Reducing power by removing 2 output tubes & impedance

[marshallguy]

Hi, I have a question,

I have a 2203 model Marshall that I'm running 2 EHKT90's instead of 4 EL34's. Normally when removing 2 EL34's one has to reduce output Z in 1/2 to match back up the O.T. load. The EL34's are 25w rated while these KT90's are 46w rated.

Well while running a tone & checking the output power I found that running into an 8 ohm dummy load, I had more power out with the Z @ 8 ohms then 4 ohms or 1/2.

Is that because the power rating of 2 KT90's = approx the same as 4 EL34's? Or the plate loading from these 2 tubes?

I just liked to know if I should I keep it 8 to 8 ohm under this configuration or cut the Z out in half or 4 ohms?

Thank you.

[DummyLoad]

Normally when removing 2 EL34's one has to reduce output Z in 1/2 to match back

No, you need to double load Z. 

--Pete

[pdf64]

… The EL34's are 25w rated while these KT90's are 46w rated.
 …

Unfortunately that’s not an apples to apples comparison. The anode dissipation limit of different valve types has been very confusing since 1958, when some manufacturers adopted the design maximum rating system. It was further confused when GEC started also publishing dissipation ratings under absolute maximum system, for their KT range.

The take away point being that the KT90 anodes are very unlikely to be twice as beefy as EL34 anodes.
And that valves will overdissipate unless the designer sets operating conditions that attempt to prevent it.
And redplating might not be apparent until the valve’s anode is dissipating perhaps twice its limit.
 

[shooter]

1/2 power to the meter doesn't equate well to 1/2 power of the ears, or wife, or dog...

[kagliostro]

Franco

[marshallguy]

Normally when removing 2 EL34's one has to reduce output Z in 1/2 to match back

No, you need to double load Z. 

--Pete

You're saying 16 ohms into 8 ohm load?

[marshallguy]

… The EL34's are 25w rated while these KT90's are 46w rated.
 …

Unfortunately that’s not an apples to apples comparison. The anode dissipation limit of different valve types has been very confusing since 1958, when some manufacturers adopted the design maximum rating system. It was further confused when GEC started also publishing dissipation ratings under absolute maximum system, for their KT range.

The take away point being that the KT90 anodes are very unlikely to be twice as beefy as EL34 anodes.
And that valves will overdissipate unless the designer sets operating conditions that attempt to prevent it.
And redplating might not be apparent until the valve’s anode is dissipating perhaps twice its limit.
 

That's very informative so are you're saying in the long about version to treat the KT90's like EL34's as far as loading and reduce output Z to match load source? 16 to 8? 8 to 4?

[marshallguy]

1/2 power to the meter doesn't equate well to 1/2 power of the ears, or wife, or dog...

Level 5 huh?

All I was saying was the power transfer was more efficient leaving the output transformer Z the same as the source load even though only running on 2  instead of 4 tubes. So, is it because of the 2 type of tubes or just something to ignore and change OT to 1/2 Z of source? 

[marshallguy]

Franco

Franco, it wasn't my ears it was the meter & scope.

[shooter]

No, you need to double load Z.

[HotBluePlates]

Normally when removing 2 EL34's one has to reduce output Z in 1/2 to match back

No, you need to double load Z. 

--Pete

You're saying 16 ohms into 8 ohm load?

Quote-Inception...


4x EL34s  ---> 8Ω Tap ---> 8Ω Load
2x EL34s ---> 8Ω Tap ---> 16Ω Load


Halving the tubes doubled the impedance on the tube-side of the transformer; so double the impedance on the speaker-side of the transformer to maintain correct loading.


Interesting graphics aside, half-power is an under-whelming reduction of loudness.  I could piss off apartment neighbors with 1/10th watt into some sensitive Celestion speakers.

[marshallguy]

Normally when removing 2 EL34's one has to reduce output Z in 1/2 to match back

No, you need to double load Z. 

--Pete

You're saying 16 ohms into 8 ohm load?

Quote-Inception...


4x EL34s  ---> 8Ω Tap ---> 8Ω Load
2x EL34s ---> 8Ω Tap ---> 16Ω Load


Halving the tubes doubled the impedance on the tube-side of the transformer; so double the impedance on the speaker-side of the transformer to maintain correct loading.

Hey HotBluePlates,

I agree. You know that’s exactly what I’ve been doing, can’t seem to multitask and state the right questions today. 

Let me try again, hopefully this is to the heart of the matter…

Ok answer this…

2KT90’s ——> 8 ohm tap —-> 16 ohm Load? Why do I ask?

I’ve been treating the 2 KT90’s like 2 EL34’s but I noticed while calculating output power by connecting an RMS meter and scope to the output jacks that if I move and set the tap the same as the load with just the 2 KT90’s in use, the power actually increases, in this case 8 to 8,  looking like a better Z power transfer match then going 4 to 8 based on previous discussion.

1khz 150mv input. Adjust MV a little less then half way.

Speaker Jack 4 to 8 ohms….. 27.6 pk/pk - 9.79 vrms.

Speaker Jack 8 to 8 ohms….30.0 pk/ pk - 10.5 vrms.

I’m interested on why that is. I would have thought the opposite unless using 2 higher power tubes verses 4 lower ones effecting it.

[shooter]

you're mixing your metaphors
the original ? was 4 tubes to 2...= 2Z


adding other tubes to muddy the water doesn't change the equation, it's still 2Z for 1/2 the tubes.


Tubes spec's determine the original, "base" Z

[sluckey]

It would be very helpful if you measure the impedance ratio of your OT. Google is your friend.

[marshallguy]

you're mixing your metaphors
the original ? was 4 tubes to 2...= 2Z


adding other tubes to muddy the water doesn't change the equation, it's still 2Z for 1/2 the tubes.

Tubes spec's determine the original, "base" Z

Yes I agree with that analogy just wondering the other part to my question….

[marshallguy]

It would be very helpful if you measure the impedance ratio of your OT. Google is your friend.

Dagnall C2668 on 38x40mm EI M6 core:
- 0-4 ohm secondary: 60 turns x 1.76mm AWG13
- 1/2 primary: 620 turns on four layers x 0.45mm AWG26
- 4-8-16 ohm secondary: 25+35 turns x 1.18mm AWG18
- 1/2 primary: 620 turns on four layers x 0.45mm AWG26
- 0-16 ohm parallel secondary: 120 turns on two layers x 0.75mm AWG21


Transformer Specification C2668 Dagnall Electronics

-Customer: Marshall Amplification Plc.
-Product: 100 watt Output Transformer - Steel Shroud Enclosure-Chassis Mounted

-Customer Part No: TXOP - 00006

-Core: Pattern 120N laminations in M6-35 grade material to BS6404 Sect 8.7

-Bobbin: Single Section moulded bobbin in 30% glass fibre reinforced Nylon 66 Core tube thickness 1.65mm

-Windings: In concentric sequential order with Input and Output windings sandwiched
for performance characteristics. Output is split and parallel connected to
improve efficiency and share load. All winding wires have enamel
covering of thermal class B minimum.

-Output 1: 0 - 40v 16 ohms @ 100W (Brown - Green)
Wound with grade 1 polyurethane wire to BS 6811
Insulated by 3 wraps of Nomex 410 paper, 0.08mm thk/layer

-Input: 0 - 207 - 414v 1750 ohms @ 100W (Brown - White - Red)
Wound with grade 1 polyurethane wire to BS 6811
Interleaved every 3 Layers by 1 wrap of Nomex 410 paper, 0.08mm thk.
Insulated by 3 wraps of Nomex 410 paper, 0.08mm thk/layer.

-Output 2: 0 - 20.33v 4 ohms @ 100W (Brown - Black)
Wound with grade 2 polyestermide wire to BS 6811

-Output 3: 0 - 28.67v 8 ohms @ 100W (Brown - Yellow)
Wound with grade 1 polyurethane wire to BS 6811
Insulated by 2 wraps of electrical Crepe Paper Tape, 0.24mm thk/layer.

-Terminations: All windings terminated to VX grade PVC; 105 deg C flexible leads

-Encapsulation: Laminated wound coil varnish dipped in Class F polyurethane
electrical insulating varnish and oven cured.

-Assembly: Two steel end shells totally enclose windings, the connecting leads
passing through PVC grommets.

-Electrical Tests: No Load Magnetisation Current
Turns Ratio +/- 3%
Induced Voltage Test @ x5 Rated Input @ x12 Rated Frequency
Simultaneous Electrical Strength Test -
2.0Kv rms Primary to Secondaries and Core

Examination, Fit & Function, Performance and Safety Checks carried
out at 100% level.

That help?

[HotBluePlates]

Dagnall C2668 ...
...
-Input: 0 - 207 - 414v 1750 ohms @ 100W (Brown - White - Red)
...

So much excess stuff, the important bits get masked.  The little bit above says "with marked loading, primary is 1750Ω."

I google "Dagnall C2668" and get a result that says the primary is 1.7kΩ.


Stripping out only the useful nuggets:

Dagnall C2668 ...
- 0-4 ohm secondary: 60 turns ...
- 1/2 primary: 620 turns on four layers x 0.45mm AWG26
...
- 1/2 primary: 620 turns on four layers x 0.45mm AWG26
...

Primary = 620 turns + 620 turns = 1240 turns
4Ω Secondary = 60 turns

Turns Ratio = 1240 / 60 = 20.667 : 1
Impedance Ratio = (Turns Ratio)² = (20.667)² = 427.111 : 1

4Ω on this tap reflects:  4Ω x 427.111 = 1708Ω ---> my googled link was more-better-closer.

If I also knew your power supply voltage, I might go to a KT90 data sheet (assuming it isn't garbage like most modern data sheets), look at the peak plate current available for the screen volts on your tubes, and determine whether "1700Ω / 4" seems like right-loading for max power with your amp's conditions.



OR... Keep doing as you do, adjust loading, measure power output.  The "optimum load" for max power will present itself, because power output of tube power amps drops off when you go higher/lower than the optimum impedance.  There was a piece of test gear whose applications included being an adjustable load impedance, allowing an experimenter to find the optimum loading to maximize power output.

[marshallguy]

...
-Input: 0 - 207 - 414v 1750 ohms @ 100W (Brown - White - Red)
...
[/quote]

So much excess stuff, the important bits get masked.  The little bit above says "with marked loading, primary is 1750Ω."

I google "Dagnall C2668" and get a result that says the primary is 1.7kΩ.


Stripping out only the useful nuggets:

Dagnall C2668 ...
- 0-4 ohm secondary: 60 turns ...
- 1/2 primary: 620 turns on four layers x 0.45mm AWG26
...
- 1/2 primary: 620 turns on four layers x 0.45mm AWG26
...

Primary = 620 turns + 620 turns = 1240 turns
4Ω Secondary = 60 turns

Turns Ratio = 1240 / 60 = 20.667 : 1
Impedance Ratio = (Turns Ratio)² = (20.667)² = 427.111 : 1

4Ω on this tap reflects:  4Ω x 427.111 = 1708Ω ---> my googled link was more-better-closer.

If I also knew your power supply voltage, I might go to a KT90 data sheet (assuming it isn't garbage like most modern data sheets), look at the peak plate current available for the screen volts on your tubes, and determine whether "1700Ω / 4" seems like right-loading for max power with your amp's conditions.



OR... Keep doing as you do, adjust loading, measure power output.  The "optimum load" for max power will present itself, because power output of tube power amps drops off when you go higher/lower than the optimum impedance.  There was a piece of test gear whose applications included being an adjustable load impedance, allowing an experimenter to find the optimum loading to maximize power output.
[/quote]

This is from EH should help probably accurate. https://shop.ehx.com/catalog/addimages/kt90eh.pdf

The plate voltage is approx. 460v depending of time of day and AC input voltage.

If I wasn’t retired I used that piece of test equipment in the past could have borrowed it. Everyone I knew is retired too by now. Best I can do is a Fluke 87 rms reading and Tektronics scope to measure pk to pk

[sluckey]

That help?

You are so full of SHIT!

[marshallguy]

That help?

You are so full of SHIT!

Please watch your language try dodo or caca as alternative words.

It’s ok for you to be a funny guy I just figured I’d be one too. Now everybody funny.

[sluckey]

That help?

You are so full of SHIT!

Please watch your language try dodo or caca as alternative words.

It’s ok for you to be a funny guy I just figured I’d be one too. Now everybody funny.

I'm dead serious!

[DummyLoad]

Treat KT90 like a KT88 - Not much more meat on the plate to be of any significance, just marketing IMO - KT120 and KT150 are different story altogether.

KT88 will want ≈4K load with 450V on to the plates. Your OT is 1.7K or 3.4K w/ 2x load on indicated secondary tap. It will work, not ideal, but won't be an issue if you bias cool enough while keeping x-over distortion in check.


--Pete

[marshallguy]

Treat KT90 like a KT88 - Not much more meat on the plate to be of any significance, just marketing IMO - KT120 and KT150 are different story altogether.

KT88 will want ≈4K load with 450V on to the plates. Your OT is 1.7K or 3.4K w/ 2x load on indicated secondary tap. It will work, not ideal, but won't be an issue if you bias cool enough while keeping x-over distortion in check.



--Pete

Roger that Pete, thanks for sharing.

[pdf64]

…
OR... Keep doing as you do, adjust loading, measure power output.  The "optimum load" for max power will present itself, because power output of tube power amps drops off when you go higher/lower than the optimum impedance.  There was a piece of test gear whose applications included being an adjustable load impedance, allowing an experimenter to find the optimum loading to maximize power output.

But ‘optimum’ for audio power output might be detrimental to valve life.
ie doesn’t adjusting the load impedance for maximum audio power output ignore the possibility for the anode dissipation limit being exceeded, perhaps significantly so, under load?

[marshallguy]

…
OR... Keep doing as you do, adjust loading, measure power output.  The "optimum load" for max power will present itself, because power output of tube power amps drops off when you go higher/lower than the optimum impedance.  There was a piece of test gear whose applications included being an adjustable load impedance, allowing an experimenter to find the optimum loading to maximize power output.

But ‘optimum’ for audio power output might be detrimental to valve life.
ie doesn’t adjusting the load impedance for maximum audio power output ignore the possibility for the anode dissipation limit being exceeded, perhaps significantly so, under load?

On one hand you would think that a better transfer of audio power of less resistance from source to speaker would be the optimal solution here which is the observation but that’s not how it’s suppose to be done.

The other hand, treat EL34’s, KT77’s, KT88’s, KT90’s, 6550’s, 6L6’s all the same if removing 2 out of 4 and reduce source impedance by 1/2 the speaker’s impedance and call it a day.

Great discussion. Anyone else have anything “positive” to add?

[PRR]

....might be detrimental to valve life. ie doesn’t adjusting the load impedance for maximum audio power output ignore the possibility for the anode dissipation limit being exceeded, perhaps significantly so, under load?

The "usual" tubes at "reasonable" voltages won't eat their guts out too fast. Unlike semiconductors, tubes won't even try to carry "infinite" current.

Sure, you can probably find tube-eating conditions.

[marshallguy]

Treat KT90 like a KT88 - Not much more meat on the plate to be of any significance, just marketing IMO - KT120 and KT150 are different story altogether.

KT88 will want ≈4K load with 450V on to the plates. Your OT is 1.7K or 3.4K w/ 2x load on indicated secondary tap. It will work, not ideal, but won't be an issue if you bias cool enough while keeping x-over distortion in check.


--Pete

Pete,

Marshall 100w amps were all set up with 4 EL34’s or 6550’s using the Dagnall O.T. EL34’s have the same Rp as EHKT90’s or 2k ohms. We know the O.T. Primary resistance, 1.7k.

Using EL34’s for or discussion, are the tube total plate loading calculated the same as 4 say 16 ohm speakers in a cabinet? Series / parallel?

That would make one side 2k + 2k / 2k + 2k = 2k plate loading total for 4 output tubes?

Pulling 2 out reduces it to 1k? So now reducing the secondary output in half to speaker load brings it back to 2k? Would you say that’s the correct math or am I off on this?

jC

[marshallguy]

That help?

You are so full of SHIT!

Not really, I was using a resistive load, trying it again with speakers brings the math and science back to proper perspective.

[pdf64]

... Unlike semiconductors, tubes won't even try to carry "infinite" current.

Indeed, but the flip side of limiting the current is that the voltage across them increases, hence dissipation follows suit.

Sure, you can probably find tube-eating conditions.

I was concerned that’s what post 1 was describing 

[pdf64]

... EL34’s have the same Rp as EHKT90’s or 2k ohms.

The Rp you’re picking up on is the load impedance for single ended (hence class A) operation at the specified anode and screen grid voltages (250V).
Your amp is class AB push pull with much higher voltages. It’s not good to pick a number from one set of operating conditions and expect it to be fine under totally different conditions.

Using EL34’s for or discussion, are the tube total plate loading calculated the same as 4 say 16 ohm speakers in a cabinet? Series / parallel?

No push pull operation is a different scenario.

[HotBluePlates]

I have a 2203 model Marshall that I'm running 2 EHKT90's instead of 4 EL34's. ...

... the power rating of 2 KT90's = approx the same as 4 EL34's? ...

On one hand you would think that a better transfer of audio power of less resistance from source to speaker would be the optimal solution here which is the observation but that’s not how it’s suppose to be done.

It doesn't work like you think/describe.

Let's start with a better KT90 data sheet (even though I still dislike it, since there's no graph of the effect of changing screen voltage).

   - Right away, plate dissipation rating is 50 watts vs 25 for an EL34.
   - If Voltage stays same, 2xPower / Same-Volts = 2xCurrent
   - Ohm's Law:  Same-Volts / 2xCurrent = 0.5xImpedance ---> 1x KT90 probably uses the load impedance of 2x EL34
   - You install 2x KT90 into your 2203 Marshall instead of 4x EL34 ---> Use the marked tap impedance for max power


Page 3 of the data sheet I linked has a graph of "Average Plate Characteristics":

   - The top-most red line is "Vg1 = 0V"
   - This curve has a "knee" at about 50v and about 350mA.

   - The graph is for a screen voltage of 225v ("Vg2=225v" in the top-left), but your screen voltage is about double at 450v.
   - For double-screen-volts, let's guess the KT90 can manage double-plate-current:  700mA.
   - The knee for those curves moves towards higher plate voltage when current is higher; let's guess our "Vg2=450v" knee is at 100v on the plate.

   - If plate voltage starts at 450v, and the knee is at 100v, the max plate voltage swing might be as high as  450v - 100v = 350v.
   - It can be shown that your push-pull Class AB amp output transformer looks like a load to a single KT90 of 1700Ω / 4 = 425Ω at max power output.
   - Ohm's Law:  (Peak) Volts = (Peak) Current x Impedance = 0.7A peak x 425Ω = 297.5v peak
   - Peak Power = Peak Volts x Peak Current = 0.7A peak x 297.5v peak = 208.25 watts peak
   - RMS Power = Peak Power / 2 = 208.25w peak / 2 = 104.1 watts RMS
      ** A push-pull pair of KT90s should be able to swing the same ~100w output as 4x EL34s given the same loading.

________________________________________________

I happened to have a graphic I drew for a different explanation of the same thing.  That is that for tube amps, shifting the loading higher or lower results in less output power.

   - The Blue Line is the "optimum loading" for the example tube & supply voltage.
   - The Blue Rectangle formed by the bounds of the Blue Line is equivalent to Peak Power Output.

   - The Brown Line is loading that tube with a higher-impedance load.  The resulting Brown Rectangle has a smaller area than the Blue Rectangle, and so represents less Peak Power Output.

   - The Red Line is loading the tube with a lower-impedance load.  The Resulting Red Rectangle has a smaller area than the Blue Rectangle, and so represents less Peak Power Output.

   - All you're doing with moving to a higher or lower load impedance is changing the ratio of voltage-change to current-change.  But we should already see that will be the case from Ohm's Law.
   - What we can accomplish with a loading-change depends on the unchanged amount of voltage available to work with, and the tube's limitations (mainly of cathode/plate current, and also plate dissipation).

[marshallguy]

... EL34’s have the same Rp as EHKT90’s or 2k ohms.

The Rp you’re picking up on is the load impedance for single ended (hence class A) operation at the specified anode and screen grid voltages (250V).
Your amp is class AB push pull with much higher voltages. It’s not good to pick a number from one set of operating conditions and expect it to be fine under totally different conditions.

Using EL34’s for or discussion, are the tube total plate loading calculated the same as 4 say 16 ohm speakers in a cabinet? Series / parallel?

No push pull operation is a different scenario.

Well even in Class AB shouldn’t’ they still be close and treated as such? It was mentioned earlier treat KT90 like KT88 which leads to several other tubes of lesser wattage but similar.

[marshallguy]

I have a 2203 model Marshall that I'm running 2 EHKT90's instead of 4 EL34's. ...

... the power rating of 2 KT90's = approx the same as 4 EL34's? ...

On one hand you would think that a better transfer of audio power of less resistance from source to speaker would be the optimal solution here which is the observation but that’s not how it’s suppose to be done.

It doesn't work like you think/describe.

Let's start with a better KT90 data sheet (even though I still dislike it, since there's no graph of the effect of changing screen voltage).

   - Right away, plate dissipation rating is 50 watts vs 25 for an EL34.
   - If Voltage stays same, 2xPower / Same-Volts = 2xCurrent
   - Ohm's Law:  Same-Volts / 2xCurrent = 0.5xImpedance ---> 1x KT90 probably uses the load impedance of 2x EL34
   - You install 2x KT90 into your 2203 Marshall instead of 4x EL34 ---> Use the marked tap impedance for max power


Page 3 of the data sheet I linked has a graph of "Average Plate Characteristics":

   - The top-most red line is "Vg1 = 0V"
   - This curve has a "knee" at about 50v and about 350mA.

   - The graph is for a screen voltage of 225v ("Vg2=225v" in the top-left), but your screen voltage is about double at 450v.
   - For double-screen-volts, let's guess the KT90 can manage double-plate-current:  700mA.
   - The knee for those curves moves towards higher plate voltage when current is higher; let's guess our "Vg2=450v" knee is at 100v on the plate.

   - If plate voltage starts at 450v, and the knee is at 100v, the max plate voltage swing might be as high as  450v - 100v = 350v.
   - It can be shown that your push-pull Class AB amp output transformer looks like a load to a single KT90 of 1700Ω / 4 = 425Ω at max power output.
   - Ohm's Law:  (Peak) Volts = (Peak) Current x Impedance = 0.7A peak x 425Ω = 297.5v peak
   - Peak Power = Peak Volts x Peak Current = 0.7A peak x 297.5v peak = 208.25 watts peak
   - RMS Power = Peak Power / 2 = 208.25w peak / 2 = 104.1 watts RMS
      ** A push-pull pair of KT90s should be able to swing the same ~100w output as 4x EL34s given the same loading.

________________________________________________

I happened to have a graphic I drew for a different explanation of the same thing.  That is that for tube amps, shifting the loading higher or lower results in less output power.

   - The Blue Line is the "optimum loading" for the example tube & supply voltage.
   - The Blue Rectangle formed by the bounds of the Blue Line is equivalent to Peak Power Output.

   - The Brown Line is loading that tube with a higher-impedance load.  The resulting Brown Rectangle has a smaller area than the Blue Rectangle, and so represents less Peak Power Output.

   - The Red Line is loading the tube with a lower-impedance load.  The Resulting Red Rectangle has a smaller area than the Blue Rectangle, and so represents less Peak Power Output.

   - All you're doing with moving to a higher or lower load impedance is changing the ratio of voltage-change to current-change.  But we should already see that will be the case from Ohm's Law.
   - What we can accomplish with a loading-change depends on the unchanged amount of voltage available to work with, and the tube's limitations (mainly of cathode/plate current, and also plate dissipation).

That’s impressive. What you’re saying is exactly what I was speculating. Now the EHKT90’s which I’m using are rate at 46w each. I’ve been calculating bias at 46w. Plates read 475v. Screens drop measured approx. 7.5ma. 46 / 475 x .6 or (.097 x 60%) = .058….58ma + 7.5ma screen = 65.5ma for 60% bias. I like a little cooler and shoot for 60 to 65%.  However if these are really 50w rated then I’m under biased. When I do the shunt current method, it is reading colder than the 1ohm cathode voltage method. Not close, several milliamps difference, think it was around 8mA. I can check again.

[pdf64]

… Right away, plate dissipation rating is 50 watts vs 25 for an EL34…

Isn’t that turning a blind eye to the different rating systems used to derive those limits?

[HotBluePlates]

On one hand you would think that a better transfer of audio power of less resistance from source to speaker would be the optimal solution ...

... What you’re saying is exactly what I was speculating. ...

When I said, "It doesn't work like you describe," that's because the words you used effectively communicate, "If we make the load smaller, we get more power."

What I tried to show with the Brown/Blue/Red diagram was that smaller-load increases power up to a point, then power starts falling off again.  You had your KT90s loaded equivalent to the Brown Line in my diagram, and you moved to the Blue Line when you observed a power output increase.  But if you continued reducing load impedance, power would fall again.

This was shown in RCA's 1936 Application Note on the 6L6.  Scroll to the last page; on the left is a graph where everything is kept constant except the load impedance.  Below some broad range of "optimum" a further reduction of load just results in less power output (and more 2nd harmonic distortion).

… Right away, plate dissipation rating is 50 watts vs 25 for an EL34…

Isn’t that turning a blind eye to the different rating systems used to derive those limits?

Things in this thread pointed towards needing "a broad sketch."  One the major issues are understood, it is reasonable to address "the fine shading."

Also, "50w KT90" vs "46w KT90" made the "broad sketch" more intuitively obvious.  That serves a valuable purpose here.


On some other forum, folks are getting into plugging only into an old Fender's Ext Speaker jack (giving a near-0Ω load) as an attenuation method.  These impedance threads have been going for over 20 years; it seems people need to touch the stove to learn to stay clear, or they need to touch the stove to figure out it's not even warm.

... Now the EHKT90’s which I’m using are rate at 46w each. I’ve been calculating bias at 46w. ... However if these are really 50w rated then I’m under biased. ...

I'd say you're learning the wrong lesson, leading to pdf64's concern:  AFAIK, Electro-Harmonix is the only company making KT90 right now.  The same single tube is called "46w" under a Design Max system, and "50w" under an Absolute Max system.  So you're just think you need to run the bias hotter, perhaps for no real reason.

If you had a scope or distortion-meter, I'd say try driving the amp to some arbitrary low-% of distortion to determine "max power."  Then repeat with whatever bias you decide to use (hotter or cooler).  You will probably find some moderate-to-low idle current provides max power.

   - Now leave everything the same, but drop drive-signal to maybe 1/3 or 1/4 of that for max power output.
   - If you observe crossover notch distortion on the scope (or hear it in a speaker), the bias is too cool.
   - Crossover distortion happens at small-signal, but you often have to be idling very cold (maybe under 20% rated dissipation?) to encounter it.  People often hear distortion & wrongly ascribe it to crossover-notch, then bias hotter than they need.

[marshallguy]

… Right away, plate dissipation rating is 50 watts vs 25 for an EL34…

Isn’t that turning a blind eye to the different rating systems used to derive those limits?
[/quote]

Things in this thread pointed towards needing "a broad sketch."  One the major issues are understood, it is reasonable to address "the fine shading."

Also, "50w KT90" vs "46w KT90" made the "broad sketch" more intuitively obvious.  That serves a valuable purpose here.

If you had a scope or distortion-meter, I'd say try driving the amp to some arbitrary low-% of distortion to determine "max power."  Then repeat with whatever bias you decide to use (hotter or cooler).  You will probably find some moderate-to-low idle current provides max power.

   - Now leave everything the same, but drop drive-signal to maybe 1/3 or 1/4 of that for max power output.
   - If you observe crossover notch distortion on the scope (or hear it in a speaker), the bias is too cool.
   - Crossover distortion happens at small-signal, but you often have to be idling very cold (maybe under 20% rated dissipation?) to encounter it.  People often hear distortion & wrongly ascribe it to crossover-notch, then bias hotter than they need.
[/quote]

Well EH muddied the waters when the say their tube is 46w rated instead of 50w. That’s a 5 mA idle current difference depending what % one is looking for. 

With the current applied voltages I’m getting approx. 75w before clipping starts to occur (clean channel @1khz). That was with 8 ohm tap out to 16 ohm 2 x 12 cabinet. Using both speakers I can only connect in series 16ohms or parallel 4 ohms.

The guts of this discussion was how to treat 2 tubes that = 4 tubes in rated power, though we know plate and screen voltages and impedance play a role in what output power would be and wether or not to set the output Z 1/2 the speaker cabinet Z or both the same.

So a few here say to treat KT90’s the same as 77’s, 88’s, EL34’s, and set output Z for 1/2 when removing 2 of the 4, and one says to keep the Z the same. The math and arguments are all good, but can you’ll reach a consensus which one?

“To be or not to be” that always confused me.

[shooter]

go play through the amp for 2 weeks, use all the knobs on all the equipment used within the system, pedals, guitars.....
every 3rd day swap out the speaker, cab, speaker count, repeat til times up.


wanna do fun math, do a parts count of the system, weight each components based on AC electronics, so  R's ~ 1, C's ~ 2.3, tube ~ 5.5...
now decide how many permutations you want to compute over a components "normal range"
create an equation that expresses the "center norm" from all datasets collected, all variations of the dataset, boom, perfection, or not 

[HotBluePlates]

So a few here say to treat KT90’s the same as 77’s, 88’s, EL34’s, and set output Z for 1/2 when removing 2 of the 4 ...

I'm not the Brightest Bulb in the Box, but I have not seen anyone except you make that statement.

Who else said to treat KT90 as EL34 or KT77?
(KT88 is an entirely different tube, closer to 6550 than EL34 or KT77).

[marshallguy]

So a few here say to treat KT90’s the same as 77’s, 88’s, EL34’s, and set output Z for 1/2 when removing 2 of the 4 ...

I'm not the Brightest Bulb in the Box, but I have not seen anyone except you make that statement.

Who else said to treat KT90 as EL34 or KT77?
(KT88 is an entirely different tube, closer to 6550 than EL34 or KT77).

Dummyload compared 88’s to 90’s, to treat them the same. The other tubes mention as suitable subs in other venues and literature that are suppose to be interchangeable but of course with changing the bias and or grid resistors too. That includes the 6L6 too. We can agree on that, right?

When Marshall’s JCM800’s were shipped to Unicord Corp. for North American distribution, they were shipped without valves. The 6550’s were then placed in and those amps that had 150k grid leak resistors and different value bias resistor. All the other Marshall’s had the EL34’s & 220k grid leaks. Not many liked the 6550’s and had them removed for EL34’s. The amps I brought back of course all had EL34’s and sold like hot cakes. I was given a combo with KT77’s installed. I was gigging for awhile with 6L6wgb’s too. That PT only had 390v plate voltage.

I never used KT88’s but when a good friend and respected local amp tech told me about the KT90’s, I decided to try them. I like them, sound great and more reliable than the EL34’s, sounded more like an oversized EL34. Plus they can withstand a higher screen voltage, 600 or 650 if memory serves me….

[HotBluePlates]

So a few here say to treat KT90’s the same as 77’s, 88’s, EL34’s, and set output Z for 1/2 when removing 2 of the 4 ...

Who else said to treat KT90 as EL34 or KT77?
(KT88 is an entirely different tube, closer to 6550 than EL34 or KT77).

Dummyload compared 88’s to 90’s, to treat them the same.

I agree with him on that.  4x EL34s makes a 100w Marshall, but 4x KT88s makes a 200+ watt Marshall Major.

So a few here say to treat KT90’s the same as 77’s, 88’s, EL34’s, and set output Z for 1/2 when removing 2 of the 4 ...

Who else said to treat KT90 as EL34 or KT77?
(KT88 is an entirely different tube, closer to 6550 than EL34 or KT77).

... The other tubes mention as suitable subs in other venues and literature that are suppose to be interchangeable but of course with changing the bias and or grid resistors too. That includes the 6L6 too. We can agree on that, right?

When Marshall’s JCM800’s were shipped to Unicord Corp. for North American distribution, they were shipped without valves. The 6550’s were then placed in and those amps that had 150k grid leak resistors and different value bias resistor. All the other Marshall’s had the EL34’s & 220k grid leaks. ...

Do we need to re-litigate the past, or speak to incomplete or inaccurate info elsewhere?

   - All tubes noted fit the same 8-pin socket, and have the same pin-out.
   
   - EL34 and 6L6/5881 are in the same general class of power-handling.
   - 6L6/5881 has about half the transconductance that an EL34 has.
   - EL34 requires only about half the drive signal and bias voltage required for 6L6/5881.
   
   - 6550 has the same general transconductance that the EL34 has.
   - 6550 could use a similar bias voltage and drive signal that EL34 uses.
   - Best-practice for 6550 required smaller grid-resistance, so grid-current didn't upset bias in a negative way.
   - 6550 could deliver a lot more output power, but the amps weren't optimized to get max output power.


EL34 and KT77 are basically the same thing, except one is a pentode, and the other is a beam tetrode trying to evade the patent on the EL34.


You could do as the North American Marshall distributor in the 1970s did, and install 4x KT90s in your amp.  You won't get all the performance they can deliver, and Unicord didn't get "all the performance" out of 6550s.  They merely needed a readily-available domestic tube, and EL34 was a Philips product in Europe.

The tubes noted above aren't exactly "interchangeable":  we've seen them plugged into the same amps, but they don't perform the same, and some of the tubes have capability well beyond the demands of the amps in which they were installed. 

If we get back to just "KT90" and "EL34" then I've said several times that 2x KT90 will replace 4x EL34 with no changes needed on the amp-side.

[marshallguy]

The tubes noted above aren't exactly "interchangeable":  we've seen them plugged into the same amps, but they don't perform the same, and some of the tubes have capability well beyond the demands of the amps in which they were installed. 

If we get back to just "KT90" and "EL34" then I've said several times that 2x KT90 will replace 4x EL34 with no changes needed on the amp-side.

HotBluePlates,

I want to thank you for all your time and in-depth information you provided on this topic.

Before we close this discussion, I have one last question. I previously was set up 8 to 16 ohms, but now using this new info, with a 2 - 12 cabinet, would you recommend going  4 to 4 ohms parallel or 16 to 16 ohm series? 

[HotBluePlates]

I previously was set up 8 to 16 ohms, but now using this new info, with a 2 - 12 cabinet, would you recommend going  4 to 4 ohms parallel or 16 to 16 ohm series? 

I have no opinion.  If you suspect there will be a difference, try it each way & decide what sounds best to you.

[pdf64]

Just to note that the 200W 4xKT88 Marshall Major used a much higher HT voltage, eg 620V, see  https://el34world.com/charts/Schematics/files/Marshall/Marshall_charts.pdf

[marshallguy]

Just to note that the 200W 4xKT88 Marshall Major used a much higher HT voltage, eg 620V, see  https://el34world.com/charts/Schematics/files/Marshall/Marshall_charts.pdf

With 2 KT90’s and 475v HT, I get approx. 75w before clipping occurs. That’s ok though because just those extra 25 watts all it took to help me cut through the mix. 

[marshallguy]

Just to note that the 200W 4xKT88 Marshall Major used a much higher HT voltage, eg 620V, see  https://el34world.com/charts/Schematics/files/Marshall/Marshall_charts.pdf

Taking a 2nd look at those charts, 520v on the screens of an EL34 isn’t going to last too long and I never measured voltages that high on my 100 waters, more around 470v. Those were 1983 amps, I worked on a friends 50w 2204 1986 head and those voltages were quite high through out, 480v HT. He complained about the amp being too bright so I lowered the preamp voltages by increasing the first node resistor and added another resistor / cap node to the B+ line and I also made a voltage divider out of the CF feeding the tone stack giving him more play in the MV. He’s been one happy camper since.

[pdf64]

… Taking a 2nd look at those charts, 520v on the screens of an EL34 isn’t going to last too long …

100 watters from the very early 70s ran crazy a high HT, with good valves they an put out over 120W clean. I suppose Mullards of the time must’ve been able to take it, but yes, more recently made EL34 will tend to struggle, especially V4&5, which due waveform asymmetry when overdriven, get pushed harder.

Conversely, 70s 50 watters could have a pretty low HT, some well under 400V at idle, as per the chart. Maybe 30-35W clean power. I think it was done like that to protect the pair of greenbacks often used for combo cabs.

[marshallguy]

 …
[/quote]
1

Conversely, 70s 50 watters could have a pretty low HT, some well under 400V at idle, as per the chart. Maybe 30-35W clean power. I think it was done like that to protect the pair of greenbacks often used for combo cabs.
[/quote]

Even early 80’s 50 watters had well under 400v at idle, which questions how calling it a 50w amp when 30-35w clean power was the best you could get. You couldn’t keep up with the volume jokers on stage. They put the higher HT in the 100 watters until some where in the mid 80’s they must have switched and both had the higher voltages.