It might help to think of the buss set-up as a single thread where the noisiest/ heaviest ac current connection is at one end and the lowest is at the other. Usually that means from rectifier through the filter chain sequence , to signal output device then driver / input device to input connector.
This will have influence over how you do the layout so that the runs are short and not crossing back and forth across the space.
The connection to the chassis should only be at one point. Some take the power supply end, some make it at the input , like right next to the RCA input jack. Some make the connection through a pair of parallel diodes in reverse polarity to each other.
You mentioned somewhere that grounds of input circuit to pot and circuit on output of pot are not connected yet they have continuity. That means you don't have control over the connections, How many are there?
Just me but I'd never put the centre -tap from the heaters on the same terminal as the input pot and plate load.
I have to stop here, so will repeat , sequence - only one connection from stage to stage in sequence from noisiest to most sensitive. The only place that gets two is where a single supply has two outputs , one to each channel (in a stereo circuit)
HTH
This will have influence over how you do the layout so that the runs are short and not crossing back and forth across the space.
The connection to the chassis should only be at one point. Some take the power supply end, some make it at the input , like right next to the RCA input jack. Some make the connection through a pair of parallel diodes in reverse polarity to each other.
You mentioned somewhere that grounds of input circuit to pot and circuit on output of pot are not connected yet they have continuity. That means you don't have control over the connections, How many are there?
Just me but I'd never put the centre -tap from the heaters on the same terminal as the input pot and plate load.
I have to stop here, so will repeat , sequence - only one connection from stage to stage in sequence from noisiest to most sensitive. The only place that gets two is where a single supply has two outputs , one to each channel (in a stereo circuit)
HTH
Last edited:
Thanks for the comments!
The challenge is my attempt at a separable chassis layout. I've got tube sockets and output transformers, along with my main TX, on the top half. On the bottom half, PSU, AC inlet, 45 filament transformer, Coleman regulators. I had some terminal strips that accept some "bullet" connectors I had that are easy to connect/disconnect. I may need to order one more to handle a more sophisticated grounding scheme.
The "bus" I came up with is a length of copper wire to which all the cathodes connect on the top then are routed, along with a wire connected to RCA + Bluetooth + pot (in) to those same elements (out). My Alps Pot came with a PCB for ease of soldering which bridges signal in and signal out common.
I see what you and Jan are saying about mixing "bus" and "star" and will think about the best way forward.
Last edited:
Hello,
120Hz is definitely PSU ripple induced or radiated to the signal path. Here are some guidelines that actually helped me in a similar situation:
120Hz is definitely PSU ripple induced or radiated to the signal path. Here are some guidelines that actually helped me in a similar situation:
- Keep the chassis fully isolated from signal ground, including the input RCA shield. Connect them with antiparallel 35A diodes (Graetz) at the PSU 1st capacitor -ve.
- Add 1k or similar grid stopper resistors direcly at each tubes' grid to prevent HF oscillation
- Connect the output transformer common (center tap) to the signal ground. Do not leave it floating.
- Test with shorted input (shorting RCA plug is handy) and open input
- Test with the volume pot disconnected, and short the 1st tube g1 directly to signal GND (through the grid stopper)
- Try to create a real star ground instead of mixed star/bus ground
- The 1st PSU capacitor carries the highest ripple current. Connect its -ve directly to the Graetz -ve (signal ground star point), do not share this wire with other component's return path
- If there is global negative feedback, the lower resistor should go where the GND of the 1st tube goes.
IF we are to believe your schematic, then your supply to the input tube is straight out of the rectifier and unfiltered. Sequence should be as in sketch below.
In addition , it is customary to filter the supply to the input tube a little better than the supply to the output tube so the taps would ordinarily be taken off in the reverse order you have shown them.
Also you have one of the two output transformers connected to the supplies for both channels while the other output transformer is not.
I gather you are using R1 as a pulse damping resistor? OK I guess but if your schematic is what's there , I would remove R2 in your schematic and make the connection from rectifier ground directly to the neg. pin of C1. You might try grounding to chassis with a clip-lead from neg of C1 to chassis.
Just clip to a screw you know has good contact with the aluminum. Do it with power off and then switch on.
Do you have bleeder resistors across any of the power supply caps?
I didn't get what you were doing with the second winding's R12 and R13 at 200 Ohms each so didn't do anything with them.
I hope something here helps.
Nice drawing program. Easy and quick to use. Fun to learn. Thanks for showing it.
In addition , it is customary to filter the supply to the input tube a little better than the supply to the output tube so the taps would ordinarily be taken off in the reverse order you have shown them.
Also you have one of the two output transformers connected to the supplies for both channels while the other output transformer is not.
I gather you are using R1 as a pulse damping resistor? OK I guess but if your schematic is what's there , I would remove R2 in your schematic and make the connection from rectifier ground directly to the neg. pin of C1. You might try grounding to chassis with a clip-lead from neg of C1 to chassis.
Just clip to a screw you know has good contact with the aluminum. Do it with power off and then switch on.
Do you have bleeder resistors across any of the power supply caps?
I didn't get what you were doing with the second winding's R12 and R13 at 200 Ohms each so didn't do anything with them.
I hope something here helps.
Nice drawing program. Easy and quick to use. Fun to learn. Thanks for showing it.
I've tried to illustrate what I have below after working on this again today. Changes shown below are not major, but include splitting 45 grounding stage point and 6N23P/Heater CT/Gyrator ground into two separate points. I am still measuring ~72 dB 120Hz noise.
Take a look at that and I'll try to identify what's still wrong based on comments made in this discussion:
Notes
Changes to Make?
Is this so wrong it's not even funny, or am I getting closer?
Take a look at that and I'll try to identify what's still wrong based on comments made in this discussion:
Notes
- Only had room to draw 1 45 with coupling capacitor + grid stopper + grid leak. They both have these elements!
- This is SET not PP.
- I have provided for a chassis connection switchable between the three ground terminals, but have only tried C2 so far. On/off makes no difference.
- Potentiometer housing is currently ungrounded. Grounding didn't make a noise difference.
- Potentiometer L - G - R connections are handled by PCB. Ground IN is continuous with ground OUT. I can remove the PCB, but used it in prototyping and was able to achieve low enough noise floor.
- It is distasteful to have the heater CT where it is--I agree, but I am not measuring significant 60Hz noise at the output of my speakers. Let's ignore that, since when disconnected, I see a spike in 60Hz noise.
- There's a bleeder resistor across C2.
- RCA are isolated from chassis.
- Disconnecting Bluetooth module makes no difference in noise. It's fed by its own PSU.
Changes to Make?
- Signal ground may need to be grounded with Gyrator - 6N23P cathode instead of at 45 cathode grounding point?
- Chassis connection to signal ground? Potentially with 10R resistor or diodes?
- It's a mistake for channels to share grounds like I am doing, and they need to be split up?
Is this so wrong it's not even funny, or am I getting closer?
You make it look easy. I am the worst drawer/illustrator you will likely ever meet.
- C3 (+) goes to OPT, then to 45 and speaker terminals. Same for C4 with other 45. I felt that I was unable to create a schematic with all actual connections without it becoming a chaotic mess. The amplifier works and I am pretty certain I have my (+) wires going to all the right places.
- R12/R13 on 6.3VAC virtual center tap for heaters.
- My understanding of Ale Bartola's gyrator is that its CCS provides substantial ripple rejection, so ripple at C2 should be fine. C3 and C4 get extra filtering and feed the 45s.
- I've got a bleeder across C2.
- 30 ohm off of rectifier (+) and (-) are to fine tune the power supply response--see my attached PSUD2 (with a few wrong values; but in the ballpark!)
I will:
- Connect C1 (-) directly to rectifier (-).
- Try chassis connection at C1 (-).
This schematic causes a ripple voltage between signal ground and chassis. May or may not be problematic.
Jan
Thanks Jan. What am I missing? Are you referring to the ground symbol at C1? Everything else is more or less bog standard isn't it?
The return current from bottom of C2 & C3 to the bridge causes ripple across that piece of wire. Since the signal ground is connected to bottom of C3, it is at a ripple potential with respect to the ground symbol at C1 which I assume is chassis.
You can avoid it by moving that ground symbol to the bottom of C3.
Jan
You can avoid it by moving that ground symbol to the bottom of C3.
Jan
Ah, I see.
I didn't intend it as chassis ground. In this case putting the ground symbol there was partly a habitual repeat of what I've learned to do with LTspice schematics , which won't work unless you put one somewhere.
I think I put it at that spot because I was thinking about how the bridge in the top schematic wasn't grounded and wanted to stress that. . . . . . . And that the 30 Ohm R2 shouldn't be there. (Or so I think, anyway)
Chassis ground is one thing I'm NOT certain of, so in builds I leave it until everything else is done and then try the circuit common to chassis connection with a clip-lead here and there to see what works best.
Thanks Jan.
I didn't intend it as chassis ground. In this case putting the ground symbol there was partly a habitual repeat of what I've learned to do with LTspice schematics , which won't work unless you put one somewhere.
I think I put it at that spot because I was thinking about how the bridge in the top schematic wasn't grounded and wanted to stress that. . . . . . . And that the 30 Ohm R2 shouldn't be there. (Or so I think, anyway)
Chassis ground is one thing I'm NOT certain of, so in builds I leave it until everything else is done and then try the circuit common to chassis connection with a clip-lead here and there to see what works best.
Thanks Jan.
What I normally do is bond the chassis to mains PE for safety.
Then I figure out the best point of the circuit to ground to the chassis via a pair of anti-parallel diodes with an R of say 12 ohms and a C of say 10nF in parallel.
That way I preserve the safety ground but avoid ground currents due to the diode threshold voltage.
Some people use 4 diodes, two antiparallel pairs of two in series, but I don't know if that makes a difference.
Jan
Then I figure out the best point of the circuit to ground to the chassis via a pair of anti-parallel diodes with an R of say 12 ohms and a C of say 10nF in parallel.
That way I preserve the safety ground but avoid ground currents due to the diode threshold voltage.
Some people use 4 diodes, two antiparallel pairs of two in series, but I don't know if that makes a difference.
Jan
Jan, can you clarify what this would mean relative to my diagrams (C1-C4)?
C3 and C4 are decoupled. They are the output stage.
- The negative terminal of pain C3 should go to C2 immediately and the 45 returns to C2?
or
- The 45 fed by C3 should return to C3, then C3 returns to C2?
After reorganizing the PSU layout itself and its position in the chassis, verifying that all capacitors were functioning correctly (I had a short at one point), replacing the choke with a resistor (no change), shielding input wiring, etcetera etcetera--I removed the Traco SMPS providing power to the bluetooth module. That turned out to be culprit! 💩 I had heard that these could be noise factories, but thought that the ~7"from its position to any of the rectification/signal circuity would be enough. Apparently not!
Thank you for all input provided.
Thank you for all input provided.
I was just reading through this thread and watching to see when someone was going ask more about this Bluetooth power supply scheme…
Glad you found it!!
Glad you found it!!
Update many months later:
Wanted to post this here for full disclosure. I feel embarrassed and want to apologize for wasting everyone's time. For those of you that have read the thread, it was pretty rocky!
In my last post, I "concluded" that the switching PSU was the culprit here. This was based upon moving it and "resolving" the noise issue. However, when I got back into the amp and attempted to tidy up some wires in preparation for shipping to my friend, moving the leads seemed to restore the noise.
At this point, I decided I didn't know what the issue was and wanted to start "from scratch" as much as possible.
Ultimately I discovered that I had swapped one lead from each of my Antek transformer secondaries for the 45 filaments. One side was ~12V and the other was ~4.5V. This was the source of the noise. This probably occurred when I moved the amp from its breadboarded stage into the chassis.
Again, I want to be "in good standing" on this forum and want to apologize for the time I wasted. Once I got the transformer secondaries sorted out, the amp was dead silent. Has now been shipped off to my friend without issue.
Wanted to post this here for full disclosure. I feel embarrassed and want to apologize for wasting everyone's time. For those of you that have read the thread, it was pretty rocky!
In my last post, I "concluded" that the switching PSU was the culprit here. This was based upon moving it and "resolving" the noise issue. However, when I got back into the amp and attempted to tidy up some wires in preparation for shipping to my friend, moving the leads seemed to restore the noise.
At this point, I decided I didn't know what the issue was and wanted to start "from scratch" as much as possible.
Ultimately I discovered that I had swapped one lead from each of my Antek transformer secondaries for the 45 filaments. One side was ~12V and the other was ~4.5V. This was the source of the noise. This probably occurred when I moved the amp from its breadboarded stage into the chassis.
Again, I want to be "in good standing" on this forum and want to apologize for the time I wasted. Once I got the transformer secondaries sorted out, the amp was dead silent. Has now been shipped off to my friend without issue.