Analog-to-digital conversion

From: sleeper75se (sleeper75se_at_yahoo.se)
Date: 2002-01-14 17:00:56


--- In buildcheapeeg_at_yahoogroups.com, "Jim Meissner" <jpmeissner_at_mindspring.com> wrote:

Hi Jim,

> I have used the LT1294, talking to it from the PC with Quick
Basic. Makes you wonder whether you need a micro processor at all?
>
> BUT, seriously, having written the PC program for the Com port,
> FFT, VGA display for Waterfall and waveform, etc there probably is
> not enough time to "handhold" the A/D converter.

Yes, that's another $4 we shouldn't save.

> My preference would be the HC11 ( $9 ) because I know the code and
> I use it in one of my products. The BrainMaster also uses it.

Ok, but AVR assembler is really easy and you personally won't have to
mess with that.

> Would compatibility with the BrainMaster be important?

Not for me. And as far as I know, the protocol is made for 8-bit
samples...(?)

> I would like to address the question of 8, 12, 16 bit converter
> choices.
>
> Having used an 8 bit converter for years, I can tell you the
> frustration with that limited range. I actually had to install a
> manual gain pot in order to get the signal within the range.
> During deep meditations I would have to increase the gain in order
> to see the signal.

That does it for me, I'll look for a good 14-to-16-bit ADC. Should be
around $20, but will probably need an external multiplexer for
another $7.

> At that gain, the eye blinks would overload the amp and cause
> clipping which causes havoc with the FFT analysis. Talk about data
> distortion! This is why I cautioned you about multiple AC coupling
> capacitors. These will create false frequency data to the FFT and
> take a long time to settle.

Yes, an overload takes about 15 seconds to settle.

> Since you and Joerg love to do simulations, look at the step
> response of an overload signal into an FFT. I don't know it your
> software can do transient analysis.

If I remember my signal-processing theory correctly, a fourier
transform of a step would pretty much be smeared all over the
frequency spectrum, with most of the power in the lower bands, so I
see what you mean.

> I would rather spend an extra $10 for the 12 bit converter.
>
> As a matter of fact I would be willing to spend an extra $50 for a
> 16 bit converter and have DC coupling throughout.
>
> Is your design flexible enough to give these options?

Yes. :-) Just change the gain and disable the AC coupling. The mixed-
signal part would have to be specifically designed, unless there are
12 and 16-bit ADC's that are pin-compatible.

I'm not sure a through hole PCB is good enough though. An all SMT
layout, would feel much better ...

/Andreas



This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:36 BST