Reference electrode for GSR

Post Reply
Gordak
Posts: 7
Joined: Sat May 03, 2014 12:11 pm

Reference electrode for GSR

Post by Gordak » Fri May 09, 2014 3:24 pm

Hello !

First, thank you for your previous answer.

I'm wondering about the use of the reference and active electrodes in the recording of GSR.
To measure biopotentials, it is important to have a good common mode rejection ratio. I found some schemes on your website. One of them is about CMRR: http://www.biosemi.com/mark8_cmrr.htm

If I understand correctly, the common mode (floating common) is the mean between E1 and E2, respect to the CMS. If this mean (E1 and E2) is more positive than the CMS, electrical current (physical) is injected via the DRL. This will increase the potential near the DRL electrode (and near CMS as well, because of proximity). This decreases the common mode, because E1 and E2 are potentials respect to CMS. If the common mode is more negative than the CMS, current is "absorbed" by the DRL electrode. 1) Is that right :?: I'm not able to understand why the CMRR is 160 dB, but it doesn't matter. If I'm not mistaken, the DRL electrode could get corroded after a while, if it doesn't manage to inject enough current to balance the CM. It would have to inject a almost DC for a while, or not ?. 2) Can the corrosion of the DRL electrode occur and be a problem :?:

About the GSR now. From what I have read, two electrodes are enough to record GSR. A constant voltage can be applied between the two electrode. Then, a small resistance is put in series with the skin and the voltage across this resistance is measured. In the Biosemi device, I have read that the current is a 16 Hz square wave with a 1 uA amplitude. So, measuring a voltage drop across a small resistance would be useless.
Instead, the potential between the two skin electrodes is measured, respect to the reference electrode. This voltage will be proportional to the resistance of the skin. But my concern is about the injected current. It must be constant through one of the electrodes only. Because the skin behaves like a capacitor and charges can accumulate there. The same about the rest of the body. 3) So... the currents across both electrodes are not exactly the same :?: 4) Tt's the voltage between the two electrodes that is measured :?:

My third concern is about the frequency of the injected current between the GSR electrodes. I have read that in some modules, the frequency was 512 Hz and in other it was 16 Hz. 5) How to know the frequency of the injected current :?:

I know it's a long message, but I needed to write it. I feel like I have found some of the answers by myself while writing it.

Could you just confirm my guesses ? It would be nice !

Gordak

Ho and if the BBcode could be activated it would be really great !

Coen
Site Admin
Posts: 954
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Re: Reference electrode for GSR

Post by Coen » Mon May 12, 2014 1:07 pm

For the CMS/DRL circuitry in the ActiveTwo system, please refer to: http://www.biosemi.com/faq/cms&drl.htm --> http://www.biosemi.com/pics/zero_ref1_big.gif (the pre-2000 mark8 circuitry mentioned in your post is more complicated because this mains powered system needed increased CMRR).

To understand the CMR/DRL circuitry, take note of the following rules:

- Common Mode voltage is the potential difference between CMS and the AD-box reference.
- No current flows via the CMS electrode and any of the measuring electrodes.
- Leakage current caused by capacitive couplings of the body with interference sources and capacitive couplings of the AD-box with interference sources flows via the DRL electrode.

From these rules, it follows that:

- No current flows between DRL and any of the other electrodes (including CMS).
- The current via the DRL electrode is determined by capacitive couplings between the body and interference sources, and between the AD box and interference sources, but not by the CMS/DRL circuitry.

The CMS/DRL feedback loop drives the DRL electrode (and body) to such a potential that the potential difference between CMS electrode and AD-box reference is minimized. This increases the Common Mode Rejection Ratio of the system with a factor of 100 at 50 Hz (40 dB) when compared with a simple ground electrode replacing the CMS/DRL (and note that the current through this ground electrode will be exactly the same as via the DRL electrode).

It can be concluded from the schematic that the amplitude of the Common Mode interference remains proportional to the impedance of the DRL electrode. So, even though the CMS/DRL circuitry in combination with battery power and fiber-optic data transfer (http://www.biosemi.com/faq/battery_and_fiber.htm) ensures very low levels of Common Mode in all normal operation conditions, a good electrode contact of the DRL electrode remains important (and more critical than the impedance of the other electrodes including CMS, that do not have to conducts current).

Corrosion of the DRL electrodes does not occur provided the electrode handling instructions are followed.

Best regards, Coen (BioSemi)

Coen
Site Admin
Posts: 954
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Re: Reference electrode for GSR

Post by Coen » Mon May 12, 2014 1:18 pm

The two GSR electrodes are driven by alternating current sources with opposed polarity and equal amplitude. The current sourced by one electrode is exactly sunk by the other. Because the current alternates (with 16 or 512 Hz) no polarization of the electrodes or permanent charging of any stray capacitance occurs.

The frequency of the GSR excitation current can be determined by observing the GSR units in the ActiView: 16 Hz systems display in nanoSiemens whereas 512 Hz systems display in Ohm. Modification to 16 Hz is free of charge for the older 512 Hz systems.

Best regards, Coen (BioSemi)

Gordak
Posts: 7
Joined: Sat May 03, 2014 12:11 pm

Re: Reference electrode for GSR

Post by Gordak » Mon May 12, 2014 2:48 pm

Thank you for your explanations !

Some unrelated simple questions first :
1) If my signal in my BDF file has 'nS' as units, can I conclude that my excitation frequency is 16 Hz :?:

2 ) I have some troubles about the digital to "physical" conversion. It is said here http://www.biosemi.com/faq/adjust_gain.htm that the resolution of the LSB is 31.25 nV. And here http://www.hum.uu.nl/uilots/lab/resources/GSR.PDF I can read that 1 uV = 1 nSiemens. So I would conclude that the range of physical values of conductance goes from -262'000 nS to 262000 nS. Which means that half the values are impossible to get (negative conductance). But I can also read "The GSR signal has an input range of 574 nanoSiemens (1.7 MOhm) to 262000 nanoSiemens (3.8 kOhm)."
Why is that 574 nS ? Is it due to the processing in ActiView or an internal resistance of the device/recording system :?:

Thank you for your answer about the feedback loop.
I have some remaining questions though... :wink:

3) In the scheme you showed me, do E1 and E2 electrodes correpsond to GSR electrodes :?: Or is the circuit a more complex than as it is shown here :?: (I guess it is, because current must be injected) From the scheme I would say that E1 and E2 can't have any current flowing through them...

4) If I'm not mistaken, the potentials E1 and E2 are measured versus the AD box reference, not the CMS. But the feedback loop injects current through the DRL electrode (not via through CMS), which makes the electrical potential of the CMS very close to the electrical potential of the AD box reference, as a result there a very difference between AD box potential and CMS potential. Is that right or am I completely wrong :?: ? I have not much background in electronics, I only try to "get the big lines".

5) I'm having a hard time to understand why the electrical current injected by one of the GSR electrodes is the same in the other. Isn't it possible that some of the injected current leaks into the DRL electrode :?: In that case it would mean that the voltage between the two GSR electrodes would be lower than without any leakage ? So, how do you make sure that the current from one GSR electrode flows into the other GSR electrode ?

Thank you for your patience, I really appreciate it !

Gordak

PS: Ho by the way, is it possible to have a labView code for GSR at 16 Hz ?
Like in this post :
viewtopic.php?f=7&t=607

Another PS question :
6) In this post viewtopic.php?f=7&t=808
You described the ActiView processing of GSR signals coming from USB2 converter before it gets written into a BDF file. Apparently, in ActiView you apply a lowpass filter, some scaling and then you check for the max - min value over a 16 Hz cycle. which gives you the conductance, and you do it with a 32 Hz frequency. But why is the GSR signal sampled at 2048 in the BDF file ? I'm confused about this as well.
7) In the same post, it is suggested that the value written into the BDF file is already converted from uV into nanoSiemens. In the BDF file, there is for each channel a 'physical' min and max and a 'digital' min and max. Does the digital value correspond to the conductance, with some scaling to do to obtain the physical conductance in nS ?

I apologize, I didn't find answers on the forum. Is there somewhere else I could find answers ?

Coen
Site Admin
Posts: 954
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Re: Reference electrode for GSR

Post by Coen » Mon May 12, 2014 5:52 pm

> If my signal in my BDF file has 'nS' as units, can I conclude that my excitation frequency is 16 Hz ?

Yes.

> Why is that 574 nS ? Is it due to the processing in ActiView or an internal resistance of the device/recording system ?

It is a limit of the measurement range determined by hardware.

> In the scheme you showed me, do E1 and E2 electrodes correpsond to GSR electrodes :?: Or is the circuit a more complex than as it is shown here :?: (I guess it is, because current must be injected) From the scheme I would say that E1 and E2 can't have any current flowing through them.

E1 and E2 in the schematic are regular (active) electrode inputs. The GSR inputs are more complicated (with added current sources for the excitation current)

> the potentials E1 and E2 are measured versus the AD box reference, not the CMS. But the feedback loop injects current through the DRL electrode (not via through CMS), which makes the electrical potential of the CMS very close to the electrical potential of the AD box reference, as a result there is a very small difference between AD box potential and CMS potential.

Correct, provided you realize that the current flowing in/out the DRL output is the same leakage current that would flow with a ground electrode instead of the CMS/DRL loop.

> 5) I'm having a hard time to understand why the electrical current injected by one of the GSR electrodes is the same in the other. Isn't it possible that some of the injected current leaks into the DRL electrode :?:

Only when one of the two GSR electrodes is disconnected (or has an exceptionally high impedance), the current will be imbalanced and the excess flows to the DRL. The two green lights for the GSR electrodes in ActiView must both be ON to indicate that the both GSR electrodes are properly conducting the excitation current.

> So, how do you make sure that the current from one GSR electrode flows into the other GSR electrode ?

Current is injected by current sources which means that current flow does not significantly depend on the impedance of the GSR electrodes.

> Ho by the way, is it possible to have a labView code for GSR at 16 Hz ?

ActiView comes complete with source code on http://www.biosemi.com/download_actiview.htm

> But why is the GSR signal sampled at 2048 in the BDF file ?

Because processing of the BDF file is more simple when all channels have the same sample rate.

> Does the digital value correspond to the conductance, with some scaling to do to obtain the physical conductance in nS ?

Yes, see "Note 2" in http://www.biosemi.com/faq/file_format.htm

Best regards, Coen (BioSemi)

Gordak
Posts: 7
Joined: Sat May 03, 2014 12:11 pm

Re: Reference electrode for GSR

Post by Gordak » Mon May 12, 2014 6:15 pm

Thank you so much !

I will have some more questions tomorrow ... You did a great job so far :D

Gordak

Gordak
Posts: 7
Joined: Sat May 03, 2014 12:11 pm

Re: Reference electrode for GSR

Post by Gordak » Tue May 13, 2014 9:51 am

Hi

I have one last question about the measurement of GSR.

> Only when one of the two GSR electrodes is disconnected (or has an exceptionally high impedance), the current will be imbalanced and the excess flows to the DRL.

The reason why the current does not flow through DRL is that the resistance is very high compared to GSR electrodes. What is the resistance put in series with the DRL electrode ? My measurements indicate a value of the skin resistance of about 100 kOhm. The resistance should be like 1 MOhm to be considered as "negligible". However, I can read "The DRL electrode is the only current return path between the subject and the AD-box. The return current is limited electronically at 50 uA" on Biosemi website.
The OA voltage is probably around 15 V or even less, so the resistance of the DRL should be around 10V/50e-10 = 300 kOhm. Is the current limited by other means than a resistance and the voltage of the OA ? Maybe a resistance with a capacitor in parallel ?

Thank you for your explanations ! They are very helpful

Gordak

Coen
Site Admin
Posts: 954
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Re: Reference electrode for GSR

Post by Coen » Tue May 13, 2014 2:11 pm

The reason that the GSR current does not flow to the DRL output is not that the DRL output has a high impedance (the DRL output has a very low impedance). The reason is that one GSR electrode sources exactly the same amount of current that the other GSR electrode sinks. So, no excess current is available to flow between the DRL electrode and any of the GSR electrdes (provided both GSR electrodes are properly attached to the body).

Current flow via the DRL is limited by a resistor inside the DRL integrator feedback loop, see http://www.biosemi.com/publications/pdf ... uction.pdf

Best regards, Coen (BioSemi)

Post Reply