Thomas Ales [M S E]
38 lines
1.7 KiB
Markdown
38 lines
1.7 KiB
Markdown
# ADC YOFF Sign Bug — sras_viewer.py
|
||
|
||
## Status
|
||
Fix applied, awaiting user testing.
|
||
|
||
## What was wrong
|
||
|
||
`DC_YOFF_ADC` in `sras_viewer.py` was `+87.04` instead of `-87.04`.
|
||
|
||
The Tektronix scope stores CH3/CH4 waveform data as **signed int8** (−128 to +127), where ADC 0 = screen center. The scope's vertical position for CH3/CH4 is set to `−2.72 div` in `sc3_aui_app.py`, which places 0 V **below** center at ADC count `−2.72 × 32 = −87.04`. The comment in the code had the formula as `-position × (256/8)` (sign flipped), producing `+87.04` instead of the correct `−87.04`.
|
||
|
||
## Effect of the bug
|
||
|
||
- `adc_to_mv` was off by 272 mV in the negative direction
|
||
- ADC −87 (true 0 V signal) → −272 mV (should be ≈ 0 mV)
|
||
- ADC 0 (screen center, above ground) → −136 mV (should be +136 mV)
|
||
- DC images for CH3/CH4 (Bias A/B) showed large negative voltages, physically impossible for DC bias signals
|
||
- RF mask threshold (`mv_to_adc`) was also broken: threshold ADC value ~+87 was being compared against pixel means clustered around −87, so nearly every pixel would have been incorrectly masked
|
||
|
||
## The fix
|
||
|
||
`sras_viewer.py` line 48:
|
||
```python
|
||
# Before
|
||
DC_YOFF_ADC = 87.04 # ADC count that represents 0 V
|
||
|
||
# After
|
||
DC_YOFF_ADC = -87.04 # ADC count that represents 0 V
|
||
```
|
||
Comment on line 47 also corrected from `-position × (256/8)` to `position × (256/8)`.
|
||
|
||
## What to verify during testing
|
||
|
||
1. CH3 and CH4 DC images show positive (or near-zero) voltages consistent with the bias signal levels
|
||
2. RF (CH1) image is not excessively masked — pixels with a genuine bias signal above the threshold should appear
|
||
3. `mv_to_adc(0.0)` should now return −87.04 (not +87.04)
|
||
4. The default threshold of 0.125 mV should correspond to ADC ≈ −87.0, not +87.1
|