Well, I don't know where to start, but here are the basics of what I have, and what I'm wanting to do....
I need to feed a voltage that is going through a triac into an ADC (10kohm max impedeance, would like to get downwards of 50ohms for speed though).
I am using a VB409 for my 5vdc (5% regulated) power going to my uP (ADC is built into it). The VB409 also has a 12~15vdc unregulated source to draw from too.
Obviously, I have avoided using a transformer (bulky, expensive, etc..) in this design so far.
Now, the uP's ADC has seperate +&- voltage references (+Vref & -Vref). Both of which must be within the uP's Vdd & Vcc. I'm planning on using a "precision" voltage reference for that (as the uP's 5vdc is only 5% regulated).
What I'm wanting to do is a digital PID controller, controlling my triac firing time dependant on the last half cycle's output to my load (a transformer). My ADC is 10bit BTW, and I'm hoping to be able to sample at 24khz, and get an RMS voltage measurement.
Okay, what I was thinking of doing, was to rectify the triac output going to my load with a full bridge, then running it though a voltage divider to give me a 4vdc span to work with (please, no isolation warnings, and they do make 0.05% resistors capable of surviving 10kv spikes). Okay, so I'll lose about 0.7 volts at the bottom end because of the diodes, I would'nt think it would affect it very much.
Now, I've been told that I could just put in a capacitor after the voltage divider to lower my impedance for the ADC port (must be under 10k, but would prefer 50ohm for faster accquistion speeds). But that will basically induce a bunch of lag into the system for one (I'm sure that would affect a PID's performance), and it won't let me do another trick I was thinking of doing. The load is inductive, so I figured that I could have the uP calculate the phase angle, via a table lookup or something, by which it could determine if there was too much current going to the load (i.e., the secondaries of the transformer are shorted closed). If I sense that condition, I could throw up an error code on the LED display, and just shut the power down to the transformer so that stuff don't start melting (the transformer is capable of putting 20 amps out at 2v, which will melt solder BTW).
This is where I'm thinking that I should use and Opamp setup instead. BUT (and it's a big one ;-), I don't have a negative supply to feed to an opamp (evidentally, so called "rail to rail" opamps "aren't really" R2R). Even a low R2R of 125mv is quite a lot, when you consider that on a 10bit ADC looking at a 4vdc signal, each step is only 0.0195volts.
Now although I don't have a negative supply.... I think I could maybe "fake it". If I hook up the VB409 to a full bridge rectifier, it puts out a 5vdc bias, but that swings (in reference to a true ground) at 1/2 negative line potential (don't worry, the user will have 100mohm plastic buttons between themselves and the live circuit). But if I "half rectify" the incoming voltage to the VB409, I get a -2.5 & +2.5 5vdc bias (again, when referenced to a true ground). Now here is where it gets tricky, at least for me any ways. Now, on a +2.5~-2.5 5vdc bias setup, a 4v fully rectifed signal, "should" travel from 2.5+ (or 0v referranced to true ground) to 6.5v. Now remember, I do have that 12~15vdc unregulated to play with (I'd power the opamp with that). I'm thinking that I would then level shift that signal down to precisely 0.250vdc to 4.25vdc, then run it though a low pass opamp filter, and then "maybe" run it through a voltage follower after that. To do this, I'd also need a 0.25 and 4.25vdc precision references for my ADC's Vref- & Vref+.
Now, the concerns I have are (1), where WILL a fully rectified (and then voltage divided down to a 4vdc P2P span) run at in referance to a true ground? (2), does the positive and negative supplies going to an opamp have to be equal, in reference to the incoming/outgoing signal? (3) Any suggestions on where a low pass filter should cut off at (24khz sampling on a 60hz SCR waveform)? (4) If I half rectifiy it like that, then I'm stuck on how to detect ZC off the mains.
Yep, it's a diller of a pickle I've gotten myself into here, ain't it ;-).
Thanks in advance, for any help given on this matter.