Menu:



Version:

November 3, 2014:
Revised: v1.2

Converting a Clock to a Frequency Counter

Maplin's LED clock kit

I thought converting a Maplin LED clock kitset to a frequency counter would be simple, fast and tidy. The final outcome was anything but that.

Introduction

Each time I travel outside Saudi Arabia where I live and work these days, I try to visit a few electronics component shops. I buy parts for upcoming projects, and I often get more ideas for new projects as I browse the shelves.

On a visit to the UK, I visited one of the Maplin stores. Depending on the store, they can be great places to browse and buy, or they can be a total waste of time with only a few odds and ends to be found in an overstocked pile of consumer chaff.

Kit corner at a typical storeAs I walked around this time, buying a few things I needed (or thought I might need), I came across a 4-digit LED alarm clock kit. An assembled clock board is shown at the top of the page. The instant I saw that the kit used an Atmel ATtiny2313 microcontroller, it sparked an idea.

At the time, I was building a new transceiver, testing some bidirectional amplifier stages. I had just completed a permeability-tuned VFO for it. (It’s described on another page on this website) I have written some frequency counter software in the past for this chip, and I thought I might be able to quickly convert the clock kit into a compact 4-digit LED frequency counter and reuse that old code.
 
Just how hard could it be? Probably take just a few hours. Whip it all up in a jiffy. This documents the result. Oh, it all worked, no problem there. It's just....Well, you'll see.

Multiplexing with LED Displays

Displaying numbers on a multi-digit 7-segment LED display is usually done using multiplexing. This method reduces the number of pins required on the microprocessor. To do this, all of the same segments in each LED digit are connected to a common output line on the microprocessor. Then, each LED digit is turned on separately and sequentially.  If this process is done quickly enough, it appears as if all of the display digits are individually driven.

In this case, this reduces the required pin count from 32 (to select 7 segments plus one decimal point independently per digit x 4 LED digits) to just 12 (to select 7 segments plus one decimal point connected in parallel plus 4 LED digit selection lines). Perhaps the diagram below will make this a bit clearer. (Another possible approach which reduces the required pin count even further is “charlieplexing” but it’s a little more complex)

LED multiplexing using common cathode LED displays

In almost every LED counter and clock design you can find, the eight segment lines (7 for the digit plus one for the decimal point in each digit) are driven from pins on a single 8-bit port on the microprocessor. My earlier software was written for this logical approach to segment connections. However, as I discovered when I opened the kit and explored this design in more detail, the Maplin clock was arranged completely differently! In this design, the display segments were each connected to random ports and pins scattered around the ATtiny2313. This was probably done to simplify the PCB layout but it makes the software significantly more complex.

Here’s the circuit diagram of the Maplin 4-digit frequency counter. Incidentally, the circuit diagram was not included in the kit instructions, unfortunately, and I couldn’t find a circuit diagram for this counter online either. I ended up drawing the schematic shown here based on my analysis of the kit. Of course, a week or so later, I stumbled across the circuit diagram online. Anyway, here is my version:

Circuit diagram of original Maplin LED clock
Figure 1 : Schematic of Maplin LED clock kitset

Aside from the strange method used to connect the LED segment connections across two different ports on the microprocessor, the other thing I noticed was that the LED display was not a generic 4-digit display. Instead, it was one specifically designed for clocks. Well, that’s hardly any surprise, I suppose, but it did present an issue for me. I had assumed each digit would have its own decimal point. That increases the flexibility of the frequency display format. Instead, the options were immediately reduced because the display only had a single inter-digit colon in the centre of the display. Sigh.

Constructing the Clock Kit

The next step was to build the kit. This was the easiest way to check that all the parts worked. It was a very easy kit to build, as expected, with a very easy to follow PCB layout. When I powered it up, everything worked just fine.

And then, of course, I started to pull it apart to start the modifications.

Hardware Modifications

Here is the schematic of the frequency counter. If you compare this with the clock circuit diagram above, you can see the various modifications required.
Hardware changes for frequency counter operation
Figure 2 : Schematic shows the revised design and changes required to implement the frequency counter

These changes included:

The RF input required is relatively high, at least 2.5Vpp. My VFO had a much lower level than this, so I added a small single transistor preamplifier. The schematic is shown opposite. The components were mounted under the board.

I also moved the 5V regulator to the underside of the board. This will allow me to bolt it to a solid heatsink (the chassis of the transceiver) to keep it cool during operation. The counter draws 70mA so there is a fair amount of heat given off during operation.

Given the number of modifications required to this clock kit, I’m certain I would have been better off just building a counter from scratch. However, you don't know that at the start of a project. In any case, by documenting this design here and providing the software, maybe someone else will find some part of it useful.

Software

I had intended to just reuse some code I’d previously written. Like the hardware, it turned out to require more changes than I'd expected, largely due to those strange segment connections. In the end, I was able to reuse some software written by others, and add some new code myself.
 

Counter operating

As usual, the source code is provided below, just in case you want to add new features or change something. The Hex file (the output file from the Atmel compiler, and the file you need if you are just going to program your ATtiny2313 directly) is also available for download at the bottom of this webpage.

Aside from simply counting the frequency, the software also supports a Push-to-talk (PTT) input to detect if the transceiver is in receive or transmit mode. In some cases, the VFO frequency is offset, usually by the IF frequency, in just one of these modes and this must be factored into the calculations before the frequency is displayed. In my case, I didn’t have an Tx/Rx VFO offset so the software simply reads the PTT input and reflects its state on the board’s large green LED.

For my transceiver, I also had to allow for an IF offset of 8.6436 MHz. My VFO is being used to tune across the 20m amateur band (14.0 to 14.5 MHz). Because my VFO produces an output 8.6436 MHz below the transceiver’s operating frequency, the counter software must add this IF offset to the frequency it measures so it can display the correct frequency.

Let’s give an example. If the VFO frequency measured by the counter is 5.4434 MHz then the counter software must then ADD the IF offset (8.6436 MHz) to this value. In this case, the display will then read 14.0870 MHz which is where the transceiver is actually tuned. Actually, the four digits on this counter (in the usual “kHz” mode) will display the last four digits of this frequency i.e. 0870,  and in the alternate “MHz” display mode, the LEDs will display 1408.

RF preamp






A
small preamp was added due
to the low RF level from my VFO
(Schematic above)


Two interrupt routines are operating in the background. These use the two ATtiny2313 internal counter/timers. Counter/Timer 1 (C/T1) is configured as a counter, and counts the input frequency. It is incremented each time a falling edge is detected on the frequency input. Counter/Timer 0 is configured as a 100mS timer. This starts and stops C/T1, and the same routine clears the counted value for the next count. This timer limits counter resolution to 10Hz, but that’s ten times the resolution of the display, so that’s more than enough.

One of the existing momentary pushbuttons is used to select the “MHz/kHz” display format. The display format is normally (and at power-up) set to XXX.X, the last digit being the 'hundreds of Hz' digit. If the MHz/kHz button is pressed, the display format changes to XX.XX where the last digit is the 'tens of kHz' digit. Of course, since we have no digit-specific decimal points with this kit’s display, you just have to imagine those on the display. Pressing the button again will revert to the original display format.

Programming the ATtiny2313

I had made yet another assumption that the clock software programmed on the ATtiny2313 could be simply erased and the chip reprogrammed with my new software. No, this kit was going to fight me every step of the way – The chip had its programming interface disabled when the clock software was programmed, to prevent copying of the code. That was annoying.

However, if you’ve done any significant programming of AVR chips, you’ll have already have encountered similar problems, especially if you program the fuses in the processor to allow the reset pin to be used as an extra I/O pin, or if you’ve accidentally select the incorrect clock settings. The microprocessor just acts like a brick and won’t accept any new program. Just like this chip was doing.

Fortunately, the solution was  sitting right on my bench. I built this Fusebit Doctor some time ago for several other projects where I was using the reset pin for I/O. It took me just seconds to unbrick the ATtiny2313. A few seconds more and the ATtiny2313 was programmed with the new frequency counter software.

Perfect.

Prototype Construction

The photo shown here says it all. It’s rough, and hardly pretty, but that was all I needed to get it going. The two short wires hanging off the board are for temporary connection to clip leads from my bench power supply. There was a DC power socket supplied in the kit, but I didn't bother to install it.

Modifications can be seen under the board. Not pretty.

If I end up using this counter in the final version of my transceiver, the board will be directly wired to the transceiver power supply rail, and the counter will be mounted deep inside the chassis. That will ensure that all of this ugly stuff will be completely out of sight.

That's probably just as well.



Underside of the modified Maplin clock PCB shows the extent of modifications required.


Downloads

Software: Four_Count.zip contains both Hex and source code (*.asm) files



Credits

Thanks must go to Andy Palm N1KSN whose software, originally used in his 80m receiver, lies at the heart of this counter. Significant portions of the original software have been changed, with some unnecessary code for this application removed and other new code added by me. Don’t blame Andy for this code not working to your expectations. Without any doubt, that’ll be my fault…

And credit must also go to Maplin and to Middlesex University Teaching Resources in the UK for the original clock and kitset design, now decidedly hacked about by my efforts here. For that, my apologies.






Want to go back to the main page? Click here to return directly.