Hi Ragon,
Please see my attached analysis: (Please visit the site to view this file)
I did not see a passband ripple above 0.003 dB. I think my step sizes of 0.1Hz from 1-100Hz, and 1Hz from 100-1000Hz are sufficient. I also checked multiple data rates.
Regarding your questions:
1) I believe the coefficients are correct. I was able to match the filter responses well. It is possible that some computational rounding is occurring though.
2) You can use either phase response. I observed the minimum phase filter to have just slightly higher passband ripple, but still below +/-0.003dB.
3) Yes. I thought they might have been incorrect too, because they did not equal to the sum of all the coefficients. However, I discovered that they are rounded to the nearest power of 2. Powers of 2 are easier to physically implement in digital circuitry...Yes, the scaling factor for stages 3 & 4 should be 1/134217728 for consistency.
Best Regards,
Chris