Announcement

Collapse
No announcement yet.

Gary's "Complete Advanced" SSG Build

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #61
    Hi Gary, RS,

    Test results 24Ah Lead Acid @ input
    I finished a couple of cycles with the 24Ah Lead Acid at the input, see attachment. As a comparison I put another test cycle set there as well with different input.
    If I had to draw a conclusion based on what I see in those two sets I’d say: All the named parameters do not have a significant impact (yet), that is:
    -24Ah Lead Acid vs 12Ah AGM battery at input
    -20magnets vs 24magnets rotor
    -8 power windings vs 7 power windings
    -1.8A input vs 2.4A input
    So I’d say at this moment the bottle neck is still somewhere else.


    Scope results
    See attachment. Since I wasn’t sure how far to zoom in / at which level you can call the signal “synchrone enough”, I just zoomed x200. This is done in Radiant mode. If the transistor switching looks “synchrone enough” to you, then this is not the bottle neck either and will continue building a new rotor with 22mm wide magnets and after that another one with 25mm wide magnets.


    Thanks Gary for spending so much time watching those DVDs again to help me. Hope to do some more watching tomorrow and will respond if something catches my eye.

    Looking forward to your input based on the above.

    Best regards,
    Rodolphe

    2020-09-01 - Attachment.pdf

    Comment


    • #62
      Hi Rodolphe,

      The traces you zoomed in on look like they are pretty well synchronized after the high voltage spike event, but they don't show the beginning of the spike or the max voltage of the spike which are the most important places that need to match. Your first trace would show that, except it is not vertically centered on the screen and the max voltage is off (above) the screen.

      If your scope is set at 1x and 5 volts on the vertical scale and you sample at every 5us instead of 200us on the horizontal scale it would be easier to see the entire initial spike before the ring-down. The spikes need to be centered in the screen to see how they compare with each other at the critical points.

      I re-posted one of my images from post 268 of the "Complete Bedini SG Book!" thread that shows what I mean. This is the one that shows the sniffer coil instead of a second power coil, but the alignment and size is what I wanted to point out in this post.






      Gary Hammond,
      Last edited by Gary Hammond; 09-01-2020, 06:27 PM.

      Comment


      • #63
        Hi Gary, RS,

        While taking the new scope screenshots, I didn’t read your instruction well enough and now see that I have not taken the scope screenshots with the horizontal deviation of 5us as you asked. I’m not at home now, so if the images are not accurate enough, I can redo it tomorrow.

        I just took a screenshot of a pulse and then a x25 close up of a peak section. Probe was at 10x (but you can see that based on the vertical scale). See attachment.


        I watched the parts of DVD 33.1 and 33.4 (=34.2). He does mention indeed there at about 1:50 (DVD33.4) that 7 coils is about the max. But I’m testing now with 7 coils, so whether 7 or 8 is the max is not so important for me right now/didn’t seem to make a difference at this point in time as you can see in my results from #61.
        3:39(DVD33.4) he talks about that the transistors need to be very closely matched, 4:00(DVD33.4) that the voltage at the base should be the same, and that the spikes should all be the same: he talks about the 200+V range, so the real top which we talked earlier about will be hard to measure. But I guess we’re looking into this right now (the transistors).


        When we’re looking at the synchronicity of the signals as we do at the moment, and when we’re trying to determine if the transistors are matched and if 100-Ohm resistors are causing a problem compared to 470-ohms resistors; What deviations between two signals tells us that the first is an issue (unmatched transistor) and what deviation between two signals would tell us the second is an issue (100 vs 470).

        Some other thoughts I mention just in case (although I doubt whether they have influence at this stage):
        -I use alligator-clips to connect my wires to the batteries (I’ve used different types, didn’t seem to make a difference)
        -I use a regular/hot charger to charge my input battery for the test cycles
        -As mentioned I’m leaving the Fan out of the equation for now, but do you have a website link Gary to where you bought yours?
        -I once accidentally swapped the cables of the output battery but had the clamps not yet “clamped”/attached yet, just touched/sparked for an instant when I touched the battery. Recently verified my diodes with the diode-test function on my Fluke, and all still gave the same value.


        Best regards,
        Rodolphe

        2020-09-02 - Attachment.pdf

        Comment


        • #64
          Hi Rodolphe,

          Some things I noticed in your screen shots:

          #1 The red trace max voltage is always significantly higher in voltage than the blue trace in every screen shot. Which color is the reference (#1 transistor) and which color is the channel that is being compared to it (i.e. #2, #3, #4, #5, #6, #7)? The voltages of each channel should be more equal than what is shown. This could be caused by either the probes not being properly compensated or else there really is this much difference in voltage.

          Have you done the probe compensation for each probe on the channel it's being used on? Each probe should have a trimmer capacitor to adjust to the scope compensation output.

          Also, If you swap probes and the highest voltage stays with that probe, then the probes are not balanced with each other. If the voltages remain the same for each transistor after the swap, then the voltage differences are real.

          #2 The max voltages for each channel vary from screen shot to screen shot even for the reference transistor. This happens on my scope when I have the horizontal time base interval set too long. The digital circuitry in the scope doesn't always capture the highest peak. But when I change the time scale to a shorter time interval, then the voltages all remain very even every time the trigger circuit fires.

          #3 The slopes of the high voltage spikes all look vertical at the time base you used. They actually have an up slope and a down slope that should line up from transistor to transistor. The curve at the beginning of each up slope should also match up from transistor to transistor. This can only be seen when the time base is much shorter then what you show.

          You can see the initial curve and the up slope and down slope in the screen shot I placed in post #62 of this thread.

          What deviations between two signals tells us that the first is an issue (unmatched transistor) and what deviation between two signals would tell us the second is an issue (100 vs 470).
          Any difference between the initial curve, the line slopes, and the max voltage would indicate unmatched transistors and/or unmatched coil windings. Since the coil is from Teslagenx, I doubt it it would be a problem.

          The exact turn on time and turn off time is controlled by the base circuit (trigger signal) if the transistors are perfectly matched. The resistor values in the base circuit will determine at which voltages on the trigger signal this occurs. You can sample the wave forms at the transistor bases to compare them just like you can at the collectors.

          If your scope has a math function you can also display the difference between the two channels. If everything is evenly matched I think this difference signal will be a straight line at zero volts. This allows you to see any differences in the traces and what they are.

          Gary Hammond,

          Last edited by Gary Hammond; 09-03-2020, 02:11 PM.

          Comment


          • #65
            Hi Gary, RS,

            Please see attachment for updated scope readings.

            Best regards,
            Rodolphe

            2020-09-03 - Attachment.pdf

            Comment


            • #66
              Hi Rodolphe,

              From these screen shots transistors 1 and 7 appear to be pretty well matched. Transistors 5 and 6 are close. Transistors 2, 3, and 4 are a little off. I haven't checked mine this way, but doubt they are any better matched.

              Gary Hammond,

              Comment


              • #67
                Hi Gary,

                Thanks for reviewing the scope readings. So based on what you see, you conclude that the transistors and resistors should not be the cause of the low performance of my SG?


                I just received new magnets, 49x22x9mm and will proceed with building a new rotor. I’m considering how many magnets I will put on there this time. Apart from our recent discussion about the 21 magnets, in post #34 you mention you had the best results with the 21 magnets (ON-time 25%*), but in post #36 you say that you had the best results with a 20% ON-time (Which would mean 17 magnets)*. Not sure how these two statements relate to each other, can you explain a bit more about it maybe?
                At the moment I’m inclined to make the On-time leading, aiming to match your 25%, that’ll bring my magnet amount to 21 magnets* too.

                *See attachment

                Currently awaiting 25mm wide magnets too, to make a 4th rotor.
                Will update here when I have some results. If any other tings come to mind that I should also verify/test, I’m keen to hear about it.

                UPDATE:
                Just read the post above here again, and only now notice that I completely missed your post #64! Reading it now.

                Best regards,
                Rodolphe

                2020-09-07 - Attachment V2.pdf
                Attached Files
                Last edited by pearldragon; 09-07-2020, 11:50 AM. Reason: Re-inserted attachment

                Comment


                • #68
                  Hi Gary,

                  Sorry I missed your post #64, responding to it now:

                  #1 The red trace max voltage is always significantly higher in voltage than the blue trace in every screen shot. Which color is the reference (#1 transistor) and which color is the channel that is being compared to it (i.e. #2, #3, #4, #5, #6, #7)? The voltages of each channel should be more equal than what is shown. This could be caused by either the probes not being properly compensated or else there really is this much difference in voltage.
                  The blue trace was always connected to Transistor 1. The red trace to the other transistors.
                  Looked for it and yes, my probes have a compensation option: https://www.picotech.com/library/app...loscope-probes
                  But no, not ever done anything with it, would need to see if I can do that compensation with the tools I have here…
                  To reverse the scopes as you proposed will be a quick verification too. I’ll try to do this tonight, but it will probably be tomorrow. (Or is this already sorted out in the measurements of my post #65? In other words, if the issue described above would be of influence, it would have shown up here as well?)
                  If when swapping the probes around the blue trace is always higher I could do a third test and use transistor 2 as a reference (or any other), to verify if Transistor 1 is just giving a lower peak than the rest.


                  #2 / # 3
                  I assume this is taken care of/solved by post #65?


                  The exact turn on time and turn off time is controlled by the base circuit (trigger signal) if the transistors are perfectly matched. The resistor values in the base circuit will determine at which voltages on the trigger signal this occurs. You can sample the wave forms at the transistor bases to compare them just like you can at the collectors.
                  Ok, so this is still a test I should do to as an extra verification.

                  Just did a quick search; seems like my scope does have a math function.

                  Best regards,
                  Rodolphe

                  Comment


                  • #69
                    Hi Rodolphe,

                    Thanks for reviewing the scope readings. So based on what you see, you conclude that the transistors and resistors should not be the cause of the low performance of my SG?
                    In your 2020-09-03 attachment of post #65 your traces look close enough together to me not to be a significant factor. They aren't perfect, but they are very close.

                    I just received new magnets, 49x22x9mm and will proceed with building a new rotor. I’m considering how many magnets I will put on there this time. Apart from our recent discussion about the 21 magnets, in post #34 you mention you had the best results with the 21 magnets (ON-time 25%*), but in post #36 you say that you had the best results with a 20% ON-time (Which would mean 17 magnets)*. Not sure how these two statements relate to each other, can you explain a bit more about it maybe?
                    At the moment I’m inclined to make the On-time leading, aiming to match your 25%, that’ll bring my magnet amount to 21 magnets* too.
                    The 25% on time is the theoretical calculated maximum on time based on magnet size and wheel circumference. The 20% time is the actual observed on time, viewed on the o-scope, while the machine is running. There are a few other variables that also influence the on time such as
                    The exact turn on time and turn off time is controlled by the base circuit (trigger signal) if the transistors are perfectly matched. The resistor values in the base circuit will determine at which voltages on the trigger signal this occurs. You can sample the wave forms at the transistor bases to compare them just like you can at the collectors.
                    Your 2020-09-07 attachment doesn't show up for me. When I click on it I get the following message.
                    Invalid Page URL. If this is an error and the page should exist, please contact the system administrator and tell them how you got this message.
                    Gary Hammond,

                    Comment


                    • #70
                      Hi Rodolphe,

                      But no, not ever done anything with it, would need to see if I can do that compensation with the tools I have here…
                      My scope has "probe compensation output signal terminals" that produce a 1 kHz, 3V peak-to-peak square wave for adjusting the low frequency compensation of the probes. This is built into my scope. Or you can also input an outside signal from a signal generator as shown in the link you posted to picotech.

                      Gary Hammond,

                      Comment


                      • #71
                        Hi Gary,

                        Just tried to compensate my scope probes. Used the AWG output of my scope for a 1kHz signal, initially (accidentally) @+/-1V.
                        Calibrated the probes, but when I connected them both to the AWG signal, there was still a bit of vertical offset.


                        Then noticed the AWG signal was at @+/-1V and re-calibrated, this time @+/-2V. Now the vertical offset between the probes seemed gone, but both probes seemed to have an offset to the 0V line.

                        Lastly flipped channels/probes, but this gave the same image.


                        See attachment for scope images of all the above mentioned.

                        Best regards,
                        Rodolphe

                        2020-09-09 - Attachment.pdf

                        Comment


                        • #72
                          Hi Gary,

                          Did some new measurements after the compensation/calibration:

                          1 Set
                          Channes A connected to Transistor1
                          Channes B connected to Transistor2,3,4,5,6,7
                          Took Screenshots of peaks and spike graph


                          2 Set
                          Channes A connected to Transistor2
                          Channes B connected to Transistor1,3,4,5,6,7
                          Took Screenshots of peaks only


                          3 Set
                          Channes B connected to Transistor1
                          Channes A connected to Transistor2,3,4,5,6,7
                          Took Screenshots of peaks and spike graph


                          Comparing the graphs, I would say that the probes account for the majority of the difference on the graph form: where the tops lay a bit further apart, they are switched around when the channels are switched…

                          See attachment

                          Best regards,
                          Rodolphe

                          2020-09-09 - Transistor collector verification.pdf

                          Comment


                          • #73
                            And some more , this time base verification

                            1 Set
                            Channes A connected to Transistor1
                            Channes B connected to Transistor2,3,4,5,6,7


                            2 Set
                            Channes A connected to Transistor2
                            Channes B connected to Transistor1,3,4,5,6,7


                            3 Set
                            Channes B connected to Transistor1
                            Channes A connected to Transistor2,3,4,5,6,7

                            Same conclusions previous post:

                            I would say that the probes account for the majority of the difference on the graph form. See attachment
                            What do you think?


                            Best regards,
                            Rodolphe

                            2020-09-09 - Transistor base verification.pdf
                            Attached Files

                            Comment


                            • #74
                              Hi Rodolphe,

                              Same conclusions previous post:
                              I would say that the probes account for the majority of the difference on the graph form. See attachment
                              What do you think?
                              I agree. The differences seem to be just from the probes and not from any circuit imbalance.

                              Gary Hammond,

                              Comment


                              • #75
                                Hi Gary,

                                The 25% on time is the theoretical calculated maximum on time based on magnet size and wheel circumference. The 20% time is the actual observed on time, viewed on the o-scope, while the machine is running.
                                Ok this cleared things up for me. I finished my new rotor but before installing it, I checked the ON time too with the scope for but by rotors I already had, then I installed my new rotor and did the same. In the attachment you can see a summary of the scope readings, compared to the calculated values. Hopefully tomorrow I’ll be running test with the new rotor.
                                (although the scope percentage ON time of Rotor3 (new one) is the same as Rotor2, I think the max rpm was higher. But this would not be what I expected since Rotor3 has 1 magnet more, although the shape is different… I’m not a 100% sure I remembered the max rpm correctly, but this will be confirmed during the new test cycles).


                                I agree. The differences seem to be just from the probes and not from any circuit imbalance.
                                Instead of the probes, it might also be that one of the channels is a bit off / that I need recalibration of my scope. For now it doesn’t matter to much for me which of the two it is. What is important is to determine whether the resistors and the transistors are sufficiently matched and so far I understood that they are and we can cross that one off the list as possible bottle neck.


                                Best regards,
                                Rodolphe

                                2020-09-13 - Attachment.pdf

                                Comment

                                Working...
                                X