Partnered Output Coils - Free Energy

[MarkE]

TK:

Yes I understand that there is a lot of power in the scope.  I just think it's worth it to think about the spreadsheet as a very powerful tool also.  All in good time, I am not suggesting anything beyond just illustrating an example of what could be done for now.

One thing about a DSO that I think is worth mastering is trying to sample your waveform at the highest rate possible.  I notice you often see that the "data buffer depth window" shows that you are looking at a tiny slice of the memory buffer.  I don't have the skills or experience, but what is always in the back of my mind when I look at those displays is how many raw data points have I captured for my waveform.  Note that what you see on the display is typically a "crunch down" of your data buffer and since I don't have the skills to drive the machine I am always in doubt.

I will put it in simpler terms:  You see a waveform on your scope.  It's taken for granted that you are oversampling the waveform.  The display is 800 pixels wide.  Are there 15,000 samples that were crunched down to render that waveform into an 800-pixel-wide display?   Or were there 942 samples that were crunched down to render the waveform into an 800-pixel-wide display?  That's were I feel uncertainty and that's where I always want to know.

Of course there are diminishing returns, 200,000 samples vs. 15,000 samples is not going to give you anything.

For spreadsheets, here is just something to ponder:  You have a highly oversampled waveform and we know the A/D is eight bits.  You can take that data buffer and filter it in your spreadsheet and completely remove the "jaggies" and turn it into what looks like a perfectly smooth waveform.  So the output from the spreadsheet could be a butter-smooth highly oversampled waveform that is say a simulated 14-bits of precision.

I am no expert here but here is my train of thought:  You want to make a very advanced high-precision power measurement.  You take two raw highly-oversampled 8-bit waveforms, and smooth them out and get two "virtual" 14-bit waveforms.  Then you do the math on the 14-bit "near-zero jaggie" waveforms and output a butter-smooth set of power and energy waveforms.

You might only want to do that once in a while when there is reason to do it.  The point being in a way you can "overcome" the jaggy 8-bit A/D by oversampling and filtering your data and then you work with these derived waveforms.

Correct me if I am wrong, but this is what a high-end sound card does with its on-board processor.  It calculates intermediate values between samples to produce "smoother audio" at a simulated higher D/A bit depth.  The sound card creates virtual higher precision A/D data and a virtual higher precision sampling clock before clocking out the data into say an 18-bit D/A chip.  In the case of your DSO, you don't really have to do any virtual oversampling because your sample rate is already super high.  But the "smoothing" and simulating a higher A/D could come in handy when trying to make more precise power measurements.  And like I said, then you get "prettier" waveform plots that are free of the jaggies.

Finally, I think this is mostly valid.  You can generate derived smoother waveforms because that is what is actually happening in real life.

MileHigh

8 bits of resolution can only be increased via averaging if there is a statistically pure random noise source of 1/2 lsb amplitude summed with the scaled input signal.  In that case to go from 8 to 14 bits requires at least 4096X oversampling.  Going the other direction is comparatively easy because the data interpolation is performed in digital hardware.  There the limitation is the linearity of the output DAC.  Audio equipment long ago went to single bit DACs, basically PWM because a stable timebase is really cheap and clean low pass filters are also relatively cheap.

Consider that the gain accuracy of the system is going to be in the +/-5% area and an 8 bit ADC is not really the limiting factor on accuracy.  Probes and vertical amplifiers both have DC and frequency dependent gain errors.   This is particularly true of passive probes operating above a couple of MHz.  Bandwidth is really nice to have, but 10X passive probes even the ones rated for 500MHz introduce quite a bit of distortion above a couple of MHz even if used properly and carefully.  This can be easily seen by driving a well decoupled modern logic gate from a function generator or even a 555 timer at about 1kHz.  All the jagged edges and ripple that can easily go on for 50ns or more is due to the transmission line effects of the passive probe.  Better probes smooth out the distortion to make it less objectionable, but that is pushing on a balloon.  To get truly flat frequency response and circuit loading requires either a transmission line probe, or an amplified probe.  Commercial versions of either are not cheap.  Amplified probes tend to be proprietary to the make of the oscilloscope.

[MarkE]

Hi Stivep. I hear you. Some people just prefer older or higher end equipment, and sometimes it could definitely
come in handy to have more accurate/precise higher end equipment, but I find for day to day
use my fairly cheap economy based DSO gets the job done. Sometimes I do wish I had a higher end scope
however when making more critical measurements. I had an older analog function generator from maybe the
seventies or eighties, which had some nice features, but it was somewhat out of balance/calibration and some
of the pots and switches were corroded and dirty and didn't work the best. I ended up replacing it with
a brand new GWInstek function generator, which was a bit expensive, but it is actually a good work
horse function generator with output circuit protection features, which comes in real handy with
the various types of circuits I use it with. It is all a matter of need and preference, sure, but
for the most part a cheaper and new lower end piece of equipment will get the job done. Some of the Chinese economy
equipment does have glitches and bugs, and I have found a couple of those with my Chinese Siglent economy
DSO scope, but it still does get the job done for most things I need it for
All the best...

Vintage cars and vintage scopes are not really comparable.    This year's fast silicon is next years has been.  While I like the layout of late 1990's Tek equipment much better than the newer Tek layouts, there is no comparison in the equipment capability.  Some of the Chinese have been very aggressive in capitalizing on newer silicon to make competitive product that they can offer at low cost.  There is a lot of low end product that is not built like an instrument that cost $20,000. or more new.  For a few hundred dollars it can't be.  But if one doesn't have the probes to go with a fast scope, the bandwidth doesn't mean much.  In the meantime, deep memory gets ever more dear in older scopes, where quite a bit is standard in new scopes.

Expensive test equipment is a lot like a swimming pool.   The best option is to have a friendly neighbor or colleague that will let you use theirs when you need or want to do so.

[MileHigh]

Hi Mark,

I defer to your expertise.  I am mostly just speculating.  I took a course in numerical analysis and I barely remember the different approaches for interpolations and stuff like that.

In very simple terms if you are digitizing a linear ramp function and you get something like this:

142
142
142
142
143
143
143
143
144
144
144
144

Would there be a standard spreadsheet function to interpolate on the data to "remove the jaggies" and turn that into a series of points with a smaller step value?  I am assuming that you could also put a low-pass digital filter on that also.

If you did that I honestly don't know if it would give you a more accurate power measurement system.  If anything though, I love the thought of taking the 8-bit A/D data and being able to display it on a high resolution monitor as a bandwidth-limited perfectly smooth waveform.  That would include standard algorithms for computer graphics displays where some pixels are darker than others (sub-pixel interpolation) to create the illusion of a smooth curve, etc.  For all I know this is an option in Matlab.

MileHigh

[Just..Sayin..]

Hi Forest,

I agree. Energy can not be created or destroyed. You and I agree with Science on this one!

Kind Regards

   Chris Sykes - hyiq.org
   To Reach New Horizons!

What about Jesus  and the left over loaves?

[Void]

Expensive test equipment is a lot like a swimming pool.   The best option is to have a friendly neighbor
or colleague that will let you use theirs when you need or want to do so.

Hi MarkE. I can't disagree with that. If I need to do any serious measurements I am going to
ask stivep if I can drop by his lab.
All the best...