eg. On a phono section, the L+R signal grounds are tied together, then when leaving the unit, L+R signal grounds are again tied toghether.
This arrangement is repeated in the preamp, and in the amp, the L+R signal inputs are once again tied together. ( However, the speaker outputs are not.)
Why are the signal grounds often tied together on entry and exit from a unit?
Why not star ground instead?
Thanks!
This arrangement is repeated in the preamp, and in the amp, the L+R signal inputs are once again tied together. ( However, the speaker outputs are not.)
Why are the signal grounds often tied together on entry and exit from a unit?
Why not star ground instead?
Thanks!
Another point with audio frequencies and return current is that the return paths vary, some will follow the path of least resistance some will follow the path of least inductance. I aslo think Delta I noise caused by track inductances should be re-searches as well as Delta I noise caused by track resistances.
http://www.x2y.com/filters/TechDay0...log_Designs_Demand_GoodPCBLayouts _JohnWu.pdf
http://www.x2y.com/filters/TechDay0...log_Designs_Demand_GoodPCBLayouts _JohnWu.pdf
I tried it two ways on my Phase Linear 400.
The grounding scheme before hand looked like this:
Here, Cg = cap ground -- the point betwee the caps, LCOg = left channel output ground (negative speaker terminal), RCIg = right channel input ground (right channel sheild), PCBg is the ground line on the PCB.
Original:
1) Cg <- LCOg
2) Cg <- RCOg <- PCBg + LCIg + RCIg
Apparently, this is not good practice, since both input sheilds and signal reference are connected to the speaker negative.
I changed it to version 1:
1) Cg <- LCOg
2) Cg <- RCOg
3) Cg <- PCBg <- (LCIg and RCIg connected independently to PCBg, no common sheild ground back to Cg)
and also version 2, as suggested by the link I posted:
1) Cg <- LCOg
2) Cg <- RCOg
3) Cg <- PCBg <- (LCIg and RCIg independently)
4) Cg <- LCIg+RCIg (common sheild ground)
In the original, the DC offsets were -2 mv / 6 mv L/R channel.
In version 1, the LC DC offset increased to 20 mv/ 6mv L/R channel.
In version 2, the CD offsets were again -2mv /6mv L/R channel.
There was a *slight* decrease in hum in version 2 over version 1. No test agains the original was possible, as the RCA jacks were replaced at this time.
Hum in version 2 is evident nearing full power, in version 1, it's evident at normal listening volume (though not detectable from listening position.).
I'm going with version 2.
But why should the LC DC offset increase to 20mv with the version 1 grounding scheme? Both sheild grounds show 0 ohms resistance, between LCIg/RCIg and Cg. There is no switching circuitry or digital parts on the PCB. Pure linear supply, no regulators, no ground planes. i did not see the relevance in the X2Y article -- sorry, I'm no EE, just a hobbyist.
The grounding scheme before hand looked like this:
Here, Cg = cap ground -- the point betwee the caps, LCOg = left channel output ground (negative speaker terminal), RCIg = right channel input ground (right channel sheild), PCBg is the ground line on the PCB.
Original:
1) Cg <- LCOg
2) Cg <- RCOg <- PCBg + LCIg + RCIg
Apparently, this is not good practice, since both input sheilds and signal reference are connected to the speaker negative.
I changed it to version 1:
1) Cg <- LCOg
2) Cg <- RCOg
3) Cg <- PCBg <- (LCIg and RCIg connected independently to PCBg, no common sheild ground back to Cg)
and also version 2, as suggested by the link I posted:
1) Cg <- LCOg
2) Cg <- RCOg
3) Cg <- PCBg <- (LCIg and RCIg independently)
4) Cg <- LCIg+RCIg (common sheild ground)
In the original, the DC offsets were -2 mv / 6 mv L/R channel.
In version 1, the LC DC offset increased to 20 mv/ 6mv L/R channel.
In version 2, the CD offsets were again -2mv /6mv L/R channel.
There was a *slight* decrease in hum in version 2 over version 1. No test agains the original was possible, as the RCA jacks were replaced at this time.
Hum in version 2 is evident nearing full power, in version 1, it's evident at normal listening volume (though not detectable from listening position.).
I'm going with version 2.
But why should the LC DC offset increase to 20mv with the version 1 grounding scheme? Both sheild grounds show 0 ohms resistance, between LCIg/RCIg and Cg. There is no switching circuitry or digital parts on the PCB. Pure linear supply, no regulators, no ground planes. i did not see the relevance in the X2Y article -- sorry, I'm no EE, just a hobbyist.
Last edited:
zero resistance between 2 points is only possible of the 2 points are the verry same ones.
otherwise You have an absolute super conductor that has no resistive, no capacitive, no idunctive properity.
otherwise You have an absolute super conductor that has no resistive, no capacitive, no idunctive properity.
What's the resolution of that scale?
What is the manufacturer's stated tolerance/accuracy for that scale?
Are you serious?
The meter measured ".000" on a 200 Ohm scale.
Looking at the schematics (HiFi Engine | Download Free User/ Service Manuals, Amplifier, Receiver, CD, Tape, Tuner, Video, phase linear service manual, last version), and at the solder side, there is a 56 ohm resistor between two different "grounds" on the left channel and a 2.7 ohm on the right channel.
In version 1 ( ie No sheild signal ground direct to power common)
the "signal ground" for the left channel is connected to the power common through the 56 ohm resistor. That adds to a 39K input impedance at the input differential pair.
In the same way, 2.7 ohms is added between the right channel signal ground and power common.
I suppose this might have an effect on the DC offset, but 22 mv is major.
The factory build connects the sheild ground directly to power common as well as some other components, like the protection circuit groung reference, and Q5.
In short, there are two "ground" traces from the PCB connected to the sheild ground at the input, then via speaker negative to power common.
I changed the path to power common from the PCB via speaker negative, to a direct connection, disconnected the shield ground from the speaker negative and ran a dedicated line from the shield grounds to power common.
I suspect that the larger DC offset may have occured when the sheild ground was not directly connected to power common, and went through the 56 ohm resistor. It is odd that the meter did not read 56 ohms between the power common and the sheild ground. The meter has no trouble reading smaller valued resistors.
Anyway, if I can find a way to post a schematic, I will. My concern is around the resistors R2.... they seem uneeded now that sheild common and the second ground bus on the PCB are directly and independenty connected to power common.
I just don't see the need for those resistors in this model. If they are supposed to split a bigger bus, then why different values on each channel? The schematic lists the values as different, it's not just a repair done poorly.
Thanks for the help!
The meter measured ".000" on a 200 Ohm scale.
Looking at the schematics (HiFi Engine | Download Free User/ Service Manuals, Amplifier, Receiver, CD, Tape, Tuner, Video, phase linear service manual, last version), and at the solder side, there is a 56 ohm resistor between two different "grounds" on the left channel and a 2.7 ohm on the right channel.
In version 1 ( ie No sheild signal ground direct to power common)
the "signal ground" for the left channel is connected to the power common through the 56 ohm resistor. That adds to a 39K input impedance at the input differential pair.
In the same way, 2.7 ohms is added between the right channel signal ground and power common.
I suppose this might have an effect on the DC offset, but 22 mv is major.
The factory build connects the sheild ground directly to power common as well as some other components, like the protection circuit groung reference, and Q5.
In short, there are two "ground" traces from the PCB connected to the sheild ground at the input, then via speaker negative to power common.
I changed the path to power common from the PCB via speaker negative, to a direct connection, disconnected the shield ground from the speaker negative and ran a dedicated line from the shield grounds to power common.
I suspect that the larger DC offset may have occured when the sheild ground was not directly connected to power common, and went through the 56 ohm resistor. It is odd that the meter did not read 56 ohms between the power common and the sheild ground. The meter has no trouble reading smaller valued resistors.
Anyway, if I can find a way to post a schematic, I will. My concern is around the resistors R2.... they seem uneeded now that sheild common and the second ground bus on the PCB are directly and independenty connected to power common.
I just don't see the need for those resistors in this model. If they are supposed to split a bigger bus, then why different values on each channel? The schematic lists the values as different, it's not just a repair done poorly.
Thanks for the help!
Are you serious?
The meter measured ".000" on a 200 Ohm scale.
Looking at the schematics (HiFi Engine | Download Free User/ Service Manuals, Amplifier, Receiver, CD, Tape, Tuner, Video, phase linear service manual, last version), and at the solder side, there is a 56 ohm resistor between two different "grounds" on the left channel and a 2.7 ohm on the right channel.
In version 1 ( ie No sheild signal ground direct to power common)
the "signal ground" for the left channel is connected to the power common through the 56 ohm resistor. That adds to a 39K input impedance at the input differential pair.
In the same way, 2.7 ohms is added between the right channel signal ground and power common.
The factory build connects the sheild ground directly to power common as well as some other components, like the protection circuit groung reference, and Q5.
In short, there are two "ground" traces from the PCB connected to the sheild ground at the input, then via speaker negative to power common.
I changed the path to power common from the PCB via speaker negative, to a direct connection, disconnected the shield ground from the speaker negative and ran a dedicated line from the shield grounds to power common.
I suspect that the larger DC offset may have occured when the sheild ground was not directly connected to power common, and went through the 56 ohm resistor. It is odd that the meter did not read 56 ohms between the power common and the sheild ground. The meter has no trouble reading smaller valued resistors.
Anyway, if I can find a way to post a schematic, I will. My concern is around the resistors R2.... they seem uneeded now that sheild common and the second ground bus on the PCB are directly and independenty connected to power common.
I just don't see the need for those resistors in this model. If they are supposed to split a bigger bus, then why different values on each channel? The schematic lists the values as different, it's not just a repair done poorly.
The meter measured ".000" on a 200 Ohm scale.
Looking at the schematics (HiFi Engine | Download Free User/ Service Manuals, Amplifier, Receiver, CD, Tape, Tuner, Video, phase linear service manual, last version), and at the solder side, there is a 56 ohm resistor between two different "grounds" on the left channel and a 2.7 ohm on the right channel.
In version 1 ( ie No sheild signal ground direct to power common)
the "signal ground" for the left channel is connected to the power common through the 56 ohm resistor. That adds to a 39K input impedance at the input differential pair.
In the same way, 2.7 ohms is added between the right channel signal ground and power common.
The factory build connects the sheild ground directly to power common as well as some other components, like the protection circuit groung reference, and Q5.
In short, there are two "ground" traces from the PCB connected to the sheild ground at the input, then via speaker negative to power common.
I changed the path to power common from the PCB via speaker negative, to a direct connection, disconnected the shield ground from the speaker negative and ran a dedicated line from the shield grounds to power common.
I suspect that the larger DC offset may have occured when the sheild ground was not directly connected to power common, and went through the 56 ohm resistor. It is odd that the meter did not read 56 ohms between the power common and the sheild ground. The meter has no trouble reading smaller valued resistors.
Anyway, if I can find a way to post a schematic, I will. My concern is around the resistors R2.... they seem uneeded now that sheild common and the second ground bus on the PCB are directly and independenty connected to power common.
I just don't see the need for those resistors in this model. If they are supposed to split a bigger bus, then why different values on each channel? The schematic lists the values as different, it's not just a repair done poorly.
Are you serious?
It looks from a partial answer to one that you have a resistance meter that reads to 199.999 ohms. That makes it a 5.5 digit DMM. It is likely that it is quoted as pretty accurate.
+-0.02% of full scale reading and +- 5digits, if still within calibration.
That gives your 0r000 reading a range for those hypothetical tolerances of 0r000 to 0r045
How does your meter compensate for test lead/probe resistances?
I can't answer for anyone else, but my two questions to you were serious.
It looks from a partial answer to one that you have a resistance meter that reads to 199.999 ohms. That makes it a 5.5 digit DMM. It is likely that it is quoted as pretty accurate.
+-0.02% of full scale reading and +- 5digits, if still within calibration.
That gives your 0r000 reading a range for those hypothetical tolerances of 0r000 to 0r045
How does your meter compensate for test lead/probe resistances?
AndrewT
You're suggesting that differences in the thousands of ohms are going to cause the differential pair to become unbalanced to the point of changing it's DC offset by 22 mv. I don't believe this is possible.
In the past, when the diff pair was unbalanced, I changed the input impedances so that a 39K impedance was changed to 60K. This resulted in a change of impedance by only 10mv.
The way this circuit is implemented, if there is no direct connection from RCA shield to power common, we *should* see the R channel shield ground at 2.7 ohms, and the left at 56 ohms.
What is possible that I've made a measurement error, and there actually are larger resistances: I seem to recall that the R channel DC offset may have moved by close to 1 mv when there was no direct shield connection to power common - the meter was switching between 6 and 7 mv. So if 2.7 Ohms created about a 1 mv increase in DC offset, then 56 ohms would create about a 22 mv increase and the problem is solved.
Jumpering R2 and removing the direct connection from RCA shield to power common will leave the DC offsets unaffected. That will create a ground bus with the shields at one end, and the power common at the other. It may be noisier.
Good construction practice means the shield ground is supposed to be connected to power common immediately upon entry to the chassis. But doesn't good practice also separate the input references from the output references on the PCB? So shouldn't R2 just be removed?
It seems that the grounding layout is trying to separate input and output references: inputs to the differential pairs and the grounded emitter are on pin 2. The others (protection circuit and collector of Q5 -- via C10) are on pin 5.
Maybe I just don't get it....The circuit is like this:
.........Q1/Q2 input
................|...................C1...........R2
in--^^^---+--------+----||----+--^^^---+---Pin 5 Gnd to RCA
.......R1..................|...............|..............|
............................|......R3.......|.............|x
............................+--^^^---+-----------+----Pin 2 Gnd to RCA
Where pin2 and 5 grounds are connected at the RCA shield. sorry for the dots
Why would any current flow across R2 in this configuration?
In the updated version, Pin 5 goes to power common, and the connection of pin 5 and 2 to RCA ground it deleted (marked with the x). Now, current will flow across R2, and unbalance the diff pair.
I think I see now how the DC offsets will be affected.
But it does not explain why R2 is different for each channel. Artifact?
Thanks for your help. I really appreciate it.
You're suggesting that differences in the thousands of ohms are going to cause the differential pair to become unbalanced to the point of changing it's DC offset by 22 mv. I don't believe this is possible.
In the past, when the diff pair was unbalanced, I changed the input impedances so that a 39K impedance was changed to 60K. This resulted in a change of impedance by only 10mv.
The way this circuit is implemented, if there is no direct connection from RCA shield to power common, we *should* see the R channel shield ground at 2.7 ohms, and the left at 56 ohms.
What is possible that I've made a measurement error, and there actually are larger resistances: I seem to recall that the R channel DC offset may have moved by close to 1 mv when there was no direct shield connection to power common - the meter was switching between 6 and 7 mv. So if 2.7 Ohms created about a 1 mv increase in DC offset, then 56 ohms would create about a 22 mv increase and the problem is solved.
Jumpering R2 and removing the direct connection from RCA shield to power common will leave the DC offsets unaffected. That will create a ground bus with the shields at one end, and the power common at the other. It may be noisier.
Good construction practice means the shield ground is supposed to be connected to power common immediately upon entry to the chassis. But doesn't good practice also separate the input references from the output references on the PCB? So shouldn't R2 just be removed?
It seems that the grounding layout is trying to separate input and output references: inputs to the differential pairs and the grounded emitter are on pin 2. The others (protection circuit and collector of Q5 -- via C10) are on pin 5.
Maybe I just don't get it....The circuit is like this:
.........Q1/Q2 input
................|...................C1...........R2
in--^^^---+--------+----||----+--^^^---+---Pin 5 Gnd to RCA
.......R1..................|...............|..............|
............................|......R3.......|.............|x
............................+--^^^---+-----------+----Pin 2 Gnd to RCA
Where pin2 and 5 grounds are connected at the RCA shield. sorry for the dots
Why would any current flow across R2 in this configuration?
In the updated version, Pin 5 goes to power common, and the connection of pin 5 and 2 to RCA ground it deleted (marked with the x). Now, current will flow across R2, and unbalance the diff pair.
I think I see now how the DC offsets will be affected.
But it does not explain why R2 is different for each channel. Artifact?
Thanks for your help. I really appreciate it.
Last edited:
Here is my suggested solution --
1) Cut the trace between pin 5 right and what should be pin 5 left channel. Connect pin 5 left and right to power common independently. (Now output references will be separated.)
2) Remove R2 in both channels -- untie input and output references.
3) Connect shield grounds directly and independently to power common. Leave both Pin 2s attached to their shield grounds.
3) Both speaker negatives remain directly connected to power common.
This separates the output references from the input references, and it separates the output and input reference from each other.
1) Cut the trace between pin 5 right and what should be pin 5 left channel. Connect pin 5 left and right to power common independently. (Now output references will be separated.)
2) Remove R2 in both channels -- untie input and output references.
3) Connect shield grounds directly and independently to power common. Leave both Pin 2s attached to their shield grounds.
3) Both speaker negatives remain directly connected to power common.
This separates the output references from the input references, and it separates the output and input reference from each other.
No, I am not.
I am asking you to check how you are measuring with your test instruments and where to look for data that will confirm whether your measurements can be relied on.
I have made no comment on what or why you are measuring.
BTW,
what is the stated accuracy of your 5.5digit resistance meter on the 200r scale?
Last edited:
Andrew,
I don't know the resolution. It's just a cheap 3 digit unit, the probes are chinese plastic bag probes. I think I paid $6 for the probes and $20 for the unit.
I was measuring to check if I understood the circuit. Either I don't understand it, or I measured incorrectly.
I expected to see 56 and 2.7 ohms when the RCA shield ground was not hooked up. I did not.... which makes me suspect that the grounding is not as it looks.
I can measure again --
Results:
Meter probes touching 0.6 ohms. (I assume my error was that I was on the wrong scale.)
Resistance between RCA shield ground and power common: 3.1 ohms (measured) 2.57 ohms (calculated as 2.7 and 56 ohms in parallel).
Therefore, the resistance between RCA shield and Power common is 2.5 ohms.
That's fine.
Thank you! I just learned something important!
I don't know the resolution. It's just a cheap 3 digit unit, the probes are chinese plastic bag probes. I think I paid $6 for the probes and $20 for the unit.
I was measuring to check if I understood the circuit. Either I don't understand it, or I measured incorrectly.
I expected to see 56 and 2.7 ohms when the RCA shield ground was not hooked up. I did not.... which makes me suspect that the grounding is not as it looks.
I can measure again --
Results:
Meter probes touching 0.6 ohms. (I assume my error was that I was on the wrong scale.)
Resistance between RCA shield ground and power common: 3.1 ohms (measured) 2.57 ohms (calculated as 2.7 and 56 ohms in parallel).
Therefore, the resistance between RCA shield and Power common is 2.5 ohms.
That's fine.
Thank you! I just learned something important!
Last edited:
- Status
- Not open for further replies.