Error corrector, unlike composite, keeps the input linked to the inverting amplifier, so it doesn't add extra layer to over processing.
Initially, it was invented for servo mechanics to reduce the error necessary in feedback amplifier.
Consider an opamp in inverting scheme. The positive input is grounded and the negative has an input resistor Ri, and a feedback, Rf. If we assume that the inputs are very high impedance, than the input current,Vin/Ri is equal to the current through the Rf which is now equal to Vout/Rf. Hence we convince the noobs that Vout/Vin=-Rf/Ri. In reality, the opamp needs some differential juice across its inputs for Vout, so the inverting input is not zero volts but a Verr which is inverse function of the opamp transfer.
To reduce the Verr, in composite, is to add a non inverting amplifier to the negative input, with EC, the necessary input juice is obtained by conveying the Verr to the positive input using an inverting amplifier.
In servo mechanics, the inverting amplifier is another integrating opamp as the frequency of operation is less than the pole of few hertz. If the opamp needs 100mv input juice for the required output, and we add an inverting conveying amp of G=9, then the Verr now is 10mv and the positive input is -90mv so the juice remains 100mv. By this we reduced the Verr, G+1 times lower.
If such circuit is to be used at higher frequencies that of the poles, the output impedance becomes negative, the 2 pole compensation makes it very difficult to stabilize but there is a masochistic example I show later the ZD50 or its copy, modulus86. John Yewen, published an idea of using not one conveying opamp but infinite number of opamps to reduce the Verr by billions of times. By adding 2 conveyors minimum, one can get 3rd order low pass transfer function, perfectly stable. The problem with this method is the precision of the open loop gain of opamps. As it varies simple to double for the most precision opamps, it needs to be adjusted manually which is not practical.
My way of conveying is much simpler. As the conveyer is working at very low voltage, it doesn't need low distortion amp, but low noise type. So, why not use a single transistor as inverting amplifier of 20-40db gain with high frequency pole and be happy with an extra zero or two in THD number. For high input impedance and low noise I chose jfet of high transconductance. It can be 2sk117/209 or doubled as 2sk170 or tripled as 2sk369. For stability, as the added pole is high frequency, adding a lead capacitor, 27pf here, is sufficient to get perfect square waves.
Initially, it was invented for servo mechanics to reduce the error necessary in feedback amplifier.
Consider an opamp in inverting scheme. The positive input is grounded and the negative has an input resistor Ri, and a feedback, Rf. If we assume that the inputs are very high impedance, than the input current,Vin/Ri is equal to the current through the Rf which is now equal to Vout/Rf. Hence we convince the noobs that Vout/Vin=-Rf/Ri. In reality, the opamp needs some differential juice across its inputs for Vout, so the inverting input is not zero volts but a Verr which is inverse function of the opamp transfer.
To reduce the Verr, in composite, is to add a non inverting amplifier to the negative input, with EC, the necessary input juice is obtained by conveying the Verr to the positive input using an inverting amplifier.
In servo mechanics, the inverting amplifier is another integrating opamp as the frequency of operation is less than the pole of few hertz. If the opamp needs 100mv input juice for the required output, and we add an inverting conveying amp of G=9, then the Verr now is 10mv and the positive input is -90mv so the juice remains 100mv. By this we reduced the Verr, G+1 times lower.
If such circuit is to be used at higher frequencies that of the poles, the output impedance becomes negative, the 2 pole compensation makes it very difficult to stabilize but there is a masochistic example I show later the ZD50 or its copy, modulus86. John Yewen, published an idea of using not one conveying opamp but infinite number of opamps to reduce the Verr by billions of times. By adding 2 conveyors minimum, one can get 3rd order low pass transfer function, perfectly stable. The problem with this method is the precision of the open loop gain of opamps. As it varies simple to double for the most precision opamps, it needs to be adjusted manually which is not practical.
My way of conveying is much simpler. As the conveyer is working at very low voltage, it doesn't need low distortion amp, but low noise type. So, why not use a single transistor as inverting amplifier of 20-40db gain with high frequency pole and be happy with an extra zero or two in THD number. For high input impedance and low noise I chose jfet of high transconductance. It can be 2sk117/209 or doubled as 2sk170 or tripled as 2sk369. For stability, as the added pole is high frequency, adding a lead capacitor, 27pf here, is sufficient to get perfect square waves.
Attachments
Last edited:
