Dear friends,
I would like to built a digital tachometer for my turntable. Not with a processor or stuff like that, but with "old fashioned IC's ( TTL or CMOS) and NIXIE tubes. Who Could Help me with a schema or article? Thank you.
I would like to built a digital tachometer for my turntable. Not with a processor or stuff like that, but with "old fashioned IC's ( TTL or CMOS) and NIXIE tubes. Who Could Help me with a schema or article? Thank you.
Difficult to do with any accuracy without using math (uP).
For high frequency signals, a counter is used with a defined gate period. The slower the input frequency, the longer the gate period becomes especially as the accuracy (to the right of the decimal point) increases. To measure a frequency of 60Hz using a 1 second gate to count pulses would produce a result 60, with 1 Hz resolution. To get 0.1Hz resolution would require a gate time of 10 Sec which produces a count of 600 and the decimal is put between the two zeros.
A turntable platter is slow moving; counting single pulses per revolution would have a gate time of 60 Sec for a resolution of 1RPM. If you use a rotary encoder to produce many pulses per rev, you will need one with thousands of pulses per rev, or do some math: An encoder with 6K PPR will give a count of 3333 with a 1 second gate with resolution of 0.01 RPM at 33 RPM and a count of 4500 with a 1 sec gate at 45 RPM:
Count=RPM*PPR*Gate/60 where gate is in seconds.
or
Gate=Count*60/(RPM*PPR)
If you have an encoder with 2560 PPR, your gate time would be 2.34375 Sec for both 33 & 45 RPM and produce counts of 3333 and 4500.
To get resolution 3 places to the right of the decimal would require an encoder with 10's of thousands of PPR or a gate time of 10's of seconds.
You could also preload a counter with 78333 and count down at 25K pulses/sec using 1 revolution as a gate. This will give you accurate results for 45.000 and 33.333, but inaccurate results the farther away you get from either of those 2 numbers. The 74C192 counter IC is a BCD ÷10 up/down counter with preset inputs and could be used for this application. 7447 BCD to 7 Seg decoders could drive the digits.
For high frequency signals, a counter is used with a defined gate period. The slower the input frequency, the longer the gate period becomes especially as the accuracy (to the right of the decimal point) increases. To measure a frequency of 60Hz using a 1 second gate to count pulses would produce a result 60, with 1 Hz resolution. To get 0.1Hz resolution would require a gate time of 10 Sec which produces a count of 600 and the decimal is put between the two zeros.
A turntable platter is slow moving; counting single pulses per revolution would have a gate time of 60 Sec for a resolution of 1RPM. If you use a rotary encoder to produce many pulses per rev, you will need one with thousands of pulses per rev, or do some math: An encoder with 6K PPR will give a count of 3333 with a 1 second gate with resolution of 0.01 RPM at 33 RPM and a count of 4500 with a 1 sec gate at 45 RPM:
Count=RPM*PPR*Gate/60 where gate is in seconds.
or
Gate=Count*60/(RPM*PPR)
If you have an encoder with 2560 PPR, your gate time would be 2.34375 Sec for both 33 & 45 RPM and produce counts of 3333 and 4500.
To get resolution 3 places to the right of the decimal would require an encoder with 10's of thousands of PPR or a gate time of 10's of seconds.
You could also preload a counter with 78333 and count down at 25K pulses/sec using 1 revolution as a gate. This will give you accurate results for 45.000 and 33.333, but inaccurate results the farther away you get from either of those 2 numbers. The 74C192 counter IC is a BCD ÷10 up/down counter with preset inputs and could be used for this application. 7447 BCD to 7 Seg decoders could drive the digits.
Last edited:
I have done it with just digital IC chips. You need a quality oscillator for each speed. You divide the frequency of the oscillator, which also divides the error, until you have a frequency that is .003% accurate. You used a hall effect sensor to trigger a one-shot which latches the counter so the speed count will display and also triggers another one-shot which clears the counter a microsecond later to begin the count again. The best table I have measured was showing a different speed each revolution at 33.332, 33.333, and 33.334. This is much more accurate than you can get with a microprocessor.
Start with a 1.5 Mhz oscillator with 50 PPM error, using 2 divide by 9 chips and you are dividing by 81 so you get a frequency of 18518.518518518 Hz, etc. a turntable using at 33 1/3 rpms takes exactly 1.8 seconds to revolve once, or should. That frequency times the 1.8 seconds gives you the 33,333.3333333333... speed. That should get you thinking.
So you are accumulating a count at a rate of ~18.5kHz over the period of one revolution using an oscillator with a stability of 50PPM?
The RR tach accumulates a count at a rate of 30kHz over the period of one revolution using an oscillator with a stability of 2.5PPM, so not only does it have almost twice the resolution, but the stability is 25x better. Dividing the 1.5MHz clock by 81 does not increase the stability, the 18.5kHz clock is still 50PPM, just as my 18MHz oscillator divided by 600 still produces a clock with 2.5PPM stability.
If you accumulate a 18518.518... Hz clock for 1.8 secs, it will produce a count of 33,333.3333 in 1.8 secs, so drop off the fractional part and put a decimal point where the comma is an viola, you have the RPM reading! Unfortunately, this is the only reading that is accurate. If the platter is turning at 33.4 RPM, the time for one rev is 1.7964 secs, the count will be 33,266.799 so your reading will be 33.266 instead of 33.4 RPM. 33.2 RPM takes 1.807 secs, the count will be 33,467.202 so your "tach" will show an increase in speed when the platter is actually slow and vice-versa.
When you accumulate a count like we are doing, you are not measuring RPM, you are measuring the period of the revolution which is the inverse of RPM. In order to convert that count to RPM, you must take the inverse (1/x), which gives you revs per second and multiply by 60 to give you RPM. Sorry, but there's no other way to do it.🙄
Is this how the PBN Audio tachometer works (I understand that you designed it for Peter)?
Last edited:
Hi Pyramid,
What you say is essentially correct. If the table is running slow it will show more "counts" and vice-versa. In practical use you want your table to run the correct speed so you adjust the speed up or down until it reads the correct speed, like 33.333. if the table deviates from this as the center count does it matter whether it is deviating up or down. And you are right that the PPM is the same but the actual error of being off the correct frequency is divided.
The main thing is anyone can build one with three speed for under $40 in you get some cheap PCBs from China, less for a single speed, and no programming required, all you need is a soldering iron.
What you say is essentially correct. If the table is running slow it will show more "counts" and vice-versa. In practical use you want your table to run the correct speed so you adjust the speed up or down until it reads the correct speed, like 33.333. if the table deviates from this as the center count does it matter whether it is deviating up or down. And you are right that the PPM is the same but the actual error of being off the correct frequency is divided.
The main thing is anyone can build one with three speed for under $40 in you get some cheap PCBs from China, less for a single speed, and no programming required, all you need is a soldering iron.
Well, that's certainly a novel approach to the problem, but I think the OP was looking for a measurement system rather than just an indicator of fast/slow; a strobe disc will give you that, and it doesn't read backwards.
The DIY Arduino project will give an accurate measurement of RPM, costs ~$50 and doesn't require any programming, you just download the provided script onto the Arduino with tools that are available on-line at no charge. Of course, the Arduino is a DIY project and not available for commercial use.
I was confused by your statement that your design was "much more accurate than you can get with a microprocessor", but I think we can agree that is not the case; even the Arduino app will produce a reading that is not only more accurate, but higher resolution than a hardware counter.
The down counter using 74192 ICs will produce accurate results at both 33 and 45 RPM with a single oscillator and the indication will not be reversed for slow/fast, but the accuracy will suffer the further from 33.333 and 45.000 the platter wanders.
The DIY Arduino project will give an accurate measurement of RPM, costs ~$50 and doesn't require any programming, you just download the provided script onto the Arduino with tools that are available on-line at no charge. Of course, the Arduino is a DIY project and not available for commercial use.
I was confused by your statement that your design was "much more accurate than you can get with a microprocessor", but I think we can agree that is not the case; even the Arduino app will produce a reading that is not only more accurate, but higher resolution than a hardware counter.
The down counter using 74192 ICs will produce accurate results at both 33 and 45 RPM with a single oscillator and the indication will not be reversed for slow/fast, but the accuracy will suffer the further from 33.333 and 45.000 the platter wanders.
John and Pyramid,
Thank you both for the replies. It gives me the hind to do further experience.
Jos
Hi Pyramid,
Having it count and show reverse of the speed is not perfect but so long as it is known and you understand it you can adjust the speed of the table accurately, does it bother me, not much. If it used a microprocessor you could read the count and display the actual speed. What is meant about being more accurate than the microprocessor is the speed of the processor has to be checked, I looked at three and right out of the box they were all running a slightly different speeds. You can definitely tweak them to be very accurate. If you are not using its clock for timing then it wouldn't matter. But I still think a 1.5 MHz oscillator with a 50 PPM error rate is plenty accurate, I don't think I can hear a .003% speed difference. And I am happy using my simple circuit so each to his own.
And good luck to JosM on your circuit, it is the playing with a design that is the fun part.
Having it count and show reverse of the speed is not perfect but so long as it is known and you understand it you can adjust the speed of the table accurately, does it bother me, not much. If it used a microprocessor you could read the count and display the actual speed. What is meant about being more accurate than the microprocessor is the speed of the processor has to be checked, I looked at three and right out of the box they were all running a slightly different speeds. You can definitely tweak them to be very accurate. If you are not using its clock for timing then it wouldn't matter. But I still think a 1.5 MHz oscillator with a 50 PPM error rate is plenty accurate, I don't think I can hear a .003% speed difference. And I am happy using my simple circuit so each to his own.
And good luck to JosM on your circuit, it is the playing with a design that is the fun part.
- Status
- Not open for further replies.