Next I added a 74HC390 dual decade counter to the circuit. This effectively is half of my 2-chip Frequency Counter. The crucial thing, is that there is a lot of similarity between the timebase of the 2-chip frequency counter, and that of the 2-chip Huff & Puff stabiliser! With a little care, the frequency counter and the Huff Puff stabiliser can be made to share the SAME timebase! This allows me to build a combined device using just 3 IC's. It is important to use a 32.000KHz crystal, the usual 32.768KHz watch crystals are NOT Ok!

The counter has two digits of Binary Coded Decimal (BCD), i.e. 8 LED's. These are much easier to read than 8 bits of binary as on the 2-chip Frequency Counter, unless you are very good at mental arithmetic. For added functionality, I decided to try to build the counter to have two resolution modes. First, 00 - 99KHz with 1KHz resolution. This is the "course" mode. Next, a "fine" resolution mode measuring 0.0 - 9.9KHz with 0.1KHz resolution.

The 0.0 - 9.9KHz mode presented some timing complications. To measure frequency with 100Hz resolution requires a counting time of 10mS. Therefore I had to somehow generate this period, without adding further IC's. To do this, I added an AND gate fashioned out of 3 diodes and a resistor, to the outputs of the 74HC4060. Effectively, they count mS and cause the counter to be prematurely reset when it reaches a count of 26. During the last 10mS of this 26mS cycle, the Q10 output (pin 15) will be high. This can be used to gate the 74HC390 frequency counter! During the first 16mS, I arrange for the output LED's to be lit, and for them to be extinguished during the 10mS counting period. This is easy: just connect Q10 (pin 15) of the 74HC4060 to the commons of the LED's. I actually do this via a series of four 1N4148 diodes such that the voltage drop across the LED's is about right.

The 74HC390 also needs to be reset prior to each new count. I do this with a simple RC differentiator, which generates a very short spike when the counting period starts. The duration of this spike is only of the order of 10 or 20nS so doesn't affect the counting at all. I make the time constant as short as possible, whilst still allowing correct resetting of the 74HC390. This was a matter of trial and error.

The new gating arrangement increases the frequency lock step slightly to 38.46Hz (1000 divided by 26).

The 00 - 99KHz mode is easy. All I do is connect the 74HC390 control signal (reset, gating, and display enable) to the 74HC4060's Q6 output (pin 4) which is oscillating at 500Hz, and naturally has the 1mS required high time already.



This design works very nicely. The VFO circuit is the same as the 2-chip stabiliser. A few points are worth of note:

1. The display does flicker sometimes - this is inherent of ANY frequency counter, they all have an uncertainty in the least significant place. Read my comments in my 2-chip Frequency Counter web page, which apply equally here.

2. Voltage regulation and smoothing are very important! I use a 10nF decoupling capacitor across each chip's supply (not shown). The VFO and crystal reference chip need their own separate +5V supply. Or very good filtering and decoupling of their supply. It is important that switching noise from the LED's does not find its way back into the VFO and destroy the stability. When I come to use this stabiliser in an actual project, I will pay particular attention to screening, supply filtering and decoupling, and so on. The importance of keeping noise out of the VFO cannot be overemphasised! I will post futher results here as they become known.

3. Use low-current LED's! I do not think an output of the 74HC4060 (e.g. Q10 pin 15) could source the current required for 8 ordinary LED's by itself. I use low current miniature LED's, the same kind I used for my Mk2 2-chip Frequency Counter. I measured the current consumption of the frequency meter IC as 6mA in the circuit as drawn. It is not as efficient as the 2-chip frequency counter, because the voltage drop is derived from a string of diodes in series, rather than by pure pulse modulation. The duty cycle is 50% in 00-99KHz mode, and a little more in 0.0-9.9KHz mode. One of those diodes can be removed, leaving three in series - that makes the display brighter and increases current consumption to about 12mA. Or a 5'th diode can be added, making the display quite a bit dimmer but decreasing current consumption of the display to about 2mA. I think that 4 diodes (6mA) is a good compromise and gives reasonable brightness.

Above, two photographs of the project. On the left hand photograph you can see a 7805 voltage regulator at the bottom left of the picture. Next to that an On/Off slide switch and red LED - note that this red LED is NOT the one in the VFO. I like to have a power indicator on my circuits when in development, to give some visual indication that power is applied. This helps me not to forget to switch the power off when working on the circuit. Particularly important on high-voltage valve equipment...

The tuning LED can be seen near the top right corner of the KANK3333 can, below the tuning knob. Of course it is not lit, since it is reverse biased.