Re: [buildcheapeeg] High resolution EEG schematic

From: Sar Saloth (sarsaloth_at_yahoo.com)
Date: 2002-03-07 21:28:09


--=====================_32583302==_.ALT
Content-Type: text/plain; charset="us-ascii"; format=flowed

At the risk of sound like a town-hall-committee wannabe, I would like to
make some comments on the schematic. There hasn't been much of that
nonsense here, so I will supply some. As can be seen below, I learned
something trying to critique the circuit, but left the comments in for
people's amusement.

Has the virtual ground circuit been tested? My concern is that such a
design can be unstable. I have seen a design that was almost identical
made with the cheap old and dirty TI Cmos op-amps (like TLC27L4). It was
obviously not an important problem as it was in a product that was made for
15 years. Most of the time, it had a small amount of ripple. However, if
you were able to give it an impulse that would upset it (like a static
shock) the circuit would go into large-scale oscillation (1Vpp on a 3.3V
supply) and would often only stop if powered down.

TI makes a part that is a "virtual ground rail", one version of which is a
constant 2.5V or so above the negative rail and the other part is a
rail-splitter. I have never used either of them. They were supposed to be
around a dollar and they are designed for this sort of application.

At the risk of theoretically losing some regulation, the virtual ground
circuit in the schematic could be made unconditionally stable by replacing
the negative feedback trace with a capacitor, adding some resistance in
series (depends on the op-amp, start around 100ohms but certain newer
opamps can use much less). The resistance should be between the output of
the opamp and the 47uF cap, but the feedback cap should go right to the
opamp output. Finally, a resistor should go from the opamp - terminal to
the 47uF cap for DC accuracy. The time constant of the RC should be chosen
just slow enough for stability. With the large 47uF cap and the ceramic
caps, the regulation of the output should still be quite good.

A better solution would be to combine the above with an opamp designed for
a capacitive load. According to
http://www.linear-tech.com/pub/document.html?pub_type=desn&document=127,
the LT1114 is only stable up to 10nF load at unity gain. Oops, that means
that you don't need to start at 100ohms in series. The precision of the
ground splitter is not important, so an expensive precision op-amp should
not be required. That same article lists amplifiers that claim to be stable
with any capacitive load. I have no experience with any of these, in fact
I didn't know they existed until right now! (I left this section in for
your amusement). Does that eliminate the problem that is the main point of
this email? That article was 7 years old, so maybe there are a lot of
opamps to choose from now that will be stable in the circuit.

>>----- Original Message -----
>>From: <mailto:sleeper75se_at_yahoo.se>sleeper75se
>>To: <mailto:buildcheapeeg_at_yahoogroups.com>buildcheapeeg_at_yahoogroups.com
>>Sent: Wednesday, March 06, 2002 6:24 AM
>>Subject: [buildcheapeeg] High resolution EEG schematic
>>
>>Hello people,
>>
>>since there are a few more members here now that are into electronics
>>(biomedical, even), I might as well post a design that I have been
>>sitting on for quite some time...
>>
>>It is a variation on the "sound-card" solution discussed previously,
>>but with a difference.
>>Rather than using the sound card's ADC, you use your own, and
>>transmit the sound digitally to the sound card via optical fiber.
>>
>>Some specifications:
>>
>>Two channels, a 24-bit (16 bits effective resolution) ADC (CS5360)
>>followed by an digital-audio interface transmitter (CS8405A) and an
>>optical transmitter (Toslink, TOTX173). There is no microcontroller
>>or any other programmable parts.
>>
>>The sample rate is 48kHz since it's based on audio circuits, so
>>downsampling in software is necessary. By doing this you can get
>>higher resolution, perhaps 18 bits. It depends on the ADC's noise
>>spectrum if it works.
>>
>>I would love to have comments from everyone, and it would be nice if
>>someone could take time to try to build it. The parts are fairly
>>cheap and the part count is very low. A ballpark figure is < $100,
>>not counting the sound card.
>>You will need a 12V lead-acid battery and a 5V linear regulator (7805
>>is fine) to power it, as it consumes > 100mA current.
>>
>>Some experience with low noise PCB design and debugging skills are
>>essential. It is just a paper product right now, so it might not even
>>work ... :-)
>>
>>The schematic is posted here:
>>
>><http://groups.yahoo.com/group/buildcheapeeg/files/hardware%20design/>http://groups.yahoo.com/group/buildcheapeeg/files/hardware%20design/
>>
>>/Andreas
>>
>>
>>
>>To unsubscribe from this group, send an email to:
>>buildcheapeeg-unsubscribe_at_egroups.com
>>
>>
>>
>>Your use of Yahoo! Groups is subject to the
>><http://docs.yahoo.com/info/terms/>Yahoo! Terms of Service.
>
>Yahoo! Groups Sponsor
>ADVERTISEMENT
><http://rd.yahoo.com/M=215002.1818248.3328688.1261774/D=egroupweb/S=1705060814:HM/A=847665/R=0/*http://ads.x10.com/?bHlhaG9vbW9uc3RlcjcuZGF0=1015468471%3eM=215002.1818248.3328688.1261774/D=egroupweb/S=1705060814:HM/A=847665/R=1>1f12338.jpg
>1f123f6.jpg
>
>To unsubscribe from this group, send an email to:
>buildcheapeeg-unsubscribe_at_egroups.com
>
>
>
>Your use of Yahoo! Groups is subject to the
><http://docs.yahoo.com/info/terms/>Yahoo! Terms of Service.

--=====================_32583302==_.ALT
Content-Type: text/html; charset="us-ascii"

<html>
At the risk of sound like a town-hall-committee wannabe, I would like to
make some comments on the schematic.&nbsp; There hasn't been much of that
nonsense here, so I will supply some.&nbsp; As can be seen below, I
learned something trying to critique the circuit, but left the comments
in for people's amusement.<br><br>
Has the virtual ground circuit been tested?&nbsp; My concern is that such
a design can be unstable.&nbsp;&nbsp; I have seen a design that was
almost identical made with the cheap old and dirty TI Cmos op-amps (like
TLC27L4).&nbsp; It was obviously not an important problem as it was in a
product that was made for 15 years.&nbsp; Most of the time, it had a
small amount of ripple.&nbsp; However, if you were able to give it an
impulse that would upset it (like a static shock) the circuit would go
into large-scale oscillation (1Vpp on a 3.3V supply) and would often only
stop if powered down.<br><br>
TI makes a part that is a &quot;virtual ground rail&quot;, one version of
which is a constant 2.5V or so above the negative rail and the other part
is a rail-splitter.&nbsp; I have never used either of them.&nbsp; They
were supposed to be around a dollar and they are designed for this sort
of application.<br><br>
At the risk of theoretically losing some regulation, the virtual ground
circuit in the schematic could be made unconditionally stable by
replacing the negative feedback trace with a capacitor, adding some
resistance in series (depends on the op-amp, start around 100ohms but
certain newer opamps can use much less).&nbsp; The resistance should be
between the output of the opamp and the 47uF cap, but the feedback cap
should go right to the opamp output.&nbsp; Finally, a resistor should go
from the opamp - terminal to the 47uF cap for DC accuracy.&nbsp; The time
constant of the RC should be chosen just slow enough for stability.&nbsp;
With the large 47uF cap and the ceramic caps, the regulation of the
output should still be quite good.<br><br>
A better solution would be to combine the above with an opamp designed
for a capacitive load.&nbsp; According to
<a href="http://www.linear-tech.com/pub/document.html?pub_type=desn&document=127" eudora="autourl">http://www.linear-tech.com/pub/document.html?pub_type=desn&document=127>,
the LT1114 is only stable up to 10nF load at unity gain.&nbsp; Oops, that
means that you don't need to start at 100ohms in series.&nbsp; The
precision of the ground splitter is not important, so an expensive
precision op-amp should not be required. That same article lists
amplifiers that claim to be stable with any capacitive load.&nbsp; I have
no experience with any of these, in fact I didn't know they existed until
right now! (I left this section in for your amusement).&nbsp; Does that
eliminate the problem that is the main point of this email?&nbsp; That
article was 7 years old, so maybe there are a lot of opamps to choose
from now that will be stable in the circuit.<br><br>
<br><br>
<br>
<blockquote type=cite class=cite cite><blockquote type=cite class=cite cite>-----
Original Message ----- <br>
<b>From:</b> <a href="mailto:
sleeper75se_at_yahoo.se">sleeper75se</a> <br>
<b>To:</b>
<a href="mailto:buildcheapeeg_at_yahoogroups.com">buildcheapeeg_at_yahoogroups.com</a> <br>
<b>Sent:</b> Wednesday, March 06, 2002 6:24 AM<br>
<b>Subject:</b> [buildcheapeeg] High resolution EEG schematic<br><br>
<tt>Hello people,<br><br>
since there are a few more members here now that are into electronics <br>
(biomedical, even), I might as well post a design that I have been <br>
sitting on for quite some time... <br><br>
It is a variation on the &quot;sound-card&quot; solution discussed previously, <br>
but with a difference.<br>
Rather than using the sound card's ADC, you use your own, and <br>
transmit the sound digitally to the sound card via optical fiber.<br><br>
Some specifications: <br><br>
Two channels, a 24-bit (16 bits effective resolution) ADC (CS5360) <br>
followed by an digital-audio interface transmitter (CS8405A) and an <br>
optical transmitter (Toslink, TOTX173). There is no microcontroller <br>
or any other programmable parts.<br><br>
The sample rate is 48kHz since it's based on audio circuits, so <br>
downsampling in software is necessary. By doing this you can get <br>
higher resolution, perhaps 18 bits. It depends on the ADC's noise <br>
spectrum if it works.<br><br>
I would love to have comments from everyone, and it would be nice if <br>
someone could take time to try to build it. The parts are fairly <br>
cheap and the part count is very low. A ballpark figure is &lt; $100, <br>
not counting the sound card.<br>
You will need a 12V lead-acid battery and a 5V linear regulator (7805 <br>
is fine) to power it, as it consumes &gt; 100mA current. <br><br>
Some experience with low noise PCB design and debugging skills are <br>
essential. It is just a paper product right now, so it might not even <br>
work ... :-)<br><br>
The schematic is posted here:<br><br>
<a href="http://groups.yahoo.com/group/buildcheapeeg/files/hardware%20design/">http://groups.yahoo.com/group/buildcheapeeg/files/hardware%20design/><br><br>
/Andreas<br><br>
</tt><br><br>
<tt>To unsubscribe from this group, send an email to:<br>
buildcheapeeg-unsubscribe_at_egroups.com<br><br>
</tt><br><br>
<tt>Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.</tt> </blockquote><br>
<font size=2 color="#003399"><b>Yahoo! Groups Sponsor</b></font> <br>
<font face="arial" size=1>ADVERTISEMENT</font><br>
<a href="http://rd.yahoo.com/M=215002.1818248.3328688.1261774/D=egroupweb/S=1705060814:HM/A=847665/R=0/*http://ads.x10.com/?bHlhaG9vbW9uc3RlcjcuZGF0=1015468471%3eM=215002.1818248.3328688.1261774/D=egroupweb/S=1705060814:HM/A=847665/R=1"><img src="cid:.0" width=300 height=250 alt="1f12338.jpg"></a><br>
<img src="cid:.1" width=1 height=1 alt="1f123f6.jpg"><br><br>
<tt>To unsubscribe from this group, send an email to:<br>
buildcheapeeg-unsubscribe_at_egroups.com<br><br>
</tt><br><br>
<tt>Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.</tt> </blockquote></html>

--=====================_32583302==_.ALT--



1f12338.jpg

1f123f6.jpg

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