Definitive hum and buzz sticky?

[Luthierwnc]

Hi Guys,

After a year off to start a new business I'm pulling some projects off the shelf. One of them is a Dumble-style amp that I couldn't get rid of both a hum and a buzz. There are enough experimental elements that I can't rule out design error as well as execution error. I only have this amp and an Ampeg Gemini restoration (if you don't count the preamp project) to put the orphans to bed.

I poked around on it today and thought I'd get smart first (better late than never). Does anyone know if there is a definitive sticky, tutorial, YouTube vid or other source, either here or on another site, that does a walk-through on chasing these gremlins? I'm talking how to spot the frequency of unwanted noises, circuit checks, measuring ripple at points in the rail, in short, the works. I'm sure someone has done it, I just don't know where. It doesn't have to be free.

FWIW, I have a decent scope, signal generator, DVM and pretty much every tool I should need. If I can't get to the bottom of it, I'm close to tearing out the board and doing it as a straight #102 -- which I've done well before.

Thanks for any direction you can offer, sh

[Structo]

I'm sure more experienced troubleshooters will post but what I do is inject a sine signal (1K Hz) into the input, the scope it at each stage towards the output.

Although if it is heater hum, that could just be in the output stages.

As some of the guys say, divide and conquer.

[VacuumVoodoo]

To start this off:
1. heater hum, if AC powered is 60Hz (or 50Hz)
2. 100/120 Hz hum is in most cases rectification ripple
3. IF when looking at idling output with oscilloscope you see 120Hz spikes on the hum waveform:
a) turn off any fluorescent and LED lamps in the room.
b) any spikes remaining are due to rectifier diodes on-off transients leaking over to heater winding in the PT
4. Are you familiar with the rule of "staying out of forbidden path"? That is the path connecting PT high voltage CT to the 1 st smoothing filter caps (-), or bridge rectifier's (-) to that cap (-). This cap's (-) connects directly to chassis ground, any other ground goes straight to the cap's (-) NOT to chassis. Nothing else gets connected to the wire cap-chassis or cap-CT. That's the forbidden zone.
5. Get Merlin Blencowe's book on Power Supplies. Follow his advice.
6. Train your brain to think in terms of current loops not voltages, it will make it easier to distinguish between dirty and clean ground: Clean ground is where signal ground reference currents flow, dirty ground is where current pulses due to HV rectification and filter cap charging go on the merry go round.

[martin manning]

See under "hum" here: http://www.geofex.com/ampdbug/ampdebug.htm

[Luthierwnc]

I probably need to go a little more basic.

I could use a hand figuring out the settings to tell the difference between 60hz and 120hz. Google images are pretty thin on how to set up the scope and what to look for in the waveform.

I'll read up on the other items as well, thanks, sh

[bal704]

120 Hz is between a Bb and B on the low E string (6th and 7th fret). If you play that note and it matches the hum you're hearing, that's a 120 Hz hum.

I think you can do the same thing for a 60 Hz on the 4th string between F and F#. It will be up an octave or two.

If I'm wrong, somebody please correct me.

[martin manning]

120 Hz is pretty close to the second fret on the 5th string (A string) on a 6-string guitar in standard tuning, 60 Hz is close to the second fret on the 3rd string (A string) on a 4-string bass.

On the scope, the period for 60 Hz is 1/60 sec = 0.0167 sec or 16.7 msec, 120 is 1/120 = 0.0083 or 8.3 msec.

Set the timebase to say 5 msec/div and note how many divisions for a full cycle of the hum waveform: for 60 Hz, 16.7/5 = 3.3 divisions.

Shown below is a 60 Hz waveform from stray EMF (the probe is just hanging in the air), timebase is 5 msec/div. On my scope the Smart Cursors do the math- the readout says 60 Hz, and notice they are 3.3 divisions apart.

[martin manning]

Temporarily disconnecting the filament string from the power transformer and powering it from an external source (batteries or a bench supply) can help isolate hum/buzz issues. The entire string or just a portion of it (preamp tubes, say) can be isolated. If this eliminates or reduces 60 Hz noise then it is likely that the filament circuit is coupling into the audio signal somewhere, which could be related to lead dress or a faulty tube. If it eliminates 120 Hz noise then it is likely that rectifier current transients are coupling into the filament circuit, possibly inside the power transformer.

A half-wave rectified bias supply can also be a source of 60 Hz noise.

Magnetic coupling from the PT to the OT can result in 60Hz hum in the speaker, and depends on the distance between and orientation of the transformers. Removing the power tubes will eliminate all sources except this one.

I'll bet we could create the definitive hum and buzz sticky right here...

[dorrisant]

I'll bet we could create the definitive hum and buzz sticky right here...

Agreed!! Great info!

Tony

[Luthierwnc]

Thanks guys, I'll have a chance to get after it this weekend and will report back. And yes, if we get some more bites on this thread maybe it should be at least a piece of a sticky. Certainly the Dumble amps are complicated enough that there could be multiple culprits for extraneous noise. sh

[Lindz]

I would love to see this thread form out into a sticky that goes into serious detail on how one would chase down hum and other issues. There are already some great posts..

Something like the steps VacuumVoodoo's comment above and or Martins comments but perhaps fleshed out a little more for a someone such as myself with some solid builds under my belt but that still struggles the odd time with noise or problems that I sort of have to shoot in the dark to fix. I just bought a scope and signal generator, but still do not really know how to best use them in such a capacity.

For example there is one set of 10 videos "the hunt for buzz and hum" on youtube that was somewhat helpful but the guy that posted it essentially starts with a noisy amp and goes through a bunch of tests and analysis in the 10 videos yet finally just ends up just adding a choke at the end to quiet it down.. not exactly rocket science, nor much insight into a more nuanced problem like a ground loop or other noise, but it was the most on topic thing I could find and still helpful

http://www.youtube.com/watch?v=SUBk_ZQPLDI

A few more detailed posts on how to chase down certain problems, or even better some video like the one I mention going through this various debugging situations would be invaluable to me and likely to others. Better yet if it shows some info on how the scope and or signal generator was set and used for the various steps. I realize this is kind of "electronics 101" for many people here that start with an electronics background but it could be really helpful to many here

[Structo]

We could make a sticky for troubleshooting.

Although one thing I fear is that it might be just enough information to get into trouble for an inexperienced person who doesn't understand a little bit of electronic theory and/ or high voltages.

But we could do the big Disclaimer thing to absolve any liability.

[Luthierwnc]

Got downstairs tonight. The noise is a shade flat of a B so I guess that makes it 120 hz. Here is a shot of the scope screen set at 2 msec per division. I have the probe right on B+2 with no input signal. The waveform looks the same further down the rail -- just smaller.

Next time I get into the shop I'll triple-check the power supply.

Thanks, sh

[martin manning]

Which B? B2, 2nd fret on 5th string is 123.5 Hz. The scope trace looks typical for B+, but at 2 msec/div I'm seeing three cycles in about 17 msec, or 176 Hz. Is there an adjustment knob on the timebase that is unlocked perhaps? What's the vertical scale V/div (i.e. how much ripple voltage is there)?

[Luthierwnc]

Here is a new shot from this morning. I used the SWP VAR calibration to stretch the pattern out to approximate 120 hz. At least, if my math is right, 8.3 ms for 120 hz divided by 2 ms per division should come in around 4 and change vertical lines.

You can't tell in this shot because of parallax but the volts/div is at .1 Again no signal with the volume about half up and the MV at 1/3. It makes the sound even with the input jack grounded.

The B was fifth string, second fret.

However incorrectly the scope is calibrated, my next foray will be to make sure all the power caps are behaving. It is the more complex Dumble power supply with series caps and resistors for each node. This reading is from right after the choke. At the first preamp tube the wave form looks the same but about half the size horizontally.

Cheers, sh