Well, all very interesting.
I found some time this evening and have remade the (already remade) directional coupler. Here's take 3:http://www.qsl.net/g0mgx/images/DSC_0082.JPG
So this model features:
- FT-82-76 Toroids
- Home bodged double sided copper box with one partition
- Shorter leads
- About 20 miles of solder consumption to make a good screened enclosure
I've taken a HF radio in FSK mode, connected it to the coupler and then connected the coupler to a dummy load. The forward and reflected ports of the coupler are connected to the AD8307 boards which have a 52.3R input impedence, across each of these I have a x10 scope probe, here is the result NOTE the Ch2 voltage needs to be divided by 10 due to my cockup in scope config:http://www.qsl.net/g0mgx/images/scope.png
So I have 20W out of the rig, 2.4 (ish) p-to-p out of the forward port and about 50mv (ish) out of the reflected port. This seems a good place to start. However, the meter design takes the p-to-p value through the AD8307 then into the sample and hold LF389 and then directly into the A to D converter in the PIC. My 20W is now being displayed as 444W!
2.4V p-to-p for a 20W RF source is WAY to high! The A to D converter is maxing out at about 30 Watts or so. The A to D converter is configured in the 16F876 to be using AD3 (thats connected to a 2.5V reference) as the +ve reference and 0V as the negative reference. So the analogue input is being compared to a 0 to 2.5V range.
The other thoughts that I have right now is that the software contains a whole pile of config tables which are based on a evaluation of the voltage presented to the A to D converters from the two channels under different known RF power levels of drive. The chances of that config being remotely accurate in my build of the meter seem to be rather slim (like nill) to me.
So I have some good progress, but the voltage out of the directional coupler seems too high to me; I'll worry about other stuff like calibration later!