Hello!
I'm building a preamp. The individual components are all finished now and I'm currently integrating them into a case.
The preamp gets its audio from a relay input selector. After amplification and volume control it feeds 4 outputs - three of them are connected to RCA's on the back panel. One of them feeds a headphone amp which sits in the case as well. The case itself is made of aliminum.
The outputs on the back give a nice clean signal. The headphone amp however produces a massive amount of buzz + hiss. (Such words are not very precise - I attached a recording of that sound, so you know what I'm talking about.) If the headphone amps inputs are shorted to ground, there is almost no noise. Once they are connected to the preamps output, things get quite noisy, though.
Sounds like a ground loop - doesn't it? Well, here is what I did to avoid noise/buzz/hiss:
Here are the details:
A side note: There is exactly one direct ground loop: The headphone jack is connected to the chassis (its thread is grounded) and also to the star point via a wire. I think thats no problem for two reasons: 1. The headphones are passive devices and are most likely not sensitive enough to react to inductions in this loop. 2. The wire always runs in very close proximity to the aluminum case. AFAIK, a ground loop only produces noise, if it forms a significant area through which a magnetic field can pass. The two paths (wire and case) are so close to each other, that this should not be the case here.
To sum up my question:
I'm building a preamp. The individual components are all finished now and I'm currently integrating them into a case.
The preamp gets its audio from a relay input selector. After amplification and volume control it feeds 4 outputs - three of them are connected to RCA's on the back panel. One of them feeds a headphone amp which sits in the case as well. The case itself is made of aliminum.
The outputs on the back give a nice clean signal. The headphone amp however produces a massive amount of buzz + hiss. (Such words are not very precise - I attached a recording of that sound, so you know what I'm talking about.) If the headphone amps inputs are shorted to ground, there is almost no noise. Once they are connected to the preamps output, things get quite noisy, though.
Sounds like a ground loop - doesn't it? Well, here is what I did to avoid noise/buzz/hiss:
- I tried to avoid ground loops by connecting all different grounds at a single star point that connects to the chassis. There are several "T"-connections in the GND path, but no loops.
- All input and output jacks are insulated from the chassis.
- I tried to separate grounds that carry current from grounds that serve as a signal reference. Don't know if I did it right, though.
Here are the details:
- The preamp is a design based on the CS3318 volume control chip. I made my own layout. I guess that this layout is okay, since I get a clean signal from the three outputs on the back panel. You can find more about that layout here: http://www.diyaudio.com/forums/analog-line-level/241798-cs3318-based-preamp-layout.html
I know its probably not perfect but it should work. - The headphone amp design is from Rod Elliot. It's p113: Headphone Amplifier
- The two analog power supplies for the preamp and the headphone amp are from Rod as well: Power Supply for Preamps
They are fed by a common transformer. - The CS3318 does not make sense without a microcontroller to actually control it. That's why I have another power supply (12V and 5V) to power the CS3318's digital circuitry, the user interface, the relay switcher and the microcontroller. This supply is based on simple 7805/7812 and an additional transformer.
A side note: There is exactly one direct ground loop: The headphone jack is connected to the chassis (its thread is grounded) and also to the star point via a wire. I think thats no problem for two reasons: 1. The headphones are passive devices and are most likely not sensitive enough to react to inductions in this loop. 2. The wire always runs in very close proximity to the aluminum case. AFAIK, a ground loop only produces noise, if it forms a significant area through which a magnetic field can pass. The two paths (wire and case) are so close to each other, that this should not be the case here.
To sum up my question:
- What could be the source of my buzz+hiss?
- How can I get rid of it?
Attachments
In the cheap but good department, I'd drill the headphone jack hole out a 64th inch bigger, and insulate it from the case with rubber o-rings. You can usually get different sizes of O-rings at the hardware store, and they will stretch a little bit. Buy two.
Hopefully the microprocessor and digital power supply transformer are behind a bulkhead from the analog preamp sections. Analog and digital should be in separate grounded boxes. You do have to tie the two returns together at one point, but you can separate them slightly with a resistor if you want, say 1k or something. Analog and digital source wires should run with as much separation as possible, and where crosses are necessary, cross at ninety degrees.
Is the heatsink of the headphone amp IC or power transistors isolated from metal by a mica washer or something? Your drawing says no internal ground connection in the headphone amp. I think there will have to be one between that and analog preamp ground, perhaps separated by a resistor. Unless the headphone amp is transformer coupled or coupled with an analog optoisolator (which is highly unlikely, they are mostly for telephone circuits, not hifi).
Hopefully the microprocessor and digital power supply transformer are behind a bulkhead from the analog preamp sections. Analog and digital should be in separate grounded boxes. You do have to tie the two returns together at one point, but you can separate them slightly with a resistor if you want, say 1k or something. Analog and digital source wires should run with as much separation as possible, and where crosses are necessary, cross at ninety degrees.
Is the heatsink of the headphone amp IC or power transistors isolated from metal by a mica washer or something? Your drawing says no internal ground connection in the headphone amp. I think there will have to be one between that and analog preamp ground, perhaps separated by a resistor. Unless the headphone amp is transformer coupled or coupled with an analog optoisolator (which is highly unlikely, they are mostly for telephone circuits, not hifi).
Last edited:
I think thats not the source of trouble. See my explanation in my first post. I don't hear any noise when the inputs of the headphone amp are shorted to ground - so IMO the problem lies in the interconnection between the headphone amp and the preamp.
Yes, they are. There is a 1mm steel sheet between the two sections. But to be a bit more precise: The preamp and input switching is in the rear right corner, separated from the rest by the steel bulkhead. The headphone amp, however, is in the front left - because thats where the headphone jack is mounted.
See the image in the attachment.
Can you explain that to me? I assume with "returns" you mean grounds. I guess I can put a 1k resistor in there, because there shouldn't be any current running across the connection, so they effectively stay almost at the same potential. Is that correct?
Yes, the power Transistors are insulated.
Regarding the amp schematics: You are right, there is a direct ground connection on the headphone amp board. I didn't see that... Okay, so I have a loop here.
Okay, I just removed the ground connection between the headphone amp and the preamp outputs. That means: the headphone amp has only a single ground connection to its power supply. The noise sounds exactly like before, except that it is much louder now.
I experimented a bit and it turned out that
- when I connect that ground cable (coming from the inputs of the headphone amp) to the RCA output jacks (which I tried to keep as the ground reference for audio signals), the noise is a little quieter than with no the ground not connected. This is the configuration that I had before your post.
- when I connect that ground cable (coming from the inputs of the headphone amp) to the ground plane on the preamp board, the noise is much quieter than with the ground not connected. But it is still there.
I attached an image of the case. I know it looks a bit messy, but hey - I'm constantly changing the wiring to get rid of that noise. So please bear with me
Thanks for your help
Attachments
The wires are loose, because I'm currently tweaking the connections to get the amp silent.
Okay, to make things a bit clearer: The loose wires that lead to the star ground are coming from the input/output jacks (the black wire); from the digital supply (brown wire); the earth connection of the mains power cable (green-yellow); the analog power supply-transformer (blue and yellow).
All the black wires on the top left (next to the power plug and the soft switch board) are carrying mains voltage and connect the mains power plug with the fuse holder and the soft switching circuit.
The red and green wires on the left are the top and bottom ends of the analog-PSU transformer and they lead to the PSU unit for the headphone amp (which sits below the amp itself) and the PSU for the preamp.
There are almost no signal pairs - except for the power connections. All the input jacks have their grounds soldered together directly - as indicated in the "schematic" of my first post. There is nothing I could twist...
I will try twisting the power cables and see if it improves things. Thanks for the hint!
Touching the opamp does not do much to the sound. When my fingers make contact with the chips pins, i can hear an increase in humm that decays within ~0.5sec. Then I hear exactly the same humm as before. The same thing happens when I release my fingers from the pins.epicyclic said:
I think the headphone amp is okay - it is silent when nothing is connected.
Yes. For the headphone amp there's a 100nF cap to the left of the chip. The preamp has caps right under the opamps and some more all over the place on the bottom of the pcb.
Thanks for your great help!
The Star Ground is connected to Chassis.
The PE (green/yellow wire) MUST be connected to Chassis.
The Star Ground can float above the Chassis.
The Star Ground needs a connection to Chassis if there are any exposed conductive parts, eg metal screw heads, metal knobs, RCA socket barrels, speaker terminals if the metal is touchable, etc.
This extra connection can be by a direct wire capable of blowing the mains fuse in FAULT condition, or by a Disconnecting Network, which MUST also be capable of blowing the mains fuse in a FAULT condition.
The PE (green/yellow wire) MUST be connected to Chassis.
The Star Ground can float above the Chassis.
The Star Ground needs a connection to Chassis if there are any exposed conductive parts, eg metal screw heads, metal knobs, RCA socket barrels, speaker terminals if the metal is touchable, etc.
This extra connection can be by a direct wire capable of blowing the mains fuse in FAULT condition, or by a Disconnecting Network, which MUST also be capable of blowing the mains fuse in a FAULT condition.
The PE connection does two jobs during Fault condition.
It lowers the voltage that the Chassis can rise above the Ground voltage that the equipment is sitting on.
It blows the mains fuse when that Fault current flows.
To do both jobs, the resistance of the PE route back to the distribution board must be low. In the UK this is specified and tested during installation.
It lowers the voltage that the Chassis can rise above the Ground voltage that the equipment is sitting on.
It blows the mains fuse when that Fault current flows.
To do both jobs, the resistance of the PE route back to the distribution board must be low. In the UK this is specified and tested during installation.
AFAIK the RCD usually triggers at a few mA whereas a "main fuse" is rated much more (e.g. 500mA) and blows much slower as well. I'm not an expert on this topic, but I'd guess its a bad sign if a mains fuse blows due to a current through the PE.
Nevertheless, can my ground star point and its connection to the chassis/PE be a source of humm?
Nevertheless, can my ground star point and its connection to the chassis/PE be a source of humm?
Suggestion ....... move the star ground connections on the AC input side of the analogue power supplies to the DC output sides . The AC input side is where the charging currents for the smoothing caps are circulating . This might go someway to curing the buzz .
For the hiss.... if you have used Rods feedback resistor values for the headphone amps you will have another 27db of gain so is the hiss still there with the volume on minimum .
For the hiss.... if you have used Rods feedback resistor values for the headphone amps you will have another 27db of gain so is the hiss still there with the volume on minimum .
hi epicyclic!
Thanks a lot for that wonderful hint! I connected the transformer directly to the power supplies and connected the star ground connections to the DC outputs of the power supplies.
This actually makes the humm/buzz AND the hiss much quieter.
I would like to understand what I just did and why it affected the noise. Here is my guess (could you please correct me if I'm wrong?): The source of noise is a voltage drop along a signal-carrying cable (or the ground reference). In my case, the ground reference is slightly different at the headphone amp and the preamp. That's one reason why I have noise. To remove the noise, I have to bring both ends of a cable to the same electrical potential again.
This can be achieved in two ways: A) A shorter physical cable length or a larger cable diameter decreases the resistance and thus decreases the voltage drop across the cable. B) Anything else that is connected to the ground link between two modules could be pushing a current through the wire - thus adding additional voltage drop that courses the two ends of the ground connection to have different electrical potentials and therefore alters the signal and adds noise.
In my case, the ground path between the headphone amp and the preamp was:
After I connected the star ground directly to the DC out of my supplies the path looks like this:
effectively removing the two supplies from the chain. This makes the cable shorter (A) and also removes any voltage drops due to currents from the power supplies (B).
Is my explanation correct?
Okay, but unfortunately I'm still not quite there... As I said before: The noise is much quieter now (still audible though), but only if I connect the headphone amps input grounds to the ground plane on the preamp board. I would expect that this causes a ground loop because there are two ground paths between the headphone amp and the preamp: 1) Through the audio input cables that connect the headphone amps inputs with the preamps outputs. And 2) through the power-cables of both. They connect in the star ground. See this "image"
However, breaking that loop by removing the ground link 1) creates noise because now the signal reference goes through the power cables. And they have a voltage drop across them due to the current that flows in them.
Breaking the loop by removing one of the ground connections 2) would make the power currents flow through the connection 1) and then the situation is basically the same.
I could use bigger wires for the power connection. That would decrease their resistance and thus decrease the voltage drop across them and make the noise a bit quieter. But is that the way to go?
PS: On man, why are my posts always soooo long
Thanks a lot for that wonderful hint! I connected the transformer directly to the power supplies and connected the star ground connections to the DC outputs of the power supplies.
This actually makes the humm/buzz AND the hiss much quieter.
I would like to understand what I just did and why it affected the noise. Here is my guess (could you please correct me if I'm wrong?): The source of noise is a voltage drop along a signal-carrying cable (or the ground reference). In my case, the ground reference is slightly different at the headphone amp and the preamp. That's one reason why I have noise. To remove the noise, I have to bring both ends of a cable to the same electrical potential again.
This can be achieved in two ways: A) A shorter physical cable length or a larger cable diameter decreases the resistance and thus decreases the voltage drop across the cable. B) Anything else that is connected to the ground link between two modules could be pushing a current through the wire - thus adding additional voltage drop that courses the two ends of the ground connection to have different electrical potentials and therefore alters the signal and adds noise.
In my case, the ground path between the headphone amp and the preamp was:
Code:
headphone amp <=> headphone supply <=> star ground <=> preamp supply <=> preamp
Code:
headphone amp <=> star ground <=> preampIs my explanation correct?
Okay, but unfortunately I'm still not quite there... As I said before: The noise is much quieter now (still audible though), but only if I connect the headphone amps input grounds to the ground plane on the preamp board. I would expect that this causes a ground loop because there are two ground paths between the headphone amp and the preamp: 1) Through the audio input cables that connect the headphone amps inputs with the preamps outputs. And 2) through the power-cables of both. They connect in the star ground. See this "image"
Code:
+---------------+ signal cable (1) +-----------+
| |------------------------| |
| headphone amp | | preamp |
| |-----+ +------| |
+---------------+ | power | +-----------+
| cables (2)|
+----------------+
| |
| star ground |
| + chassis |
| |
+----------------+However, breaking that loop by removing the ground link 1) creates noise because now the signal reference goes through the power cables. And they have a voltage drop across them due to the current that flows in them.
Breaking the loop by removing one of the ground connections 2) would make the power currents flow through the connection 1) and then the situation is basically the same.
I could use bigger wires for the power connection. That would decrease their resistance and thus decrease the voltage drop across them and make the noise a bit quieter. But is that the way to go?
PS: On man, why are my posts always soooo long
Yes i think your explanatoin is a way of understanding the situation and i would like to add that the charging currents between the transformer / bridge rectifier and smoothing capacitors is a particularly noisy area .
Im still puzzling your new situation .
An area that still concerns me is the amount of gain you have available .....22db in the CS3318 and 27db in the headphone amps which could be making the situation worse than it need be .
What is the music volume like into your headphones , do you have to reduce the CS3318 volume setting a lot when you use them.
Im still puzzling your new situation .
An area that still concerns me is the amount of gain you have available .....22db in the CS3318 and 27db in the headphone amps which could be making the situation worse than it need be .
What is the music volume like into your headphones , do you have to reduce the CS3318 volume setting a lot when you use them.
Last edited:
Okay, good to know I was on the right path. Thanks!
Regarding the gain: Yea, when I turn the volume up, its so horribly loud - I never exceeded that 0db point much because I didn't want to burn my headphones.
That means, I will definitely change the gain of the headphone amp to something lower. I found another thread that deals with the gain of the exact same circuit and I'll follow that route. The CS3318 is relatively quite, so I think I can safely use it with a lot of gain. That would mean i can decrease the headphone amps gain even further and get a higher level in the cables between preamp and headphone amp = higher SNR in the connection between the preamp and the headphone amp = less noise.
However, I'd like to take the chance and get rid of as much of the noise as I possibly can and change the headphone amps gain afterwards. Its much easier to trace the thing down to a narrow area when the noise is loud enough to hear what effect in terms of tone/character my changes have.
Most of the hiss must be coming from the wiring. When the inputs of the headphone amp are grounded, there is almost no hiss. Its almost inaudible.
Regarding the gain: Yea, when I turn the volume up, its so horribly loud - I never exceeded that 0db point much because I didn't want to burn my headphones.
That means, I will definitely change the gain of the headphone amp to something lower. I found another thread that deals with the gain of the exact same circuit and I'll follow that route. The CS3318 is relatively quite, so I think I can safely use it with a lot of gain. That would mean i can decrease the headphone amps gain even further and get a higher level in the cables between preamp and headphone amp = higher SNR in the connection between the preamp and the headphone amp = less noise.
However, I'd like to take the chance and get rid of as much of the noise as I possibly can and change the headphone amps gain afterwards. Its much easier to trace the thing down to a narrow area when the noise is loud enough to hear what effect in terms of tone/character my changes have.
Most of the hiss must be coming from the wiring. When the inputs of the headphone amp are grounded, there is almost no hiss. Its almost inaudible.
I just played around with the wiring and the ground loop that I mentioned before. Obviously breaking the ground loop at the connection 1) does not make sense (see the post above). Removing the ground connection between the headphone amp and the ground star point removes the buzz (as expected) but makes the hiss very loud.
Interestingly removing the ground connection between the preamp and the star ground reduces the buzz and the hiss to a very low level.
I would expect that any current that previously flew from the preamp directly to the star ground (and thus to the PSU) now has to flow through the signal cable and through the headphone amp - causing noise. But there is no noise - any guess why?!
This is what it looks like now:
Interestingly removing the ground connection between the preamp and the star ground reduces the buzz and the hiss to a very low level.
I would expect that any current that previously flew from the preamp directly to the star ground (and thus to the PSU) now has to flow through the signal cable and through the headphone amp - causing noise. But there is no noise - any guess why?!
This is what it looks like now:
Code:
+---------------+ signal cable (1) +-----------+
| |------------------------| |
| headphone amp | | preamp |
| |-----+ | |
+---------------+ | power +-----------+
| cables (2)
+----------------+
| |
| star ground |
| + chassis |
| |
+----------------+Not sure, actually... Both grounds meet at the CS3318. In the reference schematic they don't even make a distinction between AGND and DGND. I don't know how they are connected on the inside of the chip - but letting the digital gnd float might cause a rather high voltage between DGND and AGND and that might burn the CS3318. So I assume they should be connected at some point to keep their "difference in electrical potential" low.
I could use a resistor to separate both grounds from each other and still prevent risky high voltages between the two. Does that make sense?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.