equipment manufacutres to communicate.

We explained our cluster to them and our desire to maintain its isolation, they put a solution together and it was installed on our virtual server.

sadly we  had no end of problems trying to get things working. It was initially thought it was the virtual server was the cause of the issue, so at that point the project was shelved as we re-evaluated equipment, costs, and impacts of the project.

While scanning the normal auction sites a HP380 Server was found. This might not sound to interesting to you. But for Mat it did hit a cord. In a previous job he not only used them, but also serviced and maintained them. The seller was contacted and after an email exchange over a couple of days, we headed to Southampton to collect.

It took me over the summer of 2022 to learn enough Linux to allow me to build and install the FreeDMR software with little guidance from the FreeDMR team. By October 2022 we launched the server officially and allowed all cluster members to connect to it. Think of it as a large telephone exchange, that is able to route calls around a city (The southwest cluster) but also allow inter-city/Country calling to take place, This is called Dial on Demand. The picture (below) shows the layout of the new FreeDMR network. We maintain our Southwest cluster and also allow Dial on demand (DoD) via slot 1 on all the repeaters Southwest wide.

We thought this was it, all working and happy, then the 2022/23 energy spike hit.

The recent energy savings we had made to the repeater site by closing things down, sharing supplies and simply changing the lighting to LED was wiped out, and the server that was running 24/7 was now costing more than the savings we had made.

It was decided to go back to running a Virtual Server, effectively off loading the running costs and maintenance of a server to a commercial provider of virtual machines.

The cost is around 50% of the running cost of the physical box.

The new cloud server was built (now we are getting comfortable with Linux and command lines) and after a short period of testing it was linked up to our existing server so to allow repeaters to move over without losing connection to other cluster members that had not yet moved.

On Feb 18 2023 GB7BS was move over to the new system. If you were not aware that it was moved your user experience has not been affected. We hope that the other groups will move over as and when groups get to visit their sites to make the change to their repeaters.

One bonus to moving to the virtual server is the source network speed and bandwidth far outstrips my network source to site. The 5GHz link to Lansdown, once everyone moves to the new server, will be in effect running backwards. Instead of internet traffic going to Lansdown and the server, it will be be swapped over and only GB7BS will be transferring the link allowing it to connect to our new virtual server.