4
\$\begingroup\$

I am using the SparkFun Heart Rate Monitor which contains the AD8232 front-end chip. I want to use only two electrodes (LA and RA) without the RLD driven electrode.

I am aware that this question has already been asked here. I have read through thoroughly but I am still unsure how to modify the SparkFun board properly and why I am experiencing artifacts.

My aim is to make a heart rate sensor close to that of the Polar H10 which uses a two electrode chest strap and achieves clean, low noise signals without a driven electrode.

I first tried the 3 electrode configuration with the original SparkFun circuitry from page 28 of the datasheet:

Sparkfun unmodified circuitry

Configuration 1:

Three electrode ECG recording (Einthoven triangle on below left collar bone (left arm), right collar bone (right arm) and right abdominal (right leg driven inputs)

3 Electrode ECG Recording (Einthoven triangle on Below left collar bone (Left Arm), Right Collar Bone (Right Arm) and Right Abdominal (Right Leg Driven Inputs)

Configuration 2:

Three electrode ECG but incorrect placement. Below left collar bone (RLD), right collar bone (RA) and right abdominal (LA)

3 Electrode ECG but incorrect placement. Below left collar bone (RLD), Right Collar Bone (RA) and Right Abdominal (LA)

Strangely, placing the electrodes in configuration 2 yielded a good result, whilst configuration 1 had a significantly larger S peak than R peak in the QRS complex, in addition to added noise in the 50Hz region. The T peak is also larger than the R peak for this recording.

Perhaps the largest voltage peak, R, is too large and is being attenuated by the Op-amp circuitry, thus making the S and T peaks larger in comparison?

 No driven electrode circuitry, see page 27 of datasheet

For the second test, I modified the SparkFun board by moving the 10M Ω resistors to the RLD pin (red annotation) and then pulled the AC pin HIGH (blue annotation). These changes were done as described in AC leads off detection from page 21 of the datasheet.

After digital notch filtering at 50,100,150 and 200Hz frequency I have achieved a clean ecg with two electrodes, comparable to configuration 2 using all 3 electrodes.

However, I have had to place the Left Arm (LA) on my Abdominal (the Right Arm (RA) is in the correct position - on my right upper chest)!

Even with an unmodified board on battery power, the best position for the electrodes are the same for RA and LA with RLD placed on the Left upper chest (i.e. inverse placement of the LA and RLD electrodes.)

My two questions are:

  1. Why for both the three and two electrode configurations have I achieved the best results with the Left Arm placed on the Left/Right Abdominal? Is this due to an unwanted attenuation of the largest voltage (the R peak) due to the op-amp circuitry.

  2. What do I need to add to the external circuitry or modify to the sparkfun board to achieve clean results using electrodes placed at the lower chest region (like polar H10)?

P.S. The frontend is connected to the same GND as the ADC and the microcontroller. Nothing is galvanically isolated at present.

\$\endgroup\$
7
  • \$\begingroup\$ You might like to skim my post on how to use the 3 leads: raspberrypi.stackexchange.com/questions/96482/… \$\endgroup\$
    – tlfong01
    Apr 10, 2022 at 14:15
  • \$\begingroup\$ "Nothing is galvanically isolated at present" is worrisome, but probably survivable because of the large resistors in series with the electrodes. You may have common mode or power supply noise problems. Can you power the front end with a battery instead to eliminate the latter? \$\endgroup\$ Apr 10, 2022 at 16:16
  • \$\begingroup\$ You're also using the op amp stage very differently in your two configs \$\endgroup\$ Apr 10, 2022 at 17:17
  • \$\begingroup\$ Could be nothing more than better electrode connection to the skin in one case \$\endgroup\$ Apr 10, 2022 at 17:17
  • \$\begingroup\$ I wonder if the op amp had slew rate issues at higher gain \$\endgroup\$ Apr 10, 2022 at 17:19

1 Answer 1

1
\$\begingroup\$

Nothing is galvanically isolated at present

That's where the problem is, most likely. There's enough common-mode interference that the front-end can't do its job. An ECG front-end needs high common-mode impedance to earth and by extension to mains.

In most electrophysiology measurements, isolation isn't a nicety or "a safety thing". It's a basic functional requirement, given the high input impedances. You can do without, but it requires much more effort.

If you've just connected it to PC's ground via USB, or to a lab power supply not designed for high common-mode isolation, then your body acts like a nice common-to-differential mode converter, and the junk you see results from that. The front-end seems to act quite nonlinearly in the first graph.

I get better output slapping two 9V batteries, a couple stages of TL07x as a discrete buffer and in-amp, trimming it manually for CMRR, and transmitting using a current source to drive a LED... on a solderless breadboard. A light-to-analog detector chip's output goes straight to the scope. The output has random noise as expected from TL07x, but it has very little interference, and no visible mains feedthrough. The AD8232 should be able to do way, way better than that!

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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