V2.81, sys 1.64, FPGA 281, DFU 3.45 with Wildcat installed.
The attached document has some results of my testing. Tried to post here but the formatting gets all screwed up.
Quad measurements.pdf (18 KB)
Checked over your results, while the 2 lowest DC ranges are about as close as you can get, the rest of the DC ranges could perhaps be a bit better, with careful calibration with noise free (preferably battery source to get away from RFI line noise), recalibration with both low and high battery voltage or with charger connected, etc. But given the limitation of an 8 bit ADC, will never be MUCH better.
Consider that each step in a vertical division only has a resolution of 1/25th of the V/div range, so for example, for the 5.01V reading of 5.12V on the 2V/div range, the device can only display 4.96, 5.04, 5.12, 5.20… It’s easy for an ADC to be just below or above the threshold of a step, for example where a tiny increase/decrease would bring the display up/down to the next step. Add to this a possible step or part step in the DC baseline level, which will shift up/down a minimum of 1/2 step as the xpos is changed, and at voltage displays only “deflecting” the trace 1 or 2 divisions, it’s common for it to be 1 or 2 “steps” of accuracy off. Best accuracy will be had with the lowest range possible.
As far as voltage measurements at 1MhZ and above are concerned, it’s fair to say that high frequency compensation of the various ranges in the preamp on these devices is marginal at best, and varies greatly from range to range, and from version to version. One range in particular displays considerable peaking (though this does extend the frequency response) while others less so or maybe fairly flat with correspondingly lower frequency cutoff points.
A post a while back by bielec made some very good recommendations for changed/added high frequency compensation capacitors for the various ranges to extend and provide a more accurate response at high frequencies. A repost of this was recently made in the “DSO Quad bandwidth” thread above.
Finally, while an 8 bit ADC device will never be a precision voltage meter, compared to the 12 or 16 bit resolution of even cheap meters, it’s frequency response is much better, and gives close enough readings for general troubleshooting purposes. It’s interesting to note that an internal bridge exists on the Quad circuit boards to connect the preamp outputs to the 12 bit ADC’s inside the STM32 microprocessor, which seems to indicate the original designers had possibly intended to use this as a meter source. While these would provide much better voltage accuracy, they have very limited frequency response.
Much could be done to improve the hardware on these devices, beginning with the bypassing of the divide by 5 input attenuator. This would bring the vertical sensitivity to a much more useable 10mV/div without any changes in the preamp circuitry, noise or bandwidth performance. A couple of extra ranges would be needed to properly implement this, and while an additional opto/switch would be needed to engage/disengage the attenuator as well as a few extra components for input protection and high frequency compensation, the rest could be controlled by the software.
Great info Wildcat. I saw the mods you mentioned but not sure it will do much for me as the unit will only be a field troubleshooting device. My bench work will still have my TEK 7904A.
Would be interesting though to have a unit and mod it per your suggestion to see how much improvement.
Thanks again.