Simple Instrumentation Amplifier Circuit

Thread Starter

keane2097

Joined Feb 19, 2009
21
I'm trying to debug a larger circuit of mine by breaking it up into smaller parts and dealing with them one by one. One of the parts of the circuit is an instrumentation amplifier.

The chip is using is the INA118 for TI (datasheet here).

The pin-out of the chip is shown here:-



The chip is capable of operating with a single supply, or at least that's what the datsheet has led me to believe. However, to this point I have only been able to get it to do anything meaningful when I use a dual supply of +/- 2.5V. At the moment I'm trying to get it to work witha single supply of +5V & 0V and I'm having no luck at all.

The setup i have at the moment is as follows:-

Pin 1: Connected through two 40kOhm resistors in series to pin 8. Sets the amplifier gain.
Pin 2: Connected through a 390kohm resistor. This is the inverting input pin.
Pin 3: Connected through a 390kohm resistor. This is the non-inverting input pin.
Pin 4: Connected to the negative port of the power supply to give -Vcc = 0V.
Pin 5: Grounded to give 0V offset.
Pin 6: Output.
Pin 7: Connected to the positive port of the power supply to give +Vcc = 5V.
Pin 8: Connected through two 40kOhm resistors in series to pin 1. Sets the amplifier gain.

Using a probe (with the negative end connected to the negative power supply port) I've checked and found that Pins 4 & 7 are at 0V and 5V respectively as desired. Pin 5 is at 0V as desired. Pins 2 & 3 for some reason show a voltage of around 0.48V that I don't know the reason for. Finally, pins 1 & 8 each show a voltage of around 3.5V which i also am not sure is desirable.

I've tried grounding one of the inputs and connecting the other to both a small DC (~20mV)voltage and also to a small amplitude (~20mV p-p) sine wave. In each case I failed to see the expected, amplified and uninverted, output signal.

Can anybody see any issues with how I'm supplying power to my chip? Or any other issue that might be causing my problem?

Thanks...
 

beenthere

Joined Apr 20, 2004
15,819
Just a guess, but from the statement -
Pin 4: Connected to the negative port of the power supply to give -Vcc = 0V
and -
Pin 5: Grounded to give 0V offset.
- you don't have a common ground in the circuit. That is, the power supply ground and the signal grounds must be tied together.

A schematic would very very helpful. Also, why the 390K resistors? Where did you find 40K resistors?
 

Thread Starter

keane2097

Joined Feb 19, 2009
21
A schematic would very very helpful. Also, why the 390K resistors? Where did you find 40K resistors?
This is the overall system I'm trying to build:-



I selected the resistor values just because those are the ones in the schematic...

Just a guess, but from the statement - and - - you don't have a common ground in the circuit. That is, the power supply ground and the signal grounds must be tied together.
When you say "signal grounds" what do you mean exactly? As you can see, in the final circuit the input signals will be taken from electrodes attached to a person's chest...
 

Ron H

Joined Apr 14, 2005
7,063
I don't see how you can substitute INA118 for INA326 in that circuit. They are totally different. You might be able to use 1NA118, but you would have to modify the rest of the circuit.
 

Ron H

Joined Apr 14, 2005
7,063
The common mode input range of the INA118 is from (V-)+1.1 to (V+)-1. See spec Linear Input Voltage Range.

The ECG circuit you posted biases the body to +2.5V.

The gain of the INA326 is 5 in the design. The gain of your INA118 will be 1.625, using the two 40k resistors.
 

Thread Starter

keane2097

Joined Feb 19, 2009
21
I don't see how you can substitute INA118 for INA326 in that circuit. They are totally different. You might be able to use 1NA118, but you would have to modify the rest of the circuit.
Ok that sounds bad - I was just given the components and told to go ahead and build the circuit...

Any idea what kind of changes I'd need to make to get the circuit to work with the 118? Like would it be mainly using different resitor/capacitor values or will i need a more radical overhaul?
 

Thread Starter

keane2097

Joined Feb 19, 2009
21
The common mode input range of the INA118 is from (V-)+1.1 to (V+)-1. See spec Linear Input Voltage Range.

The ECG circuit you posted biases the body to +2.5V.
Would you mind explaining the significance of this somewhat please? Analog circuits aren't my strong point unfortunately...

The gain of the INA326 is 5 in the design. The gain of your INA118 will be 1.625, using the two 40k resistors.
This isn't two big a deal i assume - I can just change the value of the Rg resistors to get whatever gain I want right?
 

Ron H

Joined Apr 14, 2005
7,063
Would you mind explaining the significance of this somewhat please? Analog circuits aren't my strong point unfortunately...
Biasing at 2.5V puts the inputs in the middle of their common mode (linear input) range, so that any stray voltage common to the two inputs (usually 60Hz, or harmonics of same) are less likely to fall out of the allowable range.


This isn't two big a deal i assume - I can just change the value of the Rg resistors to get whatever gain I want right?
That should work.
How are you doing the rest of the circuit? Do you have the same A/D and op amps and voltage reference? Where did you connect the output of A4?

The output of the INA118 will not swing rail-to-rail, if that's what you were expecting.
 
Last edited:

Thread Starter

keane2097

Joined Feb 19, 2009
21
Biasing at 2.5V puts the inputs in the middle of their common mode (linear input) range, so that any stray voltage common to the two inputs (usually 60Hz, or harmonics of same) are less likely to fall out of the allowable range.
So do I need to alter the design to based on the difference in linear voltage range between the two devices? Considering the fact that the inputs are going to be as low as 5-10 mV will this really be a problem?

As for the rest of the circuit, I have the same op-amps - the OPA2335.

For the voltage reference for now I just have a voltage divider between my +5V and ground and I'm tapping 2.5 volts from that. I'm not ready to connect up my A/D converter yet but that has an internal reference voltage which I think I can use instead of the voltage divider in the end...
 

Ron H

Joined Apr 14, 2005
7,063
So do I need to alter the design to based on the difference in linear voltage range between the two devices? Considering the fact that the inputs are going to be as low as 5-10 mV will this really be a problem?
As I (and others on the other forum) pointed out, your differential input signal is small, but the common mode signal could be much larger. The human body is a big antenna, and it picks up stray electromagnetic radiation, especially the mains frequency and harmonics if same. Your IA should easily handle that, so long as the input voltage stays within the input linear range. I think 2.5V is close enough to center for the INA118.

As for the rest of the circuit, I have the same op-amps - the OPA2335.

For the voltage reference for now I just have a voltage divider between my +5V and ground and I'm tapping 2.5 volts from that. I'm not ready to connect up my A/D converter yet but that has an internal reference voltage which I think I can use instead of the voltage divider in the end...
What A/D converter are you going to use?
 

Thread Starter

keane2097

Joined Feb 19, 2009
21
As I (and others on the other forum) pointed out, your differential input signal is small, but the common mode signal could be much larger. The human body is a big antenna, and it picks up stray electromagnetic radiation, especially the mains frequency and harmonics if same. Your IA should easily handle that, so long as the input voltage stays within the input linear range. I think 2.5V is close enough to center for the INA118.
Ok cool, that's good news...

What A/D converter are you going to use?
I'm using one built into the MSP430FG4618/F2013 Experimenter Board from Texas Instruments. The one I'm using is on the MSP430FG4618 and it's a 12-bit ADC.
 

Thread Starter

keane2097

Joined Feb 19, 2009
21
I've been having a look at the ECG Amplifier circuit from the INA118 datasheet, shown here:-



I'm going to try and build this as well as it's quite a bit more basic and it's based on the right IC. I'm confused a little about the cylinder encircling the two arm inputs though. This looks like insulation or something, but what does that mean with regard to the output of the right-most opamp?

Also, I'm assuming this circuit is dual supply - am I right in thinking this?
 

beenthere

Joined Apr 20, 2004
15,819
The cylinder is the schematic depiction of the shielding around coaxial cabling.

The circuit is not discussed, but should run with dual rails. +/- 3 to +/- 5 volts should be enough, and provide inherent safety.
 

Thread Starter

keane2097

Joined Feb 19, 2009
21
Could anyone suggest a suitable shielded twisted pair cable for this application?

i don't have one at the moment so I'm going to have to order one...
 

beenthere

Joined Apr 20, 2004
15,819
I have it on hand, and so would use two lengths of RG-174 coax. A pair of stereo patch cables might do okay, but the shielding factor is less.
 

Ron H

Joined Apr 14, 2005
7,063
I have it on hand, and so would use two lengths of RG-174 coax. A pair of stereo patch cables might do okay, but the shielding factor is less.
A shielded twisted pair will probably have better common-mode immunity than two coaxial cables.
 

beenthere

Joined Apr 20, 2004
15,819
The trade off is being able to maintain shielding right up to the electrodes and superior flexibility. Patch cables are going to be weenie about shielding, as I mentioned. One can always check for results with stuff on hand.

Plus your heart will go into arrhythmia looking at the cost of a spool of braid shielded twisted pair wire.
 

Ron H

Joined Apr 14, 2005
7,063
The trade off is being able to maintain shielding right up to the electrodes and superior flexibility. Patch cables are going to be weenie about shielding, as I mentioned. One can always check for results with stuff on hand.

Plus your heart will go into arrhythmia looking at the cost of a spool of braid shielded twisted pair wire.
I guess you could twist the coaxes together. The big advantage of twisted pairs is that they are very good at not converting a common mode EMI field (like mains) into a differential signal.
 

Thread Starter

keane2097

Joined Feb 19, 2009
21
Ok, so I have the entire circuit working now (thank god). I'm getting an ECG signal out of roughly 1V amplitude which is what I was looking for. All I need now is a quick method of isolating the patient from the power supply for safety.

Is there a quick, easy way to do this? Like with diodes/capacitors maybe?

I'm using a dual supply of +/- ~5v
 
Top