Another LM2917 Project Some Help Please Speedometer Repair

Discussion in 'General Electronics Chat' started by beenthere, Jul 21, 2009.

  1. beenthere

    Thread Starter Retired Moderator

    Apr 20, 2004
    15,815
    282
    Note to all - this is not my thread. My credit on the header is purely accidental. I even managed to lose the OP's name.

    <OP>

    Well Newbie here. Let me state my application,and that I have read the data sheet. Also I have a limited experience with IC chip replacement. I am using a LM2917N-8 chip to repair an electronic Speedometer from a Kawasaki motorcycle. I have the chip in place on the original circuit board in a IC socket. I have 4 chips, I case I blow one experimenting. I have rerouted my printed circuit to match the circuit needed. 470 ohm resistor on input voltage as well as another zener, C-1 = .01uf, C-2 = 1.0uf, R-1 = 100K, and at R-2 I have a 500 ohm pot to ground at pins 4-7 for trimming. The Full scale voltage needs to be about 3.7v, meter coil resistance is 80 ohm.
    My problem is the input signal, it comes from a 3 wire sensor on the front wheel. 1-ground, 1-B+, 1-signal return. This has a switching voltage between .5-6.5volts, frequency is variable with wheel speed. I believe what I need to do is put the input signal through a "AC Coupling" as I know my signal is not going below ground.
    Can someone help me with how to complete this circuit for the AC Coupling? I found the circuit on the Data sheet but it does not explain the values for the resistors or the connection of the transistor, PNP or NPN? I've never built one before, and it sounds like it would be fairly simple.
    OK I found some info in the forum for this, I have a single pole DC square wave that needs to be changed to a BI-polar square wave. There is a capacitor and some resistors that need to be added to this in-put to LM2917. So still have not found a "clear explanation" of this addition. Please point me to another thread or jump in this one with some help. Thanks.

    So do I take the output from my 3 wire sensor and put it to the input of my LM2917 with a 1uf cap in series or parallel? I read on the Data sheet about inputs through a 10Ω resistor? Is this also needed?
    I have a new post about my speedometer repair using the LM2917N-8 chip.
    It has my schematic/ values. Thanks for such a great forum.

    <end OP>


    <moderator>

    That was a 5 year old thread you hijacked - http://forum.allaboutcircuits.com/showthread.php?t=276

    That is one reason why we ask that people start new threads for new questions. It saves confusion.

    I would try the cap in series with the signal. If the signal is too great in magnitude, you may have to use a resistive divider to reduce it. Can you measure the pulse train with a meter?

    <end moderator>

    <Sgt.Wookie>

    Could this thread be merged in as a reply to Motor head's 1st thread, here?:
    http://forum.allaboutcircuits.com/showthread.php?t=25880

    That would help give some continuity as to his project.

    <end Sgt.Wookie>

    <OP>

    I will wait for the replay on my new thread, Thanks.

    <end OP>

    Apologies to all for such a cruddy merge - moderator did poorly.
     
    Last edited: Jul 21, 2009
  2. Motor Head

    Member

    Jul 21, 2009
    10
    0
    So thanks for the reply and moving it to the thread I started.
    Now about changing this DC signal so it will go below positive voltage, that way LM2917 will operate.
    Will I put a 1.0uf CAP in series, or a different value? What about all this pulling up with a resistor? I see the diagram from National Semiconductor on AC Coupling. I have it posted here along with their Analog Frequency Display circuit. DO I need some resistors as well? I have seen where some circuits show a resistor in series on the input signal, 10Ω. I have also seen some branching off and going back to a positive voltage source, I really don't understand what there are up to. My first post has the info for what I'm working with, maybe some specific help please? Then I'll go buy some more hardware and give it a try. I do have a Fluke 88 that will give voltage and frequency measurements, but no scope.
    Thanks guy's as I have spent a lot of time trying to fix this speedo.
     
  3. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Welcome to the forums, Motor head :)

    Out of curiousity, what year/model?

    I'm curious as to why you added another Zener? :confused:
    And what Zener did you add (part number, please)?

    Haven't reviewed those values yet. How did you arrive at them?
    Are you using polarized or non-polarized capacitors?

    Ahhh, are you planning on sourcing current to the speedo, or sinking current from the speedo? The IC is configured as a current sink.

    Right. The comparator on the input of the LM2917-8 needs the signal to traverse ground. A simple cap and resistor should take care of that; perhaps 0.01uF to 0.1uF and a 10k resistor to ground. The capacitor must not be a polarized type.

    The transistor shown is an NPN. It may not be necessary, because it appears that you have a fairly decent signal coming out of the sensor on the front wheel.

    [eta]
    I've extracted the images from your Word files for easier viewing; they're attached. .PNG format image files are my preference for these forums; they are small in size and are not "lossy" like .JPG files, and don't require loading of big programs like MS Word.
     
  4. Motor Head

    Member

    Jul 21, 2009
    10
    0
    I'll try to answer all.
    First its a 1982 Kawasaki KZ1000 LTD K-2. Its my profile picture.
    Now the reason for the second zener and diode at the Battery voltage after the 470Ω resistor is simply they were on the original circuit, so I left them. I know the chip has a built in Zener.
    The values I selected for the C-1,C-2,R-1 are purely from looking at the data sheet and some other tach circuits. I figured this was a good starting point. The Caps are not polarized, C-1 is a Disk type, C-2 is a canister- film type I believe.
    Current is B+ to speedo meter coil and - current controlled by LM2917 to ground. That was also the original diagram.
    So back to the wave signal, A non polarized cap and resistor to ground. OK added in between the signal coming in and the pin 1 in what fashion. resistor with cap in series, or parallel? Or drop the signal through the resistor and the cap back to ground? I'm sorry but I am confused, I know you guy's are used to how they work but I am a novice and most likely in over my head. Is there a schematic I can look at to get it in my head straight?
     
  5. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Ah, nice bike. I had a 1980 KZ650 that I bought new.
    I see. Well, you should consider removing the extra Zener, as it won't help the voltage regulation; it'll make it worse.
    OK. I'm not going to run through and check them then.

    Take a look at the attached. Sensor connections remain on B+ (V+) and GND, the signal from the sensor gets connected to the cap.
    [eta]
    Note: 100nF = 0.1uF; 10nF = 0.01uF
     
  6. Motor Head

    Member

    Jul 21, 2009
    10
    0
    Ah it was a poor neglected thing when I bought it 2 months ago, skipped the cam chain and bent some valves. Rebuild/ repaint/ recover seat/ rebuild brakes/ change handlebars/ fix & Clean control switches, and still have the speedo.

    So your replies are fantastic!
    I will go get another set of CAPS at the store tomorrow and a resistor. Now would that be a 10Ω? Or a 10KΩ?
    I'll definitely need some more help with this. I'll pull out the other Zener from the input side.
    I had found that some one had opened the gauge up before but it looked to me they had done nothing. I found that the small electromagnetic coil for the analog pointer movement was burned out. I carefully pulled the movement apart and rewound the small coil only losing about an inch of the tiny wire. Then when put back together I found the circuit board supplying full current pegging the gauge, no wonder it burned out.
    I'll get back here again with some more test results tomorrow.
    Thanks again.
     
  7. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    A buddy of mine had both a KZ900 and a KZ1000 about the same vintage. For some reason he thought the KZ900 was quicker/faster. Beats me, I bought the KZ650 because insurance on anything over 700cc's was just insane at the time. It was fun riding it around in the mountains of New Mexico.

    10 Ohms would probably kill the signal input completely.
    10k Ohms (10,000 Ohms) is what I put in the schematic. All it has to do is keep the average voltage of the input close to zero.

    Yeah, I'll say! My KZ650 had a mechanical speedo and tach.
    OK, good luck. :)
     
  8. Motor Head

    Member

    Jul 21, 2009
    10
    0
    Well got the signal input set up with a .01uf non polarized Cap in series, and with a 10KΩ resistor back to ground. Following the suggested schematic, but it is still is not registering a meter movement from the 3 wire speed sensor.
    I had taken a signal from another source with some meter registration, before using the AC coupling you suggested. This is a 2 wire magnetic pick up with 2 magnets spinning on a drill attachment. This is a simple test for registering a AC wave.
    So with the AC coupling in place I loose all input at the CAP .01uf. Getting no voltage signal to the LM2917 pin #1 now that the resistor is in place. I get 1-over 5volts coming in, but loose it as it passes thrugh the AC coupling suggested. I've used a .03 volt scale on my DVOM for AC or DC volts and get nothing. So I pulled the .01uf and experimented with every CAP I had, both non and Polarized. I got the signal to go in with my test bench using the 10K and a 33uf polarized CAP. The signal is getting through again, but acts like before. Not getting pick up from 3-wire sensor.
    Great to have the help, thanks for taking time for my project.
     
    Last edited: Jul 23, 2009
  9. Motor Head

    Member

    Jul 21, 2009
    10
    0
    Using the attached schematic, I have taken the incoming frequency signal and run it first through a 10KΩ resistor with another 10KΩ on a branch to ground, also a .01uf CAP to ground. Then circuit goes straight to pin 1 of LM2917. Also on pin 3 back to ground I have replaced the 100KΩ resistor with another Pot resistor so I can try to fine tune that as well. Now on my test bench I get a great adjustable speed signal with my Drill operated 2 wire speed sensor signal, I can show 0- 80 MPH. So I put it back on the bike to see if it will respond with the 3 wire sensor and its signal. No it still will not register anything on the needle movement when hooked to the sensor on the bike. I must still be having trouble with the wave not going below 0 or its still DC?
    I'm closer but still could use a knowledgeable persons opinion on where to go next.
    Thanks.
     
    Last edited: Jul 24, 2009
  10. Motor Head

    Member

    Jul 21, 2009
    10
    0
    :D
    Well patience is what was needed, as well as some good knowledgeable advice. Let me thank Sargent Wookie as he was a lot of help and his Idea was close to what worked.
    So as the thread says the biggest problem I was having and it sounds like everyone else is getting the input signal worked out. THIS CHIP MUST SEE THE INPUT SWING A BI-POLAR WAVE SIGNAL, WHICH MEANS ABOVE AND BELOW GROUND + & - PERIOD. So my signal did not go below 1 volt = BAD, I had to change this. I found a lot of other circuits on several web sites, and a lot of people with the same issues. Many forum threads without any conclusion. So hopefully this will help out another novice trying to fix an old Speedo or Tach.
    What I used for the input signal change was to first in series bring my 1-5v wave through a .1uf non polarized CAP, then still in series a 10KΩ resistor this then goes to pin 1 of LM2917N-8 but with a branch back to ground with another 10KΩ resistor and .01 CAP in parallel. This drops the voltage to around 1 volt but the wave goes above and below ground which is what this chip needs!!!!
    Now that you have a frequency going in you should get a nice voltage signal for your analog gauge. I am still verifying the values I have for the R1,C1,C2 combinations, as well as trimming for speed accuracy. Once done I will POST the finished schematic here. Soon hopefully.
    Again great help, and hope my results help also.
     
    Last edited: Jul 25, 2009
  11. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I find it odd that the original input circuit I posted didn't work for you. It's just a simple differentiator circuit, and it would have caused the input to go above and below ground, even though your input from the wheel sensor varied from 1v-5v. The output from the differentiator would've been 2v to -2v, assuming a 50% duty cycle from the wheel sensor.

    Attached is a simulation of the two circuits with a 100Hz square wave input.
    The suggestion I gave you is on the left; it's trace on the simulated O-scope is yellow. Your circuit is on the right; it's trace is green.
     
  12. Motor Head

    Member

    Jul 21, 2009
    10
    0
    :cool: Looks like your circuit is the stronger amplitude as well. I found that I couldn't get a signal through that ,01uf cap. It was blocking out the voltage. Strange because it is the same cap that is going back to ground in parallel with the 10KΩ, or maybe its not really doing anything in that configuration. Don,t think it would be the slight difference in Values, .01uf- .1uf. I'm just happy the thing will WORK!!!!!!! These are hard to find used, and then expensive with not much if any warranty, I saw a couple on Ebay $85-$250 depending on condition. Cost me less than $20 with freight for the chips and gas to run for different resistors/ CAPS.
    Now I can "LET THE GOOD TIMES ROLL" :D :cool: etc.
     
  13. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    The idea of the cap is that it blocks DC voltage levels, but lets the effects of AC signals through.

    The 10k resistor from the input pin to ground is intended to keep the average voltage at a ground potential.

    If your input signal is a square wave varying from 1v to 5v, the circuit's output peaks at low frequencies will be roughly -4v and 4v. At higher frequencies, the signal will represent more of a square wave with peaks at around -2v and 2v.

    I'm pretty sure that your 10nF (0.01uF) cap is dead; open circuit.

    What is the radius of your front wheel? (center of axle to the surface of the tread)
     
  14. Motor Head

    Member

    Jul 21, 2009
    10
    0
    :D
    The front wheel is 13" from center of axle shaft to ground.
    That was a brand new .o1uf from the store, a non polarized ceramic disk type. Just waiting for the weather to dry up so I can take it on the road for some POT tweaking to get the calibration right. We have some fixed speed monitoring sites in town that I can use to get it set.
     
  15. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    OK, so I should've asked this before... :rolleyes: how many pulses does the sensor put out per revolution of the wheel?

    What are the markings on the 0.01uF ceramic capacitor?
     
  16. Motor Head

    Member

    Jul 21, 2009
    10
    0
    :)
    Sounds strange but I got 23 high voltage signals, and now after the battery is recharged from riding, the range is from .5-7volts. Higher than what I had measured the other day. Sure glad I decided to put a POT at pin 3 to ground so I could trim there as well, its the 1K-45KΩ POT that was the original trimmer.
    The 500Ω POT that I put in its place, grounding pins 4 & 7, is turned to the least resistance. Then the speed trimming is quite adjustable with the POT at pin 3 to ground. I am at 24mph when the speed monitoring site said 25mph. I still want to goo out with another car and see if it will match up at higher speeds like 60mph. So it reads really good, swings up real fast since it can use full Batt voltage. I still believe you and your CAP & resistor in series makes a better wave pattern, but this is working quite well and I believe I'll be able to finish the job shortly. That is unless you come up with some great recommendations for the C-1,C-2, R-1 values. I can messure the frequency from the wheel sensor again with the full Battery charge tomorrow, I can bring my Fluke 88 back from work.
    Thanks again.
     
  17. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Do you mean that you're getting 23 transitions from low to high in 1 revolution of the wheel? That does sound kind of strange. However, if it were just a single "blip" per revolution, you'd only get around 770 blips/mile, or around 16.7 blips/second at 60MPH.

    I'm asking this question because I'm trying to calculate how you could calibrate the speedo using a 60Hz or 120Hz source; easy to do using a pretty inexpensive 12v transformer and a few components. A lot safer than trying to do it while riding, too.
    Hmm. Letting a lead-acid battery get discharged is extremely hard on it. Motorcycle batteries have a pretty tough life as it is.
     
  18. Motor Head

    Member

    Jul 21, 2009
    10
    0
    :)
    Yes that is right 23 pulses from less than a volt to just over 7 volts per one rotation of the wheel. I had the key on, which runs the lights during some testing and B+ was slowly dropping. It was down to 10.8 volts when I first checked this morning, after riding around for an hour it's back to 12.8. I had discharged about 2 volts off the b+ voltage. Probably the 55 watt head lamp that shine's like a beacon at key on.
    If the weather would stop storming I could go back out and get a speed check at 60mph, but lightning and hail = no riding.
    Thanks for speeding you time on my repair project.
     
Loading...