6550's as a 6L6 replacement

[tubetek]

Whoa!
You realize that the metal base of the 6550 is connected internally!
I hope you aren't using "bearclaw" type tube retainers for the power tubes.
These will result in a short and general bad cess. The alternatives:
use EH6550 that is made with the plastic base or convert th the Marshall-style tube retainers. Don't try to go without retainers in a "Fender style tubes down" amp. HTH
Tom

[greiswig]

Whoa!
You realize that the metal base of the 6550 is connected internally!
I hope you aren't using "bearclaw" type tube retainers for the power tubes.
These will result in a short and general bad cess. The alternatives:
use EH6550 that is made with the plastic base or convert th the Marshall-style tube retainers. Don't try to go without retainers in a "Fender style tubes down" amp. HTH
Tom

These have a plastic base. Thanks for the warning, though!

[Ears]

Thank you, Ears. I'm still learning this stuff, and I consider this to be one of the parts of tube technology that I am least aware of. (and this should be basic stuff!) Terminology also sometimes screws me up...

I'm still learning too. The topic of screens is one area of vexation to me. The reason is that little is pubished in tube data and elsewhere about the management of screen conditions if you decide to operate them away from the manufacturer's recommended (HiFi) values. For guitar amps this is more of a proplem with some tubes compared with others as for simplicities sake, amongst other reasons, screens are run at plate voltage values - and for 6L6 and EL34 data sheets often provide information for this condition or conditions very close. But 6550 and KT88 data is given for 300V screens. We know screen is primary anode and has major effect (along with grid 1) on anode current, i.e. higher screen voltages mean the Vg1=0V curve of plate characteristics is higher Ia and all curves are spaced wider, more power is available. But there are down sides, in particular screen current increases. Screens have a dissipation limit , and in particular if we choose to run an amp at max power this limit is important if we don't want to melt the screens and/or cause markedly shorter operational life of the tube (exceeeding screen dissipation is a major cause of power tube failure). It is of interest to me to discover a method to figure out the screen current at max power for different screen voltages, from the already published ones. Kuehnel provides a method to generate new plate curves for new Vg2 but I haven't found the other information yet (but have only read 3/4 of his Power Amp book, maybe its in there somewhere). Can anyone help?

I think I can figure out which pins to check Vg2 (Pin 5?) and Va (Pin 3?) voltages on. If the grid resistor is the 220k that feeds Pin 5, then reducing it from 220k to 50k will raise that voltage even higher, yet that grid resistance is limited to 50k according to what I've read. So do you just have to play with the dropping string to get this into line?

Hold on here, I am unfamiliar with the bias arrangement you are using, but the limitation set for grid resistance is to prevent runaway and destruction of the tube. Valve wizard website explains it like this (more detailed analysis can be found in Radiotron designers HandBk and elsewhere):

"During normal operation the grid becomes hot (since it is close to the
cathode) and will emit a few electrons, and so will become positively charged unless a leakage path is provided to replenish the charge lost. If the grid is allowed to charge positively, anode current will increase- which increases the valve’s temperature, making the problem worse. Eventually the bias would drift out of control, and in the case of power valves this could easily cause thermal runnaway and destruction of the valve. The grid-leak resistor provides the necessary leakage path for electrons and holds the grid at a fixed quiescent voltage (in this case zero volts).
Usually we would like the grid leak resistor to be large in value, so as not to load down the preceding stage by shunting AC signals to ground. However, there is a maximum allowable value given on the datasheet. For the ECC83 this is usually given as 2MΩ, although some quote 22MΩ provided the anode current does not exceed 5mA. Almost all circuits use 1MΩ as a convenient standard value, though most preamp valves will tolerate a higher value than given on the datasheet provided they are run at low anode currents (less than 5mA say), but there is usually no need to use a value greater than 1MΩ since higher resistances increase
noise while offering little extra benefit.
Power valves will usually have two maximum values listed: one for fixed bias and one for cathode bias, which will be somewhat higher. The reason for this is that with cathode bias, anode current cannot increase to such a great extent due to the self-regulating effect of the cathode resistor, inhibiting thermal runnaway. With fixed bias this is not the case so it is important that any charge that does collect on the grid can leak away quickly, which requires a smaller value of grid-leak. These values MUST NOT BE EXCEEDED."

Note the comments on Power Amps. Despite the 50K limit given in data sheets I have seen a large number of fixed bias 6550 and KT88 ccts with 100K resistors in the grid bias cct so at the moment I don't know what the story is here. [Edit, however the input resitance to the power tube grid cct as seen by the driving cct will place limitations on how low any grid leak resistance can go, due to loading of the driver/splitter stage.]

Also, as I said, the bias reading is way off. I'm still not clear on how to get that more into line. Can you help me out there?

Not really, as above, I'll leave that to someone else familiar with the bias cct and power supply in question.

I see I can get the book you mention from pentode press...I'll see if I can get that.

Both are very good books but do assume and require you have some familiarity with EE analysis to get maximum from them. They are quite rigorous.

[Andy Le Blanc]

You stated earlier that the plate volts were running around 430v with the 6l6.
relax a bit . The conditions presented in the GE and RCA manuals for your interpretation
are "Design Center" and for a plate V of 600. Design center are limiting values
that include variations in both the tube and operating conditions. Tubes are a 20% device.
The design center for the lowly 6l6 is a plate v of 360, you see how fender ran up the plate volts.
The 6550 also has ratings when operated ultra-linear where the screen is run
up to 450v with a fixed bias. So unless your design is going to ramp up the plate voltage you should be ok.
Mashall provides a couple examples, with g1 resistors of 82k and the screen
grid run at 478v, most US imports had 150k grid resistor. But where they did
run up the plate volts for the 200watt monsters then they used a 68k.
They used 1k screen grid for the most part. Your application wont strain
the tube unless you exceed the plate dissapation rateings. Give yourself
a reliable way to measure plate dissapation (1ohm at the cathode to measure current across)
and even a 1ohm before the screen grid resistors so you can know what they'er dissipation is too.

[greiswig]

Many thanks, Andy. Since this started as a D'Lite build, I already have the 1ohmR at the cathodes. That's where I was measuring upwards of 100ma at idle at the lowest setting on the bias pot. I didn't measure the voltage, though..but assuming it's the same, 40-43 watts at idle is about the maximum plate dissipation rating, not even the 70% recommended, right? Or are you saying that even at 100ma/430v, the ratings on the tube are so conservative that it should be okay?

If not, what component in the D'Lite circuit needs to be adjusted to get the bias down to somewhere around 65-70ma? According to Doug Aiken's excellent site, "In general, as the bias voltage becomes more negative, the bias current becomes smaller, and the tube is biased "colder"." So am I reading it correctly that lowering the resistance of the bias circuit by changing the 220k grid resistors to something more like the 50k that the tube data books call for...is that how one gets the adjustment into the desired ballpark?

Also, it seems like there is still confusion on why the value of the grid resistor is crucial at 50k for a 6550, versus the 220k the D'Lite circuit uses for 6L6's. Does it only have to do with bias adjustment, or is there some other parameter that is critical?

[Ears]

You stated earlier that the plate volts were running around 430v with the 6l6.
relax a bit . The conditions presented in the GE and RCA manuals for your interpretation
are "Design Center" and for a plate V of 600. Design center are limiting values
that include variations in both the tube and operating conditions. Tubes are a 20% device.

Andy would you mind expanding on this please.
I read the words "limiting values". I interpret that as meaning the limiting value that can be safely used in the design without having to factor in supply and cct component variations etc (i.e. they are unknown). If I knew maximum limits of supply etc in the application I could use design maximum values. And if I also knew the maximum variation of tube I could use absolute maximum values. I'm I little confused.

The design center for the lowly 6l6 is a plate v of 360, you see how fender ran up the plate volts.

RCA give 6l6GC design MAX of 500V - so I wonder what design max value for 6l6 is?

The 6550 also has ratings when operated ultra-linear where the screen is run
up to 450v with a fixed bias. So unless your design is going to ramp up the plate voltage you should be ok.

This comparison is likely OK for operation under full power. In UL the screen tracks the instantaneous plate voltage and at full power the feedback limits the screen current. In tetrode connection this doesn't happen, Vg2 stays high and the sceen will draw current, the problem being we don't know how much in this condition of operation. Also for interest see PRR's explanation (in regard to biasing UL and Tetrodes - gee I wish he would grace this forum again!) https://tubeamparchive.com/viewtopic.ph ... light=ears

> says -34V bias pp tetrode vs -75V bias for p-p UL

Look at the G2 voltage!!!

> I hadn't heard of that very large difference in bias requirements before, and it seems odd to me, as UL really only affect the AC operation of the circuit,

There's one big effect on DC operation: the G2 must sit at Plate voltage. With Pentode operation, G2 voltage may be very different from Plate voltage.

It is VERY convenient to set G2 just a bit below plate voltage, and when you take medium-big power from medium-big tubes, that can be accomodated in tube design decisions.

But when you want BIG power from medium-big tubes, you go in a direction which suggests setting Plate far higher than a sane G2 voltage. 8417 likes 600V Plate 300V G2. 6146 likes 500V-600V plate 150V G2. KT88 is in this class: it has a high plate voltage rating but not a very high plate dissipation rating. We can run high voltage, but not high current. So we have to keep current down, and we do this with grid voltages.

KT88 has a pentode ("Tetrode") suggestion for 300V on G2 and a rather small G1 bias. But in UltraLinear, the G2 "must" be way up at plate voltage. (OK, there is a trick with a tertiary winding but it is a lot of trouble.)

Another way to look at it: in pentode mode, the G2 stays at 300V through the audio cycle. In UL mode, the G2 tracks 40% of plate voltage and dips low just when we need maximum current. To get the peak current, the G2 idle voltage must start high (and plate DC voltage is natural). But if G2 voltage is high, G1 voltage must be large or that silly little 40W plate will melt.

Yet another way to look at it (Ears already saw this): Triode and UL are Feedback Amps. Compared to Pentode, they have less gain and lower distortion and output impedance. The improvement is about proportional to loss of gain. UL is usually aimed at about half the gain of Pentode: enough to improve output, not enough to strain the driver. So we need twice the G1 swing, but G1 still can't go (conveniently) positive, so somehow we are going to wind up near twice the bias.

> I now wonder how ccts that switch in/out of U-L get around the differing drive requirements if they wish to maintain ability to deliver max power.

If the Pentode amp was optimized for G2 voltage nearly equal to plate voltage: bias stays essentially the same.

If the Pentode amp load and G2 voltage were optimized for Power Output: switching G2 to 40% taps is gonna cause a loss of power due to low G2 voltage at the pull-down peak of the cycle.

But the difference between Naked Pentode and Half-Triode operation, especially in low-NFB amps like many guitar amps, is more impressive than a few (dozen) Watts more or less. There's nothing like the raw sound of an overdriven pentode, love it or hate it.

Mashall provides a couple examples, with g1 resistors of 82k and the screen
grid run at 478v, most US imports had 150k grid resistor. But where they did
run up the plate volts for the 200watt monsters then they used a 68k.
They used 1k screen grid for the most part. Your application wont strain
the tube unless you exceed the plate dissapation rateings. Give yourself
a reliable way to measure plate dissapation (1ohm at the cathode to measure current across)
and even a 1ohm before the screen grid resistors so you can know what they'er dissipation is too.

You are most probably right, but I don't like unknowns and grey areas in my understanding and to me it seems some aspects designing with tube technology remain full of it. I find it both fascinating and frustrating

[austinb]

You might add a series cap in the bias ac voltage circuit before rectifier for a voltage doubler to get twice the - bias voltage.

[Andy Le Blanc]

there are three rateing systems. The oldest is "absolute maximum" its the area of operation
that must be avoided under any area of operation, certain death.
its used for transmitting and industrial tubes, not generally receiving types.
prior to 1957, "design center" these make the assumption that the tube and operating conditions are identical to the published values, as "normal" conditions.
then there is "design center", a standard adoped in 1957 by NEMA and EIA
which assumes that the tube is identical to the published data but does not provide for variations of operating conditions...

all data is based on the assumption that the tube or operation is "EQUAL" to
what ever they used to make the data.

which assumption do you trust, they are examples of succesful applications.

but the only hard limits are voltage and dissipation in watts.
as with the quoted info about UL, which is very nice.

the 50k screen resistor is based on the assumption that you are duplicating the conditions found in the data, which is not the case.
Id try 100k. and record your static dissipation figures. plate V, ma..

spend some time going over schematics that look close to your project.
these are also successful applications.

if the bias voltage is not enough you can add resistance between the bias pot. and ground.
you are correct about the bias v, its better to have control, and set it where you want
you might not always have the choice of tubes so youll want to have a good range with the bias control.

[Ears]

all data is based on the assumption that the tube or operation is "EQUAL" to
what ever they used to make the data.

Thanks Andy,
I interpret this RCA manual stuff as saying that the ratings take into account that the tube may vary, i.e. the quoted ratings allow for normal tube variation and are still safe. Are we meaning the same thing?
PS I think there is a typo in your post, Design MAX was adopted after 1957.
While we are in the manual, go to page 99 then pg 300 (in my RC-28 ) , or wherever the
" Grid No 2 input rating chart" is drawn. The text states that it must only be used for the tubes "for which it is specified in the data section".
What does the quote mean, where in the RCa manual does it give such specifications, can someone find me an example?

[Andy Le Blanc]

Its "a tube of the specified type having characteristics equal to the published values under any conditions of operation."
that leaves a lot to the consumers qualitative judgement.

the reff. I have says "adopted in 1957" but you may be right.
I second your statement about pg 300 in RC-28, its very late to judge errata
in a publication so old. but I would also like to see that clarified.

the equation might be good , use the 6l6 design center.

Rg2 = Ec2 x (Ecc2-Ec2)/Pc2

you need 270 v and the screen has a 2.5w rating, but have a 300v supply

270x(300-270)/2.5....... thats around a 3.2k resistor

seems off... but maybe it is an error.

[Ears]

Its "a tube of the specified type having characteristics equal to the published values under any conditions of operation."
that leaves a lot to the consumers qualitative judgement.

I find myself thinking that all too often.

the reff. I have says "adopted in 1957" but you may be right.

My manual states
(for Design centre) "these ratings, which include allowances for normal variations in both tube characteristics and operating conditions, were used for most receiving tubes prior to 1957."

and for design maximum

"These ratings allowances include for normal variations in tube characteristics but do not provide for variations in operating conditions. max ratings were adopted for Receiving Tubes after 1957"

by "operating conditions" I assume component and supply variation/drift.
Regardless I am not really much wiser as to the most efficient manner to use the various ratings in design.

I second your statement about pg 300 in RC-28, its very late to judge errata

in a publication so old. but I would also like to see that clarified.

Is it errata? Probably is, it certainly appears in an odd place in the book. Yeah, I would just like to know what the heck tubes they use the chart for, I've flicked though lots of entries for beam tubes and pentodes (and any tubes that have G2) haven't yet found any example of any further statement that refers to this chart (and so is specified by) and the tube in question. [Edit: now located few references, for example 6DC6 the reference is written amongst the max rating values. That's the great thing about these forums, you can make your mistakes in front of the world]

the equation might be good , use the 6l6 design center.

Rg2 = Ec2 x (Ecc2-Ec2)/Pc2

you need 270 v and the screen has a 2.5w rating, but have a 300v supply

270x(300-270)/2.5....... thats around a 3.2k resistor

seems off... but maybe it is an error.

I'ts about 9mA though the resistor, If that is referring to max power screen current for a power tube it is in the ball park. Bit high for idle though. But the chart is for voltage amps.

[greiswig]

Okay, I had a bad set of 6550 tubes, but got some good ones. They biased up fine with the stock setup; I could set them anywhere between about 20ma on up. I set these up for 42ma and played around there in the low to mid 40s.

I did not change the grid resistors from their 220k value, and I have no idea what sonic effect changing those closer to the 50k recommendation might do.

Initial findings: I don't care for them with this amp, a D'Lite with Brandon's Twin iron. They're different from the 6L6's, in not an altogether bad way. They seem less crisp than the 6L6s, and have more low mids to them. They thicken up single coils pretty nicely, actually. Very tight bottom end, and noticeably more headroom on the clean channel.

But they seem to lack some of that sizzle and harmonic character that the 6L6s provide. As I said, I don't know what lowering the grid resistors for the 6550's might do for their tone.

[Andy Le Blanc]

I remember hearing that lowering the grid resistors was a to get more grind out of a marshall.
what type and brand 6550 did you get?
where have you set the tubes plate dissipation in watts?

going from one type to another can have mixed results, double check your
loading of the amp with eye after effective impeadance that the tubes "see".
You might find that the set up respondes very differntly with a different load,
sometimes the fractional differences between differnt speakers can have an effect.

Do you have a signal generator? You can test you OPT with one for impeadance, to double check loading.

[greiswig]

I remember hearing that lowering the grid resistors was a to get more grind out of a marshall.

I was just going based on Ampeg SVT schematics that I've seen, and on tube documentation that lists the maximum grid resistance as 50kOhms. I'm not clear on whether this will have any sonic effects or whether it just affects the range of the bias adjustment. Do you know?

what type and brand 6550 did you get?

These are Tung Sol tubes.

where have you set the tubes plate dissipation in watts?

They're dissipating about 21W at idle, 48ma, when they seem to sound the fullest without starting to get tubby. The 6L6s in this unit seem to sound best at about 13W at idle, 33ma

going from one type to another can have mixed results, double check your
loading of the amp with eye after effective impeadance that the tubes "see".
You might find that the set up respondes very differntly with a different load,
sometimes the fractional differences between differnt speakers can have an effect.

Do you have a signal generator? You can test you OPT with one for impeadance, to double check loading.

Interesting...yes, I have both a scope and a sine wave generator. I've not heard this kind of test described.

[Andy Le Blanc]

check out the schematic heaven site, there's many examples, by various
makers. Theres quit a few examples at the marshall end of the site.
and others through out the site , its huge.

With the signal generator, you connect it to the secondary of the transformer
and turn up the amplitude until you get 1v at the secondary. Then you measure the resulting voltage at the primary.

and this is what it tells you, ( from the Aiken Site)

"The impedance ratio is the square of the turns ratio, which is also the square of the voltage ratio, as shown in the following equation:

Zp/Zs = (Np/Ns)2 = (Vp/Vs)2

If you put 1VAC across the secondary, and measure 20VAC across the primary, you have a turns ratio of 20:1, which corresponds to an impedance ratio of 400:1. This means that if you put an 8 ohm load across the secondary, you will get a reflected impedance of 3.2K ohms across the primary. If you put a 4 ohm load across the secondary, you will get a reflected load of 1.6K ohms.

For example, if you have a transformer designed for 4.3K : 8 ohms, you can apply a 1 volt AC signal across the secondary 8 ohm winding, and you should see 23.18VAC across the primary, which corresponds to a 23.18:1 voltage ratio or a 537.5:1 impedance ratio, which would reflect an 8 ohm load back as 4.3K."

with the plate dissipation follow your ear of course, but be exhaustive and listen to
the range that is available to you , with the scope ( and dummy load) you
can find where crossover distortion is and where it begins to effect the performance, but use your meter and do the math to set operating point.
That is the "number" that will corilate to the "published data"
you can also use the rig to find max output power, max undistorted output
and generate zero-signal and max signal plate current figures, that and a good figure on your load from your tranny test you can fully describe
the typical operating characteristics of your amp, for a given load.
and make use of other figures generated from the data youve taken.
there are some really good materials at the pentode press site to take advantage of , if you can generate your own numbers.