Improving V/Div accuracy

[cdw2000]

First, I have to say that I really enjoyed building this and am amazed at the functionality and utility of such an inexpensive kit. I do have a few suggestions for improved voltage accuracy after looking at the schematic:

1) Looking at the attenuation ladder formed by resistors R7 - R12: R9 should be 90 ohms, not 91 ohms, to get the perfect sensitivity dividers (at least with the nominal resistor values). 90 ohm 1% resistors are available, but if you cant find one, putting an 8.2K resistor in parallel with R7 results in an almost perfect 90 ohms (nominally).

2) The non-inverting amp formed by U4C introduces a potential additional voltage gain error based on the tolerances of 1K resistors R5 and R6, which give this amp stage a gain of 2 +/-2% (assuming 1% resistors are used). Removing and jumpering R5 would make the gain =1, but would eliminate the resistor tolerance error. Resistor dividers R1 and R2, and R3 and R4 on the input amps U1A and U1B could be changed to account for the loss in gain. The drawback is that the input may only be able to handle 25Vpk (I need to research this further).

3) On my kit, the 3V Vbias diode was actually at 2.88V, making the voltage at V5 be only at -0.96V and the voltage at V4 only be at 1.4V versus the 1.61 volts desired. I assume this is biased so that the signal is as close to the center of the A/D Converter scale as possible. To correct for this, I put a 4.7K resistor in parallel with R15 (680 ohms) to get about 594 ohms. This corrected the bias voltage at V4 to be almost exactly 1.61V as specified in the instruction sheet. Even if the voltage at V5 were exactly at -1.0V as specified, with R15 at 680 ohms, the voltage gain of the inverting summer node on U4B would make the voltage at V5 be 1.47V. In order to get 1.6V, R15 would need to be around 621 ohms. This all maybe moot since the firmware has the capability of auto-zeroing by pressing and holding the V/DIV button, but it would be best to have the bias be at center scale for maximum +/- headroom.

4) Replacing all resistors in the signal chain with 0.1% tolerance resistors would greatly improve accuracy, however they are more expensive and would add to the cost of the kit. Maybe offer an upgrade kit containing just the user installed resistors, but at 0.1% tolerance (and modifying the values to be closer to the ideal values)?

All comments/discussion welcome!

[jye1]

Thank you for the suggestions. They are much appreciated although we might not be able to implement them right away. We will consider in future.

For the bias 1.4V was good enough. As you said the firmware will auto-zeroing anyway and there are big enough headroom at top and bottom you won't see any impact.