I'm making an amplifier and seperate active crossover with an input selector. My first problem is whether or not to include the preamp with the amplifier or the crossover. Including it in the crossover would give me added noise rejection because the amplifier would not be as sensitive to noise in the input. The disadvantage here is that I would need to link the grounds of the crossover and amp. I plan to use no capacitors between these two. The amplifier has no caps and amplifies DC.
Now, my understanding is that a balanced line system uses three wires, ground, + and -. The reason regular wiring picks up noise is because the ground line (if it is earth ground) is 100% immune to noise but the signal line is not, therefore the noise causes a voltage difference across the wire and this is audible. Now if I want to make a balanced line system and just use UTP wire, say CAT5, can I use the transformer ground if it won't be connected to the chassis or earth ground? I'm counting on the fact that both the crossover's transformer ground and signal line will be affected by noise equally, therefore eliminating it. I'm planning not to run any ground between the amplifier and crossover. The only thing the amplifier will see is the voltage difference between the two wires. Is this "groundless" balanced line possible?
Now, my understanding is that a balanced line system uses three wires, ground, + and -. The reason regular wiring picks up noise is because the ground line (if it is earth ground) is 100% immune to noise but the signal line is not, therefore the noise causes a voltage difference across the wire and this is audible. Now if I want to make a balanced line system and just use UTP wire, say CAT5, can I use the transformer ground if it won't be connected to the chassis or earth ground? I'm counting on the fact that both the crossover's transformer ground and signal line will be affected by noise equally, therefore eliminating it. I'm planning not to run any ground between the amplifier and crossover. The only thing the amplifier will see is the voltage difference between the two wires. Is this "groundless" balanced line possible?
Balance is all about relative impedances. Think of the source and load as a Wheatstone bridge circuit. Balancing impedances will cancel out noise common to the + and - lines. Not easy to do, especially at all frequencies, since capacitances come into play at high frequencies. See white papers at:
Jensen Transformers
Jensen Transformers
Interesting...
So If my crossover/preamp has an output impedence of 330 ohms and it's - line is a direct tap the the transformer ground, the signal line will be more vulnerable to noise and this will unbalance the line, correct? To balance the line I would have to put a 330 ohm resistor between the - line and transformer ground?
So If my crossover/preamp has an output impedence of 330 ohms and it's - line is a direct tap the the transformer ground, the signal line will be more vulnerable to noise and this will unbalance the line, correct? To balance the line I would have to put a 330 ohm resistor between the - line and transformer ground?
Solid Snake said:
Effectively, yes. A lot of semi-pro audio gear uses that technique.
But if your line is going to drive an input transformer, I wouldn't bother. An input transformer will give you nearly as much common-mode rejection fed from an unbalanced source as a balanced source.
For example, Jensen's JT-11P-1 gives you about 100dB of common-mode rejection from an unbalanced source and 107dB from a balanced source.
se
Solid Snake said:
You're using a 2N2222?
You mean your source has an output impedance of 4.7k or do you have a literal 4.7k resistor in series with the base? What's the schematic you're working off of?
Common-mode means a signal (or more typically noise) which exists equally in magnitude and polarity on both lines, as opposed to differential (or normal) mode where it exists equally in magnitude but of opposite polarity on each line.
se
Solid Snake said:
Great. That's all that counts. I was just a bit surprised as I rarely see them used as much more than switches anymore.
Ah, ok. What exactly were you trying to protect with such a large value resistor?
One of the drawbacks of balancing the line with an equal value resistor on the ground line is that while it results in a balanced line, it also dramatically increases interchassis resistance and can result in much greater noise which kind of defeats its purpose.
se
Well it's used in a low current high voltage gain pre-amp with lots of NFB. They're good up to 300Mhz so it couldn't hurt to use them in audio.
I was trying to protect the laptop's audio circuitry.
When I said ground, I meant the secondary center tap of the transformer which is not connected to earth or the chassis. More of a 0v reference point I guess.
The trick of the impedance-balancing resistors will only work when you use a unbalanced output connected to a differential input or when you connect a floating differential output to a unbalanced input
To make it work, signal cable must be two wires plus ground/shielding
Shielding can be used to connect chassis grounds at both ends [taking into account that this ground may carry current so it should go directly to star grounds without making any loop with signal grounds in the circuit] or leave ground open at one end [or coupled at RF to chassis with a 100nF capacitor]
There are basically 3 kinds of noise to reject in signal lines [and also inside each circiut so you sould better put the crossover away from the high currents flowing inside the power amplifier] :
- Inductive coupling : Magnetic fields are caused by any current flowing anywhere and induce in any wire a voltage proportional to its derivative and its frequency, It's relatively easy to cancel out these voltages by using a twisted pair as a signal line so both wires get almost the same induction, shielding introduces little or no attenuation to magnetic fields so it's optional
- Capacitive coupling : Voltage variations between every two wires will casue AC current to flow between them [as if there were a capacitor], for signal lines shielding forces the current to flow in the shield preventing the signal to be disturbed an twisted-pairing forces the current to be almost the same on both signal wires so it's easy to cancel it out
- Ground loops : Any PSU transformer can be considered as a noise voltage source between mains and its ground output so if you are interconnecting the grounds of two PSUs then this ground wire will carry noise current and there will be a noise voltage drop on it. Things get worse if both sides are connected to different earths or mains at different potentials. To solve it signal has to be processed in differential mode between 2 wires [apart from optional shielding] at least in one end of the signal line or in both
A simple differential input with an op-amp, 4 resistors and 2 capacitors per channel toghether with balanced wiring [shielded twisted pair] connnecting the shield at the rignt points at both ends [leaving current flow without affecting signal circuits] and connecting the -Vout wire to signal ground at the output end will solve all three problems
PD: A very bad practice is to connect input or output connector ground to the same ground trace you are using in the signal circuit instead of connecting it to central star ground, as soon as you connect the input/output to anything there will be a noise voltage drop in your signal ground traces and most of this drop may actually be RF and may cause oscillations and lots of nasty feedback effects [the circuit would be seeing differtential ground potentiales at different points of the signal ground trace]
To make it work, signal cable must be two wires plus ground/shielding
Shielding can be used to connect chassis grounds at both ends [taking into account that this ground may carry current so it should go directly to star grounds without making any loop with signal grounds in the circuit] or leave ground open at one end [or coupled at RF to chassis with a 100nF capacitor]
There are basically 3 kinds of noise to reject in signal lines [and also inside each circiut so you sould better put the crossover away from the high currents flowing inside the power amplifier] :
- Inductive coupling : Magnetic fields are caused by any current flowing anywhere and induce in any wire a voltage proportional to its derivative and its frequency, It's relatively easy to cancel out these voltages by using a twisted pair as a signal line so both wires get almost the same induction, shielding introduces little or no attenuation to magnetic fields so it's optional
- Capacitive coupling : Voltage variations between every two wires will casue AC current to flow between them [as if there were a capacitor], for signal lines shielding forces the current to flow in the shield preventing the signal to be disturbed an twisted-pairing forces the current to be almost the same on both signal wires so it's easy to cancel it out
- Ground loops : Any PSU transformer can be considered as a noise voltage source between mains and its ground output so if you are interconnecting the grounds of two PSUs then this ground wire will carry noise current and there will be a noise voltage drop on it. Things get worse if both sides are connected to different earths or mains at different potentials. To solve it signal has to be processed in differential mode between 2 wires [apart from optional shielding] at least in one end of the signal line or in both
A simple differential input with an op-amp, 4 resistors and 2 capacitors per channel toghether with balanced wiring [shielded twisted pair] connnecting the shield at the rignt points at both ends [leaving current flow without affecting signal circuits] and connecting the -Vout wire to signal ground at the output end will solve all three problems
PD: A very bad practice is to connect input or output connector ground to the same ground trace you are using in the signal circuit instead of connecting it to central star ground, as soon as you connect the input/output to anything there will be a noise voltage drop in your signal ground traces and most of this drop may actually be RF and may cause oscillations and lots of nasty feedback effects [the circuit would be seeing differtential ground potentiales at different points of the signal ground trace]
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.