My 140 mph speedo conversion came with a 7.00 mHz crystal, which caused it to read about 35% low. Using one of the formulas available on PFF, I was certian that a 3.81 mHz would work. Here is what I just can't understand: With a 4.00 mHz, the speedo now reads about 21% high. How can this be? I assumed that lowering the frequency would bring the error down close to zero, not change it to reading too high. Has anyone else had this happen? I have no idea what frequency to try now!

Bummer, I was hoping you had it figured out and could help me out. It has been a while since I have messed with it, but I remember having simular results. Put a higer crystal in and it goes too slow, try a lower one and still too slow. I never found a pattern to it and gave up. We need to find an adjustible crystal !

Changing the crystal in my opinion is the wrong way to go. It's too hard to get it right. Also, it is very likely that there is a limit to the range the circuit will go to. It's old tech after all.

I'm working on a solution to adjust the speed pulse before it gets to the speedo. The challenge is getting a fine enough adjustment without things getting too complicated and expensive.

I'll post more as I find solutions.

DG

------------------
You Dream it Up -- I'll Build it

7.67 Mhz is the frequency, as far as I know.

when I got my crystal, I couldnt get exactly 7.67, I got the next closest frequency - dont remeber what - but it was close. and the speedo reads a little low. I switched the VSS gear in the tranny (from blue to red), and that got it pretty close. when I bought it - I had to get a batch of 5. More than happy to give ya one, if I can find it.

Has anyone ever looked in to HealTech Electronic's Speedo Healer?

It's primarily marketed at motorcycles, however if you dig far enough in to the product description it claims to be compatible with "All bikes, ATVs, cars, vans, trucks and snowmobiles which have 3-wire speed sensor or 2-wire ground switch sensor."

[This message has been edited by dguy (edited 09-01-2010).]

Yes that is exactly what I'm building but way less complicated. That unit is designed for people who change their setup frequently.

It's too big to fit in the gauge cluster, it's more complicated than it needs to be and it's almost $120. And you still need a gauge face.

I'm aiming for somewhere in the $80 range for everything.

With the fancy programmable chips out there now it's doable.

DG

------------------
You Dream it Up -- I'll Build it

Well, at least I'm not the first to experience this. Yeah Mike, I thought I had it figured out too! Well, it sounds like Dawg is working towards a solution to our dilemma, in the mean time I am going to try some frequencies between 4.0 and 7.0 mHz. If I come across some pattern, I will pass it on. Thanks

I spent some time today reverse engineering a 140 km/h speedo. I've got it pretty much figured out. Here's an image of me fooling the speedo into thinking it's hooked to the car. Full deflection occured with a VSS signal of 98 Hz (cycles per second).





I've got some test printouts I did to see if I could generate something opaque enough to look good. You be the judge:




I should have some finished overlays done this week for all to see. If not, I'll be on holidays for 2 weeks before I can work on this again.

I could use some feedback as to what people would actually be willing to purchase for no more than $100 exchange+shipping.


DG

------------------
You Dream it Up -- I'll Build it

Check out the following thread on speedo calibration:
https://www.fiero.nl/forum/Forum2/HTML/110335.html

Oliver Scholz (he lives in Germany) has created a number of interesting electronic items for Fieros and has examined the speedo also - see his conclusion to the proposed method of speedo calibration that would have been common to his tachometer calibration procedure. The response Oliver (aka fieroluke) posted on the tread was in response to a personal email I sent him. Also, check out Oliver's site for some really interesting stuff, like auto close and auto open for power windows: http://www.fieros.de/en/main.html

Having accepted Oliver's assertion that speedo calibration is beyond what is practical, and suggesting the Dakota Digital converter as the last ditch but apparently undesirable choice, I and likely many others would be very impressed if you came up with a practical solution.

------------------
Daviero - 88 N* GT

Well I've read all those links and the only part I'm not understanding is why the Dackota unit is undesirable. There is mention of harmonics buggering up the readings or some such thing.

The Fiero speedo takes the signal from the VSS which is a sine wave and converts it into a square wave right away. Internally, it's all digital from that point on. I don't see how harmonics would come into play. Yes a square wave has lots of harmonics but the digital chips don't use anything but the fundamental frequency. It's possible that the signal coming from this converter is "dirty" somehow.

With todays programmable chips, the variable division needed to pull this off is not a problem. Can it be done for the right price? Time will tell.

DG

Ok my Fiero homies, have a look at this.



This version has the font and style of the original Fiero gauges. No one will ever know it didn't come that way from the factory. Being Canadian, I made a metric gauge face of course. But fear not, the MPH version is being worked on....

Cheers,

The Dawg

------------------
You Dream it Up -- I'll Build it

Very cool. Wish I had your knowhow when it comes to electronics! I've bookmarked this so keep us posted.

Well this might be all I'll get done till after my holidays. Here's the MPH face.




There seems to be some interest in tach faces with higher red lines. I've scanned a stock tach and will be playing around with different ideas.

Stay tuned.

The Dawg.

------------------
You Dream it Up -- I'll Build it

quote Originally posted by Dawg: Well I've read all those links and the only part I'm not understanding is why the Dackota unit is undesirable. There is mention of harmonics buggering up the readings or some such thing. The Fiero speedo takes the signal from the VSS which is a sine wave and converts it into a square wave right away. Internally, it's all digital from that point on. I don't see how harmonics would come into play. Yes a square wave has lots of harmonics but the digital chips don't use anything but the fundamental frequency. It's possible that the signal coming from this converter is "dirty" somehow. With todays programmable chips, the variable division needed to pull this off is not a problem. Can it be done for the right price? Time will tell. DG

because changing the pulses to force the speedo to read a certain way will also change the odometer increments.

Unless you tap into the signal just before the speedo. The speedometer is the last to get the signal. So converting this signal at that point won't bother the odometer, trip meter or ECM.

This will never be a plug in kit I'm afraid. As you've stated, it would bugger up too many things.


The Dawg

quote Originally posted by Pyrthian: because changing the pulses to force the speedo to read a certain way will also change the odometer increments.

------------------
You Dream it Up -- I'll Build it

quote Originally posted by bc12dJohn: My 140 mph speedo conversion came with a 7.00 mHz crystal, which caused it to read about 35% low. Using one of the formulas available on PFF, I was certian that a 3.81 mHz would work. Here is what I just can't understand: With a 4.00 mHz, the speedo now reads about 21% high. How can this be? I assumed that lowering the frequency would bring the error down close to zero, not change it to reading too high. Has anyone else had this happen? I have no idea what frequency to try now!

I don't understand why the 7 MHz crystal didn't work. When I was trying to convert a 140 speedo to MPH, the 7MHz frequency was perfect for the conversion, and displayed the same as the 120 MPH did on my bench. My problem was finding a 7MHz crystal that wasn't going to cost me an arm and a leg, so I never finished the conversion.

Did you notice an error through out the whole range of the speedo with the 7 MHz crystal?

------------------
Nick D.
'88 Fiero GT 5-spd
'03 Jetta GLS TDI 5-spd

Part of the problem is that everyone assumes that all these speedometers are identical inside. It's likely that the divider circuit changed over the years. In other words, different crystals are doing the same job. So there might not be any one formula that will work.

Not only that, but these crystals degrade over the years. It takes very little change in the crystals to make large changes in the reading.

The Dawg.

------------------
You Dream it Up -- I'll Build it

What did you do those face-drawings in, and would you by any chance be willing to email me the files in some vector format?

I have a 140kph speedo that I'm going to convert to a 140MPH or other, and would love to be able to print my own face plates out, and I'm too lazy to whip the lines out on AI..

------------------


Currently in the middle of my 88 + 3800NA swap

quote Originally posted by Dawg: Part of the problem is that everyone assumes that all these speedometers are identical inside. It's likely that the divider circuit changed over the years. In other words, different crystals are doing the same job. So there might not be any one formula that will work.

Exactly. According to Oliver Scholz, who probably knows more about the issue than anybody, there are at least three different versions of the Fiero speedometer out there, with three different clock divider ratios ... which means at least three different crystal frequencies. See this thread for a more thorough discussion.

quote Not only that, but these crystals degrade over the years. It takes very little change in the crystals to make large changes in the reading.

I disagree. When quartz crystals fail they usually do so abruptly; either they work or they don't. Yes, their frequency may drift slightly with temperature and as they age, but that frequency error is usually measured in parts per million (i.e. ppm = 1/10,000 of a percent). Any change in speedometer reading will be exactly proportional to a change in crystal frequency. A 100 ppm error in crystal frequency would produce a speedometer error of only 1/100 mph, so the crystal frequency would have to drift at least 1% before you could detect it by simply looking at the speedometer.

On the other hand, components like capacitors can and do drift in value by 10% or more with changes in temperature and as they age. Electrolytic capacitors are evn worse; even new electrolytics will vary by +100/-50% from their nominal value. Perhaps that is what you were thinking of.

[This message has been edited by Marvin McInnis (edited 09-08-2010).]

Well, I come from the land of RF circuits and the accuracy in most high frequency circuits needs to be so high that small errors become significant pretty darn fast.

But I believe you are right in the is case. I stand corrected.

Having said that, I haven't taken the time to reverse engineer the divider, so I'm not clear on how the oscillator interacts with the incoming signal. Is it simply providing a clock pulse or does it directly control the division itself?

I don't believe there are any capacitors involved in this case. The tach however is another story....

The Dawg

quote Originally posted by Marvin McInnis: I disagree. When quartz crystals fail they usually do so abruptly; either they work or they don't. Yes, their frequency may drift slightly with temperature and as they age, but that frequency error is usually measured in parts per million (i.e. ppm = 1/10,000 of a percent). Any change in speedometer reading will be exactly proportional to a change in crystal frequency. A 100 ppm error in crystal frequency would produce a speedometer error of only 1/100 mph, so the crystal frequency would have to drift at least 1% before you could detect it by simply looking at the speedometer. On the other hand, components like capacitors can and do drift in value by 10% or more with changes in temperature and as they age. Electrolytic capacitors are evn worse; even new electrolytics will vary by +100/-50% from their nominal value. Perhaps that is what you were thinking of.

------------------
You Dream it Up -- I'll Build it

quote Originally posted by Dawg: ... I'm not clear on how the oscillator interacts with the incoming signal. Is it simply providing a clock pulse or does it directly control the division itself?

According to Oliver Scholz (an authoritative source), the crystal is used for the master clock in the microprocessor. The microprocessor uses hard-coded clock-counting loops to calculate meter deflection based on VSS pulse frequency. If you want to change the speedometer range scaling, you can use an external box to rescale the incoming VSS pulses (number of pulses per mile), you can change the hard-coded divisor in the firmware, or you can "trick" the microprocessor by changing its clock frequency. External rescaling has the disadvantage that your odometer will now be wrong (because it just counts VSS pulses directly), and since the divisors in firmware aren't available to us to change, the crystal is the only other alternative available to us.

Oliver and I both agree that GM probably purchased the cheapest off-the-shelf crystals that they could, and then set the divisors in the firmware to match the crystals for each batch of microprocessors they ordered. This theory is supported in part by the suspiciously-common (and thus cheap and widely available) crystal frequencies of many of the OEM crystals, e.g. 4.430 MHz.

The fact that the crystal is used for the microprocessor clock may also explain why different-frequency crystals may or may not work correctly. GM would have used the least expensive microprocessor available to them, so if the design called for a 4.430 MHz crystal they probably wouldn''t have used a 10 MHz-rated microprocessor if a 5 MHz part was available at lower cost. So what happens when you replace the 4.430 MHz crystal with a ~7 MHz crystal in a 5 MHz microprocessor? Due to normal manufacturing tolerences, some will probably work OK and others will not. With a 7 MHz crystal in a 5 MHz microprocessor (a 40% overclock), the clock rate may end up being half the crystal frequency ... 3.5 MHz (first subharmonic) instead of the expected 7 MHz. Or the uP clock may run erratically, alternating between 3.5 and 7 MHz bursts, yielding an effective clock rate significantly below 7 MHz. I've seen both happen. In early microprocessors (i.e. 1970s), the maximum clock rate and clock pulse timing (duty cycle, skew, etc.) were specifications you really had to respect; overclocking was sure to result in nonstop engineering headaches. The rule of thumb for high reliability back then was to design for a clock rate at least 10% (and preferably 20%) below the rated maximum.

[This message has been edited by Marvin McInnis (edited 09-09-2010).]

That is overcome by breaking the signal path just before the speedo. Nothing else is effected then.

For all the reasons you have just stated and more, I think the crystal method is the less desirable. One big advantage of the rescaling as you call it, is that theoretically, one could construct a variable divisor circuit that could take into account different sized tires for instance. Something not possible using the crystal method. It would be a one solution for all varieties of speedo no matter when they were made.

I suspect the sample rate will have to be fairly high to have something to divide smoothly, time will tell.

The Dawg

quote Originally posted by Marvin McInnis: External rescaling has the disadvantage that your odometer will now be wrong (because it just counts VSS pulses directly), and since the divisors in firmware aren't available to us to change, the crystal is the only other alternative available to us.

------------------
You Dream it Up -- I'll Build it

quote Originally posted by Dawg: That is overcome by breaking the signal path just before the speedo. Nothing else is effected then. For all the reasons you have just stated and more, I think the crystal method is the less desirable. One big advantage of the rescaling as you call it, is that theoretically, one could construct a variable divisor circuit that could take into account different sized tires for instance. Something not possible using the crystal method. It would be a one solution for all varieties of speedo no matter when they were made. I suspect the sample rate will have to be fairly high to have something to divide smoothly, time will tell. The Dawg

I'm liking your thinking. it would be nice to go back to the original crystal, and basicly have a "knob" to adjust the speedo, while keeping tho odometers accurate.

I did the crystal swap, and got it close, and tuned it closer by changing the VSS gear - but, I am still about 5% low. I was gonna do a final tune by running the VSS thru the 7730 ECM, which does minor speedo adjustment as well. just ends up being a long chain of adjustments - pretty chaotic.

Hi everyone,

since I'm referred to a few times in this thread, I thought I'd give my $0.02

I think the cleanest (not the simplest!) solution would be to design a plug-in board eliminating the old GM processor with a new "state-of-the-art" processor of today's technology, and make it externally adjustable with a pot. The simpler CPUs cost about a buck each.

The advantages:

- Can be made to support a wide range of speedos from 85-200 mph.
- Could be made for cheaply if more than a few are needed (circuit boards are really the cost issue b/c etching isn't for everyone)
- Doesn't disturb ECM or odometers if it's just replacing the GM CPU controlling the coil driver IC, it's a simple upgrade in that respect

The disadvantages:

- Only cheap/simple if circuit boards are made centrally, CPU programming requires special HW as well not everybody has access to
- Automotive temperature range CPUs may be more difficult to get (but absolutely necessary!)
- Installation more difficult (~7 wires) than crystal swap, but "infinite" adjustment may justify that
- GM coil driver needs 8V, not 5V if I remember correctly. Some experimentation is needed before a reliable circuit can be designed
- Lots of people want lots of different gauge faces -> the non-electrical part may be an issue


Apparently I've been thinking about that same issue, because the topic comes up regularly.
But, what do you guys think? Is this something people will be interested in/willing to perform?
I figure a finished board will cost some $15-20. Will people be willing to spend that amount? (I don't think so, but prove me wrong...)

Let the discussions begin... ;-)

$15-$20 for a board that would allow infinite adjustment to the Speedo? HELL YES! Sign me up if one becomes available.. I'll definitely go on that list.

Welcome to the 500th discussion of this topic.....

I think some version of the above needs to be done. I'm going to build something for myself and the gang at our Fiero club. I'm talking proto board and so on, nothing fancy. If there ends up being more interest than imagined then great but I wouldn't spend a lot of time developing a professional assembly line just yet. As I'm sure you're aware, the Fiero community is hard to read sometimes.

I'm of the opinion that creating this as a conversion "kit" will be nothing but problems. Yes many will be handy enough to pull it off but many more will not be so lucky. There would be endless support questions flying back and forth. Not my idea of fun....

A speedo exchange is more the plan for me anyways.

I think everyone should prove me wrong and flood this thread with interested buyers......

The Dawg

------------------
You Dream it Up -- I'll Build it

Hi Oliver ... thanks for chiming in.

I like your idea, but lack sufficient motivation to do it myself. My soldering skills are good, too, so I would be very interested. I don't think that most people (even cheapskate Fiero owners!) would be deterred by a price of $15-$20; a custom crystal alone will cost almost that much. But people who lack soldering skills (skills that are increasingly rare today) would probably prefer to pay more to have somebody else do the work. That said, you'll have to sell a lot of kits at $20 to recover your development costs. Be fair to yourself.

Concerning programmability: "Infinite adjustability" with a 10-turn pot is certainly the most flexible way to go, but I would prefer the stability and repeatability of a digital programming method like a DIP switch. Even a 4-bit switch would provide 16 ranges, and an 8-bit switch would provide 256. Since I have a fairly complete electronics shop I would really prefer to just program an on-chip variable, but that's probably way beyond the capabilities of most home shops.

Concerning gauge faces: My experimentation in this area indicates that the air-core meter mechanism and its IC driver are somewhat nonlinear, and that GM probably corrected for this by adjusting the position of the markings on the gauge face. In addition, I believe that GM also tried to compensate somewhat for parallax error by the layout of the markings on the gauge face. For example, I have a 180 mph speedometer conversion (new crystal + new gauge face) that I carefully bench tested using a high-precision waveform generator. The converted speedometer is dead-on at both 0 and 180 mph, and the markings are perfectly linear, but it reads almost 5 mph low in the important 40 - 60 mph range. Operation in the car, comparing speedometer reading vs. both GPS and your DashScan-II, verifies that this error is present in actual road use. If this nonlinearity error (and parallax compensation) is repeatable, as it appears to be, then you could rather easily compensate with a lookup table, a least-squares-fit calculation, or a spline-fit calculation in your programmed chip.

[This message has been edited by Marvin McInnis (edited 09-09-2010).]

Oliver...$15-$20...sure! I might even take 3 (don't ask)

You're probably right as far as support effort is concerned.

And the price estimate was for parts and labor of building them only, not taking into account support.
As far as development effort is concerned: I'm doing this stuff for fun anyway - if I wanted to remotely consider dev cost, almost everything I ever built would not be affordable.

Marvin, you're right, I'd prefer to flash the right value as well, but as you say, most lack the equipment. Dip switches are also something I considered, but they almost cost more than the CPU...

I have a spare speedo for experiments here - maybe I'll find some time for this now that my Fiero is back on the road after 9 months...

quote Originally posted by fieroluke: Dip switches are also something I considered, but they almost cost more than the CPU...

Understood. How about a pin header and jumpers? Far from ideal, but far less expensive than switches. Failing that, how about hard-wired (or soldered) jumpers directly on the PCB traces? Then again, a 10-turn trimpot might work perfectly and be very stable over time and temperature. It's your idea, and you obviously know what you're doing.

FWIW, the digital speedometer functionality in your the DashScan-II is as good an implementation as I have ever encountered, with multi-sample averaging, very little LSB dithering, etc. You may very well be able to improve the operation of the OEM Fiero speedometer, in addition to any rescaling.

[This message has been edited by Marvin McInnis (edited 09-09-2010).]

Solder jumpers are probably best, basically free, and you can set the max digitally.
You won't need to set the max finer than 5 mph, then 5 jumpers could set a range from 85 to 245 mph...
Plus another one for mph/km/h...
The pot solution would not be accurate over temp and humidity, and precision pots are expensive...

You could do a step-push button programming:

Like, use a gps or phone to drive a steady 50MPH and push the button which would adjust on the fly until the needle read correctly.

------------------


Currently in the middle of my 88 + 3800NA swap

Hmmm. Doesn't sound precise either. With the board attached to the speedo it'll be an effort to mount a pushbutton, and operate it while driving and watching a GPS.

But first things first: let me find the time to rig up something to control the needle with my own processor.
Once that's done, I can work out the details and make things as simple as possible...

quote Originally posted by aaronkoch: You could do a step-push button programming ...

quote Originally posted by fieroluke: Hmmm. Doesn't sound precise either.

I think that's a good idea, actually ... kinda' like you did with the switch on the DashScan-II. Use one (round robin) or two (up/down) switches to set a precise scaling variable in flash. (I would prefer using a single momentary, center-off, SPDT toggle switch ... i.e. on-off-on ... but that's just me.) The switches could be used for precise adjustment on the bench or (less desirable) in the car. The switches could be left connected (perhaps through a long pigtail) after installation in the car to allow on-the-road fine tuning, and then removed once the variable had been set. It's the best of both worlds: very flexible field programming combined with the high precision and repeatability of a stored variable.

Please keep us posted on your progress.

[This message has been edited by Marvin McInnis (edited 09-10-2010).]

I'm back from holidays and looking into this again. I'm going to order the development kit and a few chips this week. We've got it mostly figured out in our heads now. My buddy has the code written already. All I have to do is get the hardware shipped from Digikey and we'll be making a prototype.

I will report back when we get the first one working.

I'll probably end up with two versions. One version will be a plain converted speedo (150 MPH/240 KPH) and the other version will be initially set to stock tire sizes but will have the ability to adjust for larger sizes.

Again I'm thinking of $80 for a new gauge face and the non adjustable high speed mod all installed.

I will be making some for our local club first and if there's enough interest elseware I'll make a batch then.

So, if anyone from here is interested, let's see a show of hands.

The Dawg

------------------
You Dream it Up -- I'll Build it

Favourite'd.

Been looking quietly for a good way to ensure my speedo is nice and accurate, this sort of project sounds like the go. I'd be fully up for one.

------------------

quote Originally posted by Dawg: I'm back from holidays and looking into this again. I'm going to order the development kit and a few chips this week. We've got it mostly figured out in our heads now. My buddy has the code written already. All I have to do is get the hardware shipped from Digikey and we'll be making a prototype. I will report back when we get the first one working. I'll probably end up with two versions. One version will be a plain converted speedo (150 MPH/240 KPH) and the other version will be initially set to stock tire sizes but will have the ability to adjust for larger sizes. Again I'm thinking of $80 for a new gauge face and the non adjustable high speed mod all installed. I will be making some for our local club first and if there's enough interest elseware I'll make a batch then. So, if anyone from here is interested, let's see a show of hands. The Dawg

Update.

The development kit arrived yesterday and all the other bits and pieces needed to get the prototype done. I ordered a couple different chips to try out. We are doing all we can to keep this simple. Making things complicated is easy....

I've sent an email to Jeric with all the parameters and he's coding away. I should have the first circuit breadboarded some time next week. If all goes well with the testing, I'll then concentrate on perfecting the gauge faces.

I'll then move this over to The Mall to take orders.

The Dawg.

------------------
You Dream it Up -- I'll Build it

I do believe I have the gauge faces figured out. It took a ton of effort let me tell you.

I've managed to create a face with white lettering in the day light:





And orange glow lettering when back lit.





How does that look everyone? Keep in mind the images are kinda rough. They look even better in person.


The Dawg

------------------
You Dream it Up -- I'll Build it