cliffyk

In the first, I'd like to establish that in performing the exercise we have been discussing one is not calibrating the instrument, but rather adjusting the impedance match between a particular probe and the 'scope's input. The process should not even be called "calibration", as nothing is being calibrated. I realise that many do refer to the process as "calibration", nonetheless engineers and manufacturers refer to it as probe compensation, I.e. impedance matching of the probe and the oscilloscope input. While there is little to no difference in probe compensation between inexpensive and expensive oscilloscopes, one way to make a cheap 'scope (and mostly what separates cheap 'scopes from expensive 'scopes) is to cut corners on the analog signal "front-end". This almost always results in differing input impedances (and very often inaccurate vertical accuracy) between ranges. The impedance changes at different ranges is what affects probe compensation as you change ranges. This recent photo shows the probe/scope compensation to be peaked at higher frequencies, however you indicated that you "zoomed in" (changed the vertical range?) to capture the trace. I suspect that what happened was that when you changed ranges the input impedance of the 'scope also change slightly, upsetting the impedance match between the probe and the instrument's input. In precision applications (with any oscilloscope) one would adjust probe compensation at the selected range before making measurements. However, this all needs to be put in perspective. What we have here is a 10MS/s, 2MHz bandwidth oscilloscope, with a 128 x 64 pixel display--I.e. (and forgive me for saying this, I do not mean to be provocative) a toy oscilloscope. It will probably be fine for the sorts of applications I have seen mentioned in this thread and forum (although I believe an amplifier could be well into clipping before it would be readily observed on the small display). Beyond this, at the audio and low range supersonic frequencies (DC to 200kHz) that seem to be focus of the work being performed by most here, and given the 2MHz bandwidth of the unit is question, HF probe compensation is very nearly an irrelevant issue. The complete range of HF adjustment reflected in your photo series probably amounts to less than a +/- 1dB error at 2MHz (well under the 'scopes overall performance). At 20kHz the error becomes less than negligible. One last comment, in the interest of full disclosure, I do have a Velleman PCSU1000 'scope which was given to me by Velleman for some software development work I did for their PCGU1000 function generator. I mention this because while the function generator is a very competent instrument and a tremendous value, the PCSU1000 scope is only a bit better than "ho-hum"--yet another toy FPGA based oscilloscope with a PC interface. The vertical accuracy and input impedance bounce around as you change ranges, and with signals of greater than 1MHz or so the jitter makes it of marginal value. Dual channel measurements at higher frequencies (it has a 60MHz analog BW) are flawed by the instrument's use of a single FPGA operating at 50MHz for all 'scope functions (A/D conversion for both channels, triggering, timebase, and digital output). This causes a significant timing offset between channels. This is NOT because Velleman has produced an inferior or defective product, but rather manifestations of the compromises required to make an inexpensive "toy" oscilloscope.

I replied to your post in the Velleman forum, however I will report my reply here for the benefit of others. For reference, here is one of the photos you posted there: ========================================= You are getting a perfect calibration, it's the top of the waveform that matters. not the sides... The effect you are seeing is typical of digital all scopes, and is only more pronounced on the HPS10SE because of its 128 pixel wide display and 10Mhz sampling rate. While a perfect square/rectangular wave would have 0.0s rise and fall times, the reality is that there's no such thing as a perfect square wave, they all take time to change from high to low voltage and vice-versa. What you are seeing is that the calibration signal fell from its maximum level to its minimum level over a period of two samples made by the scope. One was made while the signal was high, the other while it was low, since the two samples represent different points in time the display of the falling signal is "stepped". This is the same effect as the horizontal steps seen in the top and bottom of the "peaked" and "rolled" waveforms, the difference being that the horizontal steps represent the scope's vertical (voltage) resolution while the vertical steps represent the unit's horizontal (time) resolution. Here is a screen shot from my $3200 Lecroy WJ-322 scope showing the same effect, deliberately exaggerated through selection of a slow rise/fall time wave. This was recorded at a 500MHz sample rate. In the top trace the red arrows show the steps created by changes in level that fell between samples. The bottom trace is a 10x zoom of the selected portion of the top trace, showing the "real" shape of the falling level. Note however that there are still steps, just smaller ones. Short story, there's nothing wrong with your 'scope...