From: Jim Meissner (jpmeissner_at_mindspring.com)
Date: 2002-01-08 01:51:14
Dear Andreas:
> I suppose what you are recommending here is
> 10ohms or so in series with each capacitor? NO!
In my days of designing tube amplifiers we would run into a problem called motor boating. This is a very low frequency feedback coupled through the B+ power supply line. The bypass capacitors would be large enough to short the audio frequencies to ground, but as the frequencies became lower the capacitors became ineffective. The solution was to add resistors between stages of bypass capacitors to make R-C filters. Of course by the time you got to the input stage the B+ would be down to a third or half of the power stage voltage. A phonograph preamp stage and power amplifier is similar in gain to what you are building here.
> Won't breadboarding cause a lot of trouble in itself, like high
> parasitic capacitances?
Just the opposite. PC board will have much higher (but controlled) capacitance than point to point wiring.
> I usually build on PCB directly (preparation and etching of a 2-sided
> board takes about 2 hours) and then build and patch whatever needs
> patching. That's hard to do with ground planes of course.
Hey, if you can do it that fast, great! Then do it and redo it. You may be lucky and it works the first time!
> 5) The Bessel filter may be an enormous overkill.
> I found no significant brain wave frequencies above 60 Hz.
> Do you mean that it perhaps can be omitted entirely? It is not that
> expensive...
You will need some filter ahead of the A/D to get rid of aliases. My application involved biofeedback and listening to the brain on speakers. I wanted full range frequency response. That may not be of interest to you.
> 6) Your input protection diodes should also go to a good ground.
> I used a 2N3904 for a reason. Don't arbitrarily change transistors
> unless you know why.
> What property/ies of the transistors is the most important in this
> case? Can I make a first selection using datasheets or is trial and
> error the only way? BC547 has a leakage of a few picoamps near zero
> volts base-emitter voltage.
If I were not retired I would not tell you the trade secret. It took a long time to find this solution. Any regular diode is light sensitive (generates stray voltage) and when you pot it then it is pressure sensitive and causes drift. Metal can transistors have glass seals and are also light sensitive. Regular diodes make good RF detectors. I ran into the 2N3904 solution by accident. This is a good light tight epoxy package. You will not find this in the spec sheets. The base emitter is not as good as the base collector. The emitter collector makes a good low leakage zener diode for op amp bounding. Now, the BC547 might be better ( I have no idea ), but I would try the 2N3904 first because I have had good experience with it.
Juergen P. (Jim) Meissner
Check out my Website at www.MeissnerResearch.com
Read about the benefits of the Brain State Synchronizer sounds for improving your life and health.
----- Original Message -----
From: sleeper75se
To: buildcheapeeg_at_yahoogroups.com
Sent: Monday, January 07, 2002 4:09 PM
Subject: [buildcheapeeg] Re: TinyEEG
--- In buildcheapeeg_at_yahoogroups.com, "Jim Meissner" <jpmeissner_at_mindspring.com> wrote:
> Dear Andreas: (and welcome back Doug)
>
> Since you asked for some comments, here are a few.
> These are not in any order. They are meant to be constructive!
Hello Jim,
thanks for your comments, they are very helpful to me.
>
> 1) The RF bypass caps (C1 and C53) must go to a good solid ground
> like chassis or pc board ground plane.
Acknowledged. I won't be using the shield driver then.
> 2) The inductor L2 and R56 must go in the (-) power bus. I am not
> convinced the inductors are needed if you are running from
> batteries.
Me neither, but I'll keep the inductors and then try to remove them
and see what happens.
> > It runs on four 1.5V batteries.
> Do you intend to run the op amps from +/- 3 volts and then float the
> processor? That could be big trouble!
I've built a few mixed signal devices this way before, and I always
get this 1-10mV, 400MHz ringing all over the place. It comes in 8MHz
pulses - matching the switching of the 4MHz oscillator. I'm half
expecting the same here, so its definitly time for a new approach.
Will try to add a 4.5V battery for the MCU.
> 3) The .1 uf bypass caps C27-37 etc need to be distributed
> with resistors throughout the various op amp stages.
The caps are placed like that on the board. That they are placed
together on the schematic is just for convenience.
About the resistors, I suppose what you are recommending here is
10ohms or so in series with each capacitor?
> 4) With such a small board, keeping the processor noise out of
> the input will be take some careful design and layout. This should
> be breadboarded first before committing to a pc board layout.
Won't breadboarding cause a lot of trouble in itself, like high
parasitic capacitances?
I usually build on PCB directly (preparation and etching of a 2-sided
board takes about 2 hours) and then build and patch whatever needs
patching. That's hard to do with ground planes of course.
> You might think of this as two projects, the preamp, and rest of
> the circuit.
Ok, I'll draw a distinct line between the preamplifier and the rest.
If big changes are necessary, I can just cut the board in two.
> Consider a shielded box for the input stage 1NA114 and associated
> components.
will do that.
> 5) The Bessel filter may be an enormous overkill.
> I found no significant brain wave frequencies above 60 Hz.
Do you mean that it perhaps can be omitted entirely? It is not that
expensive...
> 6) Your input protection diodes should also go to a good ground.
> I used a 2N3904 for a reason. Don't arbitrarily change transistors
> unless you know why.
What property/ies of the transistors is the most important in this
case? Can I make a first selection using datasheets or is trial and
error the only way? BC547 has a leakage of a few picoamps near zero
volts base-emitter voltage.
> 7) What is your decision to use the 1NA114 rather than the lower
> noise LT1012.
Good question.
LT1012 is superior but would make the input stage more complex as it
is a dual opamp and not an instrumentation amplifier.
Now that you have made me look at it again... Linear manufactures
LT1167 - an instrumentation amp that can handle 13kV ESD (human body
model) using only two 5k resistors at the inputs. It is a drop-in
replacement for INA114 as far as I can see, but with better
performance.
Here's the datasheet: current-noise figures are on page 9.
http://www.linear.com/pdf/1167f.pdf
> 8) Have you or Joerg actually used the left leg drive?
> I have had no experience using it. I do know why it is used for
> ECG.
I think Joerg has used it. I'm still at the drawing board. :-p
> Do you have any information that it works better than the
> traditional circuit ground reference for EEG.
The leg driver is used and recommended by Biosemi, a dutch EEG-
equipment maker:
Basically, they say that it should be used always, by everyone. :-)
A relevant question from their FAQ:
http://www.biosemi.com/faq/limit_current.htm
All of Biosemi's research articles are here:
http://www.biosemi.com/publications.htm
Quoting from one of their articles (third one in the list I'm linking
to) on the subject:
"A proper driven right leg circuit (see Fig. 2) offers a large
reduction of common mode voltage magnitude in both isolated and non-
isolated measurements by actively reducing the voltage difference
between patient and amplifier common; a reduction between 10 and 50
dB is usually accomplished. A driven right leg circuit is the most
practical way to reduce the common mode voltage if a reduction of
interference current through Zrl is not feasible.
In addition, the driven right leg circuit makes measurements
reasonable safe in a non-isolated situation (switch closed in Fig. 1
and Fig. 2) because a rather large impedance between body and ground
can be achieved by selecting a large resistor R0 (several MOhm) and a
small feedback capacitor Cfb (< 1 nF). This feature can be used to
omit isolation amplifiers in experimental situations in which safety
standards are not as critical as in clinical situations.
The main drawback of a driven right leg circuit is it being
potentially unstable (Winter and Webster, 1983). In practical
designs, a compromise between common mode suppression and possible
instability - depending on circumstances - must be found. This
problem is worked out in more detail in the appendix."
And, of course, the appendix is missing .... :)
Hmm, I think I can solve that with a trip to the university library,
they should be able to find the article for me.
> > It uses a LED for data transmission into a plastic
> > fiber. I've posted a schematic for the receiver
> > previously.
> Could you tell me the date when you posted it, so I can look at it?
Please do, but it is available in the files-section under hardware
design / PDF_and_PNG_schematics.
http://groups.yahoo.com/group/buildcheapeeg/files/hardware%
20design/PDF_and_PNG_schematics/
It's been a while since I first built it so I'd like to make a few
amendments: The 1K resistor is probably unnecessary and the
word "flawless" a bit too much - the capacitance in the photo diode
distorts the incoming pulses a bit and that can perhaps cause
problems for some receivers. If I could give more current to the
transistors, perhaps by stealing power from the PC's keyboard-
connector, it would work much better.
Also, I'm referring to a toslink-transmitter. It is designed for
digital audio and can handle a transfer rate of several megabits per
second. Since that is overkill for this application so I'm going to
try using a LED instead.
Whoa, a lot of typing there.
Thanks again for your help.
/Andreas
Yahoo! Groups Sponsor
ADVERTISEMENT
To unsubscribe from this group, send an email to:
buildcheapeeg-unsubscribe_at_egroups.com
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:36 BST