Only 1 (of 2) screen resistors keep burning up.

[munkeyboy]

I'd like to learn better diagnostic methods instead of replacing components willy-nilly.   What steps would you all recommended to investigate this issue? I'll work on gathering normal voltage readings and post back.

History:  I rebuild a Traynor yba-1 into a plexi 50w (two el34).  It came to me many years ago in a bad state.  Burn sockets, burnt components, missing components and some unknown mods.   So I rebuild it and sold it after I found it too loud for me. It has enjoyed a home for a few years as someone's favorite amp.   The guy contacted me a few months ago because it started tripping the circuit.   On opened it up, i found one of the screen resistors burnt.   No traces on the socket, but the tube was bad (red plate).   So I assumed the tube took out the resistor and I replaced both tubes and screen resistors.  I then biased, tested and sent back home.   It came back pretty soon for, what sounds to be oscillation when channels are jumpered and bright channel turned up.   I opened it up and noticed the burnt resistor is back.  Same tube socket.   The owner also uses an attenuator (tone king i think), but if that was a problem, I would think you would see the issue on both sides.   Currently the tube is not red plating.  Makes me wonder about the OT.   Makes me wonder if that is the cause of the original issue before I owned it... tho it worked well for two years after.

[sluckey]

The owner also uses an attenuator

Bingo! I bet he's running the amp dimed. It's also possible to have a defective attenuator. Just read a discussion that turned out to be a bad attenuator.

Maybe double the wattage of the screen resistors. And tell him that he's burning up his amp.

[tubeswell]

EL34s are prone to screen grid failure in Plexi type circuits, especially if you run them flat out through an attenuator all the time. I'll have to admit, the most alluring thing is they sound incredibly fantastic just before they die.

[munkeyboy]

Interesting.  I hear both sides of the argument.  Attenuator=bad or Attenuator=harmless.  But yeah, running the amp flat out might do the trick (and have nothing to do with the attenuator).

However, the guy says he only runs it 30% up the volume and he barely played it before bringing it back.   To be fair, it could have been me and the 5 minutes I was testing thru an attenuator at near 100%.   But why would just one side be burning up tho and the same side both times.

BTW, plate/screen voltages look normal at 420/413 respectively.  Both sides.

[sluckey]

It is not uncommon for the OT to have slight differences on each half of the primary. Or, it's possible that the OT has been slightly compromised.

[tubeswell]

But why would just one side be burning up tho and the same side both times.

Or maybe a coincidence that in each case the naturally 'hotter' tube was on the burny side? As you supply more current, the current will 'prefer' a naturally hotter tube (making it hotter'n'hotter)

[pdf64]

Also it's typical for the overdriven signal to the power valves to be asymmetric, resulting on one side being 'on' longer than the other, causign increased wear / dissipation on that side.
But worth checking that the socket contacts are clean and tight, and that for that control grid, the coupling cap to it isn't leaking ANY DC, and that its grid leak resistor and connections are good.
Note that at power up, the coupling caps to the power valve control grids may have full HT + bias surge voltage across them, perhaps over 500V. My view is that's not a good place for vintage caps, especially 400V types; too many times, when I've actually checked, a tiny bit of DC is getting past them, changing the bias a little. Hence by default I check them or just replace them with modern 600V types.

[tubeswell]

Also it's typical for the overdriven signal to the power valves to be asymmetric, resulting on one side being 'on' longer than the other, causign increased wear / dissipation on that side.

Yes - it is most likely this if the amp is always being run flat out through an attenuator.


https://el34world.com/Forum/index.php?topic=22271.msg237418#msg237418

[munkeyboy]

Also it's typical for the overdriven signal to the power valves to be asymmetric, resulting on one side being 'on' longer than the other, causign increased wear / dissipation on that side.
But worth checking that the socket contacts are clean and tight, and that for that control grid, the coupling cap to it isn't leaking ANY DC, and that its grid leak resistor and connections are good.
Note that at power up, the coupling caps to the power valve control grids may have full HT + bias surge voltage across them, perhaps over 500V. My view is that's not a good place for vintage caps, especially 400V types; too many times, when I've actually checked, a tiny bit of DC is getting past them, changing the bias a little. Hence by default I check them or just replace them with modern 600V types.

Good points.  Every resistor, cap, socket, fiberboard is new within 3years.  500v filter caps.  No vintage caps, but doesn't mean something hasn't failed.  I will check the leakage to the best of my ability (just a dmm, not leak tester).

I am not happy with the socket pins and will retention them.

[munkeyboy]

I also wonder if I shouldn't be using the metal oxide 1k screen R for this or if they are just crappy product I got from amazon.   Are wirewound typically better for this application?

[DummyLoad]

use a quality 2W WW or MF resistor. i wouldn't touch anything amazon sells for use in a tube amp, except for LEDs, relays, etc. - the low volt stuff. 


--pete

[munkeyboy]

use a quality 2W WW or MF resistor. i wouldn't touch anything amazon sells for use in a tube amp, except for LEDs, relays, etc. - the low volt stuff. 


--pete

2W not 5W?    Also, what is a good method to get the amount of current thru that resistor.   I measured across the screen resistor and of course at idle the current is steady.  But playing a guitar thru it the current fluxuates.  Regardless, both sides seem about the same.  At idle with volume at 50% and channels jumpered = ~4mA.   Playing hard, it fluctuations over 20 or more mA.

edit:
If reading correctly 4mA @ 420v = 1.68w.   So 20mA or more would be bad.. but maybe not because it isn't constant?

[sluckey]

Measure the voltage across the resistor. Then use Ohm's law to calculate current.

[munkeyboy]

Ok that makes sense.   So at idle, that is ~5v, with the max 35v playing hard volume maxed.  That still low watt.   I'm not seeing anything that would cause an issue, but not sure I'm capturing peak voltage.

[pdf64]

... Playing hard, it fluctuations over 20 or more mA.

edit:
If reading correctly 4mA @ 420v = 1.68w.   So 20mA or more would be bad.. but maybe not because it isn't constant?

That would be the dissipation of the screen grid itself. It’s a good idea to check it.
At full load - playing hard, you’ll probably find that the screen voltage drops somewhat below its 420V idle level.
Remeasure and reassess.

[tubeswell]

I'm not seeing anything that would cause an issue, but not sure I'm capturing peak voltage.

Also it's typical for the overdriven signal to the power valves to be asymmetric, resulting on one side being 'on' longer than the other, causign increased wear / dissipation on that side.

Yes - it is most likely this if the amp is always being run flat out through an attenuator.


https://el34world.com/Forum/index.php?topic=22271.msg237418#msg237418

[munkeyboy]

Maybe, but from what I know it is the tone king mini 30w.   That's an expensive unit and don't think he would risk running over the rating.

[tubeswell]

If you read through that other thread I linked (with the other information, including the link to the posts by Randal Aiken on the TAG forum a few years back), it explains how you get red-plating one one side of a PP output stage if you overdrive an amp with a 12AX7 LTP inverter that has a 470R bias resistor. The quality of the attenuator, and who made it, is not really relevant. But running such an amp flat out (albeit through an attenuator) all the time will most probably trigger the problem. If you swap the output tube grid leads around, and the problem then occurs in the other output tube, that will be the acid test

[High Voltage]

2W not 5W?   

Plexis need 5W screen resistors!

[munkeyboy]

If you read through that other thread I linked (with the other information, including the link to the posts by Randal Aiken on the TAG forum a few years back), it explains how you get red-plating one one side of a PP output stage if you overdrive an amp with a 12AX7 LTP inverter that has a 470R bias resistor. The quality of the attenuator, and who made it, is not really relevant. But running such an amp flat out (albeit through an attenuator) all the time will most probably trigger the problem. If you swap the output tube grid leads around, and the problem then occurs in the other output tube, that will be the acid test

Gotcha, but what I'm saying is that is that he is not running the amp flat out.  He says (i only have his word) that it runs the amp at 30% volume... not the attenuator.   Running the amp flat out would be putting "50w" into a "30w" attenuator.  If he ran the amp flatout into a 30w attenuator, I would guess the attenuator to have issues first, not the amp.   But I don't know enough about attenuators to say for sure.

Regardless, this is real good info.  I will be watching the tube plates for red more carefully and may try swapping the 470r for 820r if need be.

[HotBluePlates]

Gotcha, but what I'm saying is that is that he is not running the amp flat out.  He says (i only have his word) that it runs the amp at 30% volume...

The "30% volume" is somewhat meaningless, because we don't know exactly how much power the amp is kicking out at that point.  "30%" on a linear taper pot is the same as ~75% on a log taper pot.  And Volume-knob-setting doesn't account for more-signal at the Input jack leads to more-drive at the output tubes, regardless of exactly where the Volume control is pointing.

Scroll past the text & look at the graphs Aiken posted here.  Class AB amps have their highest tube-dissipation happening before the amp is full-tilt.


I read Aiken's replies on the thread Tubeswell linked.  To me the takeaway wasn't "you should get more-balance" but that "you should increase the load impedance to reduce the peak plate current."  Or reduce supply voltage, but that might not be the easiest route.

The Tone King Ironman is not a typical "attenuator": it is just a transformer with many taps that reduces how much voltage (and therefore power) goes to the attached speaker(s).  Unlike the Ironman & Ironman II, this one is 8Ω only.

So what load is your customer running the amp at?  Is both your rebuilt-version's output & the customer's speaker loading 8Ω?  Is the burned screen resistor the only damage? or does that side redplate, too?

Do you want to simply reduce drive to that side by splitting the plate load resistor into 2 parts (for example, 33kΩ and 56kΩ in place of a 100kΩ resistor) to "turn down" the driving voltage to that side?

As an example of that last bit, look at the 6G3 Deluxe, and the 12AX7 stage where the two channels are mixed.  The plate load is really 100kΩ + 15kΩ = 115kΩ total, but the junction of these resistors is where the coupling cap is attached.

     -  100kΩ & 15kΩ are a voltage divider:  15kΩ/(100kΩ+15kΩ) = 13% of the stage's output gets sent to the phase inverter.

Attenuated or not, you're gonna have to crank this amp through its range of output power, and see the plates & screens stay safe.  If you choose to "turndown" the drive to a side, you'll have to experimentally determine what is needed to keep things from overheating.

[munkeyboy]

The "30% volume" is somewhat meaningless, because we don't know exactly how much power the amp is kicking out at that point.  "30%" on a linear taper pot is the same as ~75% on a log taper pot.  And Volume-knob-setting doesn't account for more-signal at the Input jack leads to more-drive at the output tubes, regardless of exactly where the Volume control is pointing.

Scroll past the text & look at the graphs Aiken posted here.  Class AB amps have their highest tube-dissipation happening before the amp is full-tilt.

Good points. 

I read Aiken's replies on the thread Tubeswell linked.  To me the takeaway wasn't "you should get more-balance" but that "you should increase the load impedance to reduce the peak plate current."  Or reduce supply voltage, but that might not be the easiest route.

The Tone King Ironman is not a typical "attenuator": it is just a transformer with many taps that reduces how much voltage (and therefore power) goes to the attached speaker(s).  Unlike the Ironman & Ironman II, this one is 8Ω only.

So what load is your customer running the amp at?  Is both your rebuilt-version's output & the customer's speaker loading 8Ω?  Is the burned screen resistor the only damage? or does that side redplate, too?

The OT is 8ohm as is the customer (and my) speaker.   So the first time the amp came in for this, the resistor was burnt as well as the tube was red plating and tripping the breaker (traynor has breaker built in).  Also looked like there was some evidence of sparking at a ground point that I attributed to the bad tube.

The second time this came back a month or so later, just the resistor was burnt, but not as bad as the first time.  No red plating is currently happening.   I have since replaced the grid resistors with the wirewound type.  I am have not yet seen anything indicating an issue, but have only run it for 20min at various volumes.   Not sure I can run a good attenuator test as I only have a cheap 16 ohm resistive type, tho building or buying a 8ohm should be easy enough.

Do you want to simply reduce drive to that side by splitting the plate load resistor into 2 parts (for example, 33kΩ and 56kΩ in place of a 100kΩ resistor) to "turn down" the driving voltage to that side?

As an example of that last bit, look at the 6G3 Deluxe, and the 12AX7 stage where the two channels are mixed.  The plate load is really 100kΩ + 15kΩ = 115kΩ total, but the junction of these resistors is where the coupling cap is attached.

     -  100kΩ & 15kΩ are a voltage divider:  15kΩ/(100kΩ+15kΩ) = 13% of the stage's output gets sent to the phase inverter.

Attenuated or not, you're gonna have to crank this amp through its range of output power, and see the plates & screens stay safe.  If you choose to "turndown" the drive to a side, you'll have to experimentally determine what is needed to keep things from overheating.

The rest of this is a bit heavy.  I will need to keep reviewing this a few times.   If feels like I'll be covering something up if I need to reduce drive to one side as it was working well for 2+ years.