zoom/notes/old/notes-a6302.txt
jim bbc7dcdba4 reorg
git-svn-id: https://bucket.mit.edu/svn/nilm/zoom@7008 ddd99763-3ecb-0310-9145-efcb8ce7c51f
2008-12-05 00:43:01 +00:00

60 lines
1.9 KiB
Plaintext

Given out scaling and that n=100 on the secondary for the A6302 probe,
and a 16-bit DAC value, each LSB cancels out 300 µA of current on
the primary, for a maximum range of
DAC 0000h ≈ -9.83 A (-300 µA * 2^15)
DAC 8000h ≈ 0.0 A
DAC ffffh ≈ 9.83 A (300 µA * 2^15)
Let's use 14 bits from the DAC. The range covered by a single bit is
(9.83 A * 2) / 2^14 = 1.2mA
For 5 more bits from the probe, we need to measure 1.2 mA / 2^5 = 37.5 μA.
The probe scaling is 1mA = 10mV.
So 37.5 μA is 375 μV.
The PIC can measure 12 bits over 2.5V which is 0.61mV, so we need
to scale the probe output. Scale by (4.32) so that
1mA at probe = 43.2mV at PIC
37.5μA at probe = 1.62mV at PIC
(PIC input) = clamp(0, 1.25V - (probe current * 10 * 4.32), 2.5V)
Clamp by putting a 1K in series with diodes to VSS/VDD.
----------
DAC range is ± 10 A.
Probe range should be about 2^-14 as big, so ±610 μA
which is 0.000610*10*4.32 = 26.35 mV.
PIC measurements:
(1.25V + 26.35mV) / (2.50 / 2^12) = 2091
(1.25V - 26.35mV) / (2.50 / 2^12) = 2004
--
1 bit change at full 16-bit DAC resolution is 0.3mA.
At PIC, the voltage will change by (0.3mA * 10mV/mA * 4.32) = 12.96 mV
This means changing 1 bit at the DAC should cause the 12-bit ADC count to change by
12.96 mV / (2500mV / 2^12) = 21.234
To reset the PIC input to 2048, we therefore need to step the DAC by
((2048 - count) / 21.234)
--
If it starts off centered, the ADC value at the PIC will max out when
it gets a step of 1250mV. This corresponds to an input step of
(1250mV / (10mV/mA * 4.32)) = 28.935 mA.
Sampling rate is 8000 Hz, so this limits slew rate to 28.935 mA * 8000 Hz = 231.48 A/s
A 1A sinusoid with is sin(t*2π*60), so the max slope is max(2π*60*cos(t*2π*60)) = 376.99A/s
TODO:
- Kick sampling rate up by 4, to 32 KHz on ADC and DAC commands.
- That should up our slew rate to 900 A/s.
- Send data to PC as fast as possible; 8KHz or 16KHz or all 32KHz (?)
- Smarter recentering algorithm (!)