For me, hum and noise are the most important aspects. I may not be able to hear THD 0.1%, but I definitely can hear hum if there is any.
Below is a simple way to implement balanced input. When the resistor ratio matches between positive and negative side, it gets optimal common mode noise rejection.
I am not going to convert my amp to balanced interface, but let's see what if we use the same principle to the existing RCA unbalanced interface.
Let's replace the opamp with a power amplifier. Like this, you can see it does reject common mode ground noise.
1. Ground noise is canceled when R1/R2 = R4/R3.
2. Don't bridge the signal ground to the chassis.
3. R5 is necessary, R5 needs to be << R2 so that the negative feedback ratio won't change much when the signal ground is floating in air.
4. Be careful when adding volume pot directly to the input of the amp, it is equivalent to an out-proportional R3. The noise cancellation condition no longer holds true. The work around is put a unit buffer after the volume pot.
5. As the input impedance of the signal ground is very low, 10 Ohm in this case. I would not call this a balanced interface. It is a still unbalanced interface. However, it has the capable to reject some of the ground noise. It will be an improvement to the existing RCA interface.
PS: C2 should be equal to C1 to get optimal results. Thus, the improved version is C1 = C2 = 22u.
PPS: In some situation, it might not be easy to put extra R3 or keep R3/R4 ratio constant. For example, a volume pot. I post a compromised version at #11. Although it cannot cancel out ground noise completely, the ground noise won't get amplified, either. Thus, only the differential signal is got amplified.
Below is a simple way to implement balanced input. When the resistor ratio matches between positive and negative side, it gets optimal common mode noise rejection.
I am not going to convert my amp to balanced interface, but let's see what if we use the same principle to the existing RCA unbalanced interface.
Let's replace the opamp with a power amplifier. Like this, you can see it does reject common mode ground noise.
1. Ground noise is canceled when R1/R2 = R4/R3.
2. Don't bridge the signal ground to the chassis.
3. R5 is necessary, R5 needs to be << R2 so that the negative feedback ratio won't change much when the signal ground is floating in air.
4. Be careful when adding volume pot directly to the input of the amp, it is equivalent to an out-proportional R3. The noise cancellation condition no longer holds true. The work around is put a unit buffer after the volume pot.
5. As the input impedance of the signal ground is very low, 10 Ohm in this case. I would not call this a balanced interface. It is a still unbalanced interface. However, it has the capable to reject some of the ground noise. It will be an improvement to the existing RCA interface.
PS: C2 should be equal to C1 to get optimal results. Thus, the improved version is C1 = C2 = 22u.
PPS: In some situation, it might not be easy to put extra R3 or keep R3/R4 ratio constant. For example, a volume pot. I post a compromised version at #11. Although it cannot cancel out ground noise completely, the ground noise won't get amplified, either. Thus, only the differential signal is got amplified.
Last edited:
Hi, below 2 papers are mighty useful, if not read already.
https://www.google.com/url?sa=t&sou...4QFnoECEQQAQ&usg=AOvVaw0ChJ872ttBnVTfKnHDc5Wv
https://www.google.com/url?sa=t&sou...MQFnoECCMQAQ&usg=AOvVaw2XLPl2WNNIishj2mUekIBr
https://www.google.com/url?sa=t&sou...4QFnoECEQQAQ&usg=AOvVaw0ChJ872ttBnVTfKnHDc5Wv
https://www.google.com/url?sa=t&sou...MQFnoECCMQAQ&usg=AOvVaw2XLPl2WNNIishj2mUekIBr
"Why is C2 unequal to C1?" This type of circuit is not really balanced. The + input is 21k impedance, and the - input is 1k , if we don't look at that 10k resistor to ground. Why ? Engineering 101, the op amp - pin is virtual ground because of feedback. Only an instrumentation amp made from 3 opamps is really balanced. However, this simple approach does afford ok common mode rejection if driven from a very low source impedance.
The value required for a coupling capacitors depends on impedance. There is a 20k ohm difference between the + and - input impedances here. Again the 1k resistor at the - input sees ground potential (virtual ground), and the 1k resistor at the + input sees 20k. We do tend tend to make capacitors equal when we use this kind of circuit only to make it look better on a schematic.
Right, it isn’t.
My actual object is to use minimal efforts to mod existing amp to improve hum rejection, with existing RCA interface.
Next let's see what if connect R4 to the signal ground instead of local ground.
Below, the low frequency component is the ground noise. The high frequency component is the actual signal. You can see the actual signal is only one gotten amplified. The amplitude of ground noise stay the same. By hooking R4 to signal ground, although it cannot cancel out all the ground noise, it won't make it worse, either. Also it allows you choose C2, R3, R4 freely. It is more practical.
From this, here are some takeaways I came up.
1. Don't short input signal ground to your local ground. Tie 2 grounds together with a 10 Ohm resistor instead, that allow the signal ground float a little bit.
2. Except the speaker negative terminal, everything else should be tied to the signal ground.
PS: If you use 3-prong plug, the earth wire should always be mounted to chassis directly. Someone would like to add 2 diodes in reversed parallel to break the ground loop. I don't agree this approach. The peak current on the earth wire should be able to trip 15A breaker (in US) reliably. I don't think you could do that with diodes reliably.
Below, the low frequency component is the ground noise. The high frequency component is the actual signal. You can see the actual signal is only one gotten amplified. The amplitude of ground noise stay the same. By hooking R4 to signal ground, although it cannot cancel out all the ground noise, it won't make it worse, either. Also it allows you choose C2, R3, R4 freely. It is more practical.
From this, here are some takeaways I came up.
1. Don't short input signal ground to your local ground. Tie 2 grounds together with a 10 Ohm resistor instead, that allow the signal ground float a little bit.
2. Except the speaker negative terminal, everything else should be tied to the signal ground.
PS: If you use 3-prong plug, the earth wire should always be mounted to chassis directly. Someone would like to add 2 diodes in reversed parallel to break the ground loop. I don't agree this approach. The peak current on the earth wire should be able to trip 15A breaker (in US) reliably. I don't think you could do that with diodes reliably.
Last edited:
One of the best applications for using common-mode rejection I found was in an automotive audio system I put together. This was a LONG time back so I had an in-dash cassette player with an external amplifier. Local grounds for both. Connecting the two resulted in a very large "whine" from the alternator. I cleaned that up by using a pair of opamps wired so the cassette player's ground was connected to the "-" input of the differential amplifier. It worked quite well.
I definitely get the "modify what's there" approach. In this case I actually didn't have to modify anything, just add a pair of common mode rejection circuits.
I definitely get the "modify what's there" approach. In this case I actually didn't have to modify anything, just add a pair of common mode rejection circuits.
Balanced only works for noise/hum rejection if the input impedances looking into each side are equal.
In the very first schematic, where the resistor ratios are equal and the gain ratios are equal, the input impedances are not equal!
So if your sources have some output impedances, that destroys the impedance balance and the common mode rejection.
In this case you would need the Superbal circuit that has equal input impedances.
Jan
In the very first schematic, where the resistor ratios are equal and the gain ratios are equal, the input impedances are not equal!
So if your sources have some output impedances, that destroys the impedance balance and the common mode rejection.
In this case you would need the Superbal circuit that has equal input impedances.
Jan
My mod is not to convert it into balanced interface. It is to improve the existing unbalanced configuration.
Also, as I mentioned in the OP, as long as there is no volume pot in the middle, it works fine. If you really want match the impedance, just put a (or a pair of) unit buffer at front of the power amp. The mod would become too much for me.
You improve the existing unbalanced config for hum and noise by making it more balanced. Your circuit doesn't do that.
But if it sounds fine to you, more power to you!
Jan
Actually this idea isn’t new, it was implemented in some amps decades ago. You often see there are 2 grounds in the design, one signal ground, one power ground only for the speaker. They are tied together with a small resistor, e.g. 10 Ohm. One example is the Quad 405.
Right now, everything makes sense to me.
The idea you describe has nothing to do with your initial post and has no effect on your problem.
Discussion for the sake of discussion. Good luck.
Jan
Discussion for the sake of discussion. Good luck.
Jan
I would revert to 2 papers linked in post #3. It is about matching impedances, same as Jan said. Signals can be balanced but it's not fundamental. It is fundamental that receiver is differential and sees exactly same impedances from both nodes.
If impedances are matched well (requiring matched resistors) then it is best to hard wire all possible grounds to chassis nearby input signal sockets.
If impedances are matched well (requiring matched resistors) then it is best to hard wire all possible grounds to chassis nearby input signal sockets.
Quite right! Signals need not be balanced. The only requirement is that the induced hum and noise are the same amplitude so that the receiver cancels them. And those induced noise and hum are the same in the two leads if the impedances are balanced. QED.
If the signals have to be balanced it is usually called symmetric. And if we say symmetric, we normally mean symmetric with respect to ground, and you re-introduce ground as part of the signal (reference). Not good.
So (impedance) balanced good, (signal) symmetric no good. ;-)
Jan
If the signals have to be balanced it is usually called symmetric. And if we say symmetric, we normally mean symmetric with respect to ground, and you re-introduce ground as part of the signal (reference). Not good.
So (impedance) balanced good, (signal) symmetric no good. ;-)
Jan
Last edited: