76 Twin Reverb: grounding and fighting 100hz hum

[Lambertini]

Hello!

In the circuit of my 76 Twin Reverb, Fender used a huge brass ground plate below the pots that connects many grounds from the pots and the circuit.
This leads to about 10 different ground points to the chassis in the circuit.

On the other hand, I've read most amp builders recommend having 2 separate ground points to the chassis, one for "power ground" and another for "preamp ground".

Looking at the Twin Reverb AB673 circuit proposed by Rob Robinette, you can see these 2 ground points: one "star" ground on the left for power grounds, and a circuit ground bar joining the chassis at the opposite side, through the normal channel input jack.

What are the advantages of each approach? Which one reduces hum the most?

[tdvt]

Start here:

https://www.valvewizard.co.uk/Grounding.html

1001 opinions on grounding out there. Read Merlin's take & go from there.


Keep in mind that Fender was a production line, which undoubtedly dictated many of the methods/techniques that they employed based on time, skill & $$.

[mresistor]

Is your twin producing some hum?  Have you used the humdinger (balance) pot on the back for minimum 60hz hum?  If it has hum is it 60Hz or 120Hz? 
Why do you want to modify the vintage Twin?   


I've had many original 60s and 70s Fender amps on the bench that all had the brass plate under the pots and almost every one has been very quiet with no hum.
I don't think the brass strip is causing hum in your amp. I do always use a magnifying glass and check every single one of the solder joins on the brass plate for cracks and reflow if necessary, which requires a
large 85W soldering iron, in my case.

[Lambertini]

Is your twin producing some hum? [...] If it has hum is it 60Hz or 120Hz? 

Yes. Mainly an annoying 100hz (see pic) and a fairly small amount of 50Hz. (230V 50Hz from the wall here).

Have you used the humdinger (balance) pot on the back for minimum 60hz hum?

I completely rewired the heaters and changed the hum digger pot. The remaining amount of 50Hz is rather low. Would be happy to remove it more (like dead quiet), but it's no more an issue.

Why do you want to modify the vintage Twin?   

Because it was a dirty and noisy piece of crap never serviced for almost 50 years while heavily played. Even the chassis is warped. Plus someone has done few dirty mods, including adding ground wires that caused RF. I don't trust this amp at all.

I do always use a magnifying glass and check every single one of the solder joins on the brass plate for cracks and reflow if necessary, which requires a large 85W soldering iron, in my case.

I've already reflowed all solder joints to chassis except the power tube ones.

[HotBluePlates]

... On the other hand, I've read most amp builders recommend having 2 separate ground points to the chassis, one for "power ground" and another for "preamp ground". ...

Most of the crap you read on "grounding" is pure crap.

+1 for the recommendation to read Merlin's info on grounding.

In the circuit of my 76 Twin Reverb, Fender used a huge brass ground plate below the pots ...

Do you actually have hum in your Twin Reverb?  If yes, there are like 10 different things that could cause that.

That said, sharp techs start work on a vintage Fender amp by cleaning it.  Cleaning the tube socket contacts, cleaning the jacks and switched contacts, cleaning pots.

And a key move is removing all knobs & pots from the front panel, and lifting the brass grounding strip.  Thoroughly clean both the brass & the chassis with a scouring pad & solvent so there is zero chance for oxidation to live between these 2 pieces of metal.  There are a lot of "imperfect grounds" and so a lot of hum-causes that get cured by this one act.

[mresistor]

Have you thought about replacing the PS filter caps? they are quite old now.  Are the filter caps the silver mallorys? 100hz hum is probably coming from the power supply.

[Lambertini]

Have you thought about replacing the PS filter caps? they are quite old now.  Are the filter caps the silver mallorys? 100hz hum is probably coming from the power supply.

All caps are brand new.

For sure the 100hz is generated by the rectifier. If I clip some extra caps in parallel of the primary filter caps to double the capacitance, 100Hz drops bellow -60dB on my spectrometer app. But I don't think it's a filter cap issue. Tested several ones, same results. I'm suspecting grounding and choke. I've ordered a new choke to see if it fix.

I'm still concerned by these multiple grounds. Why not having something like this? (orange = to be deleted ; green = new connexions)

[tdvt]

All caps are brand new.

Does that include the bias filter caps?

Hard to tell from the pic.

[Lambertini]

Does that include the bias filter caps?

Any single cap has been changed: main filter caps, secondary filter caps, bias, coupling, cathodes, TMB, bright, tremolo, all.

Thanks a lot for the link to the PDF on grounding, it answers many of my questions, including the fact that Fender factory grounding was pure crap ;) At least the rest of the circuit delivers one of the best tones of all time!

Do you think this would partially match the recommandations, removing things hashed in orange, adding connexions in green?

[proaudioguy]

Wow, I think you did it backward.  Keep all the old tone caps and change all the old resisters to the same or larger wattage metal film to get rid of the noise.  The grounding scheme you propose may help a tiny bit but cleaning will probably help just as much.  If its that loud, you must have really sensitive speakers.  I have a Twin with a modern grounding scheme and its deal quiet.  I have a few other amps with the brass plate and they are pretty close.  Of course, the more sensitive the speaker is, the louder all the noise appears to be.

[mresistor]

JBL and Electrovoice speakers will definitely let you hear noise better..   


That one orange hashed black wire is the PS filter ground isn't it? That might be better going to a PS grounding point if it is.


You may want to add 1 ohm 1w 1% resistors to the power tube cathodes to ground for easy P Diss biasing and checking each individual tube in biasing.

[HotBluePlates]

All caps are brand new.

... If I clip some extra caps in parallel of the primary filter caps to double the capacitance, 100Hz drops bellow -60dB on my spectrometer app. But I don't think it's a filter cap issue. ...

Your test of adding extra capacitance actually proves the issue is insufficient capacitance.

Maybe one or more of the new caps isn't full-value, or has a poor ground connection.

[Lambertini]

Your test of adding extra capacitance actually proves the issue is insufficient capacitance.
Maybe one or more of the new caps isn't full-value, or has a poor ground connection.

I've tested these caps with a DMM and both have the right capacitance :'( I'm going to rewire the ground as this is something I haven't done yet.
Wouldn't it be possible that "something" is causing so much 100Hz that correct capacitors can't cope with it?
Or maybe further filtering steps don't do it right?
I'm strongly suspecting the choke right now... Will follow up once ground is redone.


BTW, I've removed the superflous ground wires from the doghouse, and rewired the preamp grounds to a single bus, with a single point of contact on the chassis at the normal chanel input 1.
It's totally uggly compared to the photos of some ultra neat ground buses I've seen on the forum, but it does the job for an old grandpa like this Twin.
The already acceptable 50hz is now absolutely gone. Unfortunately no effects on the annoying 100Hz :/

Still have to rewire the tremolo pots grounds but when I attempted to solder them to the same bus as the preamp grounds, the 100hz hum got a 10db boost. This is a trail to follow.

[Lambertini]

That one orange hashed black wire is the PS filter ground isn't it? That might be better going to a PS grounding point if it is.

That's what I tried to do, one single ground point for:
- HT Center Tap (orange hashed red)
- Primary filter caps ground
- Bias caps ground

Something bad here?

[Esquirefreak]

Since it's tied to a transformer bolt, make sure that the metal surfaces are clean and the keps nut is tight.

/Max

[mresistor]

I don't think I've ever heard of a choke causing spurious 100/120Hz hum., could be induced I guess - I've only heard of them having insulation breakdown with resulting shorting out.


What is this wire? would it be better to route it directly back to the back wall of the chassis then along to the switch?

[Lambertini]

What is this wire? would it be better to route it directly back to the back wall of the chassis then along to the switch?

The universal power transformer has several primaries (110, 200, 220, 230, 240V) that are normally selected with the red selector at the back (you can see the pins at the top right of the pic). But the selector is completely dead, and primary wires were cut too short to reach the power switch. The white wire is an extension to connect a primary to the switch.

I don't think I've ever heard of a choke causing spurious 100/120Hz hum., could be induced I guess - I've only heard of them having insulation breakdown with resulting shorting out.

Aren't chokes used to filter more by forming a LC filter?

[mresistor]

Yes chokes are used as filters - in this case they are smoothing/filtering the supply voltage to the screens, it is also more resistant to voltage fluctuations than a resistor.
Yet, if the choke isn't shorting out one would think its functioning properly.


If it were my amp I'd route that white wire the way I indicated and not flying over the top of other components.

[shooter]

I've tested these caps with a DMM and both have the right capacitance

caps tested with no load don't really tell you much


monitor with the amps load, see if voltage "sags".  scope to see how much ripple, ain't getting filtered, slap a new cap in parallel, see if the ripple goes down noticeably
replace as required.

[eleventeen]

As long as you are testing "everything" and doing plenty of work along the way, I might try installing the dual-100 ohm resistor synthetic center tap on the heaters versus the power trans center tap. If that's your present config.


All I can say is that if the halves of the PT are very unequal in terms of the volts delivered to the rectifier, some amps will hum like a wounded rhinoceros. Loud, gross, intolerable. If the amp has a PT center tap, it may be that a prior tech installed the synthetic center tap and ONE of the resistors is open.   

[Lambertini]

As long as you are testing "everything"r tech installed the synthetic center tap and ONE of the resistors is open.   

I'm not testing everything, just trying to get rid of the 100hz hum which doesn't come from the heaters.
Tried to arrange the grounding with no success. Now I'm stuck.
Thank you anyway.

Maybe one or more of the new caps [...] has a poor ground connection.

I've rewired the primary filter caps ground and redone all solder joints at the PS ground, no success.

I think we can conclude that this 100 hz is not a grounding issue.

monitor with the amps load, see if voltage "sags".

It doesn't.

scope to see how much ripple, ain't getting filtered

Don't have one.

slap a new cap in parallel, see if the ripple goes down noticeably

Yes it does. See this post.

replace as required.

Done with 3 differents sets. Same result.

[pdf64]

… BTW, I've removed the superflous ground wires from the doghouse, and rewired the preamp grounds to a single bus, with a single point of contact on the chassis at the normal chanel input 1.
 …

On the face of it, that seems a bad idea.
It may be beneficial to show exactly what you’ve done there.

[Lambertini]

On the face of it, that seems a bad idea.
It may be beneficial to show exactly what you’ve done there.

You have the pictures of the preamp ground bus in the message. Here's the pic of the regrounded doghouse, getting rid of 2 useless ground looping wires.
The whole move looks like it was actually a good idea as it reduces the 50Hz. Unfortunately did nothing for the 100Hz, which was the goal.

[AlNewman]

So did you only notice the hum after replacing the filter caps?

[Lambertini]

So did you only notice the hum after replacing the filter caps?

No

[sluckey]

Output tubes are another source of 100Hz hum. The tubes need to be matched and the bias balance properly set for maximum hum cancellation.

[pdf64]

On the face of it, that seems a bad idea.
It may be beneficial to show exactly what you’ve done there.

You have the pictures of the preamp ground bus in the message. Here's the pic of the regrounded doghouse, getting rid of 2 useless ground looping wires.
The whole move looks like it was actually a good idea as it reduces the 50Hz. Unfortunately did nothing for the 100Hz, which was the goal.

Just to note that unless a rectifier is bad, there should be no 50Hz in the doghouse.

I’m not keen on the screen grid node filter current being forced to share the same 0V path as that if the preamp node filter.

[Lambertini]

Just to note that unless a rectifier is bad, there should be no 50Hz in the doghouse.

The 50hz is so low I can't hear it. The problem is the 100hz.
If I clip parallel caps to the reservoir caps to increase capacitance, the 100hz drops belows -60db (still a bit annoying but acceptable). But it's equivalent to 2x200µF caps, which seems to be very high values for a Twin (original AB763 is 70µFx2, silverface is 100µFx2).
Ideally reaching -70dB would make it barely audible.

[stratomaster]

scope to see how much ripple, ain't getting filtered

Don't have one.

slap a new cap in parallel, see if the ripple goes down noticeably

Yes it does.

The above points to a filtering problem.  You don't need a scope to quantify it.  Switch your DVM to AC volts and probe each node.  And you'll see the RMS value of the ripple.  On a healthy amp expect to see 3-8v on the OT CT node,  1v or so on screens, and less than 1v for the remainder. 

[Lambertini]

Switch your DVM to AC volts and probe each node.

Unfortunately mine just displays 0.000V :/

[sluckey]

Maybe you missed this? Anyhow, it's worth mentioning again.

Output tubes are another source of 100Hz hum. The tubes need to be matched and the bias balance properly set for maximum hum cancellation.

[Lambertini]

Maybe you missed this? Anyhow, it's worth mentioning again.

Output tubes are another source of 100Hz hum. The tubes need to be matched and the bias balance properly set for maximum hum cancellation.

Sorry, I've seen your message and just forgot to reply.

I've installed a new balanced quad of TAD 6L6 small bottle 30W, bias balanced at 0.1mA between both pairs, 35mA, same 100Hz. Did the same with a new balanced quad of Tung-Sol 6L6 35W, no effect.

I also replaced the 50 years old, crackling and unpredictable "center tapped" bias balance pot (4 lugs) with a new regular bias pot (w/o center tap, 4 lugs ones are not available anymore) and retrofit bias circuit to AB763. Bias was then slightly unbalanced, 55mA / 54mA (70% dissipation), which seems ok to me, no success either.

I also tried colder bias, the 100hz doesn't significantly decrease unless bias is unusably low (<20mA).

***

I would also like to spot on one thing. I have a 3-way switch to select NFB resistor between open circuit (no NFB), 820ohm, 1500ohm.
- With no NFB, 100Hz hum is the highest -38dB.
- 1500 ohm NFB, 100Hz is -45dB
- 820 ohm NFB, 100Hz is -50dB

Also if I remove V6, 100Hz is at -57dB.

So a relatively mild level of ripples are there after the reservoir caps and before the choke, leading to an almost acceptable level of 100hz hum when feeding OT and plates, but also remain in DC after passing through the choke and the 3 others filter caps to preamp sections, getting amplified.
One trail I didn't explore at this time is to change the (yet new) 22µF caps.
As said earlier choke is also in my target.

[stratomaster]

Pulling the PI cleaned up the noise a good bit, so you know you have some level of noise induced in the preamp.  In Fenders I find the reverb recovery and final gain stage to be the noisiest of the bunch.  See what pulling v4 does for you. 

If you are unable to measure/quantify the ripple how can you make statements about how much ripple there is and where? 

The power tubes rely on common mode cancelation to negate the ripple in the supply at that node.  That's why a mismatch can lead to increased ripple noise. 

I think you should lift the dropping resistors in the power supply circuit.  That should help you isolate noise. 

Your chassis grounds in the photos look messy at best.  You may need a higher power iron with more aggressive flux to get those to be reliable connections. 

[Lambertini]

If you are unable to measure/quantify the ripple how can you make statements about how much ripple there is and where? 

100hz = amount of ripples, I can measure 100hz via a spectrometer app.

[stratomaster]

Are you taking these measurements with the chassis fully shielded or out on your workbench.

[Lambertini]

See what pulling v4 does for you. 

Pulling V1 to V5 has no effects at all, pulling them all neither.

Are you taking these measurements with the chassis fully shielded or out on your workbench.

Both, but it doesn't seems to change the 100Hz.
Pic1 amp closed, pic2 chassis on the bench.

I think you should lift the dropping resistors in the power supply circuit.  That should help you isolate noise. 

You mean unsoldering an end and lift the dropping resistors in the air to measure the effects when cutting off power to these parts of the circuit?

[shooter]

Pic1 amp closed, pic2 chassis on the bench.

pic 2 1 seems to have a "ringing" component that isn't there in 1

[stratomaster]

You really need to find a way to get a scope or at least a good RMS meter.  You can't find something you can't measure. 

And yes, I was talking about cutting power to different parts of the circuit by strategically lifting the resistors. 

[shooter]

You can't find something you can't measure.

had a guitarist say "I swear I hear something", scoped the amp, borrowed the spectrum analyzer from work, nothing!
told him If I can't see it, it doesn't exist.

[Lambertini]

told him If I can't see it, it doesn't exist.

Can you see this 100Hz?

You really need to find a way to get a scope or at least a good RMS meter.

I'm gonna try to find a way to measure ripples at various points to precisely locate the problem.