Turntable Speed Control

Simplicity is good for reliability. It's neater. I like tidy circuits.

The 81 Hz simply to do 45 rpm on 60Hz mechanisms - I mean, who plays 45's? (OK, a DJ perhaps) - does not fit the words simplicity, or neat, or tidy.

It will take 6 16-pin and 14-pin integrated circuits to divide the crystal clock frequency by the required 30340.

The 67.5 Hz required to do 45 rpm on a 50Hz mechanism is only one IC fewer. Whereas to obtain 50Hz and 60Hz for 33.1/3 rpm on either mechanism requires just 3 each.

But just think of the accuracy! That's what I keep telling myself.

Now, the accuracy also depends on everything being glitch free. And what seems easy isn't easy when you start to consider glitches. Logic circuits can seriously misbehave because of propagation delays, and some ICs are far from perfect. Take the CD4017 decade counter for example: a great solution for dividing by 10, except when you need to divide by 10,000 where 4 are required. Inside it is packed with decoding circuitry to turn binary counts into decimal, and these can glitch because of their complexity. Cascade 4 of them and think of the result just one glitch will bring?

Therefore it is better to count in binary coded decimal which removes all that decoding circuitry, but that's much harder to get ones head around. Worth the effort though.

There are quite a few albums recorded at 45 rpm these days, so it would seem necessary to include it. Surely 50/60Hz would also be an either/or choice, could it be as simple as just changing between pairs of crystals?

30340/2 = 15170/2 = 7585/5 = 1517/37 = 41/41 = 1

Neat isn't it?

And 2458000 divided by 30340 = 80.015 Hz.

Bingo!

So all we need is a 2.458 MHz crystal oscillator followed with divide by 41; divide by 37; divide by 5; divide by 2; and another divide by 2, and we get 80.015 Hz?

Well maybe... it depends on the counters, or more precisely the width of their reset pulses, which may not allow sufficient dwell on their last count outputs - i.e. 41; 37; and 5. Yes, they count the complete integers, but don't keep their last output high for long enough - that's what is meant by dwell - like how long do you dwell on a thought?

Therefore although the logic is doing 41 or 37 or 5, there is error in that the count falls short of the required time by a tiny fraction.

So now we try a count of say 38 instead of 37 to try and lengthen the time a little, but find it isn't quite sufficient.

The result, which is quite logical, provided the above logic is understood, is 81.49 Hz. It is 0.6% out. The electrical mains fares better at 0.4%. However, unlike the mains where the frequency can increase or decrease by 0.4%, the 0.6% error is rock-solid.

However, playing a 45 rpm record at a constant 45.3 rpm might not be what people want. The pitch will obviously be wrong. I need better logic!

For the technical among us, I used CD4518BC binary coded decimal counters for the 41, 37 and 5 counts. Now looking for other suitable counters.

How are you going to adjust the motor/turntable speed ( for bearing oil drag?)
Kr.