Re: [buildcheapeeg] Re: Input protection

From: Joerg Hansmann (info_at_jhansmann.de)
Date: 2002-01-02 01:20:39


Hi Andreas,

----- Original Message -----
From: Andreas Robinson <sleeper75se_at_yahoo.se>
To: <buildcheapeeg_at_yahoogroups.com>
Sent: Tuesday, January 01, 2002 4:30 PM
Subject: [buildcheapeeg] Re: Input protection

> Hi Joerg,
...
> Well, the only thing I can see is limited by low
> voltage is common mode range, but that has nothing to
> do with the protection. Can you think of anything
> else?

Lowest possible operating voltage of the used OPs.
Are they all rail-to rail ? (Or at least near to that)

...
> > For what purpose are the 1nF C's C1..3 ?
> > For shunting HF they must be connected to a low-
> > impedance node (what they are seemingly not)
>
> <confused> They don't attenuate HF?

Only in differential mode as far as I can see.
However common mode rejection of OPs decreases at high
frequencies. So HF should also be shunted for common mode.

> My intention here
> was to let the capacitor act like a short for high
> frequencies between a channel and the reference. It
> looks good on paper/in simulation.

Also for common mode HF signals ?

...
> Please consider using transistors rather than diodes.
> See the attached schematic where I've replaced the
> diodes with transistors.
> The transistors don't not protect the user very well

A forward voltage of 0.7 V together with the 15k R
would in worst case exceed 50uA. On the other hand
I do not like to increase the 15k significantly
(in combination with INA114).

> due to their higher turn-on voltage, but the diodes,
> while an effective protection, leak far too much.
>
> BAR42 schottky diodes (Vf = 0.4V supplied with
> Simetrix, that's why I used them) leaked 1nA with a
> 1mV signal. The transistors leak only a fraction of
> that, around 2pA. It gets worse when you add an
> offset. At 100mV offset the diodes let 100nA through
> while the transistors stay at a lower 20pA. Of course,
> all these calculations depend somewhat on what
> capacitors are used...

Oops, I did not realize that shottky diodes were so bad.
Are BAT48 diodes(max 40V, 0.35 A) comparable with BAR42
(max 30V,0.1 A)?
I fear they are even worse ...

Do such things like shottky transistors exist ?
(Like in the 74LS- TTL-series)

> I'm beginning to think we would need to measure the
> current through the electrodes and have an active
> emergency-shutoff...

I have had the same idea, but could not find a
way to measure the error-condition without
producing noise, degrading impedance or adding new
risk by the protection circuit.

> That is, if we should still aim for FDA/IEC
> compliance?

At least IEC, because it is easier to comply.

> I'm leaning towards just putting something
> small and simple together now, so that people can get
> started writing software

For writing SW the digital part would be enough.

> as soon as possible. FDA can
> perhaps wait a bit... and the paranoid can always
> check the amplifiers with a multimeter before
> connecting themselves.
>
> > BTW: Do you have an idea how the PSPICE AMB
> > equations for OP-amps looks like ? (gain and
> > three poles would be enough ...)
>
> Unfortunately no.
> I'm guessing here that your spice-simulator lacks
> built-in support for opamps?

It is restricted (freeware version) to 3 real op-amps
simulated on transistor level or 50 parts over all.

> You have previously
> mentioned that you were forced to simulate opamps
> using voltage controlled voltage sources.

Yes, it is a more or less accurate work-around.

> In that case, take a look at Simetrix:
> http://www.newburytech.co.uk/
>
> The user interface is a bit different, and the free
> version limits circuit size, but other than that, it
> is very good.

I will have a look at it. (At the moment I am trying switchercad,
a free and _unlimited_ PSPICE. See addendum for more)

...
> I made a little practical experiment the other day:
>
> I've built some electrodes of 18mm silver discs, and
> using two of them I measured the impedance on the
> lower arm.
>
> First it was cleaned with alcohol and salt water was
> poured onto it.

A kind of sponge or cellulose tissue that is soaked with the
saline would be better I assume.

> The electrodes (non chloridized, also
> cleaned with alcohol) were applied and pressed hard
> onto the skin. The resistance was measured with a
> multimeter.

Normal multimeters use DC current to measure resistance.
This quickly produces a galvanic reaction (like loading a
battery) and falsifies the reading.

> The results: 160Kohms when they were about 10 cm apart
> and 24Kohm when they were 1 cm apart.

Hmmm. the distance should not matter, because the transitional
resistance of the skin to the body internal normally is much higher
than the resistance of the internal.

> I would have gotten lower values

Perhaps you should use 10/20 paste wax...

Regards,

Joerg

BTW: I am just testing this here:
(could be a good alternative to my limited microsim eval version)

SwitcherCAD III
---------------
Free, high-performance, unlimited, fully functional
SPICE simulator with integrated schematic capture and waveform
viewer:

"...
This program will run all those Pspice models in public
domain or available from vender sites, but faster and
more reliably than Pspice. Somehow I've gotten the vague
impression that there is no one working full time on SPICE
internals for Pspice, not that I have any inside
information at all or am in any way connected with
MicroSim/OrCAD/Cadence. Maybe Mohi who posts here
from his Cadence address can verify.

Version 1.10h is available for immediate download(save
possible site mirroring delays) as

http://LTspice.linear-tech.com/software/swcadiii.exe

People with earlier versions installed can use the "Tools=>Sync
Release" menu option to incrementally update to the current
version and bypass all site mirroring.

This is a free, high-performance, unlimited, fully functional
SPICE simulator with integrated schematic capture and waveform
viewer. It is not shareware or a demo. It is a fully functional
freeware.

* P-SPICE compatible semiconductors models, e.g.:
- diode recombination current
- bipolar quasi saturation
- JFET impact ionization
* P-SPICE behavioral modeling, both legacy POLY() statements
and arbitrary expressions, Laplace, and look-up tables.
* P-SPICE netlist compatibility for .param, .func, ako models...
* New devices such as an arbitrary capacitance. Write an
expression for the charge, it symbolically differentiates the
expression and integrates the capacitances. I've found it
useful for doing rapid prototyping of hypothetical transistor
behavior.
* An original mixed mode simulator(not xspice based)
* A VDMOS transistor device that intrinsically handles the
power MOSFET's gate charge behavior
* BSIM 3.2.2, BSIM 4.1, BSIMPD2.2, and EKV 2.6
* Marching waveforms/cross probing of any device or subcircuit
port current
* Dynamic waveform data compression
* Device library/database(power MOSFETs, diodes, bipolars, JFETS)
* 350+ models of switch mode controllers
* Integrated schematic capture with unlimited undo and redo
* Parametric plotting(often called X-Y plotting)
* Fourier analysis(both .four statements and FFT's of simulated
data.)
* Numerous educational example circuits included.
* Free lifetime updates

This is a very easy to use SPICE/Schematic capture system. It
includes clear, on-line help but you should just be able to
mouse it out without reading any documentation. You draft your
circuit and push the "Run" toolbar button. There are numerous
examples included in the "examples\Educational" directory.

It is the World's fastest SPICE. It started out as a re-write
of Berkeley SPICE 3F4/5 to improve it's performance on machines
with faster processors than memory and to improve the timestep
control of Berkeley SPICE. Analog design engineers use it for
full-chip, transistor level simulations. Typical speed-ups
over P-SPICE 9.1 are three to six fold. It wins the industry
standard transistor-level SPICE bench mark, the MCNC Bench Mark
Suite, in speed and convergence. Of course, when using it's
own mixed-mode simulator in it's own native HDL, there's
nothing like it at all. There's been articles appearing, e.g.,
the June 18th Electronic Design[1] and there are advertisements
in various current trade journals. It's called SwitcherCAD III
and you can download it from www.linear-tech.com/software/.
..."



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