Dr Z Stangray - oscillation, and lead dress voodoo

[stratomaster]

I'm working on a Dr Z Stangray that was in for a few upgrades, issues, and a "strange alien spaceship sound".  I tried to replicate the sound a few different ways but was unsuccessful so I proceeded with the upgrades:

-Cap Job (amp was 17 years old)
-Backup diodes to gz34
-Move Standby switch to after 1st filter node
-Dual gang pot ppivm
-Inline screen dropping resistor (1.3k)
-Replace damaged power cable and rewire switch to line instead of neutral
-Dedicated safety ground
-Elevate heater CT to cathode voltage
-Cool bias to EL84s, but not by much

Prior to this the amp was eating rectifier tube and EL84s.  The standby was hot switching the rectifier to the first filter node and the EL84s dissipated north of 16w each. B+ ~345V depending on the incoming feed.

All of the incoming power rewiring went well, as did the inline screen dropper, cooling the bias, heater elevation, and MV install.  Tested functionality after each change.

The cap job and standby rewiring required the pots and jacks to be removed from the front panel to lift the board for access.  Voltages in the amp and standby tested well after the completion of the job. 

However, that's when I got the "spaceship" sound.  It was oscillating when the volume pot was at 0.  The oscillation frequency responded to pressure on the ground bus wire by changing pitch.  The amp is wired for high and low inputs like a Fender with 2x68k and 1M at the jacks.  A shielded wire from the junction of the 68ks feeds the EF86 grid. 

I can move the 68ks into a position that minimizes the oscillation at 0 volume, but this usually results in oscillation at high volumes.  There is a very narrow window of lead and resistor positioning that minimizes both.  With the addition of the PPIVM, the full range of the preamp Volume control is fair game and needs to function correctly. 

Before I dig into the voodoo, I'm seeking guidance.  My first thought is to replace the shield wire with RG316 and run it along a different, direct path with a 10k grid stopper at the tube socket.  Aside from a marginal drop in high end, is there a side effect of having the grid stopper essentially distributed apart like that with the 68ks on one end and a 10k at the socket?  Alternatively I can build the voltage divider on tagstrips at the EF86 and just land them back at the jacks with shield wire. 

Any other thoughts or suggestions?

Thanks.

[tubeswell]

The spaceship sound was something you did. Either look closely at the PPMV mod or undo it

[tdvt]

The oscillation frequency responded to pressure on the ground bus wire by changing pitch.

That is the optional Theremin feature...

[Dave]

The spaceship sound was something you did. Either look closely at the PPMV mod or undo it

My understanding of the original post was that the alien sound was one of the symptoms for which he was asked to work on the amp to start with.


Dave

[tubeswell]

The spaceship sound was something you did. Either look closely at the PPMV mod or undo it

My understanding of the original post was that the alien sound was one of the symptoms for which he was asked to work on the amp to start with.


Dave

Ah yes, on a 2nd read I see that now. Intermittent sound.


May be mechanical hairline crack in jumper/lead or dodgy load resistor - chopstick the circuit.


Or could be bad tube or heat related component failure

[stratomaster]

The spaceship sound was something you did. Either look closely at the PPMV mod or undo it

My understanding of the original post was that the alien sound was one of the symptoms for which he was asked to work on the amp to start with.


Dave

Correct.  I can replicate it now.  It has to do with lead dress at the input jacks.  Moving the shielded wire and/or the grid stoppers at the jack makes it worse/better.  I don't trust the components to remain the the safe window so I want to address the problem directly.  This led me to new wire, new route, and grid stopper at the socket instead of jack. 

The MV just controls the volume of the oscillation that is happening upstream. 

[acheld]

This led me to new wire, new route, and grid stopper at the socket instead of jack.

Makes sense.  I would do something like that . . .  This sort of symptom can be a bear to sort.

[pdf64]

A friend’s EZG50 was randomly flipping into a shrieking oscillation.
I discovered the cause to be the multiple, poorly established connections between the amp’s 0V common and the aluminium chassis. The design was using the chassis as an 0V conductor.

[stratomaster]

A friend’s EZG50 was randomly flipping into a shrieking oscillation.
I discovered the cause to be the multiple, poorly established connections between the amp’s 0V common and the aluminium chassis. The design was using the chassis as an 0V conductor.

Yes. This thing is a mess in terms of grounding and power distribution wiring.  It's lucky it sounds great when working properly, otherwise I'd advise the owner to get something else instead of having me bat cleanup. 

[pdf64]

The preamp relies on the wire across the back of the pot shells, and the pot bodies electrical connection to the chassis, for its 0V common return.
The Alpha pots used couldn’t be torqued up much.
The oscillation was caused by every pot losing its connection.
The slightest vibration or tap could send it in and out of oscillation.

My fix was a new chassis connection point for the preamp, using a high torque fastener and star washer.
Not ideal but it’s worked fine.

[stratomaster]

The preamp relies on the wire across the back of the pot shells, and the pot bodies electrical connection to the chassis, for its 0V common return.
The Alpha pots used couldn’t be torqued up much.
The oscillation was caused by every pot losing its connection.
The slightest vibration or tap could send it in and out of oscillation.

My fix was a new chassis connection point for the preamp, using a high torque fastener and star washer.
Not ideal but it’s worked fine.

I'll give this a shot.  I noticed the chassis holes are 3/8" clearance, and the Alpha pots are metric...    So all of the stress from torquing the pots down goes into dishing the washer. I replaced the washer on the Cut control with something more stout. Shafts are too short for lock washers, though.  The Volume control threads were damaged in the process of me finding all this out, so I've replaced it with a CTS with a longer 3/8" shaft.  I think I'll add conductive lock washers to it and use as the main preamp ground to start.  Keep the original intent, but improve the execution.  If that doesn't help, then I'll resort to a dedicated ground. 

I'm anticipating the need to isolate the input jacks from the chassis too in an attempt to deal with the oscillation/ground loop at 0 Volume. 

I also noticed the EF86 pins 2&7 are tied to the cathode (3&8).  While this might work on paper, I think a dedicated 0v for 2&7 would likely give better stability.  Is it common practice on these to tie the shield to cathode?

[PRR]

> EF86 pins 2&7 are tied to the cathode (3&8).  While this might work on paper, I think...

It's on the paper used to $ell the tube. If the makers had any doubt they would leave it to designers to try-and-error. I say: don't be like my dogs, barking after every squirrel-fart in the woods. **** Sorry, I thought "2&7" was G3 Suppressor. If the cathode is WELL bypassed, 100uFd, then Shield to cathode may be OK. Since G-amps like partly bypassed cathodes, shield to cathode may not be a good plan.

You do seem to have two problems. The shielded input may be a good plan. The spaceship seems more like a bad connection, as the others are saying. Probably there from before, even from the factory. It may "find a happy place" for years; I had a big tube amp and also a PC monitor with bad joints that worked well for a long time. (The tube amp finally caught fire, on the 43rd floor in Manhattan...) If not your fault it may be very hard to find. Look anyway, then run a bus as close as possible to the existing ground path but soldered RIGHT and terminated on TOOTHY washers/screws.  I don't understand why you'd want to bond the pots and lift the jacks. Crap more likely comes in at jacks, pot grounding should be totally secondary.

[tdvt]

Are there really that many pot grounds to re-route to avoid using the chassis at all as a ground plane for the controls?

I have had pretty good results using an isolated buss wire behind the pots & include the isolated input jack grounds on that same buss. Then one ground wire to a ground point on the PT end.

Doesn't really seem like much extra work... eliminates reliance on all those potentially iffy connections.

The grounded pot cases are then just shielding.

Pretty sure Doug has the metric pot to 3/8 adapter rings to improve the fit of those pots in the larger holes.

[stratomaster]

There aren't too many pots.  I've attached a photo post top-side cap replacement, screen dropper added, bias cooled, and MV install.  The power and standby (and can cap ) have been completed since this photo was taken.  You can see the ground buss on the backs of the pots.  All circuit grounds go to this bus which leads to a chassis ground on the far left. 

The input jacks are also visible with the 68ks, 1M, and factory shielded cable to the EF86 grid peaking out from the chassis on the top right.  Moving the junction of the shielded cable and 68ks around affects the oscillation. 

[tdvt]

Not many pots at all..

I often refer to Sluckeys pages for methods & ideas.

In the linked pic, check out the control buss wire running behind the pots, over to the jack. Not sure if the black wire is an isolated ground to the PT or not. Maybe he can confirm.

Anyway, this method was one of my inspirations towards managing ground loops.

https://sluckeyamps.com/6v6plexi/P-6V6_05_big.jpg

[stratomaster]

>  I don't understand why you'd want to bond the pots and lift the jacks. Crap more likely comes in at jacks, pot grounding should be totally secondary.

I don't want to do that, but there's not enough thread on the input jacks to add a toothed washer and torque it enough to make it reliably the preamp ground point.  The replacement CTS  volume pot has plenty of thread to accommodate a toothed washer.  Otherwise I'll need to replace the input jacks or drill a hole near the input for a toothed ground lug. 

Haven't been able troubleshoot with these tips yet as I'm awaiting a shock-mount retrofit for the EF86.  Once this thing is done it'll be safe, reliable, free of unwanted noise, small stage friendly, less tube hungry, and road worthy.  Unfortunately it was none of those things to start out. 

I think the strategy will be to break the buss along the backs of the pots between the volume and tone pots and tie the grounds for the input jacks, EF86 shields (new wire to 2 & 7), preamp cathode cap and resistor, and lug 3 of the volume pot to the pot casing which will be torqued with an adequate toothed washer to the chassis.  This seems the least invasive way to break the loop.

I'm hoping this takes care of the oscillation, but if it doesn't I'll proceed with the RG316 and add a 10k grid stopper at the socket.  Since I need to remove the socket for the shock-mount installation I'll redo the lead dress in accordance with the practices laid out in the document PRR attached   I might even have a socket with the central lug in my stash. 

Edit:  I added a schematic and layout I found on the web.  They don't agree with each other nor the build, but they're close enough to be serviceable.  Looks like I'll need to break the connection between PI and Preamp filter caps too and run the preamp over to the input area.  In which case a dedicated hardware ground with a fresh hole is the only way to go.  Can't ground a filter cap through a volume pot. 

[tdvt]

>  I don't understand why you'd want to bond the pots and lift the jacks. Crap more likely comes in at jacks, pot grounding should be totally secondary.

I think the strategy will be to break the buss along the backs of the pots between the volume and tone pots and tie the grounds for the input jacks, EF86 shields (new wire to 2 & 7), preamp cathode cap and resistor, and lug 3 of the volume pot to the pot casing which will be torqued with an adequate toothed washer to the chassis.  This seems the least invasive way to break the loop.

If your amp is fairly close to the drawings, it doesn't look like that much work to take the chassis out of the equation completely.

A few sets of isolation washers (on all the jacks) & a couple new wires from the input buss & from the output jacks, over to the terminal strip ground (cap can, CTs, etc.)

Then use a Merlin Hum-blocker to actual chassis ground, if needed.

I have had good success in my limited experience, approaching things as though the chassis is non-conductive & implementing intentional ground paths. 

[stratomaster]

I think the attached might get me there with one wire.  But if not I'll experiment with these great suggestions.

I've also realized why they're routing the shielded wire to the grid in such a tortuous path and why the jack mounted 68ks are so sensitive to position.  Combination of very high gain from the EF86 and the placement of the choke.  Since the chassis is aluminum I bet there's an insane field coming off that choke polluting the exposed connections at the jack and flying them while positioning them in the lowest oscillation placement was the solution at Z factory. I've included a picture of the wire as it was before I did any work near it.  They even "secured" it in place with electrical tape.

I'm tempted to put a thin piece of grounded flashing or copper foil on the bottom of the enclosure interior to see if anything changes.

[Willabe]

Have you read this on grounding;

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

[PRR]

> 68ks are so sensitive to position.  Combination of very high gain from the EF86 and the placement of the choke.

I had not seen that. Really begging for buzz. One choke lead has 50V of 120Hz on it. Also some fraction of the large signal in the output stage. It is mounted righ above the most sensitive part of a hi-gain amp. I don't think the chassis metal is a real problem, it's bad layout.

Disconnect both choke leads and replace with a 100r 10W resistor. Is that better?

[stratomaster]

> 68ks are so sensitive to position.  Combination of very high gain from the EF86 and the placement of the choke.

I had not seen that. Really begging for buzz. One choke lead has 50V of 120Hz on it. Also some fraction of the large signal in the output stage. It is mounted righ above the most sensitive part of a hi-gain amp. I don't think the chassis metal is a real problem, it's bad layout.

Disconnect both choke leads and replace with a 100r 10W resistor. Is that better?

The good news is I only measured about 7v AC RMS on the input to the choke--beefy 50uF reservoir cap.  The reason I mentioned aluminum is because I think the induced oscillation is electromagnetic from the choke body rather than from the input lead.  Though I agree the layout is poor. 

I experimented with different shielded wire, lead dress, and copper tape.  I landed on RG316 routed to the extreme upper perimeter of the chassis and terminating to the grid stoppers on the panel side rather than circuit side.  I added a strip of grounded copper tape across the tops of the plastic input jacks to provide a bit of a shield for the connection to the 68ks themselves with a pad of battery insulation paper to keep the connection from shorting.  This seems to have alleviated any oscillation at the remaining natural/possible positions of the grid stoppers, and no electrical tape necessary.  See attached image.

However the grid wire is now extremely sensitive to movement.  I know this is normal but I've never experienced it to this degree--especially with a shielded cable.  It's almost like I need to use patch cable with the conductive triboelectric layer, but I can't see how this would work as the shield is only connected at the input end.  From the factory they did use two layers of double sided foam tape to hold the shielded wires sandwiched at the sockets.  I'm thinking this did double duty of positioning and vibration damping where the cable meets the chassis.

[stratomaster]

Added some padding to the wire from input to grid and rewired the EF86 socket to have a dedicated shield and twisted heater wires well away from the grid and the effects loop jacks.  Added some small orings under the tube socket fasteners to act as shock dampeners.  The Weber retrofit needs to be better thought out, so I adopted the Matchless approach. 

There some microphony on the EF86 that rings at about 5k.  It persists with stout orings adding mass to change the resonant frequency of the tube.  My understanding is that these tubes are notorious for high frequency microphonics and not much can be done about it unless you're willing to drop the gain of the stage and add a gain stage at the input to keep the overall amplification about the same.  That's outside the scope of this repair.