CIRCUIT DIODE SWITCHED SYNTHESISER

Now I shall introduce my somple diode counter. To start off with, the MC145152-2 A-counter can have a count of 512, 1024, or 2048. This means that a 12.8MHz crystal, divided by 1024, will give you a reference frequency of 12.5kHz. You could alternatively use 6.4MHz divided by 512. Not many options there! You still need to buy a special crystal.

In my counter you can divide by ANY number from 2 to 2047. I put my hand in my junk-box and fished out a 5075kHz crystal. I can get 25kHz from that by dividing by 203. Of course, I can also adjust the oscillator a little if the frequency is not exact. 203 (decimal) is 11001011 in binary. Add leading zeros to get 11 digits, and we have a binary count of 00011001011.

IC1 is a CD4001B, quad NOR gates. IC1a is the crystal oscillator, but you can add a few turns of wire to pull the crystal a few kHz. You can also increase the 33pf capacitors to drag the frequency down a bit. The capacitor connected to the 100K/2K2 junction can be increased to 100pf, if necessary. You can also add a small inductor in series with the crystal. IC1b is just a buffer.

The 22K resistor is a pull-up to Vcc (+8v) and is coupled to the Reset (R) input of the counter. The counter delivers a binary output from 12 pins, but I have only used eleven of them. If any output has a logic “0″ (low voltage) then this will sink the bottom end of the 22K resistor to ground, through its respective diode. If you pull out a link, then that output can be forgotten: it is only the “1″s that count. If the binary 00000001011 were programmed, then the 22K will not pull up to +8v until the count reaches 11 (decimal) (000000 + 8 + 0 + 2 + 1 = 11). In this photograph I have programmed 203 (00011001011).

The 22K is not connected directly to the counter reset input. The capacitance of the PCB, and the 22K resistor make this pulse a rising sawtooth, with a rapid return to ground. The pulse would be far too short to feed any external circuits, so I have added and “IS” gate (two series-connected “NOT” gates), (or if NAND is to AND,

This circuit will later be used as the heart of my synthesiser, but it demonstrates how to make a divide-by-N stage using normal un-sexy, garden-variety CMOS. You can use this to generate 1750Hz to access repeater, using any-old-crystal

The first diode matrix resets the counter, exactly as I showed in the previous sub-project. The second matrix detects the A-count, and sets a latch. The Q output of the latch will change low-to high when the A-count has been reached. The NOT-Q output is the one we want. It will be used to set the prescaler “Modulus Control” line to “1″ so the count is 33. After the A-count has been reached the NOT-Q output will change to “0″ and set the prescaler back to divide by 32, and stay that way until the N-counter has been reset.

At first glance, a basic problem with my CMOS/Diode counter technique is that the program count will be detected several times. A count of 18 (00000010010) for example, will also occur at 00000110010, 00001010010, 00001110010, 00010010010, and so on. In this application this does not matter.

Here is a practical circuit of the A and N counters, using cheap CMOS and a heap of diodes. The “Out” pin is the Modulus Control pin to the prescaler chip. The A count can never equal or exceed the N count, and should never exceed the divide rate of the prescaler. So I have only used seven diodes, which allows an A count of up to 127. This should just about cover all the prescalers chips that are left in the market place.