A cartridge has no ground reference so common mode noise will not induce a signal at the pre-amp except at frequencies well outside the audio band.
Matching gains of gain elements is only critical if sufficient feedback at bandwidth is not used to ensure gain is set by the resistive/capacitive elements and not the gain elements themselves.
|
@kijanki instrumentation amplifiers use laser trimmed resistors internally and/or external precision resistors as well as significant open loop bandwidth to achieve high CMRR and fixed gain levels. If you mean the differential pair / long tailed pair on the input, they also need somewhat matched resistors and in op-amps they are trimmed to provide higher CM range. <<$1.00 op-amps have 100db+ common mode rejection, but not in practical circuit which would be about 70db with 0.01% resistors. Low cost practical instrumentation op-amps with high CM range readily achieve 90db CMRR. Your amps may have a few precision resistors, and/or they are calibrated.
The main point of differential connections in audio is to eliminate the noise from having ground references which are different at the receiver and the transmitter whatever they may be and how they are caused. In a home audio system, that is going to be predominantly through noise induced through the AC, with the capacitance of the power supply, even EMI caps completing the circuit.
With a cartridge, you only have a ground connection on one end. Equal currents induced in the same direction on each wire (common mode) "induce" the same voltage and cancel each other out (it is a loop), at least at practical audio frequencies. Differential noise, will of course look just like a signal. The cartridge is already "isolated".
|
It's common mode. By definition the currents must be matched. If they are not matched, they are not common mode, they are differential and then you cannot tell them apart from the signal. The coil in the cartridge also completes the loop. It is by analogy the isolated end of the transformer. If it is a powered microphone that has an active ground connection, power and signal, then yes, the benefit is there. The benefit is mainly in better shielding for a non powered microphone.
That
is true, as long as currents flow in the loop only. If one of the
wires go to input while the other goes to GND (unbalanced input),
currents in both wires are not even anymore. Perhaps, that's why
floating (for audio frequencies) balanced output, like transformer is
not enough and has to be connected to balanced input (not single
ended). Microphones also benefit from balanced input.
The differential input of an op-amp does not have a gain of 1. It's a current source into a differential pair, that will share current based on the transistors being perfectly matched, and it will convert that current into a voltage by the load resistors ... who's matching also impacts the common mode rejection of this stage.
As for
Instrumentation amps, yes they have laser trimmed resistors, but front
(two amp) differential section has always gain of 1 for common mode
signal independently of resistor tolerance.
That's not how it works. I think you are confusing something. For a 40db CMRR, you must have gain matching of gain matching between the two channels to +/- 0.5% (1% total), and you must have perfect transistor matching (which they never are).
If you set gain of 100 for
this section you get automatically 40dB CMRR independent of resistors
tolerance. It happens because each of two amps is referenced to input
of another (instead of the GND).
Typically you use op-amps with laser trimmed resistors and/or you calibrate. You of course need tight temperature tolerance matching which is easier at the IC level, since with low power draw, temp will be very consistent between the temperature co-efficient.
That
is the only way I can see, for truly balanced amps, to function without
converting common mode noise to normal mode signal. Otherwise matching
resistors and keeping them matched to some sensible number (like
60dB=0.1%) is not practical, while adding RIAA frequency correction and
matching capacitors to 0.1% is next to impossible.
Which would probably be pointless as you pointed out, you are not going to get capacitors to that tolerance. |
I perceive this would only be of benefit at frequencies outside the audio range, as within the audio range, effectively the cartridge completes the loop and common mode noise would be rejected. Do you have examples of commercial products that use this technique and are they measureably quieter?
An input transformer magnetically isolates common mode noise, and is a good solution to building a quiet balanced phono stage.
|
Sorry kijanki, I thought we were talking about discrete level or discrete within an IC circuit, and that the 3 op-amp instrumentation amplifier you showed was a different part of the discussion. The circuit shown is missing the output buffer that would be on a practical circuit and those two output resistors have to be matched as they dominate CMRR in the calculation. I now get completely where you are coming from :-) If the noise current induced are not the same in both wires, then they are not common mode, they are differential mode, and a balanced connection only removes common mode noise. As for the balanced connection - it is not balanced anymore when you
connect it to unbalanced input. Noise currents induced in both wires
will be different and noise will go thru. In a typical connection of two components, you have multiple paths for current flow which is how CM noise generates a signal. With a catridge/phono stage, you have a single loop so induced common mode noise does not generate a signal. Where it gets dicey, is when you start adding new current paths like shields. Fix your DM noise, and now create CM noise. I think this is perhaps was jcarr was referring too? |
kijanki, When you short a oscilloscope and probe to ground, you are creating a loop antenna and creating a differential signal which is not to say that common mode noise and/or common mode voltages are not a problem in measurement. That is why Picoscope makes a differential input scope, and Tek and Lecroy sell high voltage differential probes for about $2,000/each (or more). While you are "floating" you are also a big capacitor to ground which creates a path for common mode signals. Also remember that most oscilloscopes probe grounds are also connected to chassis grounds. Floating oscilloscope inputs (battery powered or otherwise), can offer most of the benefits of differential inputs for many measurements (and tend to be much quieter). And are you assuming the noise was common mode? Are you sure it was not differential mode? I had TT long time ago and always had problems with low frequency hum (single ended phono stage). Current thru this ground wire causes tiny voltage drop, seen by the input as input voltage (input is referenced to BNC GND and not to end of the cable GND). But again, we are talking a cartridge and absent the shielding and other metal structure (and antenna effects isolated by the cartridge itself), it is a loop. Absent the shield/structure capacitance to ground, there is no other current path, whatever current flows in one wire must flow in the other, and hence whatever drop you have on one wire is cancelled by the other. Transformers are already used for MC coils, so we know they work. |
Maybe a point that needs to be made, you don't need an XLR to have a balanced input connection. You can use an RCA and still have a balanced input connection. BAT incorporates its “Flying RIAA Network” differential circuit here,
which the company claims has fewer parts yet because of ease of
adjustment at the factory and dual differential common mode noise
rejection, produces greater RIAA accuracy, with typical measured error
of less than 0.1dB. |
Atmasphere,
How much advantage did you find by adding shielding of the XLR over just using a differential input w.r.t. noise?
Interesting on the differential mode filter network. When you say it it seems "obvious", but certainly wasn't when the question was first posited.
Very interesting post, thank you.
|
We are all told that cables make a difference. Absent knowing what cable or that there was a change, my experience in controlled environments says that is the cable is competent, there may not be a difference. Diminishing returns much quicker for interconnects as well versus speaker cables. Noise rejection is important yes.
|
Except with a phono input, you don't gain 6db, as it is an application specific input. It amplifies the difference between the two leads. There is no "differential" output that has 2x the signal level (6 db).
|
As a counter to that cleeds, and not taking anything away, it is also true that because an RCA connection is used, does not mean that the input is not true differential.
With great gain, comes great responsibility. But seriously, w.r.t. phono, differential gives you the ability to increase gain while decreasing common mode noise, but you are still amplifying differential noise as well as noise internal to the amp. You are not getting 6db for free. You could always crank the gain 6db on a single ended connection, which again, for a cartridge, is already pseudo-differential due to the floating source.
|
luisma, This shielding "concept" is flawed and I am going to assume comes from a lack of understanding of what is happening. The person who came up with it is giving himself a bit too much credit. https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwio...When you run a signal through one "loop", and measure the induced voltage in another magnetically coupled loop, what you have done is created a transformer. - Leave the shield floating, and you just have a piece of "metal" in the middle. It will do nothing beyond a bit f magnetic shielding.
- Ground one side of the shield, and you now have an electrostatic shield. It will break the capacative coupling from one loop to the other in the transformer.
- Add an external wire to form a loop, or complete a loop as in the experiment, and you have now created a "shorted winding" in your transformer.
So what is the problem with this? - Transformers work both ways. If it is a shorted winding to the noise source, it is also a shorted winding to the signal in the wire. That’s not a good thing.
- Since the shield is floating, it provides no electrostatic shielding which is of course also important.
- On a "normal" shield, the inner conductor and the shield are actually both "windings" to external magnetic fields, and will have similar induced voltages which can be negated with differential inputs.
- Twisted pairs, and star-quad already reduce magnetic coupling, so adding electrostatic shielding addresses both noise sources.
|
kijanki, My last post was specific to the link luisma31 posted. I am not disputing anything you said, just not sure you were aware I was making a specific response to that. |
I was not saying that shorting the shield with a loop will pick up noise kijanki. It will act as a shorted turn and will direct an external magnetic field away from inside conductors to a degree. However, it will also act as a shorted turn to the current carried by the internal conductor, which could seriously impact it's performance.
|
In the experiment presented, and the "concept" for shielding presented, their is an external magnetic field noise source. As the shorted loop is passive, and it's current is only due to the external magnetic noise field, the field it would create cannot be larger than the field already created by the magnetic noise source. It cannot effect the other wire more than the noise field would have already. It would direct the field away from the wire.
|
Oh absolutely it will do more harm than good. It acts as a shorted loop to the signal wire, which means a parasitic load on the signal wire, bigger than anything else one could do. It also provides no electrostatic ground. It is a terrible idea .... some "cable" guy came up with it :-)
|
