Hey folks, I have an ‘82 300D running an aftermarket Sanden AC compressor. It works quite well, but it has an extremely strange interaction with the tachometer. When the AC is on and the engine is between ~3000 and 3500 RPM, the tach jumps to 5000-6000 rpm. I re-capped and re-soldered the tach amp, and aside from this oddity it works perfectly. Any ideas what might be going on?

My guess is you have a weak connection ground or power on your tach amp. This can be in the fuse box, on the ground lug behind the dash on the firewall, or on your tach amp socket.

Your blower draws a lot of current and it can put noise on your 12v going to the amp. The old amps are very sensitive to noise. If your ground to the amp is no good the analog filtering on the tach amp will be rendered useless because it requires a solid ground for the pickup filters to work.

Try your hazard or signal lights. They can also put noise on the tach if your grounds and power are restricted.

__________________
79 300TD “Old Smokey” AKA “The Mistake” (SOLD)
82 240D stick shift 335k miles (SOLD)
82 300SD 300k miles
85 300D Turbodiesel 170k miles
97 C280 147k miles

Good call on checking the ground bolt behind the dash. I’ve had a few other weird electrical gremlins with the gauge cluster and a loose ground connection would make sense.

Also, to be clear, the problem is specifically happening when the AC compressor is on. I can run the climate control on EC mode all day and the tach is fine.

yk:

Can you tell us what the nature of the tach signal is (its wave form), and how it is distorted by bad grounds & induced currents?

Hi Frank,

I’ll refer to this thread and its schematics.
https://www.peachparts.com/shopforum/diesel-discussion/206808-how-tach-amp-works.html

There is a filter at R1 for the input bias current to the sensor and transistor Q1 that needs stable power. If the 100uF capacitor is dry you’ll get spikes from things like the turn signal inductive fly back or brush noise from the blower getting through here.

Bad grounding is like putting a small series resistance on C1 which will also limit its ability to attenuate noise. Same principle on C3 which is the signal input to the LM1815.

The ground path for the amp is pin 2 which is a wire that leads back about two feet on the harness to the firewall lug. This wire has some inductance and if you are trying to filter away high frequency pulses series inductance along this wire will work against you. So there is series resistance on a dirty ground and series inductance on the long wire to ground. Since the low pass filters are reliant on a free path to ground through their parallel capacitor at high frequency this is a weakness in the original design.

The magnetic signal is actually pretty small. I think old pokey measured it and it was less than 50mV. A little noise on this signal will fool the zero crossing detector in the LM1815 as additional crank counts thus making RPM rise erroneously.

The sensor signal amplitude rises as rpm rises then falls off again near 5000rpm. I think this is a dynamo effect - the faster the pin moves by the coil the more induced voltage. Up to a point. Then as rpms get fairly high the pin spends less time by the coil and coil impedance makes the signal decrease. This may be why the spikes get bad at high rpm.

Forgot to mention the horn also puts a lot of noise on the cars circuits. It’s the nature of the horn, a big inductive load that is being switched in and out of the car at roughly 200Hz. It makes lots of spikes. All these effects are simply the points ignition effect…you have a coil carrying current and you suddenly break the circuit…then bam your voltage shoots way up. The original circuit anticipated all this noise but its mitigation is contingent upon solid grounds and fresh capacitors.

__________________
79 300TD “Old Smokey” AKA “The Mistake” (SOLD)
82 240D stick shift 335k miles (SOLD)
82 300SD 300k miles
85 300D Turbodiesel 170k miles
97 C280 147k miles

After a read-thru of the posts from 2007, it would appear that the TD signal that drives the tachometer is a square wave of 5-6V, with a duration of 3ms.
Is that TD signal consistent across multiple MB models & engines (gas, as well as diesel)?

I don’t know. The gas cars use the signal from the coil I believe. They’re a completely different circuit.

Once the signal goes digital (output to the gauge) it is nearly incorruptible. The sensitive side is upstream of the LM1815 which is the unamplified side.

__________________
79 300TD “Old Smokey” AKA “The Mistake” (SOLD)
82 240D stick shift 335k miles (SOLD)
82 300SD 300k miles
85 300D Turbodiesel 170k miles
97 C280 147k miles

OP - I thought about this awhile and there may be something else going on here. It is possible that some stray magnetic field is getting out of your compressor clutch and putting noise on your crank sensor.

I’m not sure how you test for this or eliminate it.

Make sure your tach sensor wire hasn’t been zip tied to your new compressor clutch wire. I don’t think this should matter in theory but I can see how an installer would use the existing wire to stabilize your new compressor wire by zip tying them together. Long runs of parallel wires are a way to get cross talk.

Make sure your sensor is clean. When the sensor gets caked up with grease it will attract iron filings because it’s magnetic. You may be picking some magnetic field up off the magnetic compressor clutch. Cleaning all the gaps will discourage magnetic field lines from getting to the pickup head.

If you want to do a science experiment you can remove your sensor from the block and place it somewhere else in the engine bay. Your tach should flatline but it will help you identify if the source of noise is the Sanden clutch’s close proximity to your sensor. For instance if you move the sensor head up to the valve cover and your needle still jumps at 3000rpm with the compressor on you know the noise is not likely magnetic interference to the sensor head.

In theory (we all know how good that is) the magnetic field in the compressor is DC and shouldn’t cause any time varying behavior. And the field from the magnet (also static) should be contained by the steel plate in the clutch. But I’m thinking there may just be some field whipping around on the pulley face that just puts enough signal out to trip the sensor coil which is only a few inches away.

The routing of the sensor wire is also close to the AC compressor. It may be possible that the Sanden magnetic is strong enough and the wire is close enough that you are seeing cross talk at high rpm. High rpm may be doing a dynamo effect…faster spinning of the crank means more current induced in the neighboring wires.

The likely culprit is stray magnetic field getting into the sensor pickup which consists of many coils of copper wire in the little black head. Perhaps you can fashion some magnetic shielding out of mild steel sheet metal. Even zip tying a mild steel bolt to the sensor head may divert enough stray field away from the pickup.

Good luck tracking this down. This is going to take some tinkering to figure out.

__________________
79 300TD “Old Smokey” AKA “The Mistake” (SOLD)
82 240D stick shift 335k miles (SOLD)
82 300SD 300k miles
85 300D Turbodiesel 170k miles
97 C280 147k miles