4B15 Introduction to Bioengineering Electrodes and Transducers Lecture 3: Transient Response of Electrodes 3.1 Transient Specification In simple analogue interfaces and amplifiers it is practically impossible to obtain the linear phase characteristic which is required to prevent distortion of a signal. In an attempt to cater for this, early standards for electrocardiographs gave an indication of the phase distortion which was acceptable. They did this by stipulating that the phase shift introduced into the signal by the recording system should be no more than that introduced by a single-pole, high-pass filter having a cut-off frequency of 0.05Hz. This left some ambiguity surrounding the effect this would have on an ECG signal, which is generally viewed in the time domain for diagnostic purposes. In an effort to quantify the distortion in the time domain more recent standards have indicated this in terms of the transient response to a narrow pulse as shown in Fig. 1. The distortion is specified in terms of the maximum undershoot from the baseline at the end of the pulse which can be tolerated and the maximum recovery slope from this undershoot allowable. The specification is given for an input rectangular pulse of amplitude 3mV and duration 100ms. The maximum undershoot from the baseline at the end of the pulse is limited to 100μV, while the maximum recovery slope is limited to 300μVs-1. These values apply to the overall distortion in the ECG introduced at any point in the recording system and therefore must be taken as being referred to the input so that they are independent of variable amplifier gain. Fig. 1 The Transient Response Specification for ECG Recorders 1 3.2 Transient Analysis The same model can be used as before for the electrode amplifier interface as shown in Fig. 2, with the sinusoidal source replaced with a pulse generator intended to provide the pulse shown in Fig. 1. RP1 iin RS1 _Π_ CP1 RP2 VS Rin Vin Amplifier RS2 CP2 Fig. 2. A Model of the Front End of the Recoding System Again, the input impedance of the amplifier is taken as purely resistive and the polarisation potential is neglected. Previously the steady-state response of this set-up was described in the frequency domain as: π (π + π π ) π½ππ π =πΆ π π½πΊ (π + π π ) π· In order to evaluate the transient response the term ‘jω’ is replaced by the Laplace operator ‘s’ so that: π (π + π) π½ππ (π) =πΆ π π½πΊ (π) (π + π) where: π= π πͺπ· πΉπ· π= π π = πΆπͺπ· πΉπ· πΆ πΆ= πΉππ πΉππ + π(πΉπΊ + πΉπ· ) so that: π (π + π) π½ππ (π) π(π + π) (π + π) π =πΆ =πΆ = π π½πΊ (π) π(π + π) (π + π) (π + π) π 2 If the input pulse has an amplitude Vm and a duration T as shown in Fig. 2, Vs(t) Vm 0 T t Fig. 2 Input Test Pulse it can be conveniently defined as a sum of two step functions so that: π½π (π) = π½π π(π) − π½π π(π − π») This has the Laplace transform: π½π (π) = π½π π½π −ππ» π½π − π = (π − π−ππ» ) π π π The input voltage to the amplifier is then given in Laplace terms as the product of the transform of the input pulse and the transfer-function of the electrodeamplifier interface so that: π½ππ (π) = π½π (π) π½ππ (π) π½π (π) This gives: π½ππ (π) = (π + π) π½π (π − π−ππ» ) (π + π) π The proper polynomial fraction in the Laplace operator ‘s’ can be simplified by partial fraction expansion: (π + π) π¨ π© = + π(π + π) π (π + π) (π + π) π¨(π + π) + π©π = π(π + π) π(π + π) (π + π) (π¨ + π©)π + π¨π = π(π + π) π(π + π) 3 Comparing coefficients gives: π¨+π©=π π¨= ; π πΆπ = =πΆ π π π¨π = π π©=π−π¨=π−πΆ so that: (π + π) πΆ (π − πΆ) = + π(π + π) π (π + π) The voltage at the input to the amplifier is the given in Laplace form as: πΆ (π − πΆ) π½ππ (π) = π½π (π − π−ππ» ) [ + ] π (π + π) This expands to: π½ππ (π) = π½π (π − πΆ) (π − πΆ) −ππ» πΆ πΆ (π ) − π½π (π−ππ» ) + π½π − π½π (π + π) (π + π) π π Taking the inverse Laplace transform gives: π½ππ (π) = π½π πΆπ(π) − π½π πΆπ(π − π») + π½π (π − πΆ)(π−ππ )π(π) −π½π (π − πΆ)(π−π(π−π») )π(π − π») which simplifies to: π½ππ (π) = π½π {[πΆ + (π − πΆ)(π−ππ )]π(π) − [πΆ + (π − πΆ)(π−π(π−π») )]π(π − π»)} When this expression is examined it can be seen to agree with the response shown in Fig. 1 to the input rectangular pulse whereby the pulse when present experiences some ‘droop’ in its transfer to the amplifier input, then some undershoot of the baseline at the end of the pulse, finally followed by an exponential recovery to the baseline. When t=0 we have: π½ππ (π = π) = π½π {[πΆ + (π − πΆ)(ππ )]π − [πΆ + (π − πΆ)(πππ» )]π} = π½π There is no initial attenuation of the pulse since we took R S << Rin previously, which essentially ignores the small amount of attenuation caused by RS. 4 When 0 ≤ t < T we have: π½ππ (π) = π½π [πΆ + (π − πΆ)(π−ππ )] This starts at a value of Vm when t = 0 and decays exponentially towards a value of αVm. The exponential decay is governed by the pole time constant αCPRP. In practice, the voltage never reaches its final value as the time constant associated with the electrode components is much larger than the width of the pulse T of 100ms. When t = T at the end of the pulse: π½ππ (π = π») = π½π {[πΆ + (π − πΆ)(π−ππ» )] − [πΆ + (π − πΆ)(ππ )]π} π½ππ (π = π») = π½π {[πΆ + (π − πΆ)(π−ππ» )] − π} π½ππ (π = π») = π½π (π − πΆ)(π−ππ» ) − π½π + πΆπ½π π½ππ (π = π») = π½π (π − πΆ)(π−ππ» ) − π½π (π − πΆ) π½ππ (π = π») = −[(π − πΆ)(π − π−ππ» )]π½π This describes the value of the pulse with the level of droop prevailing at the end of the pulse at t = T with a subtraction of Vm volts as the pulse terminates. This gives the value of the undershoot from the baseline at this time. Finally, when t > T the voltage at the amplifier input is described as: π½ππ (π) = π½π [(π − πΆ)(π−ππ ) − (π − πΆ)(π−π(π−π») )] Manipulating gives: π½ππ (π) = (π − πΆ)π½π [π−ππ − π−π(π−π») ] π½ππ (π) = [(π − πΆ)(π − πππ» )]π½π π−ππ π½ππ (π) = [(π − πΆ)(π − πππ» )]π½π π−ππ π−ππ» πππ» π½ππ (π) = [(π − πΆ)(π−ππ» − π)]π½π π−ππ πππ» π½ππ (π) = −[(π − πΆ)(π − π−ππ» )]π½π π−π(π−π») This describes the exponential recovery of the undershoot from the baseline which is again governed by the pole time constant. 5 3.3 Maximum Undershoot Limitation The value of the undershoot from the baseline at the end of the pulse is given above as: π½ππ (π = π») = −[(π − πΆ)(π − π−ππ» )]π½π This can be expressed as a fraction of the pulse amplitude simply as: π½ππ (π = π») = −(π − πΆ)(π − π−ππ» ) π½π The specification shown in Fig. 1 gives the undershoot limit as 100μV for an input pulse amplitude of 3mV. This corresponds to a fractional value of 0.033 or 3.3%. In order to fulfil this requirement: (π − πΆ)(π − π−ππ» ) < π. πππ However, p = z / α so that: ππ» (π − πΆ) (π − π− πΆ ) < π. πππ A power series expansion for the exponential is given as: ππ ππ ππ π = π+π+ + + +β― π! π! π! π Using only first order terms the above inequality approximates to: (π − πΆ) ( ππ» ) < π. πππ πΆ (π − πΆ) ππ» < π. πππ πΆ (π − πΆ) π» < π. πππ πΆ πͺπ· πΉπ· Typical values for the electrode components for modern disposable, adhesive electrodes have been given previously as RS = 50β¦, RP = 200kβ¦ and CP = 0.5µF. With T = 100ms = 10-5 s we have: (π − πΆ) π. π < π. πππ πΆ π. π × ππ−π × π × πππ 6 (π − πΆ) < π. πππ πΆ π − πΆ < π. ππππΆ πΆ> π = π. πππ π. πππ Then: πΆ= πΉππ > π. ππ πΉππ + π(πΉπΊ + πΉπ· ) which requires: πΉππ > π. πππΉππ + π. ππ(πΉπΊ + πΉπ· ) π. πππΉππ > π. ππ(πΉπΊ + πΉπ· ) πΉππ > ππ. π(πΉπΊ + πΉπ· ) With the values of RS = 50β¦, RP = 200kβ¦ as before this means: πΉππ > ππ. ππ π΄β¦ Many ECG amplifiers use an input impedance of 10MΩ which is recommended in the AHA standards. For the values of equivalent electrical components given, this is inadequate to meet the maximum undershoot limitation. A value nearer to 20MΩ would be needed to ensure that the requirement is met. Electrodes with higher values of components will need even higher amplifier input impedance. 3.4 Maximum Recovery Slope Limitation The profile of the exponential recovery of the voltage after undershoot at the end of the pulse at the amplifier input is described by: π½ππ (π) = −[(π − πΆ)(π − π−ππ» )]π½π π−π(π−π») The slope of this profile is found as: π π½ππ (π) = [(π − πΆ)(π − π−ππ» )]π½π ππ−π(π−π») π π 7 When t = T this has a value: π π½ππ (π) = [(π − πΆ)(π − π−ππ» )]ππ½π | π π π = π» If the exponential term above is approximated by the first order terms of an expansion as before the expression becomes: π π½ππ (π) = (π − πΆ)ππ π»π½π | π π π = π» With p = z / α = 1 / αCPRP this becomes: (π − πΆ) π π½ππ (π) = π π»π½ | π π π = π» πΆ (πͺπ· πΉπ· )π π The pulse defined in Fig. 1 has the properties Vm = 3mV and T = 100ms and with the values of the electrode components as before, RP = 200kβ¦ and CP = 0.5µF: (π − πΆ) π π½ππ (π) = π × π. π × π × ππ−π | −π π π π π π = π» πΆ (π. π × ππ × π × ππ ) which gives: (π − πΆ) π × ππ−π (π − πΆ) π π½ππ (π) = = × π × ππ−π | π π π = π» πΆπ ππ−π πΆπ The limit of the recovery slope as specified in Fig. 1 is given as 300μVs-1. This requires: (π − πΆ) × π × ππ−π < π × ππ−π π πΆ so that: (π − πΆ) < ππ−π πΆπ Inverting gives: πΆπ > πππ (π − πΆ) This requires a value of α which is very close to unity. In this case α → 1 so that α2 ≈ α. Then the requirement can be approximated as: 8 πΆ > πππ (π − πΆ) πΆ > (π − πΆ) × πππ πΆ(π + π) > πππ πΆ> πππ > π. ππ πππ This means: πΆ= πΉππ > π. ππ πΉππ + π(πΉπΊ + πΉπ· ) πΉππ > π. πππΉππ + π. ππ(πΉπΊ + πΉπ· ) π. πππΉππ > π. ππ(πΉπΊ + πΉπ· ) πΉππ > πππ(πΉπΊ + πΉπ· ) With the values of RS = 50β¦, RP = 200kβ¦ as before this means: πΉππ > ππ. π π΄π This places significant demands on the design of the amplifier input stage. 9