Best oscilloscope choice?

[ayeaye]

Yes these Lissajous were really beautiful. This was an old tektronix with vacuum tubes, i guess, some of these are very precise instruments.

From my little knowledge, i try to explain how these old repetitive sweep oscilloscopes work, also called recurrent sweep oscilloscopes. Because these who may think about buying such old oscilloscope, should have some idea of what they get.

They made recurrent sweep oscilloscopes already in the 1940's and maybe even before. These made in 1940's look almost exactly like Eico 460, though they sold Eico 460 until 1976 or such.

These oscilloscopes were generally made before the triggering oscilloscopes. All i know about how they work, they say they feed a small amount of the input signal, to the horizontal sweep oscillator. I don't get a clear idea of how it works from that, but i understand that this somehow enables to synchronize the horizontal sweep oscillator, with the external signal, so the horizontal sweep frequency becomes some harmonic of the input signal frequency.

I noticed that recursive sweep oscilloscopes have a scale on the horizontal scale fine tuning knob, which triggering oscilloscopes almost never have. So i guess, you change the horizontal frequency, until luckily the signal on the screen stops for a moment. Then from that scale you can calculate frequency, though it is likely not very precise. Or some maybe enabled to find a Lissajous pattern with the external signal. Or you find frequency with an external function generator and a Lissajous pattern, if you luckily have one.

Then you switch to internal synchronization, turn the horizontal frequency knob again, and luckily can get the signal to stay still on the screen. Then, knowing the frequency before, it is possible to calculate the times. From the oscilloscope's settings at that time, it is not possible to find frequency, because the frequency is unknown, and depends on the frequency of the signal. Some simpler oscilloscopes, i understand, had a very slight internal synchronization, so no switch to switch it off, but they synchronized at the frequency very near the signal's frequency, so with some rough approximation, it was also possible to find the frequency.

It is also possible to synchronize with the external signal, like Eico 460 can do that. This is why on these scopes, the input is called external sync, not external triggering. And on this sinometer oscilloscope, it is also ominously called external sync. That way it should be possible to compare two signals, when the frequency of the external sync is known. Because it should synchronize always the same way then, the synchronization only depends on the external sync, and not on the input signal. Thus it should also be possible to measure a phase difference. Or that can be done with Lissajous, which is difficult or impossible though, when the shape of the signals is too complex.

Thus it may be possible to measure time characteristics with the recurrent sweep oscilloscopes, but this is much more difficult, inconvenient and much more imprecise, compared to the triggering oscilloscopes. If precise measurements of time characteristics are necessary, then these oscilloscopes may be useless for the task. The recurrent sweep oscilloscopes, when they were good, were very good for certain tasks, but not for others. This is why they were sold a long after the triggering oscilloscopes were widespread, as a "maintenance class" oscilloscopes.

This sinometer oscilloscope also may be a recurrent sweep oscilloscope, not a triggering oscilloscope. So this may be an additional pain, in addition to ways how it is a scrap otherwise. I'm telling this because i know that it is difficult to convince some to not to buy that sinometer oscilloscope, as it looks so appealing. This oscilloscope is an example of the new wave of chinese scrap products which start to enter the market. Be cautious about these, as they often are not what they look, or are believed to be. I mean, some of the cheap things they make, are still useful, but at one point they may become useless, or likely will, considering how the things develop further, and more "innovations" are added.

[TinselKoala]

Determining frequency using the RM503 precision low-frequency oscilloscope:

http://www.youtube.com/watch?v=teXXF0a_WoI

Displaying a stored waveform from an analog oscilloscope:

http://www.youtube.com/watch?v=HcHg5gGB5xk

[ayeaye]

Displaying a stored waveform from an analog oscilloscope:

http://www.youtube.com/watch?v=HcHg5gGB5xk

Yeah that's clever, but... I figured out something else. It is possible to use gschem, to draw waveform, and as it is a vector graphics editor, then its file is easy to use for all kind of calculations, like with using python. That way it is possible to easily calculate power, from the waveforms of voltage and current. Power is still V * I, but all is about calculating the average power.

And, pretty sure there is software, that is capable of transforming a picture into a vector form. Or calculations can be done directly from a bitmap form as well, but better if it were a bit normalized, like lines made to a single pixel. But all that is a too high tech for me, i stick to gschem, and do it manually.

[TinselKoala]

Back in the "good old days" we would tape some tracing paper on the scope screen, transfer the scopetrace with a pencil, then cut out the waveform areas carefully with scissors and then weigh them on an analytical balance. This method can be surprisingly accurate if you do a little calibration testing beforehand.

[ayeaye]

This is not an easy task, to take an image, and transform it into curve, which then can be used for calculations. I tried to find a software which transforms an image into such curve, i couldn't find. I tried to transform the picture in gimp, the difference of gaussians, etc, give an outline of the trace, two lines, but none are one pixel thick, and it is impossible to eliminate all the background dots. Thus such image cannot be used for calculations. So still it has to be done manually. Like load the image to gimp, then create a transparent layer, and draw to it one pixel dots. Then save that layer as xpm, which then can be converted to a table of numbers, and then some curve fitting program can be applied to it. There are many these. This method can provide very accurate results, and very nice graphs, but it's quite a work.

So i still stick to this manual drawing with gschem, mostly quite rough approximation is enough. Or when one wants to do it better, one may do it with inkscape or such, where i think too it is possible to add a transparent layer, and save it hopefully in some very simple vector graphics format, which then can be used for calculations, or then as a data for curve fitting.

Now does anyone understand what i'm talking about?