I meant that F(in) = F(out) when it shouldn't have. Both chips have decoupling from V+ to GND pins. I consider that as indicating it is picking up noise from the signal generator somewhere, perhaps in the F:V output voltage. I don't consider it a major issue, since when the rest of the circuit is cleaned up, it will probably disappear.

John

 

Update

I read some more and played a little. Here are some settings that work to the extent of giving a nice output, but it is not linear. You'll need to mess with the scaling factors. V+ was 14V throughout. All values were as shown in the datasheet with exception of the comments below.

FVCFigure 5 (12 Hz = 1.3V; 115 Hz = 13.2V)

R\(_{s}\)= 14K
C\(_{\scriptsize B}\)= 10 uF electrolytic (reduces ripple on output)
R\(_{\scriptsize B}\)= 100K
C\(_{\scriptsize O}\)= 0.1 uF
R\(_{\scriptsize O}\)= 6.8K

VFC Figure 2

R\(_{s}\)= 10K or 8.2K (changes scale)
C\(_{\scriptsize B}\)= 0.1 uF (works up to 0.47 uF)
R\(_{\scriptsize B}\)= 56K
C\(_{\scriptsize O}\)= 0.1 uF
R\(_{\scriptsize O}\)= 15K
R\(_{\scriptsize L}\) = 5.6K

These values give a ratio of 3.4 to about 5.0 output:input frequency. Unfortunately, it is not very linear. The tine low (t\(_{\scriptsize low}\)) is about 2 mS. Time high varies with the frequency. You can go lower on RS, but when I went higher, the range was decreased at the top end.

Good luck.

John

 

First change was to get rid of the ripple in the output of the F:V converter. I changed Cb from 1uF to 10 uF. Now that everything is running, if I go back to 1 uF, the output gets into sync with the input frequency. The larger capacitor may affect response time. Eventually, when you add the Op Amp, it may not be necessary.

I added a 10uf on Cb (on the F:V) and that straightened it right out. The response time was pretty sluggish, so I experimented with different values for Cb and found that 4.7 worked best. anything less and I had that oscilation. anything more and the response time was worsened.

Second change was to add a filter capacitor (0.01 uF) from the output at pin 3 of the V:F converter to ground. This got rid of spiking in the output. You might go smaller, but a larger one causes too much distortion.

I did this too. I didnt see much improvement to the spiking, but I did see some.

with these changes I find that my duty cycle is close to 50% throughout the range. response time isnt great, but I would attribute some of that to my scope and meter.

Ive got plans to test this out on the bus tomorrow afternoon so I will have some real world results to report then. Thankfully my protostation has a 9v battery supply in it, and Ive made sure my circuit operates from it just fine.

I have noticed that the linearity is a little bit off as you said. so what I did was dial it in at 50hz to exactly 1:2 and then the output is a little high at first, and a little low at the full scale instead of totally wrong at one end.

thanks again for all your help on this. I dont think I could have got this far without you.

 

I did this too. I didnt see much improvement to the spiking, but I did see some.

I changed that capacitor to 0.001 (i.e., 102) for the same reason. With the stable circuit, I agree, it doesn't make much difference.

Ive got plans to test this out on the bus tomorrow afternoon so I will have some real world results to report then.

I will be interested to find out how that bus does when its cranking out 140 Hz.

I have noticed that the linearity is a little bit off as you said. so what I did was dial it in at 50hz to exactly 1:2 and then the output is a little high at first, and a little low at the full scale instead of totally wrong at one end.

If the linearity problems aren't solved with an op_amp, and if you can live with an exact 1:2 ratio, there are easier ways to get that.

Please be sure to give an update.

John

 

My friend Brandon already built a vr6 bus and he thinks that it will top out around 115 with the present gearing. he even raced it once.
http://www.youtube.com/watch?v=3HaVBHyREd4

hes helping me build another one just like it out of my 78.

 

I guess second in a two man race isn't all THAT bad.

The VW buses I remember from the 60's burned up the engine at 60 mph. I had a 1965 VW 1300 (license VAC 365, California). One of the last real VW's. Bought it new at $1700. Blew an exhaust valve at 70K miles, and I overhauled the engine in our apartment's garage. Dropped it down onto a pair of roller skates. I lifted the rear of the car and my wife pulled it out. Just 4 bolts held it to the transaxle. Great memories. Have fun. John

 

thats exactly right. I found that out the hard way when I took my 68 on the turnpike and kept it at about 75 for 30 miles. ceized it right up.

4 bolts to take it off, and I threw the whole thing in the back of my subaru WRX and was able to close the hatch. - and I didnt even have to fold my seats down. that total rebuild with performance everything cost me only 1500.

 

its testing day! we are heading over to Brandons to hook up this circuit, and we will follow a car going 50 mph and dial in the output ratio. it will be great to finally find out what the exact ratio will be for his tire size and instrument combination.

Wish me luck!

 

A beter way & acurate use a GPS in the vehicle to check speed. (if you dont have one maybe borrow one)

 

Thats true Debe. since we didnt have another car with a working speedo present, we decided not to bother with the actual traveling speed. instead we made the adjustments without moving the bus at all. since we know what the actual circumference of the tire is and the amount of pulses per wheel rotation that the speed sender emits, we could calculate the exact speed the car will run at 100hz. I used a drill to spin the speed sender, then we just adjusted the pot till the speedo ran at the required speed.

As it turns out, the final ratio of input to output is actually about 1:1.77. not 1:2+ as I first guessed.

 

I suspect there is a pretty good relationship between tire circumference, rotations, and speed. For greatest accuracy, is it necessary to adjust the measured tire circumference to something equivalent to the pitch diameter/circumference in a flat-belt drive system? Also, is some estimated value for slippage considered?

In other words, maybe the actual ratio when calibrated against GPS or other accurate system will be closer to 1:2.

John

 

the stock speedometer is tuned from the factory based on the stock tire size.
we know exactly what the dimensions of the current tire is, so we are able to deduce the exact amount of wheel rotations per mile per hour, so once we spun the speed sensor with a drill at the exact rpms needed to simulate a 50 mile per hour travel speed, we were able to adjust the converter to produce that speed on the cluster.

stock speedometers are calibrated to read up to 5% fast. this is the amount of error allowed by law. car manufacturers exploit this variance so that consumers believe they are getting better mileage and speeds. I believe there is a lawsuit or something in the works to stop the auto manufacturers from doing this. Look it up.

 

Ford Australias policy was to make the speedo 5 percent fast to avoid litigation from customers for speeding. There instrument panels were made by Bosch. Police instrument panel speedo was different.

 

update: got my design all worked out and layed out in EAGLE.

optimized for a compact single sided board only 1.3"x2.5"

I even threw in some power planes for good measure.

Etching should happen tomorrow. Hopefully THIS TIME I dont melt my eyeballs out with the potassium chloride developer...

 

Good luck. Since you mention developer, I assume you are using the photoresist method. Have you done that before? Have a clear head when you print and transfer so you don't end up with a mirror image of what you want.

BTW, the active ingredient in the developer is probably potassium hydroxide, not KCl; although, some KCl may be there for other reasons if you are using a commercial developer. I make my own developer from solid KOH or NaOH. Be sure to use rubber gloves and something to protect your eyes.

John

 

It looks like your Vcc supply plane is shorted to your ground plane in the vicinity of RSIG2.

Did you run Erc in the schematic and Drc in the PCB?

Better add supply labels to your Vcc and GND rails; that will help to prevent errors from biting you too badly on your board. Look in supply-1.lbr and supply-2.lbr. I prefer the VCC and GND labels in supply-2.lbr.

You might want to add some mounting holes in the corners.

 

It looks like your Vcc supply plane is shorted to your ground plane in the vicinity of RSIG2.

Did you run Erc in the schematic and Drc in the PCB?

Better add supply labels to your Vcc and GND rails; that will help to prevent errors from biting you too badly on your board. Look in supply-1.lbr and supply-2.lbr. I prefer the VCC and GND labels in supply-2.lbr.

You might want to add some mounting holes in the corners.

Wow! GOOD EYE! I missed that HUGE short. guess I made my polygons too close together.

good idea about the mounting holes and hookup symbols. Ill get right on that.

 

Good luck. Since you mention developer, I assume you are using the photoresist method. Have you done that before? Have a clear head when you print and transfer so you don't end up with a mirror image of what you want.

BTW, the active ingredient in the developer is probably potassium hydroxide, not KCl; although, some KCl may be there for other reasons if you are using a commercial developer. I make my own developer from solid KOH or NaOH. Be sure to use rubber gloves and something to protect your eyes.

John

yup. I will be trying the photoresist method here again. the chemical developer for the photo resist is actually potassium carbonate, mSDS here http://www.mgchemicals.com/msds/english/4170.pdf

Ive done it before with some obd2 adapters I made with mixed results. and yes, I melted my eyeballs because I wasnt careful.
and I use ferric chloride to etch the boards. - altho, my FCL is pretty tired.

Ive also done the "Iron on" method before, - mixed results on that one too. but I read an instructable where someone printed the image onto a glossy magazine page (cosmo) and said it transferred best, so Im considering trying that method too.

 

When you say mixed results, what happened? Was the resist hard to remove completely, which would interfere with etching, or did too much get removed so the traces got too thin?

There are different ways to approach each problem. BTW, I use Injectorall boards (DigiKey). Potassium carbonate has almost no effect on them, which is why I use the stronger alkali.

John

 

When you say mixed results, what happened? Was the resist hard to remove completely, which would interfere with etching, or did too much get removed so the traces got too thin?

There are different ways to approach each problem. BTW, I use Injectorall boards (DigiKey). Potassium carbonate has almost no effect on them, which is why I use the stronger alkali.

John

yes to all above. my original forays into making my own boards used the iron on laser print method from glossy white staples photo paper. its a terribly time consuming process, and has lots of defects. broken traces, distored images, etc.

then I used presensitized boards and sodium hydroxide to develop them, and either too much resist is removed or not enough. plus my etchant is old and my parts have to stew a lot longer than I would like, and that might be causing problems as well. lots of broken traces, or washed out completely.

Next I used the film that you use a laminator to put on the board and the developer made specifically for this by the same company.
http://www.mouser.com/ProductDetail/MG-Chemicals/416DFR-5/?qs=fFKnDw9uGJa5rbAs3tEmyg==
datasheet http://www.mgchemicals.com/downloads/pdf/specsheets/416dfr.pdf

since I dont have a laminator, ive been using an iron. that is probably part of my problem too. again, broken traces, insufficient or over developing.

I Do have a proper UV exposure lamp. (not a black light) and I am exposing them for the reccomended time through two layers of LASER printed transparency film so my image is very opaque.