Search This Blog

Monday, 20 February 2017

Really? Wow!


How about this then?

I seem to be having a phase of collecting old 'scopes. I kind of love them, especially the old Tektronix devices.

There's a You Tube channel that I subscribe to by W2AEW where he does loads of fab and groovy things in general, but when I saw this, I just had to have one!

Using an Arduino Uno, a very primitive resistor based DAC and a bit of code, we end up with this, here on the scope I just repaired, a 2465:

Just fantastic!

This is the W2AEW actual video itself including the links to the code:

The Arduino and the DAC look like this on my bench:

and here running on an old 465 'scope:

Local conditions.

Wednesday, 8 February 2017

Tek 2465 Teardown


I have a Tektronics 2465 'scope that has some issues with it's timebase:

The sweep A is all out of bonk (that's a technical expression) - in the image above there's a 10MHz signal applied and at a 50ns sweep time per division we should see 5 sinewaves - we see 8 :-(

Sweep B is correct at 20ns however at 20ns we see two complete transitions of the signal.

I've started to take the 'scope to bits and will use this blog entry to document things as I go. Here are the initial shots of the 'scope in bits:

So here you can see both U700 and U900 removed, they are the main hybrid components in the Sweep A and B circuit. I'm going to try the 'scope with these two swapped round as they are the same component but for Sweep A and B:

So, swapping the two hybrid devices U700 and U900 made no difference; I now need to focus initially on this bit of the circuit:

I've also looked in more detail at the exact problem that I have. It seems all sweep speeds EXCEPT 5ns are impacted. I've annotated two points in the diagram above and have taken the voltage measurement at these points at all sweep speeds:

You will also note the consistency in the delta at all sweep speeds above 5ns - this is the same right up to 500ms but I just got bored writing the same thing over and over.

So, why is the voltage at the first test point so dramatically different at 5ns (the only sweep speed working) than all the others. Could it be that the B Sweep hybrid actually does the 5ns sweep on A? The B sweep can never be slower than the A sweep so there is no need for both to implement 5ns - I don't know if this is true or not.

I pulled R724 and in the process the end snapped clean off - but it measured 35.55K (it should be 35.3K) so I dont think that will be the problem. However, I've had to order some bits as this is a very non standard resistor value.

I also had a look at C712 which I have replaced. I also need to check that the +42V line is as it should be.

Once I have the replacement bits I'll update this post with my findings.

Some more pics while I wait for the bits:

Now the bits have arrived I have replaced R724 (made from two resistors in series) plus C712 and BINGO - the fault has gone and the sweep speeds are back on the money. The unit needs calibration now :-(

Final update: I've been through the step-by-step calibration routine. To access the routine you set a jumper on the control board and then press the delta T and V buttons together and whilst holding them down stick a chicken down your trousers (actually you press the slope button). Then you go step by step through the calibration routines and using the other test gear I have hear I seem to now have the 'scope very much on the money.

Local conditions.

Tuesday, 7 February 2017

A Lawful what?


Back in the spring of last year we removed a load of trees from our property and also had some fairly extensive works undertaken in the garden. The net result was that the antennas at the property became much more visible from the road.

Now, one of my neighbors who is particularly grumpy, decided to question the legality of the mast(s) and antenna(s) at the house with our local council.

In accordance with the planning laws, they all should be subject to planning permission. However, theory says that because they have been in place for in excess of four years the council are unable to take enforcement action. Unfortunately, the onus is on me to prove they have been in place for that time.

Thanks to this blog and my obsession with taking photographs, I was able to generate a very comprehensive document containing a load of google dated photographs stored in the cloud showing each of the antennas and masts in place with a photo upload date.

Finally, after a long bureaucratic process, my Lawful Development Certificate has been issued.

So my grumpy neighbor can go stick his complaining!

Local conditions. 

Wednesday, 18 January 2017

Fiddling with a Transistor


I found myself scratching my head the other day trying to remember the difference between Class A, B and C amplifiers and how that impacted a transistor bias.

To help me, I've drawn this quick schematic:

Now, you will see that R1 has a "*" next to it; that's because we are going to play with this value.

The transistor base will switch on when the voltage on the base is greater than 0.6V.

So, if we place a 10K resistor where R1 is, we place the base at DC voltage of about 1.6V. If we now apply a 2V peak to peak signal to the base (in my example it's a 100KHz signal), then the lowest point of the signal the base will be at 1.6V minus half the voltage peak to peak = 0.6V. Therefore the transistor will be switched on for the full input signal cycle.

The voltages on TP1 are in yellow and TP2 in blue. Here the first example with 10K as R1:

You can see that all of the input signal is above the dotted line (which is at 0.6V) and therefore the transistor is biased on throughout - this is Class A operation. Harmonically it looks like this:

Now, lets change R1 for 39K which will place the base at DC voltage 0.6V and apply the same 2V peak to peak signal. This time the transistor is only switched on for about half the cycle and the resultant voltages look like this:

Here we have 180 degrees of the input signal turning on the transistor - this is Class B operation. Harmonically it looks like this:

Finally, if we remove R1 completely, then the base of our transistor is at DC ground (being pulled low by the 4K7 resistor). If we now apply our signal we see this:

So the transistor is only on for less than 180 degrees of our input signal - this is Class C operation.

Harmonically it looks like this:

Now, you may wonder what use Class B and C are? Well, in an audio amplifier - none at all! You would hear all that distortion and harmonic content and it would sound completely awful. However, from an RF perspective, we can easily remove the harmonics from the output and retain just the target frequency - they all seem pretty much the same now - agree? And Class B and C amplifiers consume much less power (because the transistors aren't turned on the whole time) so are far more efficient. We just need a suitable low pass filter at the output and we can "reconstruct" our signal.

Interesting, egh?

Sunday, 15 January 2017

Have you Really? Wow!


This isn't really radio related, but I'm scribbling here to document my steps (well, thats what a blog is for - isn't it?).

At home we have a really pants copper telephone line based Internet connection provided by Sky. Now, nothing Sky can do really as its the infrastructure that makes the connection speeds so slow. We have also been having issues with Wireless speeds generally and also often hit a "bonkers" (as far as I can work out) restriction on the number of devices that can connect to my network. For some reason the DHCP works fine up to 16 devices, but as soon as device 17 tries to connect it fails.

Often this will be my Spectrum Analyser but also it happens to be the visiting sprogs laptop - all very inconvenient.

So today I've bought a TP-Link router with 5GHz and 2.4GHz wi-fi to replace the Sky provided box. The man in the shop told me this was impossible, as Sky will not release the username and password needed to connect to their network.

So, hacking hat on, here's how I've done it:
  1. First, I downloaded some software called Wire Shark from here
  2. Then I installed the software on a PC connected with a wired connection to my Sky router.
  3. Start Wire Shark recording network traffic and then re-boot the router so we capture the network traffic as the router starts up.
  4. Once re-booted, stop the software capture.
  5. use the filter in the Wire Shark software to look only at "udp.port == 67"
  6. In the search results look for "DHCP Discover" traffic and expand
  7. In the expanded area you will find "Option (61) Client Identifier"
  8. Right click and find the option to copy as printable text
  9. Paste the results into a text editor (I used Notepad++)
  10. Low and indeed behold, there is your Sky username and password in the format "=username|password"
I then quickly connected to the router and noted down the MAC address from the "Broadband Port" section of the config screens.

So, armed with this information I plugged in my new shiny TP-Link router, selected "Sky Broadband" as the ISP and entered my username and password. Bingo! I then checked the help on the router and changed (or rather cloned) the MAC address of the Sky router - this is so that from an external perspective you can't tell that the router has been changed.

Result: Much faster wi-fi and no restriction on the number of devices I can connect.

Local conditions.

AVR dude - really?


Over the past few months or so I have been playing with kits from QRP Labs based on the genius of Mr Hans Summers.

I was using the VFO I made back here to clock my newly built Signal Generator from here and found some rather odd behaviour.

When I investigated this I found that the VFO output switched off every second:

I posted on the QRP Labs forum and very soon had a reply telling me this was an issue that had been fixed in a later version of the firmware.

So now I needed to re-program the AVR processor using the hax file available - I hadnt done this before.

So, here you will find that you can use an Arduino as an AVR Programmer - now I have lots of those. On the Arduino site you can see how to configure the processor on a breadboard:

I added a 7805 regulator and powered the breadboard contents from my local bench PSU. I connected the grounds together of the Arduino and breadboard, but not the power lines! So the Arduino is powered from the USB cable and the breadboard separately.

So assuming you already have an Arduino IDE installed and working here are the steps:

  1. Start Arduino IDE and load the ArduinoISP sketch and upload to your Arduino (the sketch is in the examples folder and my board is an Uno)
  2. Wire the Arduino board as the above diagram - I used a 20MHz crystal and 2 22p capacitors
  3. Download avrdude from here (I put my download into a local directory C:\Users\Mark\Downloads\QRP VFO)
  4.  Download the hex file you want to program from the QRPLabs forum files area 
  5. Open a command window in Windows (type cmd in the search box)
  6. execute the command "cd C:\Users\Mark\Downloads\QRP VFO" (replace my target with your local directory)
  7. then we need two AVRDude commands (replace COM13 with your Arduino COM port):
    1. To set the fuses: avrdude -P COM13 -b 19200 -c arduino -p m328 -v -e -U efuse:w:0x04:m -U hfuse:w:0xDF:m -U lfuse:w:0xF7:m328
    2. To program the flash: avrdude -P COM13 -b 19200 -c arduino -p m328 -v -e -U flash:w:VFO.s1.02c.hex (replace with your hex file)
Here's what it all looks like on the bench (which is rather a mess):

Here is our very beautiful Miss Luna Cat who has been with us since July 2015, and is now, finally, letting me stroke her. She is asleep in the shack on the bag that came with my recently acquired frequency counter:

 Local conditions

Saturday, 31 December 2016

The VLNA, Oh Boy!


I think I have finished the build of the VLNA for 13cm; having constructed the project I am wishing I had bought a ready built and aligned unit. This was the hardest project to date, mainly because of the very tiny components. I really struggled to get the input matching network parts in place - and have no test gear to actually measure the return loss or noise figures of the final build.

It looks like this:

You can see the input network components in the image above; it kind of looks like the instructions from G4DDK @DXING, but I have no means to actualy measure or test the input return loss.

Lets see.