touching Circuit Reduce Noise

[asadi.siyavash]

Hi, I build Simple ECG circuit it works well. my problem is that when I disconnect cables from body output display 50Hz Noise with full scale peak to peak voltge 0-5volt. when I touch Metal part of circuit like USB output Noise reduce. so how I can solved my problem that when cable are disconnect output goes to 0volt?
any help would be greatly appreciated

 

[BradtheRad]

When left floating, sensitive input wires can pick up ambient magnetic noise (including mains hum). Your amplifier then applies high gain, causing your output to be a 50 Hz square wave, rail to rail.

The input might stabilize, if you pull it up or down, with a very high resistance. Maybe 10 or 20 Mohm. High enough so that it does not affect your sensor input.

 

[asadi.siyavash]

Yes this is what I did. I pulled it up with 10Mohm but Noise is visible in the output and also Signal ratio decreased dramatically.

 

[Audioguru]

You posted no schematic so we do not know if you used an instrumentation amplifier with balanced inputs, if you fed the common-mode interference signal from the patient's leg to cancel common-mode interference, if your circuit has a 50Hz notch filter and if shielded audio cables were used to the probes.
Oh yeah, is your circuit a mess of long tangled wires on a solderless breadboard?

 

[asadi.siyavash]

No, I designed PCB. my circuit is like this circuit
+ I used 50Hz Notch filter at the output.
my cables are shielded.

 

[Audioguru]

The circuit you copied from the datasheet of the AD8220 should have excellent performance. When it is not connected to a patient then the inputs are floating since they need the body to feed them their bias voltage.

 

[asadi.siyavash]

but I added bias voltage with Pull up inputs, but again I have noise at the output

 

[BradtheRad]

I think it's a remarkable accomplishment, to construct a homemade electro-cardiograph. You must have done a lot of things right, to make it sensitive enough to work properly.

To solve the output issue, it looks as though it would be by disabling the output when it is too great amplitude.

The instrumentation amplifier arrangement is no doubt very sensitive, and you might lose more than you gain, if you try to add additional components near the input side.

I suggest you sample the output somewhere at the right half of your schematic, and if the amplitude is greater than 4V, then disconnect (divert, attenuate) the signal further downstream.

 

[andre_luis]

I believe that the problem could be somehow related to the equipment power grounding and shielding of the case on what the board is mounted. Also, would be interesting if we could take a look on the PCB layout design.

 

[asadi.siyavash]

Hi thanks for replays. I want to solve my problem with analog design, it is not good for me to clear problem instead solve it.
Also I don't want solve it digitally I want to solve it with analog design.
I am agree with you Andre what is your solution?
One thing that is really weird for me is that there is no connection with body but when I touch metal part like ground,VCC or other part noise goes to zero.

 

[andre_luis]

I´m not aware about safety standards on ECG equipments, or even if this would suffice, but I presume that grounding the cable shield would reduce the noise.

 

[BradtheRad]

One thing that is really weird for me is that there is no connection with body but when I touch metal part like ground,VCC or other part noise goes to zero.

Isn't this what happens with most equipment? Especially audio circuitry. At least that's how I find it. I believe it's the effect of grounding, or maybe 'partial' grounding.

It seems to confirm Andre's suggestions about proper grounding (post #9).

There's a chance the noise will go away if you clip a wire between your circuit and earth ground.

In the US we have power cords which contain 3 wires. One is earth ground. It is not live. The custom is to connect enclosures to this wire.

Your region may have 2-wire mains AC. I believe these are both live high voltage. It is not safe to connect them to low voltage circuits.

However if you have 3 wire mains AC, then one might be earth ground. If you find this is the case, then you might connect this to your circuit, but only if you can determine that is the correct wire.

 

[asadi.siyavash]

Hi,
I separate grounds with ferrite bead I test my circuit again. My ECG Cable is 1 meter three wire 0.5 meter of shielded other 0.5 meter of cable separate each lead and this parts of cable aren't shielded. So when I found just when noise source like body is near to this part output goes to be noisy. in this time when I touched metal parts of circuit body noise reduce so cable noise reduce and output is not noisy.
So major problem is that cable is not fully shield. using pull down resistor doesn't solved problem. I can't solved my problem yet so I want use your idea. but I hope find better solution.
yes we have two wire AC Power and I can't do this.

 

[andre_luis]

It seems that you are dealing with a set of conditions that do not help much, the worst being the fact that has already designed the PCB board, and this would be an impediment to any substantial changes in the circuit.

Again, not sure if this violates some safety issue, but in this case, I believe you could consider doing some kind of grounding of the place where the patient is lying down, so that the stretcher could perform a kind of shield to the surrounding electromagnetic field.

 

[D.A.(Tony)Stewart]

C is an active common mode reference.
This fails when the wires are not twisted pairs or shielded & the common mode injection of stray 50Hz E-Field is unbalanced.

(A & B ) common mode = -C output to reduce common mode with active guard method.

Touching shield or CM ground applies large capacitance shunt to stray e-field and thus reduces noise.

If you want isolated floating inputs, then CM rejection must be balanced.

1) Check the gain of the C feedback loop, it may be too high.
2) Good shielding is essential on all leads. with either coax or twisted pair to CM ground.

 

[asadi.siyavash]

it is violates grounding of the place where the patient is lying down. but I can change board design, I want to design new PCB. so if you have any idea for changing schematic and PCB.
Really thanks andre_teprom .

- - - Updated - - -

C is an active common mode reference.
This fails when the wires are not twisted pairs or shielded & the common mode injection of stray 50Hz E-Field is unbalanced.

(A & B ) common mode = -C output to reduce common mode with active guard method.

Touching shield or CM ground applies large capacitance shunt to stray e-field and thus reduces noise.

If you want isolated floating inputs, then CM rejection must be balanced.

1) Check the gain of the C feedback loop, it may be too high.
2) Good shielding is essential on all leads. with either coax or twisted pair to CM ground.

Hi is it possible explain more about C. I can't change Cables but I can change my circuit. I think with using isolator like Common Mode Filters/Chokes decrease noise e.g ZJYS81R5-2PL-51T-G01. and also transfer data with optocoupler like TLP521. what is your idea about this way?

 

[andre_luis]

I can change board design, I want to design new PCB. so if you have any idea for changing schematic and PCB

You did not provide any information about layout design of the PCB.