Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

hum & noise in amp

Status
Not open for further replies.

Othello

Member
I built a guitar amp. It consists of various circuit boards (pre amps, equalizer, second tube preamp, tube power amp, various power supply modules etc.) mounted in an alu chassis. I also have a very sensitive AC level meter, down to -120dB, to measure noise etc.
All the individual components work and now I wired the whole thing up and I am disappointed about the noise level.

The chassis is grounded via a grounded outlet, all the other components are connected via coax with each other-so they have a common ground but I have not connected that signal ground to the chassis ground.
In addition to the signal coax cables there are other control cables for power for the op amps, dc for tube heaters, switches etc. which connect the modules.

The overall noise and hum level increases as I connect more and more modules and I think I am picking up noise via the power and control cables and would love to hear a general outline as to how to best wire up a project like this to keep noise out as much as possible.

Uwe
 
Connect the signal ground to the chassis ground.
 
Yes I know, this is what people say and recommend.
And since I am building it right now there is nothing easier than to take a clip lead and do just that and observe, while I am doing it, my meter ( or listen in a pair of headphones) and see what the change is.

But I am not overly impressed by the changes and therefore I am trying to understand this problem on a more fundamental level.

What does my coax cable, assuming I picked a good quality one, shield against? The electric field? The magnetic field? Both?

What about all the unshielded spaces, like the "vast expanse' of open circuit boards? I mean I use coax whenever there is a substantial space to be crossed like 3" or such, and then I have an 8" circuit board which receives no shielding??

Uwe
 
I built a guitar amp. It consists of various circuit boards (pre amps, equalizer, second tube preamp, tube power amp, various power supply modules etc.) mounted in an alu chassis. I also have a very sensitive AC level meter, down to -120dB, to measure noise etc.
All the individual components work and now I wired the whole thing up and I am disappointed about the noise level.

The chassis is grounded via a grounded outlet, all the other components are connected via coax with each other-so they have a common ground but I have not connected that signal ground to the chassis ground.
In addition to the signal coax cables there are other control cables for power for the op amps, dc for tube heaters, switches etc. which connect the modules.

The overall noise and hum level increases as I connect more and more modules and I think I am picking up noise via the power and control cables and would love to hear a general outline as to how to best wire up a project like this to keep noise out as much as possible.

Uwe


This is an all too common occurrence in multi circuit systems, where you feed the signal from one subsystem to another. In addition, you have a power supply, or multiple power supplies for different voltage levels in your systems.

You may have done some of this stuff, but since this is a forum, here are some ideas:

Signal integrity at the source: The mV signal from the pickups is very susceptible to noise. The front end is critical and needs special treatment. Use shielded wire from the entry point of the signal in the chassis to the first circuits. If the signal goes directly to the 1st circuit w/o any wires (desirable), then is the first stage properly bypassed, with bypassing caps as close to the 1st stage as possible? If the 1st stage is a dual supply, are both rails bypassed?

From one system to another: When you jumper from the pre-amp to the main amp, use shielded wire, as short as possible.

Power supply: Use a toroidal transformer if possible. If not, use a shielded transformer. Be sure that you have enough filtering in the supply. Keep the wire lengths short from the supply to the circuits that require them. Bypass the power supply feeds at the circuit boards that the supply wires feed.

Grounds: Ground the supply with a large gauge wire directly to the chassis if possible. If you are using a 3-prong line cord, the ground needs to also be bonded to the chassis. A "star" ground topology may be desirable if you suspect a ground loop. This means that you bring all subsystem grounds to a single point on the chassis, i.e. don't daisy-chain the grounds.

Digital circuits: If you incorporate any digital circuits that have clocks, use a separate ground wire and segregate from the analog circuit ground. This is especially important in mixed signal A/D & D/A circuits. A star ground topology is paramount. When you finally tie the grounds together at the star ground point, use an RF bead or toroid choke on the digital ground wire at that point.

Ground planes: If you created your own boards, ground and power planes really help in providing shielding in the circuit, as well as provide a low impedance return to the supply. Some miss this 2nd important point.

Shielding: It may be a good idea to physically shield the front end from the rest of the circuit, with the use of a small aluminum box inside the chassis.

Feedback: If the feedback network is not properly handled, it can cause hum or in worst cases an oscillation. Pay careful attention to where the feedback is physically routed. A Zobel network also helps on the output stage in this regard.

Some basic testing on your part will also help. Does the hum show up on an oscilloscope? Does it appear regardless of the load? Does it appear with no load? Does it appear with no guitar plugged in? Does it go away when you turn off the fluorescent lights in the room? You may want to selectively eliminate subsystems until the hum goes away or appears (divide and conquer).

Anyway, all that I can think of before I had my first cup of java this morning. Feel free to add more folks! These are some things that I've observed over the years and have helped me.
 
The coax shields against noise on the wires, primarily electric but also magnetic. The electric shielding is pretty straightforward, the magnetic shielding effect is a little trickier - it's from the AC magnetic field trying to induce currents in the shielding and failing because the conductive shield appears like a short circuit.

use a shielded transformer

This got me one time. I was pulling my hair out trying to trace a noise source, and it turned out to be MAGNETIC interference. Little coupling transformer, wrapped a piece of copper foil around it and it went away. Look for this in your design, Othello.

What about all the unshielded spaces, like the "vast expanse' of open circuit boards? I mean I use coax whenever there is a substantial space to be crossed like 3" or such, and then I have an 8" circuit board which receives no shielding??

Your use of coax is the same as what I do, except I'll go even an inch (use the lightweight, easy-to-work-with stuff, it's just audio frequency) for a high impedance input to something. You have to ground the shields in each of these little jumpers, of course.

pre amps, equalizer, second tube preamp, tube power amp

How many preamps have you got in this thing? What's your total gain like? Could you post a schematic, pics?
 
This is an all too common occurrence in multi circuit systems, where you feed the signal from one subsystem to another. In addition, you have a power supply, or multiple power supplies for different voltage levels in your systems.

You may have done some of this stuff, but since this is a forum, here are some ideas:

Signal integrity at the source: The mV signal from the pickups is very susceptible to noise. The front end is critical and needs special treatment. Use shielded wire from the entry point of the signal in the chassis to the first circuits. If the signal goes directly to the 1st circuit w/o any wires (desirable), then is the first stage properly bypassed, with bypassing caps as close to the 1st stage as possible? If the 1st stage is a dual supply, are both rails bypassed?

From one system to another: When you jumper from the pre-amp to the main amp, use shielded wire, as short as possible.

Power supply: Use a toroidal transformer if possible. If not, use a shielded transformer. Be sure that you have enough filtering in the supply. Keep the wire lengths short from the supply to the circuits that require them. Bypass the power supply feeds at the circuit boards that the supply wires feed.

Grounds: Ground the supply with a large gauge wire directly to the chassis if possible. If you are using a 3-prong line cord, the ground needs to also be bonded to the chassis. A "star" ground topology may be desirable if you suspect a ground loop. This means that you bring all subsystem grounds to a single point on the chassis, i.e. don't daisy-chain the grounds.

Digital circuits: If you incorporate any digital circuits that have clocks, use a separate ground wire and segregate from the analog circuit ground. This is especially important in mixed signal A/D & D/A circuits. A star ground topology is paramount. When you finally tie the grounds together at the star ground point, use an RF bead or toroid choke on the digital ground wire at that point.

Ground planes: If you created your own boards, ground and power planes really help in providing shielding in the circuit, as well as provide a low impedance return to the supply. Some miss this 2nd important point.

Shielding: It may be a good idea to physically shield the front end from the rest of the circuit, with the use of a small aluminum box inside the chassis.

Feedback: If the feedback network is not properly handled, it can cause hum or in worst cases an oscillation. Pay careful attention to where the feedback is physically routed. A Zobel network also helps on the output stage in this regard.

Some basic testing on your part will also help. Does the hum show up on an oscilloscope? Does it appear regardless of the load? Does it appear with no load? Does it appear with no guitar plugged in? Does it go away when you turn off the fluorescent lights in the room? You may want to selectively eliminate subsystems until the hum goes away or appears (divide and conquer).

Anyway, all that I can think of before I had my first cup of java this morning. Feel free to add more folks! These are some things that I've observed over the years and have helped me.



Analog, you just made my day, your check list will be a valuable starting point for me.

I have of course tried to lay out the circuits and the chassis with some of these considerations in mind. But with 10 or 11 circuit boards, many using 3 different voltages and associated switching, and a big power transformer (non toroidal), a choke and an output transformer there is no way I could think of to optimize everything.

But I can already see grounds for improvement or experimentation.
The boards containing ICs and using dual supplies all have bypassing networks close to the ICs, but none where the power enters the board.

I build and test the amp in stages and measure each time the overall noise figure, so I have the least experience with the entire amp, mostly I am testing subsections and I do not know much yet about changes with or without a load.
And all my testing is done with a 200 Ohm termination on the input, I did that after I saw what an adverse effect a cheap guitar with its wiring has on noise.

And yes, I do know about the fluorescent light and such, and no, there is no digital component in this amp.

I think I will start experimenting with the grounding of the power supplies and install bypass capacitors.

Which type is best suited? And what values, there are different requirements for the audio boards and the power supplies where only DC is wanted (my filament supply is DC also)?

Uwe
 
200 ohms sounds pretty low, I seem to remember a guitar pickup is more like 10k.

This is why I asked about the total gain. You may be swamping the tiny signal from the pickup with that 200Ω resistor, then compensating with a lot of gain, which will add noise.
 
This got me one time. I was pulling my hair out trying to trace a noise source, and it turned out to be MAGNETIC interference. Little coupling transformer, wrapped a piece of copper foil around it and it went away. Look for this in your design, Othello.

How many preamps have you got in this thing? What's your total gain like? Could you post a schematic, pics?



I will try this transformer shielding.

The amp is a 2 channel guitar amp, a preamp followed by two channels with equalizer boards followed by tube preamps for distortion and clean sound. Then both channels connect to an echo board followed by a board which contains all the preamp out, power amp in connections as well as a small headphone amp and then a 30 Watt power amp. And the 4 power supplies.
It's not for the gain that I use these amps, its for proper isolation between stages. Most of these stages have an amplification of 1.


AND I DON'T EVEN PLAY THE GUITAR!

It's for my brother and I am working on the schematics, it will be a 6 squarefoot plan!

Uwe
 
200 ohms sounds pretty low, I seem to remember a guitar pickup is more like 10k.

This is why I asked about the total gain. You may be swamping the tiny signal from the pickup with that 200Ω resistor, then compensating with a lot of gain, which will add noise.

I built a very small op amp preamp/line driver in the guitar which has very low output impedance.

Uwe
 
I built a very small op amp preamp/line driver in the guitar which has very low output impedance.

That's a good way of doing it, get that output impedance down before it even gets to the cable.

It's not for the gain that I use these amps, its for proper isolation between stages. Most of these stages have an amplification of 1.

You might try using a 1:1 signal transformer for isolation, or just cap-block the inputs. Each active stage of amplification, even at unity gain, is going to add some noise. You've got, what - three or four dual-channel inline amps before the main stage? Eliminate some of those unity-gain amps and see if that will reduce the noise.
 
Last edited:
That's a good way of doing it, get that output impedance down before it even gets to the cable.



You might try using a 1:1 signal transformer for isolation, or just cap-block the inputs. Each active stage of amplification, even at unity gain, is going to add some noise. You've got, what - three or four dual-channel inline amps before the main stage? Eliminate some of those unity-gain amps and see if that will reduce the noise.

I have this level meter, it's made by Sennheiser and I brought it from Europe, it measures down to about -120 dB and you can connect a headphone to what I assume is a very sensitive instrument amp in there. And the headphones allow you to listen in on the signal.
And it is quite clear if you use this tool, what is the hiss of amplification and what is shortwave like noise or hum in its various forms.

The amps individually are all fine. And yes, the tube amps add more noise, but all of that is NOTHING compared to the hum I am trying to eliminate here.
If I will be left with only the amps 'natural' hiss I will have the quietest guitar amp on earth!!!

Uwe
 
Could be a ground loop - you know about those? Search on "ground loop" if you don't. This is why mikebits suggested a star grounding pattern, also called a single point ground. Ground loops are a common source of 60hz hum.
 
Could be a ground loop - you know about those? Search on "ground loop" if you don't. This is why mikebits suggested a star grounding pattern, also called a single point ground. Ground loops are a common source of 60hz hum.

See analog's post earlier, he made a few very good suggestions on how to work on this issue.
 
I made a distortion analyser with oscillators that has many circuit boards. I connected the boards together with shielded audio cable and connected the shield to ground only at the receiving end since the circuits all had the same ground. Then there was no ground loop and the hum, noise and distortion was -100dB.
 
I made a distortion analyser with oscillators that has many circuit boards. I connected the boards together with shielded audio cable and connected the shield to ground only at the receiving end since the circuits all had the same ground. Then there was no ground loop and the hum, noise and distortion was -100dB.

There you go!
I am glad you bring this up.
I own a professionally made piece of audio studio equipment where all the shielded wires are connected at one end only.
The thing works beyond any specs I could reliably measure-beautiful!

If I connect a bunch of circuit boards with coax, am I not already creating a daisy chain of ground wire??- something the 'all ground to a central point' advocates have never sufficiently explained to me.

The technique audioguru mentions makes all the sense to me, but it would require some organized approach, e.g. which ends do get connected to ground and which are left floating??

Uwe
 
I checked my power supplies and could improve a few noise figures by adding bypass capacitors, they reduced the higher frequency noise sometimes several dB. (I used 0.1mF ceramic type).
The noise component was mostly in the 'fractions of a milivolt' range.

But my filament supply seems not to be ok, there I measured several milivolts ripple and I don't think this is right. I do not understand where that much ripple should come from, the regulation of the voltage regulator should be better, but I don't know what could be wrong. (see diagram included. By the way what can I use on a Mac to produce circuit diagrams, I am embarrased about the yellow pictures I have to upload...)

Uwe
 

Attachments

  • DSC00972.jpg
    DSC00972.jpg
    56.1 KB · Views: 152
Why do you have such a complicated voltage regulator that has its Mosfet barely turned on since its gate voltage is clamped too low?
Why not use an LM317 adjustable regulator and a couple of current-booster transistors if they are needed?
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top