What’s the best VPN solution?

Many roads lead to Rom. Probably, the most common VPN solutions are, SSH-Tunnel, PPTP and OpenVPN. I personally think that OpenVPN is the best option, it’s robust and secure. However it’s a bit more difficult to setup. On the other hand, PPTP is easy to setup and if encryption is enabled, it should do the job just as well, at least in non critical infrastructures. Finally the Secure Shell (SSH) can also be used to create an encrypted tunnel between two computers. It uses on-board means, is properly integrated in the Operating System and provides state of the art security.

What’s needed?

In order to establish a VPN connection you need to connect from your local computer to another endpoint, which usually is a server. in another post I’ve described how easy it is today to setup a server in the Amazon EC2 cloud. Another solution could be your home router, in case it’s using a Mini Linux Kernel (like DD-WRT or Tomato).

SSH tunnel on Mac OSX

With the help of SSH, a SOCKS proxy can be created with just one command.

ssh -i "your private SSH key file" -D "port" "username"@"server"

example:

ssh -i mykey.pem -D 8887 tobias@myserver.com

That’s it. Now you just need to route your traffic through the local SOCKS proxy you’ve just created. Therefore open in Mac OSX:

System Preferences > Network > Advanced > Proxies

and enter the data as shown in the picture below (localhost & port).

SSH tunnel on Windows

On Windows, I suggest to use an SSH client called Putty. Putty comes with a nice Graphical User Interface so you actually don’t have to use the command line at all. I found a nice tutorial. Instead of copying it, I rather link it. It explains how to use putty to create an SSH tunnel.

Don’t forget to register your SSH key within Putty. Otherwise the server might not accept your connection!

Advanced: DNS through SSH

In case you are worried about sending your DNS lookups over the local WiFi, you might want to tunnel them as well through the SSH tunnel. It’s not that easy, because SSH is based on TCP and DNS lookups are UDP. Since this is an advanced topic, I would like to refer to another website which explains in detail how to tunnel UDP packets through a TCP connection. If you are really concerned about the DNS lookups, I think it might be the right time to think about a true VPN tunnel with PPTP or OpenVPN, both solve this problem.

Troubleshooting

• In case the tunnel doesn’t work, try to establish a normal SSH connection
• The standard port for SSH is 22
• Make sure you have the private key on your client
• Check the firewall on the server (Port 22 TCP needs to be open)
• Check the Security rules on Amazon EC2 (Port 22 TCP needs to be open)

Why would you need your own server?

A dedicated server with root access opens a whole new world of applications, even if you are just browsing the internet. Here are some use cases:

1. Webserver – If you are hosting a website, it might be a good idea to move it to your own host, where you have full control.

2. Code / file repository – If you are going to collaborate with colleagues or friends on a project, it is crucial to have a common file repository. I’ve been using Subversion (SVN) for a long time, and it has proofed to be a viable partner.

3. Save Internet browsing – Is the local WiFi network at your favorite coffeehouse encrypted? If not, I recommend to route all your traffic through a Virtual Private Network (VPN). If you have your own server, it’s quite easy to set up a VPN.

4. Streaming Netflix – Are you outside the United states? No problem – you can route the traffic from US based server through the VPN connection to your Computer, TV or iPhone, wherever you physically are.

5. Experimentation – instead of installing your stuff locally into Virtual Machines, try a server. With Amazon EC2 it’s faster and more convenient.

These are just a few examples, to give you an idea about what you can do with your own server. Of course, the application of it is just limited to your imagination!

Why Amazon EC2?

Amazon has not only emerged from a bookseller to a wholeseller,  it has also created one of the world’s most powerful IT infrastructures in order to deliver it’s own content in a fast & responsive way. Fortunately, Amazon has decided to sell these IT services to the public, as well. The are called Amazon Web Services (AWS) and consist by today of a wide variety of different services. For example, all the mp3 files of my podcast are stored and delivered through Amazon Simple Storage Service (S3). Another service is Amazon’s Elastic Cloud Computer (EC2), which makes it really easy to create a virtual Server. Within a few mouse-clicks you can add almost infinite hard disk space or multiply the amount of CPUs, as you like (and as you can pay)

Best: During the first year, a microinstance is with 10Gbyte of Harddisk is completely free!

All you have to do, is to register your own Amazon AWS account.

Let’s get started

After creating your Amazon AWS Account, you need to register for EC2. The management console allows you to select predefined images (AMIs). There exists a wide range of predefined images. Feel free to select your favourite Linux distribution and follow the installation instructions. After less then 5 minutes, your Server instance is configured and ready to lauch:

A minute later, your Server is up & running. Now you can connect yourself to the server and start configuring it.

Make it even more convenient

A more convenient way however is to use an additional service called Bitnami. Bitnami allows you to preconfigure your server individually through a polished webinterface. Instead of installing the LAMP Stack (Linux, Apache, MySQL, PHP) by hand, Bitnami will do this for you. You just need to decide which kind of Linux Distribution you prefer and what you need on top (e.g. WordPress, Drupal, Joomla, etc). You just configure the server though Bitnami – it will be hosted directly on the AWS infrastructure, within your AWS Account. Here is a look the configuration looks like:

Depending on the size of the instance and hard-disk, bitnami will show you automatically the estimated monthly cost. But again, if you are a new AWS client, a microinstance will be free for one year!

Here is a video which shows you how easy it is to set up your own server:

If you are setting up a remote server on Amazon AWS I suggest to use bitnami. The preconfigured images are very good and they provide a useful Wiki which guides you though the first steps. Bitnami will for example register a DNS record for you automatically so that you can access your server directly by SSH or through your webbrowser.

Costs

Again, Amazon will give you a Micro Instance (2 cores & 613 MB Ram) one year for free. Uptime of your server is billed hourly. By today, an online hour for the instance costs 0,02 USD (free for the first year).  Incoming traffic is free, outgoing traffic costs a little bit (0,1USD / GByte).

I’m using the Iphone App called Cloud Service Manager to conveniently startup, shutdown or monitor my AWS instances.

Final words

It has never been easier to set up your own server. All you need in an Amazon AWS account and your server will be up and running within a few minutes. There is plenty of good wikis and help available, however some basic Linux administration skills won’t harm!

All is needed is a plastic box, one or two tube sockets, a few plugs and a fan

The plugs on the front allow a good access to the individual pins of the tubes. This makes troubleshooting more comfortable.

When NOS tubes are reconditioned, heating voltage is applied for a period of at least 24 hours. Even if just a few watts of heat are produced, a constant airflow ensures that the tubes stay cool. Any 12V fan will do the job.

Thanks to Matthias, DK4YJ for sharing this pictures.

Can you believe how mad I am?

Above, both damaged tubes

Customs drilled a hole in the heat sink and thereby also damaged severely the inners

This is the other tube. Here they tried to open the tube with some kind of screwdriver or plumber

The anodes of the broken 4CX800 tubes; note: you see the damages from drilling on the left anode

A look on the ceramic insulator and the anode heatsink

And finally another look into the heatsink

Did I already mention how angry I am about these incompetent customs officers????

Listen or Download the Interview

Subscribe in iTunes

Please consider subscribing to my Podcast feed in Apple iTunes

Or subscribe to my podcast feed in any other Media Player

Shownotes

• Stu, K6TU’s Blog with three detailed articles on the Contest-Knob
• Kevin, K6TD’s website
• K5FR’s DDUtil – Interfacing FlexRadio Hardware
• FlexControl product website at FlexRadio Systems Website
• Microchip MPLab Integrated Development Enviroment for PIC Microcontroller

Interview Transcription

DH1TW: Hi everyone, this is Tobias, Delta Hotel One Tango Whiskey, with another episode of my SDR, Software Defined Radio Podcast. This is the place where interesting fellow hams come to talk about their projects, share their experience and knowledge with the goal to motivate you and to help you get started with your own project- even today.

 Over the past two years, I was working on PowerSDR-UI, a modified version of PowerSDR with the aim to improve the haptic- let’s say, the real knob user interface of FlexRadio’s SDRs. The way I took was to integrate an off-the-shelf DJ console which is available for little money but provides a lot of knobs and buttons. Almost at the same time, Stu, Kilo Six Tango Uniform, was working independently on a similar solution: The contest knob which is now manufactured and sold by FlexRadio Systems and in the name FlexControl. Today, Stu is here to talk about the FlexControl and the path he went from the first idea to an industrial manufactured product. So, without further ado, welcome, Stu.

K6TU: Hi, Tobias, How are you?

DH1TW: Thanks, I’m fine. Thanks for coming here on the show tonight. Before we get started, I always ask the same question to all of my interviewees: Who are you and how did you get started in Ham Radio?

K6TU: Well, I’m, Stu Phillips K6TU, originally from the UK, where I held my first call sign back in 1973. I got in to Ham Radio as an SWL many, many years earlier actually. And, it’s been a hobby that has stuck with me. I was always fascinated by propagation, being able to talk to people in other countries and just the technical intricacies and, the opportunity to experiment with Ham Radio.

 DH1TW: Good. Do you have any technical background? Any engineering degree?

K6TU: Yes, I am electrical engineer by training and I spent the bulk of my operating career actually running software development for organizations. The last one of those I actually ran all of the software development for CISCO Systems up until 1997.

 DH1TW: Cool. So, then a few years ago FlexRadio Systems brought out their contest. Well, actually, not only contest radio but the Software Defined Radio platform and there was something missing. We had to operate the radio with a mouse and the keyboard but somehow it wasn’t the same feeling than operating a radio with the knobs. And, tell us a little bit about how you came up with the idea to develop the Contest-Knob.

K6TU: Okay, sure. Uh, so, in the middle of 2009, I decided to upgrade my station. I had had a Kenwood Radio for many years and had found that I was hitting some of the limitations. Particularly, living here in Silicon Valley, it’s a pretty high RF environment, particularly in contests and the Kenwood would- the receiver in the TS-2000 would often be limiting for me. So I looked at a couple of different choices and decided actually to buy a FlexRadio 5000, which I did. And, one of the things that I had started doing- I had always been an active contester but I would say more on the sort of casual basis. And, a couple of my really good friends here in the Silicon Valley that encouraged me to take contesting a little more seriously and so that I began to do. So I operated one contest. Actually, the first contest I operated, I think with the Flex-5000 was the California QSO Party in 2009. And, quickly found that there was a real workflow issue.

So, I use Write Log as my contest logging software and obviously if you’re trying to work a lot of stations quickly you really want to keep all of the window focus on the logging software, and really not have to move the mouse outside of the logging window. And so, at that time, I had no other way of tuning PowerSDR other than using the mouse. And so, I was obviously taking the window focus backwards and forwards between PowerSDR and Write Log. And, several times found myself typing into, typing for example a pull sign into PowerSDR. Which, if you have the keyboard shortcuts enabled you can suddenly find yourself 10 MHz off frequency and whoever you were working you just lost.

And so, I decided that it really was critical to be able to have a way of separately tuning the radio independent of the mouse. And so, the window focus could remain solely on Write Log. So, that was the initial idea if you will. It actually developed a tuning knob some years earlier around a very expensive optical shafting coder. I’ve operated a remote control station for actually pretty close to the last 10 years now. And so, I was well familiar with what was needed but I decided that there was a much tighter integration needed on the PC side than I had done in the past. So that really was, if you will, the genesis of why I decided to build what ultimately became the FlexControl.

 DH1TW: Considering the user interface of a classical radio we have a lot of knobs. We have apart from the main frequency dial, we have volume, we have RF Power, we have the RIT, we have XIT, Noise Blanker, whatever. If I’m looking onto the FlexControl, it’s just basically one tuning knob and a few buttons. Was this always a design criteria just to have just one know or did you also play with the idea to include more knobs, more functionality?

K6TU: Well, from the very beginning, I wanted to keep it simple. I was very focused on trying to build something with a cost of goods, built of materials- if you will- that was something that could be economically manufactured if that ever became of interest. Being a product development guy by background, I’ve always approached these things, as you want the minimum in the product that you can sensibly use. And so the original concept when I started looking at contest operation is there’s really two different modes. There’s what a lot of contesters call a “run” mode, which is when you are running on a single frequency, and you’re calling CQ and people come to work you. The opposite is “search and pounce” where you are tuning up and down the bend looking for multipliers or people that you haven’t worked.

So, I decided from the beginning that I wanted to be able to control a minimum number of functions using the knob. I actually enlisted the input from a very good friend of mine, Kevin, Kilo Six Tango Delta. Kevin has a lot of experience in contesting and in fact was the one who had persuaded me or encouraged me to take contesting more seriously. So, in the beginning we had focused really around a knob with a center push switch on it. And of using that pushing switch really as the way of being able to control backwards and forwards between different functions. So, we wound up with this initial specification that was jus the knob with the switch on the shaft encoder that senses the rotation. And, of using that switch to implement almost like mouse clicks. So, a single click, a double click and then a long click where you just press and hold the switch down. And, of using those to create two different modes. So, one mode for “running” and the second mode for “search and pounce” and then, being able to toggle between two different functions in each mode.

So, for example, if you’re running a single Operator 2 Radio (SO2R) in a contest you actually want to be able to control, typically the RIT on the radio or on the frequency that is your primary run frequency. And then, be able to control VFOB or the second radio’s VFO on the radio that you are using on another band to find multipliers. And so, this idea of a mode and within it being able to toggle between two different functions was a way of with using a single knob to say, “Okay, now I was to tune the RIT” or “Now I want to tune the other VFO.” And so, the whole concept became around the very simple user interface with a set number of functions that would control the shaft encoder switch. So, the original knob actually did not have shaft encoder switches in the design at all.

DH1TW: Okay, so basically just the big knob?

K6TU: Yah, that’s correct.

DH1TW: Sounds pretty much like an Apple design…The Apple Philosophy.

K6TU: Yah, I guess Silicon Valley infects everybody with the same thoughts. I mean really, if you’re going to try and build a project and if you can keep it simple and intuitive, then it becomes much easier for people to adopt and use. And, the more complicated you build a product- I mean, if you think about a lot of the hand held radios that are on the market for Hams there are so many functions in them and they’re buried in menus, you almost have to carry the user manual around with you if you want to do anything other than normally change frequency. And so, the whole concept behind the tuning knob was just to keep it simple.

DH1TW: Yeah. I made exactly the same experience. I set up for the DJ console an email reflector where everybody who was interested and can join. And I’m quite often, I receive some requests for “Why couldn’t you add this functionality” and “If I press this button combined with the other one…” I mean, okay, this might satisfy the wish of one or two particular users but I think I don’t want to increase too much the complexity of the product because the DJ console itself is like; it’s like way more complex from an optical or a haptic perspective than the FlexControl.  And, I also believe that a certain amount of complexity is just to high it might be counterproductive. So, you don’t want to overload the interface either. And, I’m always admiring Apple for their great use of interfaces and, I also like the simplicity of the FlexControl. It’s like if you can bring down the basic, the most needed functions into one knob and three buttons you can really see that there were some serious thoughts behind the product.

K6TU: Yeah, that was very much the idea, was to keep it straightforward so that it was intuitive for people to use but at the same time the more functionality you add to a product, eventually you affect the price. And, from the beginning, Kevin K6TD, and myself we had both thought that there was an opportunity perhaps to turn this into a product. We certainly had enough interest from people locally, independent of whether they were using a FlexRadio or not in having a separate tuning knob. And so, right from the very beginning we were very focused on functionality because we had a particular price in mind that we wanted to be able to hit.

DH1TW: A funny story by the way. A friend of mine, Delta Fox Three Charlie Bravo, he developed his own remote control software for the Yaesu FT-2000 and he also started with the DJ Console. And lately, I think one or two months ago, he wrote me. And he told me, “Hey, do you know anything about this FlexRadio knob? Do you think I can use that in my software?” And I said, “ Yeah, sure.  It’s just opening, you just need a serial port and then you can implement it. And exactly for the same reason he just asked. I don’t need the other 25 knobs. I’m just fine with the main tuning knob and 3 functions. So, yea that’s right, your product the FlexControl is not only used on the FlexRadio Systems, it’s also used now on other platforms.

Okay, so, you also had- if I understand this right- from the very early begin a production, a series production in mind. Is that right?

K6TU: Not entirely. The goal from the very beginning was to keep it simple and keep the design so that we could constrain the cost. Both Kevin and I had the experience of running a business out of our garage about 10 years earlier. We had built a business that was selling network time clocks. Now, if you’re familiar with the Internet network time protocol, we had a self-contained appliance that was a GPS receiver and a small microcomputer that provided a stand-alone and TV Strathmore box. And, we had sold a lot of them and eventually that business had got to the point where we were looking at one another going, “So, which of us is going to quit their main job to go run this company?” And, we decided that the market opportunity just wasn’t large enough in order to do that. And so, eventually we wound that business down.

So, one of the things that was very clear, we had very much in mind that if there was ever anybody interested in enough of these we would find somebody else to undertake the manufacturing because we weren’t going to do it. Neither of us saw this as a- you know the FlexControl has turned into a great product and sells well. But, as a single product it wasn’t sufficient to build a company. And both K6TD and I have what we call day jobs. So, you know, we had in mind that there was likely going to be interest so we both- Kevin’s actually the VP of Engineering for a very successful company here in Silicon Valley we’ve worked together professionally for very many years and been friends now for a very, very long time.

So, it was something we had in the back of our mind and its just kind of a basic disciplined approach to building something which is “Let’s keep it simple” and keep in mind that perhaps this thing winds up with other people wanting it. I think both Kevin and I have had a lot of experience with prototypes that became products. And often, with not very happy results. So we wanted to kind of do basic good engineering practice from the beginning.

DH1TW: Okay, so let’s come back to the Contest-Knob. And, let’s go a little bit inside the knob. What do we find inside? A microcontroller I suppose?

K6TU: Yes, it’s actually a microchip Pic. Microchip has a very wide range of different single chip microcontroller solutions and we chose one that had an integrated USB controller as well as all of the standard peripheral stuff like timers and digital IO pins. So it’s very much intended to be a stand-alone embedded single chip solution system with a shaft encoder. And, basically a crystal, you know, canned crystal oscillator. That’s pretty much the contents of the hardware design.

DH1TW: So when you started developing the software what kind of problems did you encounter? What did you have in mind? Like you collect- you just count the number of impulses from the rotary encoder and you translate that to a CAT command and send it to PowerSDR right?

K6TU: Well, not exactly. So, when we started this, obviously, although there is a USB hardware controller on the chip you actually have to have a software stack that implements the USB protocol layers and so the majority of the software that runs on the pick is actually to deal with the USB interface. We decided from the beginning we didn’t want to tie the knob to a particular CAT stream and there were a lot of reasons for that. And, we had in mind this was a design that could be used on things other than PowerSDR so the basic idea was to have a very simple serial protocol. The USB appearance to the system looks like a standard communications port. And so, we took the approach of using a very simple S Key command stream up and down the PC and the knob that does look like a CAT command insofar as its all in text and is delimited by a semicolon. But at that point it’s really its own basic command stream. So the software that’s on the pick aside from handling the USB interface, obviously it monitors the shafting code or it monitors the switches. It has capabilities- even the original design without the switches, has the LEDs on. So there’s- it’s timing and interrupt driven approach so the USB is all handled under interrupt control and all of the knob functions are handled based on timers.

DH1TW: Okay. And, when you connected the first prototype to PowerSDR what kind of problems did you encounter?

K6TU: Actually, very few. So the, when we started development I bought an off the shelf board that had the same pick processor on it that we were building into the design. And so, the very first prototype of the knob was actually the shaft encoder on a perforated board with a couple of ribbon cables coming out of it to this off the shelf development board. And I had contacted Steve, Kilo Five Fox Radio (K5FR), who is the designer of DDUtil the data decoder utility that is just a complete godsend in terms of doing station integration with the flex.

So I had got to know Steve before I started the knob design actually. I had purchased a linear amplified and it was one of the ones that was on Steve’s list of things to support. So I had contacted him saying, “Hey, you know, I’ve got one of these amplifiers on its way. What’s it going to take to do the software support for it?” And Steve was very kind and said, “Well if you’ll help me debug it, I’ll put the software support in.” And so, he did it. And so, the two of us had got to know each other as we were going through that debug cycle of getting this amplifier supported by DDUtil. So when I had started the idea of the know I called Steve up and said, “Hey, what about the idea of putting the PC side support for the knob into DDUtil.”  And so, that’s actually how we started. So Steve was involved with the original protocol design of what messages went backwards and forwards between the knob. And he actually had a different version of a pick development board- fortunately with the same processor on it. So, he built again this kind of ribbon cable approach while Kevin, K6TD, began the stand-alone board development. So the actual software development was pretty straightforward.

We had the initial prototype of the ribbon cable version of the knob actually up and working in about 3 weeks. It was very quick. And so, that was the point where Steve started the development of the knob support into DDUtil. So we got all of the software design done and we had the knob working. And, I think from the beginning the idea with DDUtil was DDUtil manages the CAT interface to PowerSDR and over the years Steve has done a lot of work to minimize the amount of commands that go backwards and forwards over the actual PowerSDR CAT port. In common with any radio, the servicing of the CAT stream has limitations to it because the radio has to do all of the other things it’s doing. And so, you really want to make sure that you’re efficient because otherwise what we were really focused on from the beginning was when you stop turning the knob you expect and you demand that the radio stops tuning. And so we really wanted to make sure that there was no lag. And, at the beginning we came very, very close. We had the original knob was actually did not have any lag to it. We basically- whenever we detected a rotation of the shaft encoder we would map that into a step up or a step down command to PowerSDR. But, one of the things that Kevin suggested from the beginning was that we sense acceleration on the knob. So, the idea was that the faster you turn the knob, the radio should tune faster. So, this was a capability I think that was in the ICOM 756 PRO3. Where as you tune, as you turn the knob faster, the radio would rather than tuning it whatever your current step size is, it would move the step size up.

And so, I implemented an acceleration algorithm on the knob. And that will turn round to the PC and say, “In the last interval I saw the following number of steps.” And we use that in DDUtil to be able to dynamically resize the step size. And that required multiple CAT commands and that’s when we found ourselves with a cue. So, fortunately, over the years of development of DDUtil, Steve had got to know Bob Tracy who had done the original PowerSDR CAT design. And we were able to get Bob to implement some- two new CAT commands that allowed us to send a single CAT command into PowerSDR that said “Tune a different step size.”  And so, we were really focused from the very beginning on making sure that there was no cue. The whole idea here was: You turn the knob, the radio tunes, you stop tuning, the radio stops. And so, there was a huge amount of effort in terms of the acceleration algorithm, how the CAT commands were mapped in order to be able to do that.

DH1TW: Tell me a little bit more about the acceleration algorithm. In my simple mind, I would just count the amount of impulses over a defined period of time. Then, depending on the amount of impulses DDUtil looks up the corresponding frequency step in a table. So, for example, if we count let’s say 5 impulses within 50 milliseconds our table then might say this corresponds to a step of 10 kilohertz, which then will be sent through DDUtil to PowerSDR. Is that right?

K6TU: Yeah. That’s pretty much it. It is a, it’s a time based algorithm. I mean acceleration is really the rate of change of velocity over time, right? And so, we implemented a very, a very simple integration algorithm on the knob. And then, that’s why I said in the original description the USB is all interrupt driven and everything else is based off of a clock interval. And so, it was very simple to interface to that. One of the challenges of shaft encoders, particularly mechanical ones such as the one that’s used in the FlexControl is being able to effectively debounce the switch contacts. And so, all of that debouncing is done is software. There’s a fair amount of subtlety to it. I wouldn’t say that the code is especially complicated but there was a lot of learning- I mean I have a digital oscilloscope that will also serve as a storage scope. And so I was using that to monitor the contact bounce on the shaft encoder. And so, there were some particular cases where you’d see some pretty complicated bouncing patterns, particularly when the knob was being turned slowly. And so, the actual contact changes when the knob is turned quickly are actually pretty clean. The irony was that when you were tuning the knob very slowly –say, for example, you’re trying to fine tune a CW signal or something like that, so you’re moving the knob very slowly backwards and forwards, that generated some very interesting debounce problems. And so, there was a fair amount of thought that went into, “Okay, so what does the algorithm need to look like in order to be able to make sure that you don’t get false tuning effects?”

DH1TW: Yeah. These are the small but nasty things, which come up during the development. They’re not mentioned in the date sheet and it takes quite a time to first identify them and second to create the proper work around it. During university I worked on a similar project. I created the user interface for a frequency generator. And, after a few tests with various rotary encoders I decided to go with the optical encoders. And because of my experience, they bounced significantly less and therefore provided much cleaner signals. But of course, they were also much more expensive and I only had to create one prototype. But, I still remember the hours spending in front of the digital storage oscilloscope, debugging the systems and adopting the software and microcontroller. Because, under all circumstances what you want to avoid is to upset the user, right?

K6TU: Yeah, very much so. And, the vision that we had from the very beginning was, I think, best described, as we wanted the tuning to be quote-on-quote “silky smooth”. So, we wanted it to be exactly like tuning a regular radio. And, we got there. I mean, it certainly- it was- we got a lot of valuable feedback from folks and I’m actually very happy with how it turned out.

DH1TW: You already mentioned a couple of names: K6TD Kevin, K5FR Steve, and Bob Tracy. Who else was involved in the contest knob or respectively the FlexControl Project, in terms of development, beta testing, and etcetera?

K6TU: Sure, so the development was done- so: Kevin, K6TD did the hardware design and actually took care of getting the initial PC boards produced. And then I assembled, well, I guess between Kevin and I; we assembled probably 6 or 8 of the initial boards. They’re surface mount parts so you know, microscope, and a magnifying glass at least and a very fine tipped soldering iron. We didn’t flow- we didn’t heat flow the boards. So, Kevin was responsible for the hardware, I did all of the knob software development and Steve did the integration of the PC side knob support into DDUtil. And so, it was the three of us initially. Steve had got some help from Bob Tracy that was really invaluable.

And then, we decided that we would build probably 10 or so beta units. And so, we relayed the board. We had actually sent one of the original prototypes to Lee, W9OY, Whiskey Nine Oscar Yankee, who has an SDR Radio blog. And Lee actually, was the one who came back and said, “Gee, this is really great. But, man, it would be great if there were a couple of switches on it.” And so, I got this feedback from Steve and I groaned and like “Oh no! The next thing someone’s going to want the kitchen sink on this thing.” And so, we agreed that after a lot of discussion between Steve and Lee and myself they convinced me that it would be really great to have 3 switches so that it would be able to control other capabilities. So, the original support I added the three switches- the so-called auxiliary switches. And, the original implementation of them was “Okay so, somebody presses the switch… big deal.” So Steve used that to map different capabilities.

So we sent this to Lee and Lee came back with, “Well, gee. This is great but could you implement the same click detection that you did on the shaft encoder so that with each button I could control 3 functions?” And, at that point it was like, “Oh no! The kitchen sink is really coming. I can hear it now.” So, I implemented the single double and long click detection on each of the auxiliary switches so that we had built about 10- I would say more- alpha units this time that were small boxes with a knob, with the 3 LEDs and the 3 switches. And, we had sent those to a number of different folks around the world who had given us really good feedback. I mean, I think by this point there really wasn’t anything that came back that affected the design or the concept. It was just really more comments to Steve in terms of what kinds of capabilities people wanted to be able to adjust using the knob. So, at that point I’d also sent one of the prototypes actually the original green board prototypes to Greg, the VP of Sales and Marketing at FlexRadio. And I said, “Hey, you might be interested in this.” And that was actually what had prompted the original discussion. So that was really the development team, you know. Kind of Lee was there as the initial sort of non development team guinea pig and came back with the switches and then ultimately the click detection for the auxiliary switches as well.

DH1TW: And, well, Kevin and you were both living in the Silicon Valley. And Steve, I think was living in Texas. Lee probably also not on the West Coast. How did you communicate with each other? Through telephone? Did you meet on the air?

K6TU: No, we didn’t meet on the air. In fact, it took a long time before Steve and I even had a QSO. That was the irony of it. I would say that the majority of it was done by email and the occasional Skype or phone call. Lee actually lives in Florida and I’ve never spoken to him. It was all email communication. So, email worked really well for “Here’s the spec, what do you think?” “Hey, this is what I’m seeing.” “What explains this?” So, I think Kevin, Steve and I had a couple of 3 way conference calls but to this day I have never met Steve in person.

DH1TW: Speaking about development tools, since the main FlexControl development team consisted of just Kevin and you, you didn’t have the necessity for any project management tool or centralized repositories, right?

K6TU: No, you’re correct. I mean, we weren’t really working with multiple people on the same code base. So we didn’t use an online repository. The source code is relatively small, in terms of its size. So I would email zip files basically of the project. I used the microchip integrated design environment really as the primary tool and then with a C compiler from Source Boost. So really, that was the development chain of emailing zip files to Kevin and to Steve if they needed to reprogram their particular units.

DH1TW: How long did it take you guys to develop the FlexControl? I mean from the beginning when you took the first time the soldering iron in your hand or respectively wrote the first line of code, to the moment you had done industrial fabricated product in your hand?

K6TU: So, from the very beginning I probably started in early January of 2010. And, by the April by- yeah by mid or late April of 2010 we had the majority of the units built and all of the software was working. Steve had done the integration with DDUtil. We built the alpha unit I would say probably around September or October of 2010 and it was about that time that I actually started the discussion with FlexRadio about whether they would be interested in manufacturing what became the FlexControl. So you know, this was a part time effort. I had actually used the original ribbon cable prototype I think for the RTTY WPX Contest in the early part of 2010. I can’t remember whether the WPX RTTY its either in February or March. So the original design was up and running I would say from the point where I started writing the very first code to having the first working knob was about three weeks. And again, you know this was sort of evenings and weekends kind of work.

DH1TW: Can you estimate the amount of labor, the amount of working hours the whole team put into the FlexControl Project? Just give me a rough estimate.

K6TU: Oh lord. Hundreds. Without any question. I mean, I know Steve spent a huge amount of effort designing the support into PowerSDR then ultimately coming up with the control tab that’s in DDUtil. Laying out printed circuit boards even with an order router is a time consuming thing even on a small board. Yeah, there’s several hundred hours worth of effort went into the design development of this.

DH1TW: That’s what I expected. Sometimes you wouldn’t expect this amount of work just by looking at the product itself. And I also think that sometimes we easily tend to underestimate the amount of work put into such a product.

K6TU: Yeah. Building a product, I mean, you can build one of anything in a few hours in your garage- and I’ve done that many times. But to build a product to the point where it can be manufactured, where it can be tested that it just goes together and works, even something that looks as simple as the FlexControl there’s a lot of design, effort and thought that goes into building a product.

DH1TW: So, after the first prototypes were built up, tested and were working you sent one of the prototypes to Greg, the VP of Marketing and Sales at FlexRadio Systems. By today, the FlexControl is obviously available for purchase as an industrial manufactured product. Tell me, what happened in between. So from the time you approached Greg to the time they shipped the first unit and, what was your involvement?

K6TU: Yes, I had contacted Greg with this idea and said, “Look this is something we had developed” and I had sent him one of the original prototypes and Greg is an active VHF contester himself and he’d wound up using one of the prototypes in a VHF contest. And said, “Oh wow, this really helps on the work flow.” And so, we had gotten into a discussion about whether it made sense for Flex to actually have this as a product. Kevin and I had decided right from the very beginning that we would not manufacture the product. It’s just taking orders, dealing with order fulfillment, projecting how many units you want to build and the whole order cycle of getting parts and boards built and put in to boxes. It was more logistics than we were prepared to undertake.

And so, it was one thing for us to build a handful or a dozen prototype units for our own use and for the use of some people we knew had similar requirements but it was a completely different one. So, I had started this discussion with FlexRadio. You know, obviously they have their own development schedule, their own product plans and so on. So part of it was just getting to a point where Flex felt that they had the time and capacity to do this.

And so, what we ultimately did was we sent them the schematics and the board layout and one of the guys at Flex- well actually two of the guys at FlexRadio got involved. One did the mechanical design of the box and we actually relayed the board in a different way. Again, primarily with a focus on manufacturability. And so, I worked with them to bring up that third revision- if you will- of the hardware design. And, it was actually, I would say, very much flawless. It’s a very- we had put a lot of effort in the very beginning into making this a very simple hardware design. And so, we never had any issues with any of the boards either, that we built, or, that flex built. They all came up and worked first time. So the process of working with Flex even remotely on the hardware design was actually really good.

DH1TW: Just by curiosity… do you know how many FlexControls have been sold by today?

K6TU: I don’t

DH1TW: Okay.

K6TU: I’ve seen a lot of them. So. But I don’t know what the numbers are.

DH1TW: Yeah. Maybe I should bring Greg here also on the podcast and ask him about the numbers. Okay, Stu, I think we are almost through the interview. I think it was a very fruitful one. And- but tell me, what are your future plans? Do you have any projects you’re working on?

K6TU: Not at the moment. My primary focus at the moment is on operating the contests. We’re right in the middle of the contest season and I wound up taking the role as the contest chairman for the Northern California Contest Club in the middle of this contest year. And so, we’re all just pretty busy keeping focused on competing in contests and operating.

DH1TW: Oh, me too, me too. I am currently preparing the CQ160 Contest and we want to put up a two element vertical array and four beverages for listening. And therefore, I am currently redesigning my beverage distribution box, which allows us to use to listen on the full beverages independently on the 3 lowbands 160, 80, and 40.

K6TU: Yeah. That’s really good. I envy you the space that’s required.

DH1TW: Yeah. That’s right. We are actually very fortunate at the station where I usually operate from Echo Delta One Radio (160m) that’s located in Central Spain about a hundred kilometers north of Madrid. We found an agreement with the farmer, with the owner of the fields close to the station and in winter time we are allowed to put up there all beverages. And, yes, it’s a significant difference listening on the vertical or being able to have full directional beverage system. And yeah, for transmission I am planning to put up a two element vertical array just to get 3dB more towards the United States. So we’ll see, maybe we’ll have a contact in the 160CW Contest. It would be cool.

K6TU: Well, I just got my 160 transmitter antenna back up literally yesterday so I will definitely take a listen for you.

DH1TW: Okay, Stu, where can the listeners find more information about you and your product?

K6TU: Well they can go to- I have a blog, if they look up K6TU through QRZ.com there is a pointer to my blog and there’s a lot of information on a variety of different technical topics on that blog including 3 kind of in-depth pieces on the development and evolution of the Contest-Knob.

DH1TW: Awesome. If they want to get in touch with you, if they have questions, how can they contact you?

K6TU: Then my email address is also available through QRZ.com as well.

DH1TW: Perfect! So, I want to thank you again very, very much, Stu, for coming here onto the show. For sharing your knowledge and, the experiences you made. And if you, my dear listeners, want to know more about the Contest-Knob called the FlexControl stop by at Stu’s blog and if you’re already using the Flex Control maybe drop him a little line to appreciate the time and the effort that he and the team invested into the development. So, thanks again and take care.

K6TU: My pleasure, Tobias.

The effect of adding radials

The first tests were conducted with a single vertical and the 25 radials which we have always used on 160m so far. Find below the impedance chart:

Now, after adding another 50 35m long radials the impedance on the same vertical looks like this:

The S11 / VSWR curve is sharper (as a proper vertical should look like) and the impedance dropped from 20 Ohm down to 15 Ohm. This means that we reduced the losses by 5 Ohms!

80 radials in place

Below is the simulated (NEC2) impedance curve of our vertical. (17m tall, a 35m long, slooping T-hat and MININEC ground (13mS and epsilon 26)).

The picture above shows the radiator of our 160m array

Above the simulated SWR curve is shown

And finally the simulated impedance is shown as well

Even if the real antenna impedance is likely to be a little bit lower as simulated (14.2 Ohm), I’m happy with the measured 15 Ohms.

Making the Reflector work (or not…)

After having set up correctly the radiator it was time to install the reflector. The reflector was another vertical, 18m tall with an almost horizontal T-hat. At the feedpoint an additional roll inductor was connected between the reflector and ground. The roll inductor (0.5… 15,8µHenry) was used to adjust the reflector for a maximum front / back ratio.

Macbook + DG8SAQ Vector Network Analyzer + Reflectorbox

In order to measure the front / back ratio, we installed a probe (4m tall vertical, terminated with 75Ohm) in a distance of approx. 250m behind the reflector in the field. While transmitting with low power on the vertical (array) I used the DG8SAQ as a Spectrum Analyzer in order to optimize the reflector length for optimum f/b.

Unfortunately for still unkown reasons, the F/B did not exceed 3dB. This is of course very poor.

I’m still struggeling for reasons… Probably the most likely reason is mutual coupling with the 22m tall tower (with a 2el 40m yagi) just behind the reflector.

Coupling between radiator and reflector

Next I measured the coupling between the radiator and the reflector. The chart above shows the feed impedance for the two cases

1. Reflector connected to Ground (red)
2. Reflector not connected (black) -> invisible to radiator

This measurement confirmed that at least some coupling is happening between the two elements. The simulation shows the following SWR and antenna impedance for the array:

The chart above is the SWR curve of the 2el vertical array simulated at the feedpoint of the radiator

And this chart above shows the simulated antenna impedance

Impedance of the reflector

I also measured the impedance of the reflector (while have the radiator disconnected).

The chart above shows the SWR and impedance measured without the roll-inductor. The self resonance of the reflector without roll inductor is located at 1.87 MHz.Unfortunately, the antenna impedance is 25 Ohm, which means that the earth screen of the reflector is still quite lossy.

In the chart above, the roll inductor (2,6µHenry) make the reflector resonant at 1.81 MHz, approximately 20kHz below the radiator.

Conclusions

Unfortunately we couldn’t verify yet the desired antenna performance. It is not clear what is the reason for the low front / back. Two possible reasons could be:

- Too little coupling between the two antennas

- Too much looses on the reflectors ground network. More radials must be installed in order to reduce the losses

- The 23m tall tower behind the reflector is resonant on approx 1.8MHz and distorting the antenna pattern.

… investigations to be continued

In the pictures below, the measurement setup can be seen. I’m using a high precision USB Network Analyzer made by DG8SAQ (90dB dynamic range), a Mini-Circuits BNC Calibration kit and some Mini-circuits attenuators.

A few notes on measuring amplifiers:

• When measuring active components on a network analyzer DC must not be applied at any time at the Network Anlyzer’s input terminals. Just with a few Volts you can brick your (expensive) measurement devices. A good way to be make sure than DC reaches the VNWA’s terminals is to use a coupling capacitor.
• Amplifiers can easily overload your VNWAs input terminals. Caution is needed, especially if the gain is unknown, as in my case. My VNWA allows me to reduce the output power in 0.1 dB steps, therefor I started with -40dBm and added a 20dB attenuator at the receiving port, just to make sure that no overloading will happen.

Measurement setup – Regulated Powersupply (with current limiter), USB Vector Network Analyzer and the W7IUV amplifier

A closer look in the VNWA and the W7IUV Beverage Amplifier

A detailed picture of W7IUV Beverage Pre-Amplifier. Note the 20dB attenuator at the output port

The amplifier shows a satisfactory behaviour. Over the whole shortwave band  (1MHz … 30MHz) the gain (S21) is 20dB, decreasing to approx. 8dB at 100MHz. Also the input impedance (S11) is good from 1 MHz to 30MHz.

In overall I’m happy with the gain and the input impedance. Of course a few other important measurements (e.g. IP3) are missing. Since I don’t have the necessary means to measure the IP3 I’ll accept the amplifier as is. For the contest I’ll put the Amplifier behind a Lowpass / Bandpass filter just to make sure that no unnecessary intermodulation products are generated.

]]> http://www.dh1tw.de/w7iuv-beverage-pre-amp-2n5109/feed 2 _IGP8042 w7iuv Beverage amplifier W7IUV Beverage Preamp Messung DK1NO Beverage Preamp 100MHz http://www.dh1tw.de/w7iuv-beverage-pre-amp-2n5109?utm_source=rss&utm_medium=rss&utm_campaign=w7iuv-beverage-pre-amp-2n5109 http://feedproxy.google.com/~r/dh1tw/Mnut/~3/3rIsSPznu4c/2el-vertical-array-for-160m http://www.dh1tw.de/2el-vertical-array-for-160m#comments Tue, 10 Jan 2012 06:40:18 +0000 Tobias http://www.dh1tw.de/?p=1619 In CQWW 160m we are planning again a serious participation. This time we want to errect a 2el vertical Array for 160m towards the US. In this post you will find some detailes regarding our unique situation and some design thoughts.

ED1R Contesting site

Even while having more space than the average ham, at ED1R we are restricted when it comes to lowband antennas. Our 80m and 160m antennas have to be installed the day before the contest and be removed the night after the contest. Fortunately the friendly neighbors allow us to use their  fields during the weekends. Here is a 3D model of the ED1R contest station. Note the two (brown) areas which mark the fields we can use for our 80m / 160m antennas.

Hight is everything

When it comes to 160m, vertical antennas are hard to beat. During the last year we used sucessfully (subjectively measured) a 15 tall inverted L-Antenna. The L-Antenna is a poor man’s T-Antenna. The reason is because the L-Antenna has a rather significant high angle radiation which is usually not desired. On the other hand, the equally long horizontal wires of a T-antenna cancel effectively the high angle radiation.

On 160m a Lambda/4 radiation is almost 40m tall. This results in an antenna radiation resistance of 36 Ohm. Unfortunately it is impossible for us to errect a 40m tall antenna. The maximum height is defined by our 18m tall Spiderbeam poles. The main problem with verticals lower than Lamdba/4 is that the antenna radiation resistance decreases. With a low antenna impedance it is extremely important to have an excellent ground (radial) net. Otherwise, most of the power will be lost in the earth.

There are several ways to make a 18m tall antenna resonant on 160m. Here are some of the more popular designs:

1. Adding an inductor at the feedpoint
2. Extending the antenna with a horizontal wire (L-antenna)
3. Adding a slooping T-hat
4. Adding a horizontal T-hat

No Pain, no gain

Antenna pattern

Comments welcome

 Read on for pictures, online log & QSL information.

Operation

During my stay I tried to work on all bands from 10m down to 80m. I had great pileups on 15m and 20m from JA and the US. In the evening I enjoyed very much the operation on 80m CW. During the time of the European sunrise (06:00Z – 08:00Z) I worked EU, NA, JA and ZL at the same time!

Online Log & LoTW

Thanks to everyone who gave me a call. The check your callsign in the online log below. The log will soon be submitted to LOTW as well.

QSL Cards

In case you are interested in a QSL card please send it to my address or through the buro via my German Callsign DH1TW.

Pictures

Here are a few pictures of the operation:

Albertos great Station with Monoband antennas from 10m to 160m. Usually the group of HD2A operate from the station. If you hear them the next time, give them a call!

DH1TW operating CW on 80m with a Kenwood TS-450 and a SB-220.

My host Alberto, HC2AQ (right) and me (left)

Disclaimer & WARNING

WARNING! Do not perform ANY work on your amplifier if you are not exactly sure what you are doing. The amplifier is using high voltages up to 3000V which is LETHAL. Touching the wrong part inside the amp can KILL YOU. Again – if you are not absolutely sure, contact RF-concepts or your local dealer. Perform everything which is written here under your own responsibility. It has worked for me, it might work for you, but it might also not work for you. This article claims no professional advice.

AGAIN: With the slightest doubts, please don’t experiment with high voltages. Look for professional help from your next dealer.

The broken amplifier

While I couldn’t figure out the root cause of the tube’s failure, the symptoms where quite obvious.

The most likely theory is that one of the tubes was damaged during the shipment. The broken tube then caused the remaining good tube to overheat and finally burn out during the commissioning tests.

During the comissioning, I noted three strange behaviors of the Alpha91b (across all bands):

1. The tuning peak became extremely sharp; slight readjustments of tune/load would cause the amp to go into error-modus
2. The tuning meter wasn’t really useable; Tuning had to be performed with the max-output indicator; Once max output was reached, the tuning LED would always remain 3-4 LEDs right to the center.
3. Max output was limited to approx. 1000 Watts

Damage Assessment

Initial measurements

Ordering a matched pair of GU74b / 4CX800

Gettering / Conditioning GU74b / 4CX800 tubes

Schematic of Alpha91b power supply board (Heating is directly connecting to J3)

Alpha91b Power supply board without any modifications

Alpha91b Power Supply board with CX5AO method to only apply heating voltage. HV connector is disconnceted!

Alpha91b Power Supply Board - CX5AO method - different view

New Chimney

Applying HV & power

Thanks to…

More resources