My thinking was to add the bandpass filter before the op amp, so that the op amp would not be processing unwanted frequencies, needlessly. However, if I add a bandpass filter on both phases the signal, whatever mismatch there is between the component values in the two halves of the filter will be amplified. Correct? If so, what are the typical solutions, or filter arrangements.
But you haven't understood the beauty of differential design.
Why are your input components connected to ground?
The first four resistors can be replaced with one (prolly around 68k, I couldn't be bothered doing the calculation).
The 1n caps can be replaced by a 500p cap.
My thinking was to add the bandpass filter before the op amp, so that the op amp would not be processing unwanted frequencies, needlessly. However, if I add a bandpass filter on both phases the signal, whatever mismatch there is between the component values in the two halves of the filter will be amplified. Correct? If so, what are the typical solutions, or filter arrangements.
You have it right. You can filter the differential signal before the difference amplifier, but if you filter the common mode signal, then mismatch in the filters will convert common mode noise to differential noise.
The general solution is to only implement EMI filtering before the difference amplifier, and then filter the signal after the difference amplifier.
THAT Corporation have some really amazing design resources with explanations and details, why are you not going with one of their designs?
"THAT Corporation Design Note 140"
Thanks, David. It is really great for you to join in. I don't know much about EMI filtering. Could this be as simple as a Ferrite Bead on the mic cable?
Leave out the snubber, by the way the impedance you have chosen for this is not suitable for an opamp, it puts far too much load on the opamp.
Through RPB2 of 10K the opamp output sees no capacitive load that makes it difficult for the opamp.
If you still want some protection against e.g. longer output cables, then add directly to pin 6 of the opamp a 47 Ohm series resistor.
Your output impedance on the left side of C4 is around 10K, so the high value of C4 is not needed there at all.
Also, because of the high output impedance you may not load the circuit lower than say 150K because otherwise your tilt points will shift,
A better solution is a buffer opamp that is DC coupled and leave C4 out at that point.
Then your filter is exactly as you want it, independent of the load.
If you apply the buffer opamp, again a series resistor directly to the opamp output against generation phenomena and maybe a coupling capacitor.
The Data Sheet does not clearly state that C1/2/3 are simply to filter noise, as would any other decoupling cap. I believe this is their function. Do I understand correctly?
Is there significant benefit to using a precision resistor array for the biasing resistors, R1/2, to warrant its approximately $8USD price?
- "Rake" symbol. This ground is true Earth. It is provided by the XLR Input Connector's case. It provides the exist for (only) C3. Do I understand correctly?
- "Inverted Triangle" symbol. This is the null point of the +/-15V supply. All grounds except C3 return to this point and exist viaPin3Pin1 of the XLR Connector. Do I understand correctly?
Benta did not expound on why he suggested that trim pots not be used to tweak the filter. I take it that there are two reasons...
- The setting is not critical.
- Trim pots are subject to thermal drift.
Do I understand correctly?
I would like to add David's EMI filter. See Reply #10/Figure 18.9, of the document that he provided.
- I believe it consists of the pairs C3/100pF + R7/100K(not 100R) and C4/100pF + R8/100K(not 100R). Do I have 100K, not "R," correctly?
- Which Ground does the EMI filter return to, Earth or Null? Earth is shown.
- Exactly where is it inserted into the circuit? After the biasing resistors RBIAS1/2?
QuoteIs there significant benefit to using a precision resistor array for the biasing resistors, R1/2, to warrant its approximately $8USD price?
Mismatch between the shunt resistors at the input and the source impedance limit common mode rejection, so the input resistors should be reasonably matched. 1% resistors are typically considered good enough, but 0.1% is not much more expensive.
Quote- "Inverted Triangle" symbol. This is the null point of the +/-15V supply. All grounds except C3 return to this point and exist viaPin3Pin1 of the XLR Connector. Do I understand correctly?
Correct, and there should be a single ground connection between the chassis and that point.
Do you need a filter cutoff closer than 5%?
If you still want some protection against e.g. longer output cables, then add directly to pin 6 of the opamp a 47 Ohm series resistor.
Kind regards,
Bram
Mismatch between the shunt resistors at the input and the source impedance limit common mode rejection, so the input resistors should be reasonably matched. 1% resistors are typically considered good enough, but 0.1% is not much more expensive.
I matched a pair from my stocks and they are nearly dead on.
Correct, and there should be a single ground connection between the chassis and that point.
I thought that the circuit ground was to be fully isolated from Earth ground. So, I am going to repeat back to you your instructions, for verification for certain.
At this location,
1) the PCB Board's +/-15Vsupply Null Point Return,
2) the Aluminum Case Project Box Chassis,
3) Pin#1 of the XLR connector
4) and the Metal Case of the XLR Connector
are all to be tied. Do I understand correctly?
QuoteCorrect, and there should be a single ground connection between the chassis and that point.
I thought that the circuit ground was to be fully isolated from Earth ground. So, I am going to repeat back to you your instructions, for verification for certain.
At this location,
1) the PCB Board's +/-15Vsupply Null Point Return,
2) the Aluminum Case Project Box Chassis,
3) Pin#1 of the XLR connector
4) and the Metal Case of the XLR Connector
are all to be tied. Do I understand correctly?
Ideally there is a single point ground where the chassis ground and circuit ground meet. In practice this becomes a problem when multiple external connectors are present, so some compromise has to be made. Having the chassis completely separate from the signal ground is not something I would recommend, but maybe some audio equipment does it? Maybe someone in the forum knows.
Just to recap where we are, these are the/some/all? open questions...
- If Earth Ground and Circuit Ground are ever to be tied (at the chassis.)
- Calculation of the resistor value of the Gain Pot.
- The purpose and value of the resistor that Bram suggests should be added to the output. Or, the use of an output buffer op amp and its circuit design.
I have sent PMs to Benta and Bram asking them to join back in, particularly regarding the above issues upon which they have previously commented.
I am learning a lot and having much good fun. Thank you to everyone!