Re: [buildcheapeeg] Re: modular EEG amplifier pcb

From: Joerg Hansmann (info_at_jhansmann.de)
Date: 2002-03-04 02:11:36


Hi Andreas,

----- Original Message -----
From: sleeper75se <sleeper75se_at_yahoo.se>
To: <buildcheapeeg_at_yahoogroups.com>
Sent: Saturday, March 02, 2002 6:46 PM
Subject: [buildcheapeeg] Re: modular EEG amplifier pcb

...
> Could you please post the latest schematic (v1.14, so the designators
> match the PCB)?

I have just uploaded the recent (hopefully final) 1.15 version
(modEEGamp1_15_std_*.*) on

www.jhansmann.de/eeg

> Anyway, here goes:
>
> * The two input stages (protection circuitry and instrumentation
> amps) have different layouts.

Yes. This is partly due to the different areas available on the pcb.

> Since the ground planes add a fair bit
> of capacitance, the HF-properties would be more similar if the
> layouts matched.

HF should not be a problem if it is sufficiently bypassed
(and shielded if necessary).

If the circuit were a HF amplifier that should match another
stage I would agree.

> * IC7 has no ground plane that prevents capacitive coupling between
> the traces under it and the chip substrate.

The chip substate is IMO on V- or AGND. Why should capacitive coupling
be a problem ?

> Looks like Eagle has
> messed up though.

You really like EAGLE, do you ? :-P

The reason was the set value of isolation spacing.

> Aren't the ground planes on the top and bottom the
> same net?

No. Top is AGND2 and bottom plane is VREF/2

> * Linear recommends for its LT1167 that the HF-chokes (C202,C203 etc.
> on the PCB) are placed as close to the input pins as possible. It's
> probably safe to assume the same for INA114.

HF should be rejected before any nonlinear parts that could rectify
the HF and produce DC offsets. So I have placed the bypassing Cs before
the protection circuit.

> On channel 2, the traces
> going from R205,R206 to the inputs, are several centimeters though
> they don't have to be.

You have to connect the inputs with the sockets of the casing anyway.
By lining the inputs up at the left side of the pcb these shielded cables
will have the same length (and C)

> * I read somewhere that you should always use 45 degree turns and
> avoid 90 degree turns. Don't know if it is relevant here as this is
> an LF-design.

Done. It is better for EMC.

> * Design change suggestion: Why not separate analog and digital
> signals into two different cables? If nothing else, it would be
> nimbler, and easier to shield if necessary.

I am not quite sure if the assignment of digital and analog
channels will stay as they are. So the distinction between
analog and digital is not so fixed as it seems now.

Regards,

Joerg



This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:39 BST