Bias Voltage on a Single Wire Knock Sensor?

I just took some scope readings on the knock sensor of a 98 Mazda 626 2L and I have a few questions.

It's a 200ms screen, the blue trace is the knock sensor at 500mv / div, the yellow is the CMP at 1v / div. Engine is idling very rough at about 500rpm with the brights, defrost and AC on:



You can see matching ignition system interference in both signals, but other than that interference the knock trace seems to be a flat line at about 3.5v.

This is a single wire knock sensor grounded through it's threads to the engine block. I have two questions:

1. I don't understand how a single wire sensor like this could have a bias voltage
2. This engine is idling so rough during this capture that I'd expect to see some activity, yet it's pretty much flat line. Why?

You can see the full three minute capture here .

The voltage is there to lift the signal from the knock sensor above ground voltage. The voltage output from the piezo knock sensor would otherwise be a true AC signal. Iit will also serve to allow the PCM to check if the knock sensor is connected.

OK, so this looks like a pretty normal KS waveform to you?

The Snap-On Troubleshooter says that you'll get a P0325 Knock Sensor Fault code when the signal voltage goes below 1.2V or over 3.7V after engine startup. :huh: Your 3.5V is a little bit too close to 3.7V for comfort.

Are you getting any trouble codes? If this is anything like the Nissan sensors we're familiar with, then the PCM is providing a 5V signal on the knock sensor wire. The sensors resistance to ground will then pull that 5V down to 2.5V when plugged in. Higher voltages indicate an open, lower indicate a short.

I'd be tempted to lightly tap on the block with a small hammer and watch the knock sensor signal, just to see if it's capable of producing signal.

It's probably 3.4, since the high point is the ramp from the ignition system interference, but yeah, based on what you're saying it seems borderline. There were some oil and water leaks in the back of the engine for a long time, do you think maybe water / oil got into the sensor threads and may be causing a bad ground to the block?

I'll try giving it a knock, and maybe even removing it and try cleaning the threads.

As a side note, when looking at KS waveforms on IATN I could find very few with any apparent bias voltage. One I found is this 1990 Subaru Legacy:

Definitely let us know what you find. I don't know that the oil leak would necessarily cause this issue, but it's definitely worth a shot. Just take care when threading it out - I dunno if this sensor goes into a water jacket or not. :silly:

The bias voltage thing seems to be hit or miss with different makes/models. Most GM small blocks have no bias, so the only way they test the sensor is to deliberately induce a knock and look for a response. :lol: But some older GM V6's do use a bias, and even share one bias across two sensors.

Well I took a three minutes video capture of my screen while testing the knock sensor which is up here .

Basically, with KOEO, it's showing about 3.28v on the wire. With the noise from the ignition system the scope is usually reading between 3.09v to 3.49v while running. I started tapping on various places with a small ball peen hammer with the scope set to single shot mode, KOEO, and the only place I could get a response was on the exhaust manifold near the #2 cylinder. The KS is installed on the opposite side of the block next to the #2 cylinder position. Here's what it looked like:



As you can see, I barely get any response at all, 3.02v low and 3.58v high. It may be I'm tapping too far away, or not hard enough, I've never done this before, but you can hear my vigorous tapping in the video. I tried tapping with the engine running also and couldn't see anything, the response, if any, was totally swallowed by the ignition noise.

I see that the waveform that I found for the Suburu shows way more response, but I think he was also tapping on the block right next to the sensor, which I can't really do due to space constraints.

Tyler wrote: The Snap-On Troubleshooter says that you'll get a P0325 Knock Sensor Fault code when the signal voltage goes below 1.2V or over 3.7V after engine startup. :huh: Your 3.5V is a little bit too close to 3.7V for comfort.

Are you getting any trouble codes? If this is anything like the Nissan sensors we're familiar with, then the PCM is providing a 5V signal on the knock sensor wire. The sensors resistance to ground will then pull that 5V down to 2.5V when plugged in. Higher voltages indicate an open, lower indicate a short.

By the way, I've never seen a knock code on this vehicle, although it's not OBDII since it's a European model. Also, couldn't higher voltage also be caused by a bad ground between the sensor and the block?

Watched the video, thanks for that! Considering that you can't smack on the block next to the sensor itself, I think that's a reasonable signal amplitude. You could try to feed a long screwdriver or extention through the manifold runners to get near the sensor if you wanted, but I'm not sure I'd worry about it.

The high-ish bias could definitely be from poor sensor ground. BUT, remember that I don't have a good spec on this system. What you're seeing may be normal. :blink:

Took a few more measurements today. Unplugged the sensor at KOEO and got 4.59v. Thought I'd try to do a kind of virtual voltage drop test by measuring the voltage and amperage then calculating the resistance, but I found the current was too small to measure even with my ohmmeter in line with the circuit and set to microamps. Apparently, the WSM specifies a resistance of 560 kilo Ohms, and I measured 553 kilo Ohms. That gives 0.000005857 amps at 3.28v.

Actually, I'm not sure why I couldn't pick up the amperage, as that comes out to 5.9 microamps which my ohmmeter should be able to measure.

Also not sure why my 5v ref is only 4.59v and if that's a problem or not.

It also isn't possible to accurately measure voltages with a meter or a scope on high impedance circuits this applies to both the sensor and the circuit in the ECU

Sensor resistance around 550,00 ohms.
Normal Oscilloscope 1,000,000 ohms
Good quality digital Voltmeter 10,000,000 ohms
Reading the sensor output with either an oscilloscope or to a lesser extent a multimeter will put aelectrical load on the sensor pull down the sensor output voltage.

I see what you're saying, but doesn't that mean that any measurement taken on something like a knock sensor is going to be thrown off? Yet, people measure knock sensors. Is it just that they're aware of the problem and it's taken into account somehow?

Every electrical or pressure measurement you can do is invasive it alters what you are trying to measure in most cases on cars the effect is so tiny to be noticeable.. Single wire knock sensors are piezo devices the piezo micro phone inside is tough little device when they tend to stop producing any response, so if you see any reasonable response to a tap test it is assumed to be working.