INA321 for ECG - few questions

[20nyg07]

ECG circuit design - few questions



Well my first question is kinda broad, but, could I build this circuit exactly as is and receive an acceptable signal?

Also, Im not quite sure how the shield/shield driver work. Would the shield need to be connected to the output of the OPA336?

Also, I was initially planning on using RC LPF/HPF cascaded (Before I came across this schematic on the datasheet), but I assume there would be an advantage to sticking with the active filter?

 

[moffy]

You would have to assume it would work as it's on the datasheet of a reputable IC manufacturer. TI?. The shield is driven by the output of the OPA336 which is the common mode voltage. You would have to shield your electronics well, preferably inside a metal box. The supply to the instrument amplifier (if it's non isolated) as well as that to the OPA336 would have to be medically rated for safety. Perhaps some batteries, not on charge when connected to a person. With regards to the filter, try it and see.

- - - Updated - - -

You would have to assume it would work as it's on the datasheet of a reputable IC manufacturer. TI?. The shield is driven by the output of the OPA336 which is the common mode voltage. You would have to shield your electronics well, preferably inside a metal box. The supply to the instrument amplifier (if it's non isolated) as well as that to the OPA336 would have to be medically rated for safety. Perhaps some batteries, not on charge when connected to a person. With regards to the filter, try it and see.

 

[ZekeR]

but I assume there would be an advantage to sticking with the active filter?

I don't see any advantage to an active filter (as shown) other than avoiding loading effects... But perhaps I'm not savvy enough with signal processing to see it.

I'm curious about the two bottom op-amps. Vr is set to 2.5V(nom), which is pulled slightly up or down by the input common-mode. It is then buffered to drive the shield (kinda makes sense, although I don't see much difference to just grounding the shield since Vr is virtually constant) and then the difference between the shield voltage (=Vr) and Vr (=Vr) is amplified for the RLD. Wha? Normally people obtain RLD by letting Vr remain constant, then integrating the difference between Vr and the input common-mode, thus driving the common-mode to Vr. This circuit doesn't do that.

Additionally, when people have an active shield, the goal is to have a single shield for each signal cable and to drive the shield to the potential of its conductor (thus eliminating the LPF effect caused by the signal's source impedance (skin is high Z) and the shield's capacitance). Simply driving the shield to the common-mode doesn't fix this issue—although it does solve another issue, specifically that common-mode noise can get passed through if the two paths have different impedances (if the skin contact is different, etc). But this circuit doesn't even drive the shield to the common-mode, so whatever.

I don't know... something just smells fishy about this ECG schematic. It probably still works, but something tells me it's in spite of its engineer rather than because of him. Considering that the datasheet mostly describes applications that don't use a RLD or an active shield, I'm guessing the engineer never tried it. Thankfully, for a lot of applications which involve heart rate monitoring (like exercise machines), nobody really cares if it works or not.