From: Joerg Hansmann (info_at_jhansmann.de)
Date: 2002-01-23 20:35:22
Hi Andreas,
----- Original Message -----
From: sleeper75se <sleeper75se_at_yahoo.se>
To: <buildcheapeeg_at_yahoogroups.com>
Sent: Wednesday, January 23, 2002 12:56 PM
Subject: [buildcheapeeg] Re: virtual ground problem
>...
> So, the total resistance that the virtual ground amplifier sees,
> excluding the filter, is somewhere around (50k || 10k || 1M || 330k
> || 360k) = 7.9kohms, but this can easily be doubled if the second
> gain stage is modified.
It must be changed anyway, because the first stage simply can not
have G=1000. (New schematic is under development...)
Because of the additional second channel the total is 7.9kOhms/2 .
Making 3.95kOhms.
I think this is no driving problem for the virtual ground splitter.
Below 200Hz its output impedance is better than 10mOhms.
It can easily drive up to 10mA.
> Now, this leaves us with the filter. What happens if the filter
> capacitors
They are all effectively in series with over 300k resistors.
So they have neglectable impedance compared to e.g. 10k
> are connected to the negative rail (or "real" ground)
> instead of virtual ground? There is never any DC current passing
> through the capacitor, so the DC level on the ground-side is quite
> uninteresting, right?
Better not. With DC bias on the Cs you will produce DC leak currents.
Also mechanical shock resulting in temporary C-changes would produce
AC voltages.
> My question: If the filter caps are connected to the negative
> rail/real ground how much decoupling is really necessary for the
> rest?
The filter caps are not the main burden. So why change their ground
connection ?
> Could not a regular opamp handle this all by itself without a
> 47 uF decoupling?
The 47uF (tantalum, could also be 100uF or 33uF) with its ESR (e.g. 500mOhms) is _absolutely_ necessary to provide enough phase
reserve for stable operation. (The right ESR-C combination inserts a zero or
HP-element that provides about 90 degrees more than without)
A 100nF bypass C alone (or more ,distributed at the opamp power pins)
would leave the rail-splitter oscillating (or at least ringing)
Regards,
Joerg
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:37 BST