Anyone know what happened to Bob (Techguy512)?

[H-townbenzoboy]

Bob was a member here had a website called mbelectronicupgrades.com. He lived in Austin and sold electronic upgrades catered to late 70s and early 80s MBs like digital tach amps and programmable wiper speed relays. He last logged in here in 2015, and his website went dark around the same time. I had a cell number for him but I tried it today and it's no longer his phone number. I figured one of the other old-timers here kept in contact with him outside the forums and knew of his whereabouts? I know some owners who'd be interested in some of his products.

[Jarod]

I believe I asked this some time ago as well and I attempted contacting him via various means, and got no response.
I'm planning on following in his footsteps building a better tach amp.
The programmable wiper relays are easy, you can adapt one from an older VW with a few jumper wires that have ends that fit the MB socket

[H-townbenzoboy]

I think his last name is Emerson, but Google isn't helping me much in that regard. Of course, I could be wrong about his last name. If had a package from him with his return address, I threw it away long ago.

[250 Coupe]

During my tach amp research, I couldn't find him either. Since the chip used in the original amp is available in both through hole and SMD, I started drawing up a PCB layout but gave up when I figured out my issue is the pickup and not the amp.

I was able to test the tach itself with a 5VDC square wave

Going the microcontroller route seems like a cool idea, I've never actually completed a design as it turns out I'm a terrible programmer with little idea of what I'm doing.

Michael

EDIT, the picture is a bit fuzzy, the scope is showing that this is a 48 Hertz, 5 volt peak to peak square wave.

[Jarod]

I've been gathering info on how the tach amp works, Knowing 48hz equates to 3k rpm is something I had been unable to find.
I acquired a spare cluster with tach and wiring so I'm planning on wiring it all up to a microcontroller, I'm br far not the best at designing circuits and programming but I've fumbled my way through a few custom PCBs and some adaptations like making a wireless driveway alert send me emails when someone pulls in my driveway, so a tach amp seems fairly simple.

I plan on making a setup likely on a breadboard that will plug in and digitally sweep the tach and play with it, then move to actually amplifying the signal from the pickup and making a functional replacement, and from there I have to figure out what the best option is to make it accessible to others.

[250 Coupe]

One of the things that confuses me, due to my poor math skills no doubt, is that there is one pulse from the sensor per revolution which by my calculations comes to 180KHz at 3000 RPM. Clearly, I'm not understanding something.

I did review the video I took running the signal generator up and down and it looks like 1K Hz is pretty close to 6000 RPM.

Lets see if this link to the video works.
https://photos.app.goo.gl/dwKmuy9rRsa3awst5

Michael

[Felching]

I have these from contact with him back in the day. Haven't heard from him since that 2015 range, I'm assuming he has passed.

[H-townbenzoboy]

I found a personal earthlink e-mail address for him and sure enough, it bounced back. Going by the e-mail address, his last name began with an "M", so that may be why I'm not finding anything on him now. I thought it was Emerson, guess I was wrong. Edit: Found out his last name is Myers. I'll see what that turns up.

[tangofox007]

Originally Posted by 250 Coupe

One of the things that confuses me, due to my poor math skills no doubt, is that there is one pulse from the sensor per revolution which by my calculations comes to 180KHz at 3000 RPM. Clearly, I'm not understanding something.

That's what happens when you multiply when you should be dividing.

3000 rpm = 50 Hz.

180 KHz = 10,800,000 rpm.

[H-townbenzoboy]

I found out via a post on ClubGwagen that he was fighting cancer at some point. Sounds like he may have lost that battle. https://clubgwagen.com/forum/viewtopic.php?t=6933

[ykobayashi]

There’s almost enough info in the image to reverse engineer this thing. The TO92 at the 1 o’clock position looks like a 5v linear regulator. A 78L05 to knock down the 12v battery to 5v for the Atmel controller. Q1 is a little MOSFET or BJT to boost up the coil signal to something friendly for the Atmel input.

The “program” is just reading an input and assigning it to an output.

I looked at this design back in the day and wondered why he even used the microcontroller. He’s basically using it as a wire using a line of code to do it. Kind of a waste...but, it probably is priceless to control the diagnostic lights. Those will save you a tech support call. It did seem like overkill and it adds $3 to the design + programming (burning).

I think an easier way to do it is use a dual op amp to threshold the input with some hysteresis and then buffer the output signal to the tach. My weakness is how to make an amp input for the coil but some googling with make that clear.

I used to be an embedded systems designer ten years ago.

Is this really important to recreate? I just touched up the solder on mine (82 Sd) in 2010 and went on with life.

Looks like these amps are NLA now?

Edit - ok like usual I spoke too soon without doing my home work. The LM1815 that MB used is a tricky variable hysteresis adaptive feedback amplifier. I bet the Atmel risc microcontroller is sampling the signal with its adc and doing adaptive digital thresholding on the signal. The problem with out crank sensor is it puts out a different voltage depending on speed. This needs to be compensated. I was thinking you could just take the input, amplify and threshold and POW, you have a signal you can feed into the tach. If it were that easy MB would have done that.

So it looks like Techguy used a diode clamped BJT transistor input (three pin device at 6 o’clock) to get a signal the microcontroller could accept. Probably cribbed the circuit from the original unit. You can kind of see it on the reverse engineered schematics. It looks the same.

Then he samples using the ADC and does thresholding with some code. He adds adaptive hysteresis based on the inputs to get rid of noise.

It may be easier to just lay out a circuit with the LM1815 and be done with it.

[ykobayashi]

Ok, I’ve spent some time reading the LM1815 data sheet. I also visually reverse engineered the red circuit board in Techguy’s image.

Here is what I understand. Techguy appears to have reverse engineered the LM1815 circuit input stage. It’s a PNP transistor used to put a bias on the rpm coil. From what I can tell it is the same circuit at pin 7 one one side of the coil. Instead of running the other side of the Coil pin 9 to the LM1815 he runs it into probably an ADC port on the Atmel controller.

Once on the Atmel controller he does some signal processing to the pickup signal. The problem, according to the LM1815 data sheet is the signal is noisy and varies in amplitude while the speed of the engine varies. Move the magnet by the coil faster you get more voltage - Faradays law. So you cannot trigger your pulse circuit just by saying everytime the wave form crosses zero give a high pulse and send it to the tach.

You’ll get false triggers from noise because depending on the relative location of the coil and magnet as it whizzes by,, the location of the zero crossing can be all over the place. So the LM1815 does a simple trick. It finds the peak of the signal then triggers on the first zero crossing it sees after that. It then ignores all other zero crossings till it sees a new peak. I guess this is the “algorithm” he programmed in the Atmel chip. For some reason just finding the peak isn’t robust enough.

Or maybe it wasn’t as easy using old school analog circuits. Maybe it’s more accurate to detect the zero crossing.

So basically the Strategy is to sense the peak, open a window to look for the first zero crossing, then close it up till you see the next peak. This eliminates zero crossing noise which is around when the magnet is far from the coil. This is all shown in pictures in the data sheet for the LM1815 at Ti.

https://www.ti.com/lit/ds/symlink/lm1815.pdf

Too bad we cannot get at the original design files. Even though it’s a pretty basic project it’ll probably take a couple of weeks of work to recreate the code and pcb. I recall the Atmel Atiny tools are free and simple. The hardest part of the PCB is measuring out the pin locations for the original tach amp for a proper fit. Then sourcing those pins...I’m not sure who makes them.

[250 Coupe]

Originally Posted by tangofox007

That's what happens when you multiply when you should be dividing.

3000 rpm = 50 Hz.

180 KHz = 10,800,000 rpm.

I can't even calculate my fuel mileage right on the first go. I almost always get a decimal fraction the first time but do get it right the second time.

The funny thing is, considering I'm partially color blind, making reading resistor color codes a chore, and that I don't math good, I do have a 2 year degree in electronics. I can fix things that used to work but probably shouldn't create new designs, just saying...

Michael

[250 Coupe]

@ykobayashi
I wondered why he went the microcontroller route since it's such a simple circuit but figured it was more for the adventure than about repairing a broken one. There's a guy at work that has a dozen or more microcontrollers and Pi's around his house doing tasks that maybe didn't need doing.

Michael

[250 Coupe]

@ykobayashi
He may have reused the original pins and scanned the white nylon disc to trace the pattern and dimensions from.

This is what a depotted amp looks like. I thought mine had bad solder and suspected the electrolytic cap but am now pretty sure it's the pick up as I see no signal from it on a scope. I did verify that I have the bias voltage on the board but it's hard and messy to actually measure the bias current with the amp installed.

Michael