6.283 said:
Wasn't thinking of that. So, if R1,2,3,4 are 56K the opamp can handle the load fine.
Check out the linked material. Maybe start with the last one. That should lift you to the next level 😉
I've done what you're suggesting, not with the power amp directly but by adding a 'diff amp' ahead of the power amp. Remember that source signal ground, destination signal ground and power supply ground are not all the same and it takes remarkably little signal to 'rattle' the grounds and add differential noise. If you''re concerned about wildly differing grounds tie the chassis together through small value resistors -- 10-100 ohms work well -- to keep them all on the same page. The diff amp trick on unbalanced inputs is used in some commercial products so you're not breaking any new 'ground'. I've also put the diff amp on the source side to subtract the destination noise from the source signal plus noise. It works well there too.
Bottom line is you cannot win against Ohm's law regardless of how big a ground wire you use but you can subtract the junk out and win that way.
G²
Thanks for those, they were very informative. I'm thinking maybe I should just get a balanced-input chip, since matching the resistors to within 0.005% is going to be pretty impossible, and without that the CMRR is too low to be useful.
stratus46 said:
Glad to hear that I'm not completely crazy with this idea 🙂
But this balanced input will only give a benefit if you have a balanced source and a balanced connection.
Resistor matching to 0.1% is so easy it's hardly worth omitting.
Comparing to 0.01% takes a lot more care, but is still possible with low resourced amateur's equipment.
0.005% (1part in 20,000) would be right at my limit of comparison accuracy using a 50,000 count bench DMM.
My cable runs aren't long enough to warrant a balanced source. I only need one side to be balanced to kill the ground loop, so I'd rather do the amp than buy a new DAC.
I found a balanced receiver opamp with integrated resistors that are laser etched to 0.005%. SSM2143.
You're mistaken about using diff amps unbalanced. They work very well that way.
Why bother with precision resistors or matching? Burr-Brown (T-I) INA134 and Analog Devices (SSM2141) have laser trimmed units, pin compatible by the way, ready to plug and play.
G²
Although 25K is quite something in front of a 26dB+ amp.
That's why I use the LM3886 as the diff amp directly and buffer the input signals with voltage followers. This way the resistors around the diff amp can be much smaller but the common mode impedance at the input stays high.
The only snag is that the 3886 needs a min gain of 10 to run stable. So you get away with 1K and 10K resistors around it (or 1K and 20K depending if the input opamps already add 6dB gain or not).
All this comes at the cost of some matching, though as pointed out. But I managed to build a diff amp with >100dB CMRR (below 100Hz). And the only tools were an average digital multi meter and three or four 9V blocks to power a wheatstone bridge that was used for the matching.
That's why I use the LM3886 as the diff amp directly and buffer the input signals with voltage followers. This way the resistors around the diff amp can be much smaller but the common mode impedance at the input stays high.
The only snag is that the 3886 needs a min gain of 10 to run stable. So you get away with 1K and 10K resistors around it (or 1K and 20K depending if the input opamps already add 6dB gain or not).
All this comes at the cost of some matching, though as pointed out. But I managed to build a diff amp with >100dB CMRR (below 100Hz). And the only tools were an average digital multi meter and three or four 9V blocks to power a wheatstone bridge that was used for the matching.
what benefits of balanced operation are retained when a diff amp is fed from an unbalanced cable?
Retained: Well, obviously you first have to establish a balanced transmission. So this way you can convert from unbalanced to balanced virtually just like with an audio transformer that is fed from an unbalanced source.
As a result the common mode voltage is nullified. That is the volatge that spoils the signal, audibly or not.
The lower the output impedance of the source and the higher the CM input impedance, the better the conversion works.
Or in other words: the closer is the performance compared to a real transformer, which is a purely differential device in itself.
What have I missed?
He starts with an unbalanced Source and an unbalanced connection.
What balanced impedance benefit is there by using a diff amp input stage set up to expect a balanced impedance connection?
Last edited:
Who "he", minifly ?
If you refer to that paper drawing: A diff amp is differential although one part is connected GND.
If you refer to that paper drawing: A diff amp is differential although one part is connected GND.
Because it kills the ground loop because there is no low-impedance loop for circulating currents.
Please tell me where my thinking is wrong:
1) The ground loop is caused by a low-impedance ground connected with a large area loop that has current induced into it by EMF causing large circulating currents.
2) The noise from the ground loop gets into the signal line because of the voltage drop due to the V=RI loss in the ground side of the signal line.
3) If I use a differential amp the impedance on the ground side is high enough that essentially no current flows, so there are no V=RI losses.
4) If I use a differential amp, then it only uses the difference between the two signals, so any EMF induced voltages from the area between the signal cable and the ground won't be passed into the amp.
a good fraction of ground loop current can come from line input RF filter X/Y caps too
Self's books go a little deeper but he still has useful stuff online
Ground Loops
Balanced Line Technology
Self's books go a little deeper but he still has useful stuff online
Ground Loops
Balanced Line Technology
The key is 'differential'. Assuming your source signal is operating correctly, you have a signal and shield/ground (earth) with no noise on the signal _relative_ to its output ground. However, there may be noise differences between the source ground and the destination ground. Many times - but not always - the signal ground will 'track' between the units so everything is quiet as expected. Sometimes - often in cars - this won't work out as wanted. The differential amp simply accepts the noise and subtracts it out of the signal and all is quiet as desired. No, its not a balanced line as you say BUT the noise can be eliminated easily. Try it.
G²
G²
- Status
- Not open for further replies.