Need help with automotive tacho converter

[Fity]

Hi All,

I have just replace the engine in my car from 4 cylinders to 6 and I would like to get the tacho to work properly.

The 4cyl engine had a hall effect sensor with 4 pulses per crank rotation. The 6 cyl is a reluctor type I think and produces 30hz @ 600rpm, 60hz @ 1200rpm and 90hz @ 1800rpm.

Can anyone make a suggestion on what would be a good way to convert the signal?

Thanks

 

[jpanhalt]

Are the pulses the same shape, magnitude, and duty cycle? Can you confirm that the indicated reading is based on frequency and not something else like duty cycle?

If it is a simple 4:3 frequency conversion, that can be done analog using using two frequency to voltage (F/V) converters. Simply do and F/V conversion then set the other chip to do V/F conversion with scaling. There have been several projects here and elsewhere in the past 2 years that did that.

If the conversion is exactly 4:3, then you could multiply the incoming frequency by 4 (2 doublings) followed by a divide by 3, which can be done with a couple of flip-flops as shown here (https://www.falstad.com/circuit/e-divideby3.html) or by a modulo 3 counter (3 counts in = one count out).

If you can narrow down the uncertainties, I (and others) can give you more detailed links.

John

 

[Fity]

Thanks for the reply John.

The tacho is expecting a square wave signal 0 to 5 volts. The new motor produces a sine wave signal. I did not check the voltage, just the frequency.

Is a F/V to V/F conversion the best? How did you you come to 4:3?

Thanks.

 

[Fity]

I should add that I don't think the duty cycle matters to much, the tacho counts pulses I think., but I could be wrong.

 

[jpanhalt]

F/V and V/F works and is analog. The projects mentioned actually used it for fractional changes to the speedometer caused by changing tire sizes.

In your case, I think digital will be easier, as the ratio is exact. Converting sine to square is not a problem.

At 1200 rpm, your 4-banger puts out 4800 pulses per minute (i.e., 4 per revolution). That is 4800/60 = 80 Hz. Your 6 cylinder puts out 60 Hz. So, 80/60 = 1 1/3 or 4/3.

I have to run now and won't be back for about 5 or 6 hours.

John

 

[Fity]

Thanks again John. I can't view that link on my phone, I will look at it on my pc later. I was thinking that a digital speedo corrector might work.

I was thinking I would need to reduce the pulses not increase them. Good thing I asked the question.

Troy

 

[Fity]

I found mention of a F/V to V/F for frequency multiplier/dividers on this application note.

https://www.electro-tech-online.com/custompdfs/2012/04/00795a.pdf (page 11)

Is that the sort of thing you would recommend?

The TC9400 data sheet shows sample circuits for the F/V and V/F, but none on how to feed on into the other. Would it be as simple as joining output to input with the resistor dividers as in the application notes above.

https://www.electro-tech-online.com/custompdfs/2012/04/21483d.pdf

Troy

 

[jpanhalt]

Here is a link to the article I mentioned earlier: https://forum.allaboutcircuits.com/showthread.php?t=48943&page=9
It is multi-page, but on that page the OP presents his almost final design. You can read through it for more background. I believe the final device worked well for him.

I do not have experience with the chips you mention. My experience is just with KA331, NJM4151, and RC4152 -- all of which are quite similar. I suspect almost any F/V chip will offer similar functionality.

I would also recommend reviewing the LM2907/lm2917 chip's datasheet and a related application note here.

You seem to have decided on the analog approach. If that is so, why do you prefer it over a digital solution?

John

 

[Fity]

The TC9400 was just one of the first hits I got from google. The LM2907 was another option I found, but I had not looked at yet. I will look into the KA331.

I have not decided to use analog. I am open for any suggestions. I am all for ease with accuracy.

 

[alec_t]

The tacho is expecting a square wave signal 0 to 5 volts.

If the tacho is just counting the square wave pulses and isn't fussy over their spacing being uniform then once you've converted the sine wave from the sensor into a square wave perhaps this would do the job.
The circuit uses an XNOR gate to generate a pulse for each edge of the incoming square wave. The counter and logic remove every third and sixth pulse from the resulting pulse train. So for every 3 incoming pulses six are generated but then 2 are eliminated leaving 4.

 

[Fity]

Thanks Alec, I am guessing that the tacho is not fussy over the spacing of the square waves. I have no way to test such a signal, but as it does expect the frequency to vary with the revs I can only assume that the pulse burst will be ok.

The circuit that you designed requires a square wave input, how would I go about changing from sine to square?

Troy

 

[alec_t]

how would I go about changing from sine to square?

Depends on the sine wave voltage. It might be as simple as rectifying and clipping, or you may need more advanced signal conditioning (e.g. amplification).

Edit: BTW the IC count can be reduced in my circuit by using an XNOR gate from the 4077 package as an inverter, in place of U3d.