kit-projects AGC with uBITX v5 - need help #v5
Rob French (KC4UPR)
I've posted a few times on here about my kit-projects AGC. I'm thinking that I've never really had it working right, and am hoping I can get some help. Here's my setup, and here are my symptoms.
I've got a uBITX v5 with the kit-projects AGC, preceded in the RX path by the kit-projects BCI filter. I've taken a photo of the AGC board, below. As I've noted in a previous post, I also have a G4HUP buffer tapped off of the Q12 emitter. In addition, I've got an audio buffer and preamp for a line out tapped off the VOL-H. So I can imagine there is some signal loss. However, that still doesn't (in my mind at least) explain what I'm seeing.
Also, I do have a 1K audio taper pot wired up as an RF gain control. It appears to work fine, although I always leave it at its max setting (if I decrease it, signal strength decreases appropriately; I just haven't found occasion to use it in practice). The RF gain pot is wired to the AGC via two lengths of RG-174. The shields of both are soldered to the pot body, which is grounded to the chassis. The shields are NOT grounded at the AGC end. So I think I have installed this properly, though I understand this has been a source of problems for some folks.
The red, green, and blue wires are wired to the switch per the kit-projects instructions, as is the black wire (which is NOT ground, but rather corresponds to the purple wire in the instructions). I've got a connector that goes to the Raduino A7 ADC for the S-meter, and the yellow wire is the audio input.
I will note that I have no real way to calibrate this thing. However, I have done a couple of different things to attempt to "force" some AGC action, with no luck. I will note that I am generally using 40 meter FT8 (7.074 MHz USB) as my "test signal". There are some very piercing signals, that the AGC doesn't seem to do much of anything with.
Basic Setup: AGC audio input from VOL-H ("per the instructions")
In this case, I generally see very low signal strengths, even with strong FT8 signals. The KD8CEC firmware reports ADC values in the range of 20-50 (I understand that this has been divided by 4). Using my DVM to measure the DC voltage at A7 / the control voltage, it looks to fluctuate between 0.1-0.3 VDC. As I understand it, this DC voltage is way too low to cause any AGC action.
Next Setup: AGC audio input from my line out
I have a TL082 configured with one half as a basic buffer, and the other half as a 20dB (100x?) amplifier (reference the TL082 preamp on this page http://rason.org/Projects/opamps/opamps.htm). I normally adjust the gain control (a trim pot) to provide a reasonable audio level for my sound card--this is actually the whole reason I got the AGC in the first place, to control the audio for digimodes.
Here's where it gets a little weird. If I adjust the gain to 0, then the A7 ADC registers about nothing... pretty much what I expect. However, as I increase it, the control voltage increases above what it was in the "basic" case, but it really never gets above 0.3-0.5 VDC even on those same strong FT8 signals. Even if I max the gain control on the preamp output.
The strange thing here is that if I max the gain control and run that to my sound card, the sound card at its 0dB gain setting (which is line-level) is completely clipped/flat-topped. So I would assume this is somewhere around 0.7 Vrms on the audio, which according to the ND6T write-up on the AGC, should yield about 3 VDC on the control voltage... but I'm still only seeing 0.3-0.5 VDC! And no AGC action that I can discern.
Last Case: Direct DC control
This initially led me to think that the audio wasn't getting properly amplified by Q1 (the 3904, or whatever it is) in the AGC circuit. So I tried applying 3 VDC directly to the "Common" pad on the AGC, with the switch in the "Slow" position. When I do this, I see the expected value on the Raduino ADC (reports ~155, i.e. (1024/4)*(3V/5V)). However, again there is no apparent AGC action (and in this case, I would have thought it was forced to do it).
So, any thoughts? What am I missing, or what did I break?