Nova Tachometer Calibration

Post Reply
User avatar
Spacenut
Posts: 1205
Joined: Thu Feb 20, 2014 9:43 pm
Location: Hampshire

Nova Tachometer Calibration

Post by Spacenut »

Image

So, you’ve built yourself a tachometer, or perhaps more realistically, bought an aftermarket item to fit to your car. How do you check the calibration is correct?

This was the question I was asking myself about the tachometer in the Green Machine. I built (some would argue cobbled together :D) the tachometer from two sets of plans, using the LM2917 precision tachometer IC and the LM3914 bargraph driver.

Image

The bargraph has 19 LEDs and is designed to show 8,000 rpm at full-scale. This means each of the 19 individual segments represents an engine speed increment of 400 rpm. The bargraph is driven by the input voltage from the LM2917, with full-scale deflection (FSD) at 2V.

The LM2917 application circuit I used (from good old Maplins, sadly no longer in business) has the following calibration formula:

Vout = Fin x Vcc x VR1 x C2

Where Fin is the input frequency (in this case taken from my ignition points), Vcc is the supply voltage (9V linear regulated), C2 is a fixed capacitor (0.01 microfarad) and VR1 is a variable resistor, which I determined to be 105 Mohm for this application.

So much for the theory, but how well does it work in practice? Well, up until yesterday I wasn’t too sure – the bargraph display worked well enough, and seemed to be giving the right sort of values, with 1 segment illuminated at idle (indicating 1,000 rpm) and peak power rpm (5,200 rpm according to the factory manual) illuminating 11 segments, or 5,000 rpm on my scale. But that is hardly an accurate calibration, and as I have started to explore the upper reaches of the rev range, I decided I needed a more objective calibration yardstick.

Where adjustment is provided, the usual aftermarket instruction would be to compare readings with a known, calibrated tachometer, which of course I don’t have.

As an aside, the instructions for calibrating the Dakota Digital speedometer were utterly ludicrous, advising you to drive behind another car traveling at a steady 30 mph while fiddling with the calibration switches on the back of the display! As you can imagine, I came up with a much safer, and ultimately more accurate means of achieving this, which I detailed in “Dashboard Blues” some years ago. But back to the story…

So, what do you do if you don’t have another car with a calibrated tachometer to use as a reference? Well, seeing as I have an old smartphone which I can connect to the Internet via my WiFi router (it doesn’t have a SIM card), I thought I would check out what applications are available for doing this sort of thing… for free, naturally :D

I found a number of strobe tachometer applications and I guess this would be your best bet for accuracy, but I didn’t want to faff about with highlighting the moving parts of the engine, or looking for suitable viewing angles, which in my case involve either removing the bulkhead panel to expose the front of the engine, or climbing into the engine bay and peering down the inspection hole in the gearbox bellhousing at the flywheel – so I went down the audio spectrum analyser route instead…

My first stop was Spectroid by Carl Reinke, on Google Play. This audio spectrum analyser produces a graphical display of the audio input from the built-in microphone, and has various features like peak hold and a waterfall display for monitoring spectral content for repeating patterns etc. Very cool…

Image

For this particular exercise, I needed to monitor the frequency of the engine exhaust, and use that to determine the engine rpm. For a 4-cylinder engine the following formula is used:

RPM = f/2 x 60

Where f is the measured exhaust frequency in Hz.

With each of my tachometer display segments indicating 400 rpm increments, I drew up a list of expected exhaust frequencies…

Image

I then headed out to the garage to try out my new toy!

Initial results were disappointing, but this was simply because I was sat in the drivers seat with the phone and at engine speeds below 1,500 rpm the exhaust note was not loud enough to be picked up above the other engine noise (valve clatter, oil and water pumps, the fuel pump ticking away, various overtones and harmonics etc.).

Slightly confused, I abandoned the test and went back to Google Play for another look around. There are some really nice apps available if you are willing to pay for them, but in keeping with the spirit of this job (free is better), I downloaded RPM Gauge from APKFab instead…

Image

As this app is more focused on automotive applications, I was hoping that it would be more successful at differentiating the exhaust note from the other sources of engine noise, but I actually got very similar results to Spectroid. However, knowing that any readings below 1,500 rpm would probably be inaccurate, I decided to check the calibration from 1,800 rpm to 3,000 rpm (3 to 6 display segments) instead, which would allow the exhaust note to dominate. Having 4 consecutive data points would also allow me to check the linearity of the frequency-to-voltage calibration (which is not adjustable), providing confidence that the calibration would remain accurate to 5,000 rpm and beyond. Limiting the sustained engine rpm to 3,000 would also draw fewer complaints from the neighbours!

So, with the instrument pod cover removed and a watchmakers screwdriver on hand to adjust the variable resistor, I fired up the engine and RPM Gauge and started my calibration run…

At an indicated 1,840 rpm I set the tachometer display to show 3 solid segments (with this design a segment will flicker if the engine speed is between 400 rpm increments). At an indicated 2,220 rpm the display showed 4 solid segments. At an indicated 2,620 rpm, the display showed 5 solid segments, and at 3,020 rpm the display showed 6 solid segments, just as it should.

Very satisfied with the calibration and linearity checks, I fired up Spectroid and held the engine at a steady speed with 4 solid segments showing. The spectrum plot showed a peak frequency of 73 Hz, which equates almost exactly to the calibrated engine rpm of 2,200!

Happy that the tachometer calibration is as good as it can be (certainly no worse than it’s analogue contemporaries from the 1970s), I brought the engine back to idle to check the indicated rpm, which according to the factory manual should be between 850 and 1,000. The reading was a little high, flickering occasionally above the first solid segment, but I am happy to leave it at that. I switched off and refitted the instrument pod cover.

All set for the next big adventure!

Lauren

Post Reply