Personally, I don't like the dobeck, but it does work.
The dobeck is the equivalent to a PC-V+AT, and a gauge that allows targeting certain mixtures at certain RPMs. It can work to get a tune fairly close, but it doesn't have the granularity of a proper fueling table. Think of it as a old school interface to modern technology, like adjusting a digital carb.
Carpenter pipes on a Ramair bike are going to be extremely close to the 80% peak duty cycle hard coded into the ECU at peak RPM. Also, any tune you get will not be perfect no matter who writes it. As such, I'd suggest using a good base TuneECU tune that will get you close. If you can get it on the dyno to get tweaked for your bike do so, if not, getting a PC-V+AT, can get you really close to perfect. PC-V has the added benefit of allowing you to use the last 20% duty cycle the Keihin ECU locks you out of.
Don't underestimate how much more flow your new parts add, there are L and F tables changes that are required or you'll be in the 18:1 AFR range high RPM and 16:1 range low/mid rpm.
The other option is mail your ECU to Carpenter to get flashed, when doing so, they will be converting you to Tuneboy, This is probably the best option for a rock solid base tune, but costs, I think $200 (license fee for Tuneboy).
If a local tuner can directly write TuneECU or Tuneboy, that's your best option. Otherwise, I suggest a PC-V+AT plus a good base tune. For tuning the PC-V using AT, you don't want to "accept trims" too many times though. The method I use is leaving 20% authority for the AT, go on a long ride covering many different RPMs in different gears, accept those trims. Lower AT authority to 10%, long ride, accept trims. Then go into the trim table and find all the negative values, add 5 (so if it's commanding -16, change it to -11). Set authority to 5% plus and minus, enjoy your bike.
By gear and 250 RPM intervals are very much worth doing, and ONLY the PC-V gives this ability.
@warp9.9 pointed out recently that the Bosche 4.2 and 4.9 consumer grade widebands tend to be off, and hes absolutely correct. The further from 14.5:1 afr the reading is, the more the variance. In every test I can find, they are wrong in the safe direction though, meaning, the o2 sensor may give you a reading of 13:1, but in reality it could be anywhere from 12.1:1-12.9:1. Most consumer grade sensor's Ive seen testing on (not the garbage one showing the LC-1 ad dead accurate), show Bosche based sensors tend to run between 0.1 and 1 full point off, always to the "reading leaner than reality" direction.
In all honestly, the AT feature with some thoughtful care, is so accurate I only use the datalogging to check its work rather than write my own tables now.