ECG circuit help please

[20nyg07]

I am looking into making the above circuit and have a few questions to ask.

First, where can I buy a shielded cable? Preferably not online as I would like to have it before the weekend. Could I just buy say like a shielded RJ45 cable from a tech store and cut off the ends? Also, do the electrodes need to be grounded? In other words, should I get a shielded cable with 2 or 4 conductors? If I need 4 conductors, should I get 2 pair of twisted wires or does it matter?

Also, I was thinking about safety of this circuit, should I put some optocouplers after the inputs to prevent electric shock? If I should, would I need to amplify the signal? Or any other ideas for safety precautions?

 

[tgootee]

Can you please post a larger schematic?

 

[20nyg07]

Here we go.

 

[tgootee]

You will need at least three conductors. But you'll only need two of them to be inside the same shield, according to your schematic. And you could probably do it with two separate single shielded conductors, if you simply connected the shields together at the circuit end (to the "Shield Drive").

Shielded dual twisted pair might be good. Or maybe you would want the shielding to go all the way to each arm. So you might want to experiment with using a piece of the shielded cable for each of the three electrodes, and driving all of the shields in parallel (or at least the two for the arms), with your "Shield Drive" output from the lower middle OPA336. (The patient end of each shield would be unconnected.)

You'll probably want the kind of shielded cable with a "drain wire" built in, for the shield connection. You don't say where you live, but in the USA, places like Home Depot aYnd Radio Shack and probably many other chain stores would carry shielded twisted pair cable, with a drain wire, for telecom and computer networking and other uses.

Simple ECGs use the differential signal from the two arms, for the display, and use the leg as the ground.

ECG circuits depend VERY heavily on the common-mode rejection of the instrumentation amp, which usually needs a high gain. The signals from the heart that make it to the electrodes' wires are on the order of 1-5 mV in amplitude. But the body has common mode voltages swinging 1 or 2 Volts. And the body is also a wonderful antenna for the 60 or 50 Hz from the power system. Apparently you need a CMRR of at least 100 dB, to get <= 1% error in the output.

I didn't immediately grasp how the lower part of your circuit works. But it looks like it drives the shield at the mid-scale voltage of 2.5 V, and it looks like it might center everything around that voltage.

Here are the first couple of google results that caught my eye:

http://www.cisl.columbia.edu/kinget_group/student_projects/ECG Report/E6001 ECG final report.htm
http://www.scientificamerican.com/article.cfm?id=home-is-where-the-ecg-is&page=1

I'm sure that there are lots more good ones, if you can't figure something out.

Make sure that you use a small e.g. 0.1 uF cap from right at every power oin to ground, plus a 10 uF cap from every power pin to ground.

 

[20nyg07]

Thank you tgootee, The AD264 shown on the first site is much better than the INA (or seems to be), but would the INA321 with a CMRR of 94dB and an adjustable gain prove worthy? Considering I have it avaiable.

And just to double check, I should put the optocoupler after the branch with 1M resistor but before the 10k resistor (This following the INA321)

Lastly, your final sentence "0.1 uF cap from right at every power oin to ground, plus a 10 uF cap from every power pin to ground." is a little confusing to me. You say to put 2 caps from the power pin to ground.

 

[Audioguru]

The arms on a human are too far apart to use a single shielded cable. Use one shielded cable for the left arm and use a second shielded cable for the right arm so that the unshielded parts of the cables are very short.

Coax cable for TVs is too stiff. Use flexible shielded audio cables.

Your schematic does not show the shield connected to the opamp output but I think the shields should be connected to 0V.

The circuit must be isolated from the mains electricity or battery powered to avoid an electrocuted patient.

 

[20nyg07]

Aha, I didn't even clue into the two far apart lol. But thank you, that makes sense.

Connecting the shields to 0v makes sense, but this schematic was taken from the INA321 data sheet. I understand the need for the optoisolation (would the 4N25/4N26 do the job?). The safest way (or at least to me) would be to connect the electrodes directly to the optocouplers before the rest of the circuit, but from what I have read (including one of tgootee websites) they say you need to amplify the signal before sending through optocoupler so thats why I suggested after the INA321.

 

[tgootee]

<snipped>

Lastly, your final sentence "0.1 uF cap from right at every power oin to ground, plus a 10 uF cap from every power pin to ground." is a little confusing to me. You say to put 2 caps from the power pin to ground.

Yes. If you don't want to analyze the decoupling caps' requirements and configuration, then for low-power ICs like you are using it's usually safe-enough to connect a 0.1 uF X7R ceramic cap extremely close to each power pin (and then to a nearby ground), and also connect a 10 uF electrolytic cap from each power pin to ground. So each power pin should have a 0.1 uF and a 10 uF in parallel to ground. Or you can use more than 10 uF, for the electrolytics. Their value depends on how much current might be needed, and for what length of time. The 0.1 uF could be anything up to 1 uF or so, as long as they are physically small and are connected within a couple of mm of each IC power pin.

There are at least three functions of those caps. They act as a small point-of-load power supply, to supply any fast transient currents that the power pin demands. Without them, those currents would have to try to flow through the power supply and ground conductors and then voltages would be induced across the parasitic inductance and resistance of the power and ground rail conductors, themselves, creating spikes or noise on the rails. Also, the 0.1 uF (or other small-sized cap that might be used) bypasses (shorts) high frequencies to ground, helping to prevent instability due to a hidden positive feedback loop for high frequencies that works through the power supply rail, for most transistor-based amplifier circuits. I think that those are the main reasons the caps are needed. Without them, your circuit might not work at all, or might behave erratically.

But DON'T connect the grounded ends of those caps to the same ground-retrurn conductor that is used by any of the input signal ground-reference points! If you don't use a ground plane, then you would need to have separate ground-return conductors from each of the different types of ground-returns, that all go to a single-point "ground". Look up "star grounding". The reason for that is: If they SHARE a length of conductor, then any ground currents will induce voltages across the inductance and resistance of the ground conductor itself and similar voltages will then appear back at the non-psu ends of all of the ground-return conductors that shared some length of conductor with the offending ground currents. If some of those conductors are connecting amplifier input ground reference points, then those unwanted voltages (at the grounded end of an amp's input resistor, for example) would arithmetically sum with the input signals of the amplifiers. Not good.