No announcement yet.

Need advice from electronic guru's (again)

  • Filter
  • Time
  • Show
Clear All
new posts

  • Need advice from electronic guru's (again)

    Iv tried this along time ago - did not pan out but want to ask again.

    have a new crank proto-type im testing on my bike dyno, crossed this bridge along time ago and settled on a solution that was acceptable as I tried the electronics alternative with no good results,,,

    the goal is to stabilize the dyno voltage spikes that it goes through due to the pedaling forces being pulsating in bi-mechanical nature....

    this is esp. important because all I have is a digi-meter and need to be able to read the last few digits with accuracy instead of them jumping all around - have tried the capacitor approach complete with diode and resistor combo's to just direct a "sample" of the voltage and filter it and then get a reading -- mixed success but generally not good enough, so I did what a mechanic would do - I built more of an electro-mechanical device and it actually worked very well, well enough to record down to 1/10th of a volt with stability...

    Don't laugh but here's the unit below,,, this coupled with a heart rate monitor and my cadence meter and I have a pretty good dynamometer, it's just two perm mag tape deck motors and one is the motor and the other is the generator that's sole purpose is to power the volt meter,,, ends up being just a fraction of actual voltage but no worries as it does provide a stable guide - the real dyno gens power a ford taurus rad. fan that keeps me cool while testing - ambient air temp and all kinds of other factors are figured in...

    Note the glued in heavy washer installed into the little gauge generator pulley for more stability...

    Thing is - is this unit just shot craps --- I think it's been too much of a labor load for the motor, it got warm and I believe I either fuqued the brushes or the windings are shorting because my voltage is way low and not as stable...

    here's the main dyno itself...

    Your mission - should you choose to accept - is to help me design a Cap system that will work because iv gutted out my last tape deck player and do not have any little 12 volt motors around,,, but what I do have is a couple old computers with the 110 to 12 volt power boxes,,, can something be done with them?

    can I kinda isolate a Cap with a diode or two and then use a resistor to feed it and get the reading off of it?

    How can I build a virtual mechanical "flywheel" in the electronics world, as you can tell not may area of expertise... thanks

  • #2
    There is the electronic equivalent of a flywheel- it's an op amp that charges an output capacitor to the peak voltage on the input. You get to decide how long the charge on the output cap lasts- could be a fraction of a second or minutes. The circuit would be called a voltage follower or current amplifier, and would charge the cap through a diode. A resistor across the cap determines how fast the voltage decays. Make it longer than the integration time of a digital voltmeter and you should be able to get a workable reading. I'm guessing that you won't need to see variations second by second- probably more like 5 seconds or more.
    I seldom do anything within the scope of logical reason and calculated cost/benefit, etc- I'm following my passion-


    • #3
      Originally posted by darryl View Post
      I'm guessing that you won't need to see variations second by second- probably more like 5 seconds or more.
      You guess very close,,, a mean average every 5 seconds would be just fine as my cadence meter is shorter intervals than that and if i kept watch on it most of the time like I do it means the voltage output and therefore "drag" is fairly stable too...

      gutted out an old computer power converter, think there's enough variety in here to make it happen?


      • #4
        I suppose if you powered a filament bulb from your dyno you could use a simple photo cell connected to your meter to read the brightness of the bulb which would vary with speed of the dyno. Hopefully the thermal inertia of the filament would smooth things out and all with few parts.


        • #5
          Somehow it seems to me that you are over-complicating it. You have an output voltage from the dyno that you used to run the tape deck motor that runs another tape deck motor to give the voltmeter a reasonably stable input voltage. It seems that just putting the voltage from the dyno through a diode and into a capacitor would give a slowly enough varying voltage for the voltmeter to handle. Perhaps what is needed here is a current limiter in series with the output from the dyno so the peak voltage doesn't charge the capacitor as quickly. A current limiter that would work for this is very simple- one three terminal voltage regulator and one resistor. I'm guessing you won't care what the actual output voltage comes to be as long as it doesn't vary too quickly. If the output from the dyno is more than about 6 volts at any test speed, then this will work.

          This circuit is so simple that I can describe it without needing a diagram, if you're interested in trying it. You can probably find a suitable resistor and capacitor in that old power supply, but you likely won't find a 317 regulator in there, which is what you'd need.
          I seldom do anything within the scope of logical reason and calculated cost/benefit, etc- I'm following my passion-


          • #6
            A diode and cap should do the trick. If the meter value was varying too much with that setup then it indicates that you didn't use a large enough value of capacitor. Or that your meter has a lower than optimum input resistance and it's bleeding down the voltage.

            The type of capacitor you use helps a little too. A solid tatalum cap will have a lower self leakage than a cheap electrolytic. But even so unless the electrolytic cap is a really bad one it should hold a charge without much of a drop for at least a while.

            As a test try hooking the cap to the meter. Then touch a 9v battery set to the correct polarity to the cap leads to charge the cap. As the battery is pulled away watch the voltage on the cap with the meter. If it drops too rapidly for your needs then you need a better or larger value cap to work with the input resistance of that meter. Most likely a larger value. Testing this way totally isolates the meter and cap from any other issues and will tell you if the meter is bleeding off the charge fast enough to make the small numbers unusable.
            Chilliwack BC, Canada


            • #7
              Have not tried the three terminal regulator before - tried all kinds of combo's with diode resistor and cap...

              You can see some pretty big caps in the power supply I took a pic of but those caps must be for the 110 side as they are rated 200volts

              on the other side is a 3300 uf 16volt cap must be for the 12 volt supply, is this a good cap to start with? it's the biggest one there would it not provide the greatest buffer for saturation? I will look for a 317 reg. does it have 317 numbers on it? snowed in today - nothing better to do so thanks for the help Darryl U 2 BC


              • #8
                Well, the modern way would be to feed the sample voltage into a micro-controller, read the voltage, apply whatever filtering you want programmatically, and then output said voltage.

                That would cost you, oh about $4 for an Arduino. Now, if you wanted to dispense with using your voltmeter, you could put an LCD panel shield on said Arduino. That would be another $5 or so, though it would give you 2 lines and some buttons for whatever as well. Oh, probably need a wall transformer for power. That's another $2, shipping included.

                In the end, you'd have a nice LCD readout you could mount on the handlebar that did whatever you managed to code around. Average power, peak, low, rate of climb, calories per hour, whatever. Once you get into the digital world, it's all just little bits of code, much of which is already written and freely available. You just have to figure out enough to tie it all together.

                There is a learning curve, but compared to a lathe it's not that bad. And, it won't chop your fingers off if you mess up



                • #9
                  You shouldn't need anything like a 3300ufd cap for a voltage holding circuit. Something more around 100ufd should do just fine. A 3300ufd cap takes some serious current to charge it up. Mind you that in itself may aid in stabilizing the voltage. So it might be worth a try.

                  A resistor in series with the cap to slow down how fast it charges can work towards making the reading more steady. But the resistor by itself is enough of a current limiter. You don't need a voltage regulator as well. Also the regulator needs voltage to run so it'll also cause a voltage drop. And that might become an issue. Easier to just use a resistor to limit the current. That'll slow down the charging rate of the capacitor.

                  But at the same time the current limiting resistor also needs to be selected so it works with the input resistance of the meter that is leaking down the capacitor. MOST meters up way up around 10 or 20 Mohm input resisitance. But not all of them... if you're using a seriously cheap meter it may have a much lower input resistance value. You would want to read the manual to find out what the spec is.

                  You can get a feel for what resistor to use for an input current limit resistor. The RC time constant for a capacitor and resistor combination is given by R x C=t (greek "tau" actually). The value in ohms x farads. And for a series resistor and cap the time to reach up around 95% of the final value of the charge is 5t. Now you want a fairly long "t" so it stops the value rising too fast. Let's start with .5 seconds for "t". If you're using a 100ufd cap then the resistor you want to use with it would be .5/.0001=50K That combo would take 2.5 seconds to come up to a stable value. That seems a little too long to me. So perhaps start with more like 25K or 32K. And if the meter doesn't bleed it off too fast it'll hold that value to where you probably want to bleed it off with some other longer RC resistor. To avoid splitting away too much voltage you'll want to keep the bleed resistor up at 10 to 15x the value of the inflow limiting resistor so maybe 500K? Anything like a 470K to 560K should do the trick. You might need to play with these resistor values a little. But in the end you should be able to tune the values so that you have a fairly stable reading.

                  So you would end up with a diode and then the 25K resistor in series with the far end connected to the + of the cap. And across the cap in parallel you'd have the 470 to 560K bleed off resistor. That make sense? So fast but not TOO fast to charge up and quite slow to drop away the value.
                  Chilliwack BC, Canada


                  • #10
                    What you may want is a simple R-C integrator, which is just a resistor in series with a capacitor. The time constant (in mSec) is simply the resistor (in kOhms) and the capacaitor (in uF). So for a 1 second TC a good combination might be 10 uF and 100 kOhms. You probably don't want to use a resistance much higher than that, maybe 500k maximum, and it would be best to use a "good" capacitor, perhaps ceramic (multilayer) or Mylar or polypropylene, or a tantalum electrolytic.

                    It seems you also have a problem with your dynamo. If it has brushes and a permanent magnet and generates DC, there will be commutating noise consisting of high voltage "spikes" of short (microseconds) duration. If these get to the voltmeter, they can cause erratic readings and possibly damage. Worn brushes and a dirty or pitted commutator can make the problem worse. If you are getting inconsistent and lower than normal output, I suspect such a problem, and it requires opening the device and repairing it by cleaning or resurfacing the commutator, and replacing or cleaning the brushes.

                    A small ceramic or film capacitor (with low impedance at high frequency) across the brushes can minimize the noise at its source. Locating it away from the motor/generator means the energy will travel along the wiring and possibly cause problems with radiated electromagnetic noise, which can couple into the meter even without leads attached.

                    Another good way to deal with inductive spikes is to use a series inductor, such as the toroidal common mode choke in the power supply you show. You can use the windings in parallel or series on one lead from the generator to deal with "normal mode" noise rather than "common mode" for which the CMC is designed. The power supply also most likely will have some small chokes consisting of a bar or donut of ferrite with wires wrapped around it. One of these may be all you need.

                    It is really helpful, and perhaps necessary, to use a scope to diagnose and fix such problems. You can probably get a usable old analog scope on eBay for less than $100, or you can buy one of the cheap USB or AVR based digital storage scopes with color screen for $20-$50. You should get one with at least 1 MHz bandwidth, to see the narrow noise spikes.

                    Good luck!
                    Paul , P S Technology, Inc. and MrTibbs
                    USA Maryland 21030


                    • #11
                      317- it will say LM317, or NE 317, or some other letter prefix before the number- but I can almost guarantee that you won't find one in that power supply.

                      3300 uf could be a bit large for this application, but then again it might smooth out the voltage variations enough for the circuit to be usable. You might find a 1000 uf in there- likely anything down to about 100 uf could be useful.

                      Short of finding a regulator ic, you could just use a series resistor. Your circuit would then be a resistor and diode in series, feeding a cap with its negative lead grounded, resistor in parallel with the cap. Meter reading taken across the cap. Dyno + voltage goes to the first resistor, and - voltage to the negative of the cap. As a guess I'll say the first resistor would be between 22 ohms and 100 ohms, the cap between 220 and 1000 uf, and the bleeder resistor across the cap between 470 and 2200 ohms. Pretty much all this could be found in that power supply.
                      I seldom do anything within the scope of logical reason and calculated cost/benefit, etc- I'm following my passion-


                      • #12
                        The diode and capacitor will make a peak hold circuit, which may be useful but perhaps not what is wanted. With a bleeder resistor it would be a "fast attack slow decay" filter, which will give a long-time average of the peaks.

                        Adding a regulator may give a steady output from a varying input from the dyno, but I don't think that is what is wanted.
                        Paul , P S Technology, Inc. and MrTibbs
                        USA Maryland 21030


                        • #13
                          Do you need any sort of calibration?

                          Do I understand that this dyno puts out a voltage?

                          If so, is it AC or DC? I'm not clear from the description, sounds like DC maybe

                          The basic thing is that for a resistor and cap system to get an average, there are two resistors.

                          one in series with the input, and usually one connected to discharge the capacitor.

                          The input one, along with capacitor value, sets the "charge time", the time it takes for the output to respond. when the input changes. The biigger the resistor and capacitor values, the longer. You can just multiply the values, resistor value times capacitor value. With megohms and microfarads, that comes out in seconds

                          The result is the time for the system to charge to 66% of the input voltage. That is called the "time constant" of the circuit.

                          The discharge one sets the time the system holds the value. Same equation.

                          If the input comes from a diode, you NEED the discharge resistor. If the input comes from a source that will accept current, such as a regular DC generator, you can use just the input resistor and it will average perfectly well.

                          From your comments, you may want a time constant of as much as 10 seconds. Maybe even more, if you want the meter to hold steady and change slowly.

                          With a DC source, no diodes, 10 seconds would be 1 megohm and 10 microfarads. Or 100k ohms and 100 microfarads. Charge and discharge time will be the same.

                          If it DOES come from a diode, you will need to change that somewhat. Charge and discharge will be diifferent. You may want to go with something more like 330K input, and 150k for the discharge, using 100 microfarads. That will change the voltage, but the effect will be the same.

                          Keep eye on ball.
                          Hashim Khan


                          • #14
                            Clearly from all the replies some of us are on the same page as to what the circuit should look like but there's some wildly varying suggestions as to the actual values for caps and resistors.

                            In the end it'll take a bit of trial and error to find the values that provide the limit on the input that slows down the charging and the value of the discharge resistor to slow the drop off.

                            I'd also suggest that for any regular digital multi meter used as a "speed gauge" that you consider the idea of masking off the last number. That one is always going to dither a lot.

                            Another option is to consider using a regular bicycle speedometer on some part of the arrangement. Programming these includes a step where you input the length of a wheel's circumference so the rotation time corresponds to the proper speed. And these speedometers have some suitable smoothing included so the readout doesn't wander badly. Doing this may require some additional pulley so that the wheel with the sensor magnet on it spins at some RPM that simulates a regular bicycle wheel in some sort of valid gear. Or you might be able to drive it off the pedal cranks directly if it'll take the equivalent value needed. I tend to think it'll need some form of gearing though. Otherwise you would need to input a circumference that would be suitable to an old style penny farthing bike. And I don't think the speedometers will let the circumference value go that high.

                            Note that this does not require a full size bicycle wheel. Just that you include the same sort of ratio you'd have in your most often used bicycle gearing.
                            Chilliwack BC, Canada


                            • #15
                              So cool you guys seriously thanks for the help --- I just gutted the low voltage side of the board and I got a range of caps that should do the trick --- I been charging them like BC said with a 9 volt battery and then watching the state of decay whilst connected to the meter!

                              it's cool think im going to get it this time, check out all the caps and also got an adjustable variable resistor that says "102" on it but the board empty space it came off of says "vr 401"

                              Im thinking if I use the right cap I won't have to use a bleed down resistor? just let the meter do that as it takes some consumption to run it??
                              and if I wire it right do I even need to use a diode?

                              this is actually very exciting to me because I don't know much about it so it's like a mad science experiment lol