-1
\$\begingroup\$

I Have designed a circuit to acquire the EEG signal from electrode within 0-20 microvolt. My design consists of AD620 (gain = 2), -> LPF (4th order G=2.515, Fc= 0-40Hz),-> HPF (4th order G=2.515, Fc = 0.1 to 40 Hz), -> Two stage amplifier with G=180. But I am getting a constant output at AD620 of 1.28V without or with input connected. And the output after LPF filter HPF, and two stage amplifier is found to be 4.78V constantly. I am not getting why this error persists. I want to acquire EEG signal to make it in acceptible limit of ADC of 0-5V.I want to give the acquired EEG signal to 24 ADC and then to ARM Cortex M3 for further processing. Image 1 consists of Instrumentation amplifier , LPF ,HPF (4th order filter of 0-40Hz) Image consists of AD620 ->LPF4thorder -> HPF 4thorder -> output to amplifying stage two stage with G=180 each

\$\endgroup\$
  • 2
    \$\begingroup\$ Can you post some schematic? Your circuit description is not sufficient to start. \$\endgroup\$ – Adam Calvet Bohl Dec 5 '16 at 9:39
  • 1
    \$\begingroup\$ It sounds like you haven't got the signals DC biassed correctly. You need to post a schematic of what you are doing, including all components in the signal path. Your apparently constant 1.28V at the first output might be a problem, or it might be you just don't see the small signal there. If one schematic is too big, then you may break it down into first amplifier and second amplifier, and in fact this partition is a good way into understanding the problem. But don't leave out any components. \$\endgroup\$ – Neil_UK Dec 5 '16 at 9:45
  • 1
    \$\begingroup\$ Your words tell me you know what you are wanting to do but without a circuit there is no answer. Show all power rails and decoupling capacitors - don't scrimp on this bit. \$\endgroup\$ – Andy aka Dec 5 '16 at 9:56
  • \$\begingroup\$ You probably know this, but just in case openeeg.sourceforge.net/doc \$\endgroup\$ – Dirk Bruere Jul 23 '17 at 20:27
1
\$\begingroup\$

I can see a problematic schematic but I'm going to take a stab as to what is wrong further down the line.

enter image description here

Firstly the problematic schematic; there are no values on the power rails shown - you need to do this and, in addition, the negative supply to the AD620 is not connected. This is enough to make the circuit not work at all.

I'm assuming that the negative pin is connected but you are likely suffering from the inputs not staying within the common-mode range required by the AD620. What the data sheet tells you is this; the inputs MUST be within the range of -Vs+1.9 volts to +Vs-1.2 volts. The reference input is even more constrained.

This is your likely problem even if you were to connect power to the -Vs pin of the AD620.

Moving on to the next issue, the sallen key filter is connected-up incorrectly. For instance, C_2 should be connected to the right of R3 and C_1 should be connected to the right of R_3 and not directly on the output of the AD620 - this is asking for trouble.

\$\endgroup\$
  • \$\begingroup\$ Sir I have given +-5 V to AD620 and Reference pin of AD620 to GND. And my input from electrode is 15-20 microvolt from EEG electrode. \$\endgroup\$ – user131517 Dec 5 '16 at 10:16
  • \$\begingroup\$ I have made the PCB and checked the connections as per schematic with all VCC, VEE connected. \$\endgroup\$ – user131517 Dec 5 '16 at 10:23
  • \$\begingroup\$ But you haven't connected -Vs to the AD620 on the schematic so checking the PCB with the schematic won't indicate this to you. You have something wrong with -Vs on the AD620 - I can't conclude anything else because your schematic shows no connection. Neither have you shown where the inputs connect to and why is pin 5 (reference) connected to -Vin? You need to explain all of this. \$\endgroup\$ – Andy aka Dec 5 '16 at 10:54
0
\$\begingroup\$

Note that for +/-15V power, maximum output voltage swing for the TL074 is to +/-12V. The chip needs about 3V of headroom. With your 5V power supply, an input offset voltage of 10mV on the TL074, and the gain near 200 on one stage, you may well be saturating your TL074. This is aside from any of the input issues Andy was talking about.

In terms of debugging, I'd recommend boosting your power rails to +-15V to check for simple headroom problems. I would test each stage of amplification and filtering separately, or at least go from left to right, verifying that each stage is working, one at a time. To do this, I'd recommend building and testing a high pass filter, maybe using your last stage, and passing the output of each stage to that, so you can see signals more easily on a scope.

Also, you're unclear about what exactly your input is. Are you using electrodes and a person? If so, you need a third electrode to reference your subject. You should have about 500K of series resistance between this electrode and ground, for subject protection.

After all of this you still need to add 2.5V to your signal somewhere to bring the voltage into range for an ADC. I suggest using a precision voltage reference to generate a reference voltage, and use it EVERYWHERE in place of GND, instead of grounding -- and this includes the patient reference, but you REALLY need that 500K series resistor.

\$\endgroup\$
  • \$\begingroup\$ Yes I want to take the signal from person with the help of electrodes. \$\endgroup\$ – user131517 Dec 6 '16 at 3:40
  • \$\begingroup\$ According to your say, I connected only AD620 with Rg of 51K on Breadboard. But the same problem persists. The connections were as follows: Pin 1 & Pin 8: Rg=51K. Pin 2(inverting) : Gnd, Pin 3(Non Inverting): case I-Floating, CaseII-Gnd, Pin 4: -5V, Pin 5(REF): GND, Pin 6(OUTPUT): , Pin 7:+5V. But still I am getting 4.75V output constantly without any other circuit. \$\endgroup\$ – user131517 Dec 6 '16 at 11:01
  • \$\begingroup\$ Stop using a person to test on. What if you ground the inputs? \$\endgroup\$ – Scott Seidman Dec 6 '16 at 11:42
  • \$\begingroup\$ What gain should 51k give. \$\endgroup\$ – Scott Seidman Dec 6 '16 at 12:00
  • \$\begingroup\$ What are case I and caseII? \$\endgroup\$ – Scott Seidman Dec 6 '16 at 12:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.