Tachometers for conversions

[e*clipse]

Hello everyone!

I thought I'd start a thread about adding a tachometer to a conversion.

In general, it shouldn't be too difficult. One just needs to be sure the signal put out by the ECM matches what the tachometer is expecting.

I am attempting to connect the stock tachometer on my Toyota pickup to the output signal from the VW ECM. Hopefully some of what I learn here can be useful for other conversions.

To start, I'd like to quote a friend:

The tach signal was one of the very first digital circuits in vehicles, long before computers came along. By digital I mean 0's and 1's. There is either a voltage or there isn't a voltage at that coil terminal. Minus any electronic component induced voltage ramp-up or decay it is a square wave. When the "points" are closed the path to ground is complete, no voltage. When the "points" are open the path to ground is incomplete and there is a voltage.. Presence of the electronic ignitor and lacking any points doesn't change how the circuit functions as far as the tach is concerned.

Most tachs work, as I understand it, essentially as a "voltage duty cycle" and display as a voltmeter. They measure the voltage at the negative coil terminal, reference it to battery voltage, and display the percentage as RPM's. So a tuned resistor would skew the displayed "voltage."

Ok, so the tachometer is expecting a 12V square wave with a frequency relative to the amount of cylinders in the engine as an imput signal.

The VW ECM has a tachometer output pin. On my 1Z's ECM, it is pin 2/68. Be careful not to confuse it with 8/68 (engine speed sensor).

To see what the output signal of this pin is, I connected it to my oscilloscope:

My Laptop can interface with a PC oscilloscope.

I connected the oscilloscope probe to 2/68 and connected the oscilloscope ground to the truck's instrument ground line.

At Idle, the output is:

A clearer picture:

The output is a square wave signal with a 50% duty cycle. The amplitude is 14V. Notice the markers: 4 pulses correspond to 7.53 Hz, or 451.8 cycles per minute. Since the engine was idling, one would expect 903 RPM. In other words, there are two 14V pulses per revolution.

When you rev the motor, the pulses move closer together:

Since the amplitude is right, now all I need to do is match the pulse frequency or adjust the tachometer's output with a resistor.

 

[markward]

I like the Laptop interface. I am driving my stock gasser vanagon tach off the of the ECU from an ALH TDI. The stock vanagon tach can only be accurate around a specific rpm in my application. Not sure why, but better than a dead instrument. I have adjusted my Tach to be accurate at 3000 rpm verified against my VAG COM. At idle it reads 400 rpm low. Around 2000 rpm it is off about 200 rpm. My guess is the ECU is sending a square wave that the electronic ignition tach is not translating properly. Nice write up.

 

[e*clipse]

It sounds like yours may be calibrated for an engine with more cylinders.

On any four stroke engine (diesel or spark ignition) there is one firing (either through the spark plug or injecting fuel) every two revolutions.

Thus, a four cylinder engine would have :

4 cylinders_________ = 2 firings/ revolution
​

2 revolutions / firing

6 cylinders_________ = 3 firings/ revolution
2 revolutions / firing

Therefore, a 22RE engine (or any 4 cylinder motor for that matter) would have a igniter signal of two pulses/ revolution, just like the inductive sensor signal coming out of my VW ECU.

 

[e*clipse]

I wired the ECM's pin 2/68 into plug IK2 in my Toyota truck. Plug IK2 is in the footwell of the passenger side of the truck, where the ECU used to be. The pin is on the bottom left, looking @ the male plug. It is a black wire, and connects to plug D, pin 7 on my instrument cluster.

VERIFY YOUR SPECIFIC WIRES!!! I found this info w/ a 1993 FSM and verified it with a continuity check.

I have two tachometers to mess around with. The tachometer from my 1991 pickup with a V6. It has a potentiometer to calibrate it for the specific engine. I've heard this has enough range to account for 6 cylinder to 4 cylinder swaps.

Unfortunately, when I tested them this evening, that one was DOA.

The other tach is from a 1993 4-runner with a V6. This one worked.

Here are the approximate rpm values:

Tach VAG-COM
600 903
1000 1449
1500 2245
2000 2982
All of the tachometer values are off by approximately 2/3.

Since this tach has no calibration resistors I have two options:
1) A Dakota Digital adapter
2) An older (Pre 1992) Toyota tachometer from a vehicle with a 22RE engine

 

[markward]

I wired the ECM's pin 2/68 into plug IK2 in my Toyota truck. Plug IK2 is in the footwell of the passenger side of the truck, where the ECU used to be. The pin is on the bottom left, looking @ the male plug. It is a black wire, and connects to plug D, pin 7 on my instrument cluster.

VERIFY YOUR SPECIFIC WIRES!!! I found this info w/ a 1993 FSM and verified it with a continuity check.

I have two tachometers to mess around with. The tachometer from my 1991 pickup with a V6. It has a potentiometer to calibrate it for the specific engine. I've heard this has enough range to account for 6 cylinder to 4 cylinder swaps.

Unfortunately, when I tested them this evening, that one was DOA.

The other tach is from a 1993 4-runner with a V6. This one worked.

Here are the approximate rpm values:

Tach VAG-COM
600 903
1000 1449
1500 2245
2000 2982
All of the tachometer values are off by approximately 2/3.

Since this tach has no calibration resistors I have two options:
1) A Dakota Digital adapter
2) An older (Pre 1992) Toyota tachometer from a vehicle with a 22RE engine

Guess I got lucky with the stock vanagon tach. There is a potentiometer on the back of the tach for adjusting, but the adjustment is across the range. That is why I set it to be accurate at 3000 rpm. I spend most of the time going down the highway and like to see the tach correct around 70mph. Idle is not as important. I was content to not have a tach originally, but the circuit foil on my original cluster was shot. I bought a later cluster with a tach. Decided to hook it up. I had some other problems and tried a dakota digital unit for the signal from the alternator and could not get the VW tach to work, but it would drive a VDO tach.

 

[Rockwell]

Since this tach has no calibration resistors I have two options:
1) A Dakota Digital adapter
2) An older (Pre 1992) Toyota tachometer from a vehicle with a 22RE engine

I have an '86 toyota pickup with a 1.6TD, and try as I might, I cannot get the tach to work. I have the Dakota Digital adapter. I tried 2 or 3 tachs and 2 alternators with the 'W' output and not a twitch from the tach. I am losing hope. If you try a pre '92 at least make it a post '89, my pre '89 does not seem to work

 

[shortysclimbin]

Why not use an op-amp circuit with frequency compensation?

 

[e*clipse]

@ Rockwell:

Dumb question: Did you isolate the circuit? To make SURE I got the right wire in the harness, etc, I took the tach out of the instrument cluster and wired it directly. I could then check all voltages, signals, etc.

BTW, did you do a write up on the Pirate board? That truck is amazing!

@ Shortysclimbin:
How does a frequency compensation op-amp circuit work?

I'm looking into building a circuit w/ a 12 series PIC. They are pretty cheap and capable.

I wonder if there'd be any interest for a SIMPLE, cheap pulse translator that allowed something like +/- 200% frequency adjustment. It would use a 12V square wave input signal and put out a 12V square wave.

 

[e*clipse]

Regarding the op-amp suggestion.

As I understand it, a frequency compensation circuit allows an op amp to produce a more stable output at increased gain or phase margin. This comes at the expense of bandwidth.

Electrically speaking, this is a simple (slow) problem. Even an eight cylinder engine running @ 8000RPM will produce a square wave (digital) frequency of 533Hz. A 4 cylinder TDI running at 4500RPM will only produce a 150Hz square wave.

We need to keep the input-output amplitude the same and alter the frequency.

These numbers can be handled by an 8-bit computer w/ no problem.

 

[jimbote]

check out my solution for a tach driver using an inductive pickup...super simple....about the third post down starting...if you wanted to do this using a V6 tach you could weld three nuts 120' apart....
http://www.vwdiesel.net/forum/index.php?topic=3741.30

 

[e*clipse]

Jimbote
check out my solution for a tach driver using an inductive pickup...

That would work. (in this case, a 4 cylinder motor in a car w/ a V6 cluster - same as my problem.)


I'm thinking there could be a use for something a little easier & cheaper, as long as one is capable of intercepting the signal from two wires. Of course the device would need power, so we're talking about 4 wires.

It seems to me the problem is pretty straight forward and common for anyone who does a motor swap with an electronically contolled engine. In fact, anyone who changes the tire size or gearing on their car with an electronic speedo could also use something like this.

At this point, I'm just curious if anyone would be interested in a simple, cheap, pulse translator. It could be used to adjust speedometer speeds or tachometers. I figure it could be sold for ~ $30.
​

To keep it simple and cheap, it would need a square wave signal that maxed out at 1khz. (In other words, the teeth on the flyweel would not work) Any 12V igniter signal would work, and most active inductive pickup signals (like the VW motor signal, the Toyota speedometer signal, or the VW ECM output) would work. (3 wire pickups are generally active and produce a square wave)

The output could be changed some percentage (I'm working on it) from the input. Setup would be super simple: Increase%, Decrease%, and lock.


If people are interested, I'll pursue it. If not, I've found a used one for that price. While I can purchase the parts inexpensively, it will take some time to write the code for the microprocessor and design circuit boards, etc.

 

[Vince Waldon]

The guys running non-TDI VW diesels that have to use the W-terminal on the alternator to drive the tach would love you.

The W-terminal signal about 600% faster than what a gasser tach is expecting and the only commercial solution is a $100+ frequency-conversion box from Dakota Digital. Diesel tachs for pre-A4 VWs are becoming fewer and farther between, and lots of folks are doing gasser -> diesel swaps.

 

[e*clipse]

Vince Waldon
The W-terminal signal about 600% faster than what a gasser tach is expecting

Good point.

Is the W-terminal signal a square or sinusoidal signal?

So it seems like a range of -600% to +200% would work for most situations.

It also seems like accuracy of 2% > 5% would be close enough.

My idea would be to average the signal for about 4 pulses (2 crank revolutions for 4 cylinder engines to calculate the signal. There would be a tiny lag due to this.

Ideas?

 

[Rockwell]

@ Rockwell:

Dumb question: Did you isolate the circuit? To make SURE I got the right wire in the harness, etc, I took the tach out of the instrument cluster and wired it directly. I could then check all voltages, signals, etc.

BTW, did you do a write up on the Pirate board? That truck is amazing!

Yes I isolated the circuit and took the tach out too. Maybe my Dakoda converter is bad? I am watching this thread and hoping for a solution.
I did mention it on Pirate but my build up was on Zuwharrie, it was the one with the 2wd pickup. I have hardly driven it and you already have me thinking about swapping the IDI for a TDI!

 

[e*clipse]

The build I saw on the pirate board was amazing. The guy completely took apart the WHOLE TRUCK and repainted each part. It was a crawler rig (naturally - it's the pirate board) and the guy pointed out for crawling you don't need a lot of power, just low end torque and low low gears.

I guess I would sum up the engines like this:
1.6L - simplicity > for folks who hate dealing w/ wiring looms, computers, and multiple sensors.

1.9L TDI - TORQUE, power and ability to modify. Must have no problem with wiring looms, computers, etc.

If you look around here a bit, you'll see some mind-blowing small diesels.

Before I hijack my own thread . . .

Have you tried putting a meter on the output of the signals? Do you have access to an oscilloscope?

I have a fairly simple digital meter that I use to check everything before wiring it together. With the engine running put the meter between the W output and ground. In DC mode, there will be a lot of hunting, with no final number. In AC mode it should settle on some number lower than 12V. I don't know what the signal out of the W terminal is. Vince Waldon said it's about 600% faster than the signal put out by the VW ECU.

If you have an analog meter, you may be in better luck. It will read an AC signal - the inductance in the meter will cause the meter to settle on the RMS value of the voltage. Again this is a value less than 12V.

Those values are relative to the duty cycle of the output signal. If the frequency is increased, (revving the motor) you will most likely see no difference in the output.

If there is something coming out of the alternator, measure the Dakota digital's output in the same way. Use its ground (which should be connected to the "instrument ground" of your truck. The grounds should be tied together at some point, but the instrument ground allows guages to operate without large fluctuations due to heavy loads.

When I measured the output of the VW ECU using the method described above, I got an AC reading of 7.78V. In some ways, the value is not critical, it's just a verification that you're getting something.

 

[Vince Waldon]

Good point.

Is the W-terminal signal a square or sinusoidal signal?

Square:

Really old fuzzy scope but you get the picture.

From an AC volts perspective you get just over 8V:

I'd say you've got the right idea in terms of range... might wanna expand to -700% just to be safe... and I doubt one rev of delay would be noticeable, particularly if it smooths the needle out.

 

[babyrattler]

I would be interested in a simple low cost unit. I am working on a conversion right now. I am putting a 1.9TDI into a 1977 CJ-7 that had a 304 V8 in it. I have many issues to overcome yet so getting the tach issue out of the way would be great. I want to keep the Jeep dash looking as stock as possible so if I can use the factory V8 tach that would be awsome. I really don't want to spend $100 on the DD unit. I'm in if you find a solution. Great thread!!

 

[Scott_DeWitt]

I guess I would sum up the engines like this:
1.6L - simplicity > for folks who hate dealing w/ wiring looms, computers, and multiple sensors.

1.9L TDI - TORQUE, power and ability to modify. Must have no problem with wiring looms, computers, etc.

Don't forget the 1.9 Mtdi.. A bit more expensive due to the cost of the pump, however it is for the most part wireless.

 

[e*clipse]

Vince: Thanks for the screenshot - that's a great help.

I'm thinking about a multi-turn potentiometer for adjustment. It's probably the most intuitive, quick adjustment possible.

Clockwise: increase ratio
CCW: decrease ratio

A "mode" switch would give two options:
1) "set" - This mode would allow the input/output ratio to be adjusted with a potentiometer. An a/d converter would translate the signal. This would allow -700% > + 300% for full range. While in this mode, response will be slow because the microprocessor is doing a lot more.
2) "lock" - This mode would store the value, ensuring stable fast operation

Wiring:
+ : 12V
- : Instrument ground
input: Signal from tach output on ecu, sensor, or W terminal
output: modified signal for tach or speedometer

Input and output signals will be 12V square waves.
Input is high impedance, so it ***shouldn't**** effect other devices reading the signal.
Output will be fused and current limited.

There will be protection circuits to ensure if wires are connected incorrectly, nothing will be damaged. I know how wiring harnesses are: it's VERY easy to wire things incorrectly.

Simple enough?

 

[e*clipse]

Because my tachometer did not respond to throttle input, I thought it was dead. I ordered another one off E-bay, which cost almost $100 w/ shipping.

It was dead too.

Why did one of my tachometers respond, and the others not? I could have two identical blown up tachometers, but that would be odd.

I started looking closely at the tachometer’s circuit board. I noticed protection circuits: R-C Filters and diodes to ensure the current flows the correct direction. The IC on the circuit board was labeled “BA6730B.” What is the IC’s job? Also, there is a trimpot on the upper right corner of the circuit board that is used to calibrate the tachometer.

I looked for data sheets for the IC: BA6730B > no luck.
However, I did find a similar IC, the LM2907. This IC is a frequency to voltage converter.

Then I remembered that my AC voltage from the VW ECU output was about 1V lower than the reading Vince Waldon showed from the “W” output of the alternator. Why did some inputs get the tachometer to respond, while others didn’t? When I had my voltmeter measuring the ECU’s tach output, there was no voltage difference when I revved the engine. The tachometer must be looking at the frequency, not the input voltage.

Perhaps the input for this IC is a comparator. A comparator is a device that compares two voltages, and puts out a voltage depending if one of the voltages is higher or not. This would be good if the input was a sin wave.

The circuit board for the Toyota tachometer has a voltage divider on the input from the igniter. This is the pair of resistors on the lower left corner of the board. The output of the voltage divider goes to pin 14 of the IC.

There is also a voltage divider that uses the supply voltage as its input. The output of this voltage divider is pin 17 of the IC.

Perhaps pin 17 is the voltage reference and pin 14 is the input. If the signal at pin 14 exceeds the voltage at pin 17, the comparator would provide a “high” value for the rest of the circuit.

To test this hypothesis, I measured the voltage at pins 17 and 14 while the engine was running.
Pin 17: 2.118V
Pin 14: 1.85VAC

Pin 14 is lower than pin 17, and there is no response.

I also measured the voltages at the meter’s inputs. For some reason, there are three inputs, which I don’t understand. Looking at the back of the circuit, I measured the three meter connections with ground as a reference.

Left side: 7.95V
Center: 4.51V
Right side: 4.77V

Note that the center and right side are very close. Perhaps the meter measures the difference between the center and right side inputs.

Perhaps I could increase the voltage out of the voltage divider. Generally, high input impedance is good, so it would be best to increase the resistance of the lower resistor in the voltage divider. Also, most IC’s function at around 5VDC, so something needs to change the car’s 12V to 5V. In this case, a pair of resistors is arranged as a voltage divider and a Zener diode ensures the input voltage does not exceed 5.1V. As long as the input does not exceed 5V, the IC will probably be safe. If the circuit is functioning properly, the voltage at either the center or right side of the meter connections should change.

I removed the 15Kohm resistor and replaced it with a potentiometer. (see the big, ugly potentiometer in the pic ) The plan was to start with approximately 15kohms on the potentiometer and increase it until the input voltage at pin 14 exceeded the voltage at pin 17. At that point there should be a change at the meter input connections.

Success! When the potentiometer was adjusted to 27.3kohm, the voltage at the center pin increased to 5.61V. The voltage at pin 14 was 2.675V, which was higher than pin 17’s 2.115V.

Because 27.3kohm is not a standard resistor value, I increased the potentiometer’s resistance to 33kohm. The response from the tachometer circuit was the same.

After soldering a 33kohm resistor in place of the potentiometer, I reassembled the tachometer to give it a try.

It works! After adjusting the potentiometer, the tachometer now reads correctly at idle and every other speed (referenced to VAG-COM output).

In other words, neither of the tachs were dead, they just needed a higher threshold voltage for the tachometer circuit to work.

 

[e*clipse]

At this point, I got greedy. What I would REALLY like is a tachometer that read full scale. The Toyota’s tachometer reads up to 8000 RPM, so a reading of 2X the actual value would work nicely. In other words, 8000 RPM on the tach would correspond to 4000 RPM actual speed.

I tried adjusting the potentiometer for this, but no luck. It would inhcrease to 1500 rpm at idle, and I needed 2000 rpm.

The potentiometer is 100kohm total resistance. Double that would be great. I happened to have a multi-turn 1Mohm potentiometer in my parts bin. I replaced the stock potentiometer with the new one and adjusted it for 50kohm to start.

While the engine was running it was easy to adjust so that the tach read 1800rpm at idle.

As I revved the throttle, the tach responded accurately, (verified w/ VAG-COM) Here it is at 2500 RPM:

So, now the tach is working exactly like I want. I’d like to fix the numbers on the display scale.

I don’t need to buy a Dakota digital converter or make something like it.

Cost: less than a dollar for a resistor and a trim-pot. (oh, and time to figure this out)

Now the question is: should I bother making a cheap, inexpensive signal translator?

 

[e*clipse]

Ok. . .

Well, I was excited.

I realize this is just so much EE gobldeegook. . .

Most people would probably have little interest in tracing down a problem like this or debugging their tachometer.

So, I got a new toy. Without going into more EE gobldeegook, I'll just say it allows me to simulate various tachometer and speedometer signals.

Then I'll be able to make some little plug n' play circuit.

Of course, it'll prolly cost more than a resistor.

 

[Frankendiesel]

I was excited too! Just didn't post. Kuddos my fellow geek. Good to see another engineer tinkering out in his garage (where we work best, no?)

Is your new toy a sig gen by any chance? Arb Gen?

Travis

 

[e*clipse]

Why yes, it is a signal generator.

So here are some shots of the signal generator driving the tachometer:
The tachometer was made for a 6-cylinder engine.

2000rpm* 3pulses per revolution / 60 seconds per minute = 100 hz

3000rpm* 3pulses per revolution / 60 seconds per minute = 150 hz

I was curious about how signal amplitude and duty cycle affected the tachometer output.

To find out how small of a signal the tachometer could read, I reduced the signal generator's output amplitude:

At this point, the tachometer is still reading correctly: 2000 rpm.

Here is where the signal dropped. At this point, the tachometer reads zero.

To test the duty cycle, I adjusted it from a range of 33% to 60%. There was no change in the tachometer's output.

It even works with a sin wave! The tachometer reads 2000 rpm with this input signal. This confirms the theory of the input comparator.

So, this tachometer from a 1993 toyota 4-runner is pretty robust. The signal can vary quite a bit in amplitude, duty cycle, and waveform. The tachometer will still accurately display the rpm.

The tachometer sitting on top of my oscilloscope appears to be dead. Even with a good input, it displays zero.

So now I have some reasonable idea of the tachometer's requirements and how far the signal can deviate. Next step: get a microprocessor to recognize the signal.

 

[Frankendiesel]

Nice job. I like the PC O-scope BTW. So did you ever come up with a quick design to make a speed sensor convert-o-matic? I'd be interested in buying one if the price is right. Or just buying the plans would work too.

Travis

 

[e*clipse]

I've got things roughly worked out for hardware. (which I recieved today)

I plan to use a PIC's built in functions for most of the work, with some extra bits to provide power and input/output signals. I have an 18-series pic sitting around for experimenting and developing the code.

From my perspective, most of the work is in designing and building the first one and most of the expense (for electronic stuff) is purchasing single items and duplicates (in case you break something.) Unless you have a good way to make circuit boards and have the parts laying around, I doubt you'll save money building it yourself. Only by purchasing multiples will I be able to get the cost down to the target I set.

Just my 2cents.

 

[G60ING]

Good information! Linked to the Swap Thread

 

[jjordan11]

Very interesting, I'd like to convert the signal from the Audi's 5 cylinder engine to my TDI's 4 cylinder's signal. Anybody have any idea as to what would have to be done. I'll have to tinker a little bit, I'll see when I get the Audi going again.

 

[CFM]

Wow, good work! When I did my Saturn TDI swap, my origional idea was to have a 5000-rpm tach face made and use the Dakota Digital box to calibrate it, but I haven't had a chance to experiment with it yet to determine if it will allow for that. Still, the tach does read fairly accurately with what I have set up now, so I really can't complain. I really like what you did, that gives much greater latitude in calibration and mixing/matching tachometers.

 

[Frankendiesel]

Yes, PCB's seem to be the biggest expense, as I'm sure you know. Do you use a joint like PCBexpress? They seem to have good prices. If we can get a number of folks here to lay down some dough, you should be able to order a couple of panels of PCB's. Is that what you had in mind?

If you can get an adapt-o-matic for somewhere around $20, I'm in. I don't remember what your price goal was...