Check your AFM boys and girls (re. hi rpm miss)

#61

My quick “I’m at work” response.

I should have added that I didn’t mean capacitance like how capacitors store charge. More like how any material that can hold an electric charge has capacitance. What I was trying to explain is why the conductive areas are shaped the way they are. I get the green areas and the resistance adjustments made by the laser etching. (edit: to clarify, what I get is that the laser etching is used to tune the resistance. obviously I don’t get the overall “why” part, though.) What I don’t get is why the conductive traces have the odd shapes that they do. I don’t think it was by accident, but I couldn’t really come up with a good explanation other than providing a tiny bit of capacitance to help “fill” the local area of the sweeper as current is being drawn. The pad shapes are a more pronounced in the older AFM:

The other thing I realized this morning was that I was concerned that the sweeper rate might not change fast enough to induce a voltage change due to air flow changing. What I didn’t think about was the effect that vibrations might have on the sensor itself. These kinds of higher frequency vibrations would certainly get closer to potentially causing signal spikes. That said, if signal smoothing was the goal, I’m pretty sure there would have been simpler ways to smooth out the signal without affecting the responsiveness of the sensor. shrug

Som

#62

Another reason those conductive areas are shaped the way they are could be simply that it was in some way the cheapest possible layout for them to construct.
Whomever made this board, their first mission was to make a profit, the second was to make it work. At least that is my jaded view of things.

#63

Totally with you on the jaded view part of it. I would expect the “cheapest/easiest” option there, too.

That said, the shape of the traces seem too deliberate to me to suggest this was the “easiest way” to do it. I could have seen it as a way to increase tolerances in the manufacturing process if, say, the resistive elements were being laid onto the board separately and had a wide margin of placement error. Then you’d give it some wide pads to land on and then correct away the differences with the laser etching.

But… the pads have all these little quirks/extensions beyond that pads that don’t seem to contribute to helping tolerance issues. Particularly the 2 large tracks around the top of the images that end nowhere.

As I’m writing this, another possibility I’m thinking is that the pads could be there to help with the automated testing/calibration processes during manufacturing. Like robot arms placing ohmmeters around the resistive areas to measure before laser etching.

That doesn’t seem to explain it all, though. Like how wide/tall the resistive tracks on the edges are vs. the tracks in the middle.

Maybe I need to do more circuit bridge research. :slight_smile:

Som

#64

As a youngster, I worked for an outfit for a while that made multilayer circuit boards. Som, since you’re in San Diego, you probably drive by the building all the time.

The big conductive strips don’t look unusual to me. This device is a big clumsy analog sensor. There’s probably a fair amount of current flowing so it needs big traces to keep the resistance down and sensor accurate. If you look at the design from the standpoint of “need big traces” you can see how the designer used the real estate pretty logically.

Traces that don’t go anywhere. This isn’t necessarily a single layer circuit board. There could be multiple layers inside that aren’t visible. So a trace that appears to stop could pass inside the board to other stuff, or simply pass thru to the other side of the board.

#65

That’s definitely something I hadn’t thought of – going 3 dimensional. I was so in the mindset that this was gonna be “as cheap as possible” that I thought for sure it was a single sided/single layer board. Another reason I didn’t think of multi-layers was that I didn’t think there was a way (at least using commonly available manufacturing methods) to connect multiple layers on a PCB without vias. Are there? (were they available in the 80s?)

Wide traces for minimizing resistance makes sense, but I wonder… if you’re trying to get to that level of precision and sensitivity, would the capacitance you’re adding by going with the wide traces start causing a problem with ringing/signal oscillations? As I understand it, there’s a limit to the benefits of reducing resistance in exchange for capacitance.

The way I think about it is that there are electrons in the trace. A voltage difference induces those electrons to flow in order to try and reach an equilibrium point. As the electrons start to move, you have an inductive effect. This inductive effect is like momentum. So the electrons build up momentum. When the voltage changes, or if the voltage difference starts approaching zero, the electrons need a finite amount of time to react.

Depending on the impedance vs. capacitance you have, you get all those fun oscillation/dampening situations:

  • critical damping (impedance/capacitance at proper levels to ensure a quick arrival at the signal voltage without overshooting/oscillating)
  • underdamping (capacitance is too high for the impedance and your signal overshoots the target/equilibrium point and oscillates before reaching equilibrium)
  • overdamping (impedance is too for the capacitance so your signal is slow to arrive at the equilibrium point).

(I could be muddling terms like voltage/charge/etc, but that’s basically the gist of how I understand it)

To me, there’s a (somewhat) similar analogy to how intake runners are designed for the engine. The air/fuel charge substituting for the electrons, the runners themselves substituting for the traces. The cross-sectional area of the runner is your resistance and the length of your runner is your capacitance. For a given RPM, you’re attempting to maximize the amount of charge that flows through without getting overshoots (back pressure being introduced on the intake side when the valve closes and there’s still charge momentum going towards the cylinder) or undershoots (pressure in the runner decreasing before the valve closes and getting some of the cylinder’s charge back into the runner). What you’d be going for is the “critical damping” situation where the runner’s charge pressure matches the cylinder’s charge pressure so the charge velocity is exactly zero when the valve closes. Again, very rough analogy, and being able to tune this with valve timing makes this aspect of the runner’s dimensions less critical in terms of concerns about damping, but still.

Yet another possibility is that the board was designed to for multiple possible configurations. Maybe the resistive material is laid on in a subsequent step to the traces and in some configurations some of those loose end traces get bridged by a resistor. Though I can already see one pair of pads that don’t seem to line up enough to support this theory.

By the way, I enjoy guessing/testing/guessing/blabbing about it, as you can probably tell, even if I never find “the answer”… so feel free to ignore. I won’t be offended. :slight_smile: That said, I haven’t looked at some of this stuff for years… so anyone who feels so inclined to correct me, I’m all ears. :slight_smile:

Som

#66

Re. multi-layer boards in the 80’s. Yes, they were around. I was working for Dibble Electronics there in San Diego ~1985. We could do up to 14 layer boards.

Sorry, but I think the “capacitance that you are adding” exists only in your imagination. I don’t know your background so I apologize if I’m patronizing. A capacitor adds resistance (technically “impedance”) to the circuit as a function of changing V. If V is changing pretty slowly, you’d need a helova lot of capacitance in order to have any effect. The visible traces are so far apart that they would have capacitance for squat unless the change of V was in the Mhz range.

I don’t think that the sweeper arm moves fast enough that there’d be much need to smooth the signal by adding capacitance or inductance. I don’t think the design of the circuit board adds any capacitance at the rates of V change it’s going to see.

Likewise, inductors react to changing I (current). If I changes fast, the inductor adds resistance It’s really impedance, but in DC applications it’s easier and reasonably accurate to think of it as resistance.

Re. your examples. I’d go for more simple ones. Capacitors don’t like changing voltage. If voltage changes fast, the capacitor acts to resist that change in voltage. If the change of voltage is not so much, it takes a lot of capacitance to get the job done.

Inductors don’t like changing current. If I changes fast, the inductor acts to resist that change. If the change is not so much…

In each case the “acts to resist that change” takes the form of adding “impedance” to the circuit. Normal folks don’t have much of a feel for impedance so it’s easier to think of it as resistance, except that the resistance is a function of rate of V or I change. Fast V change means more resistance, for example.

#67

AT WOT, the AFM becomes a switch to tell the ECU to tune the motor with the factory installed map. Unless said switch is unstable, there should be fluctuation in signal. So, theorize that the switch is fluctuating and calling for a different fuel/air mixture intermittently (trying to go to the O2 sensor signal) and you have a really unhappy motor. I wonder what would happen if we just jumpered the WOT switch and tried that?

#68

My background, as far as education, is computer engineering – which is a hybrid of computer science and electrical engineering. That said, most of my professional work has been on the programming side than the hardware side.

The capacitance I’m talking about definitely isn’t made up. I think we might be talking different terms, though – capacitance just means the ability to hold an electric charge. Anything that conducts electricity has a capacitance. Capacitors are components designed to store more charge (have higher capacitance) than the material itself could hold on its own, so to speak. I’m not referring to capacitors, and I’m not suggesting the board layout is meant to create some kind of capacitor component.

You’re absolutely right that the type of circuit we’re talking about might not move fast enough to be affected by the tiny amounts of capacitance we’re talking about. The discussion of trace capacitance and impedance was mostly relevant when discussing high speed (100MHz+) bus design. When you have switching at that rate in a digital circuit, even the tiniest amount of overshoot/oscillation could register as an incorrect 1 or 0 on the other side. So in that case you had to match impedance to the capacitance in your traces. Often this required adding resistors serially to your data lines. But the resistance had to be perfect. Too little and you still get ringing. Too much and the signal can’t switch fast enough.

I get that that was a discussion about high speed digital circuits – and this clearly is not a high speed digital circuit. In slower digital circuits, it doesn’t matter if there’s an oscillation when a signal switches because there’s plenty of time for the signal to stabilize. What I wasn’t (and am still not) sure of is if the concept still potentially applies in what’s supposed to be a fairly precise mechanical/analog sensor application like this. It’s possible it doesn’t… but, again, I’m just spitballing theories to try and explain the trace layout.

Regarding the multi-layer boards, totally on the same page about multi-layer boards being around for decades. What I was curious about was if it was done without vias. Connections between layers in a board are typically done with a drilled hole + plating (vias). Kind of like a rivet. (that explanation is just for anyone else happening to read the thread, btw) The circuit in the AFM doesn’t have vias in the areas where the traces go off into the middle of nowhere. I wasn’t sure if there were ways to connect layers without the drilled holes that were common in the 80s. I’ve since done some more searching and a concept of “solid” or “filled” vias appears to be a thing, which I didn’t know about. Not sure how common it is or why they’re used, but that would be a good explanation for the random traces.

Som