Measuring power with analog oscilloscope

[forest]

For AC I use 1:1 transformer with safely fuse. I made that from two identical step down transformers ,by disassembly and assembly both primaries on one core.

[ALVARO_CS]

@forest
I use to measure AC voltage with a DMM previous to connecting the scope, to check if the voltage is in the safe range of probes or scope.
In the setup I want to check out now, which is a small proof of concept, it is OK. My doubt was if I may connect my probe and g clip equally (interchangeably) at both leads of a generator coil.
regards
Alvaro

[ayeaye]

For these who don't know what gschem is, gschem is a schematic editor. You likely already know such editor, when you use any software to draw circuit diagrams, to simulate them with any version of spice, or create printed circuit boards. The schematic editor you use there is likely very similar to gschem. This is a very simple editor.

The matter is that i wrote this python script only for gschem, as i cannot write it for all numerous circuit simulation programs there are, or then for all vector graphics editors there are (such as inkscape, heh). The output file format of the schematic editor of many circuit simulation programs may be identical to that of gschem, or then compatible, but i cannot try them all. I leave it to you to port it to any other circuit simulation program, or then to inkscape, if one wishes. I saw no reason to make it for inkscape, as who draws any circuit diagram, uses a schematic editor anyway, why use some other thing. Though inkscape may be more convenient for drawing these traces, gschem is a kind of rudimentary what concerns vector graphics, but then it's simple.

A video i found about gschem, so that you can see what it is  https://www.youtube.com/watch?v=m00pvtFB0oo . Don't be afraid at all, it is very simple.

Measuring power with an analog oscilloscope is though not very precise. Is it possible to achieve a 2% precision? Maybe, with great effort.

[ayeaye]

Forest, yes, when you actually draw it on the oscilloscope's screen image, then it's not that simple. I included a screenshot to show how it actually looks like. There, the gschem window is on the gimp window where the oscilloscope's screen image is open, and is half transparent. The grid's scale space has been scaled down once in gschem. You see the voltage trace is actually drawn.

Open the oscilloscope's screen image in gimp. I had it 640 x 480, as i used the simplest webcam. The first thing, i increased the canvas size to 1024 x 768, and moved the image to the center. This is because when using the transform tools, the image goes over the size of the original image.

Open also the base sch file for drawing the traces (save it from the text i provided and name, say, power.sch). Better make both windows, gschem and gimp, full screen. Zoom it so that the upper trace grid is nice large, move it to the center. Scale down the grid space for higher precision.

Then the oscilloscope's screen image has to be transformed by the perspective tool, tools -> transform tools -> perspective, to compensate the wrong position of the camera. If you hold the camera exactly at the center of the screen, and turn it so that all lines are perfectly horizontal and vertical, then you save a lot of trouble. But it is very difficult to do by hand, you need a tripod, and even tripod is not enough.

Then put the gimp window on the gschem window. Select the scale tool in gimp, tools -> transform tools -> scale, and make the gimp window half transparent. Then when clicking at the center of the image, you can move the image. Move it so that the beginnings of the x axis are exactly at the same point. Make sure that in the scale tool dialog, maintaining the ratio is selected (x and y are connected). Then hold down the left mouse button and drag, changing the scale so that the right end of the x axis on the oscilloscope's screen image, exactly matches the right end of the x axis on gschem. Then click to the center of the image again, and move it so that both the x and y axis exactly match in gschem and gimp. Then scale the image. You may also do image -> flatten image in gimp, to make the rest of the canvas nice white.

Then make the gimp window opaque, put the gschem window on the gimp window. Select add -> line in gschem. Make the gschem window half transparent. Make again sure that the axis exactly match, like by moving the drawing in gschem. Click to the beginning of the x axis, and start the first line, move to the end of it, and click again. Then click to the end of the last line, and draw the next line, by the trace on the oscilloscope's screen image. To the end of the one cycle.

Drawing the lines is actually the easiest part there, and goes very fast. All this preparations and transforms, is what makes it difficult. But in case of pixel counting, all that should also be done, plus edge detecting and preparing the surfaces for flood fill. So doing it with pixel counting may not be easier at all, and the precision may even be worse, due to imperfections of edge detect and filling.

One uses what fits one best, of course, but when drawing the traces just manually with my method, with enough experience, a rough estimations of power can be found i think certainly much faster than with the pixel counting. Because all the image transforms can be omitted then, and even taking picture of the oscilloscope screen by camera. If this picture is not necessary to show that the trace drawings in gschem correspond to the image on the oscilloscope's screen. But for that the picture can be taken much less carefully, almost any picture or showing it in a video, will do then.

[ayeaye]

Please add here details of how to do it by pixel counting, if one thinks that it has to be done by pixel counting.

For pixel counting, the multiplication of the traces has to be measured, then the areas under the curves have to be flood filled, like the areas above and below the x axis with different colors. Like with gimp, or photoshop, likely the traces have to be made to more or less single lines, with edge detect. Then the pixels have to be counted separately above and below the x axis, like with ImageMagick. Then the number of pixels below the x axis has to be subtracted from the number of pixels above the x axis. That number has to be made to absolute. Then that number has to be multiplied by the scale of the power (how much power corresponds to one pixel in vertical), and divided by the number of columns of pixels, which is the width of the image, if precisely scaled and cropped. This is as much as i can say about how to do it by pixel counting. One should find out oneself, whether it's actually easier or better or not, than my method by using gschem , for ones's purpose or necessary precision.