6922 vs 6SN7

[rdf]

Hi Chris. From a quick perusal the entire 1960's Scott tuner line employed essentially the same 12AT7 topology. Not knocking the 12Axx line in general but even the beginnings of the start of the hint of an inkling of a notion yesterday's engineers are always right is an itch I can't help but scratch.

[Miles Prower]

Originally posted by rdf
I agree with the principle above in general and have taken the Tubelab approach of poring over curves when checking out sales and picking tubes for their curves, not their reps. Some amazing unknown performers are still waiting for the right circuit. (Best example so far was a $2 triode-diode-diode that swung 160 p-p at <0.5% pure second coupled to a cathode follower, no feedback. :P)

That's how I do it as well. Ignore the folk "wisdom" and let the characteristics speak for themselves. Uncovered a few good audio performers that way too.

[anatech]

Hi rdf,
Please, widen your search. Both in time and manufacturer.

I will point out one iffy case. The Fisher 400 control amplifier. This unit uses feedback to correct drive problems. It even inserts a balance control in the loop. Questionable, but it still sounds good.

I believe that the 60's were an era where the quality began to suffer. It's interesting to note that the electronics were still far superior to the loudspeakers in use at that time. Possibly this fact was not lost to the engineers and accountants?

Not knocking the 12Axx line in general but even the beginnings of the start of the hint of an inkling of a notion yesterday's engineers are always right is an itch I can't help but scratch.

Scratch away!
They may have made cost cutting moves, but in general they at least knew how these circuits worked properly. Even if they didn't always make the altruistic move that we may think they should have. How often does a technically perfect design actually hit the design floors today? This is not because the engineers didn't know any better, well not every time anyway!

If you look back to power supply design for instance. How many products caused lights to flicker or rectifiers to arc when turned on? Yet they managed acceptable ripple levels and could do much better if you study some test equipment. Most of todays designs are severely lacking in the area of power supply design.

I think they understood how to apply the different tube types to the job at hand fairly well. Remember that most 6DJ8 tubes were microphonic back then.

-Chris

[anatech]

Ignore the folk "wisdom" and let the characteristics speak for themselves. Uncovered a few good audio performers that way too.

Yep.

How could you disagree with common sense?

-Chris

[tubelab.com]

I hesitate to join in any "my tube is better than your tube" discussion, because, as already pointed out, each of these tubes have their own virtues, and disadvantages. These two tubes are different enough that the choice between them would depend on the application. There is one more thing that has not been mentioned yet. Many 6SN7's show some signs of microphony. This would preclude their use in low level stages that are followed by lots of gain.

This brings me to the "Tubelab approach". I have collected about 100,000 tubes over the past few years. I have started evaluating the ones that I have in large quantity. I start with the published curves, which often don't match the actual tubes. Pay attention to the original application, often tubes designed for "gain controlled RF or IF amplifiers" are nonlinear at high signal levels. If you have any broken tubes, take them apart and study the construction. Look at the uniformity of the grid windings. Some are purposely wound in a non uniform maner, for gain controlled applications. Finally, test, and test some more. Test the tube in the chosen circuit at the expected signal level.

I have also collected a lot of RF and audio test equipment over the years, but today I do most of my testing with a computer containing a high end sound card (at least 24 bit 96 KHz) and some analysis software. You can measure the individual harmonics, and see (on the FFT display) how they change with drive level, bias current, and plate voltage. I also use a scope and a variable power supply. The variable supply is essential for finding the best operating point for each tube. You should also test several samples of each tube in different brands if possible. Some tubes are highly variable. If a CCS load is a possibility, then test the tube with a CCS load and a resistive load. Some poorly regarded triodes really shine with a CCS load. Some don't like it at all.

I have not had much time to work on this lately, but the data will wind up on my web site sooner or later.

Back to the thread, 6922 or 6SN7. What do you want the tube for. Phono stage, 6922. Output tube driver, 6SN7.

The 12at7 is a very high distortion producer, as well as limited bandwidth. Take a look at the curves from the links below. And actual measurements have been performed.

The 12AT7 was originally designed for VHF amplifier service, bandwidth, OH 100 MHz or so. It does have a reputation for excessive second harmonic generation. That can be tamed with CCS loading, and about 10 mA of tube current. I have measured the 3db point in driver application (driving a triode wired 6L6GC) at 102 KHz, more than enough bandwidth. The disrortion of the complete amp is about 1% at 5 watts (with a KT88).

[anatech]

Hi tubelab,

I hesitate to join in any "my tube is better than your tube" discussion, because, as already pointed out, each of these tubes have their own virtues, and disadvantages.

Welcome to the fray! Since you have been quoted, I expected you to arrive.

I don't think you will find many people who will disagree with you on these points. You're not old but you are involved with far more tubes than one man should legally be allowed to own! More power to ya!

You do have an excellent web site there. Thank you for making it available to the world.

-Chris

[Positron]

"Way back I thought the same about the ECC81. Now I changed my decision. Look THIS thread please."
Yep, the Ecc81 has approx 10k ohms Rp, mu 60, and gp of 1.6pf.
-----------------------
Hi Anatech,

"I personally do not care that much for the cheap little 6DJ8. It was designed as a UHF tuner tube in cascode operation. It was never conceived as an audio tube. So it started life as a cheap TV tube. Yes, it can perform well and I will not take that away from it. It does well when designed with properly, but then again, so do the other tubes."

Hey it was designed for minimal sideband production, so low distortion is required. In fact the E88cc is very low in distortion.
So stop the crap.

"Mr 6SN7 kicks the 6DJ8 when you need large voltage swings. BTW, the 6DJ8 was designed with high plate current and low plate voltage in mind."

And I clearly mentioned that in a diffy circuit, the E88cc can produce 70 volts p-p with .05% distortion, with the same gain, and with a lower output Z than the 6sn7. But the response was concerning Miles post, which was preamps, and the misconception that only 6sn7s are any good.
Sy added the comment about large swings. Let's get back in focus, shall we.


"It's not happy when used at the higher plate voltages and lower currents that were typical of audio designs."

That is BS and you know it. (Maybe not since I see your next statement below.) Don't try that cheap trick 'not happy' crap with me.
If you were to check the JJ tech sheet, the tube can be run at 220 volts at less than .8 watt plate dissapation and they sound great. I run considerably less plate voltage. I have been doing it for years, and the tubes have been lasting for years with no problems.

Unfortunately, you really don't understand much, or else you have nafarious intentions behind the charade. See next.

"Hmmmm, high voltage and low current. Sounds like a high impedance circuit to me. That would be Miller time I guess."

Well, a typical tube tech (not even an engineer) would be laughing right now. If one simply inspects the plate curves of the E88cc, one will find the Rp remains almost constant until the plate current reaches down about 2 ma, maybe 2.5 ma.

So just any tube won't do, as you would like to deceive the audiophiles into believing.

So your last few comments just demonstrated how ignorant you really are about tubes and designing.

-------------------------

Hi Sy,

"Steve, regarding bandwidth, I don't follow your point. Let's look at the worst-case input capacitance, 40pF. Further assume that the tube is used at an amp input and is being driven by 5k of source impedance (a pretty high number). The f3 is 850kHz. I suspect that's far enough above the human hearing range to not matter."

5K may be accurate, maybe higher depending on RL and the capacitance will probably be higher, but what the heck.

Listen to it, then add another 40pf, so approx 80pf total, and see? (You won't be able to duplicate a second way I do it and I check for discrepancies.) You have only lowered the F3 (3db down) to approx 425khz. See if it makes a difference?

Frank Van Alstine claims to hear differences up to 6 mhz, although I doubt that high. By the way, we haven't counted the IC capacitance yet.

"I've measured the distortion spectra of a LOT of ECC88 diff amp circuits. Let's see, you've got 100V pp; I assume that's a differential number? So each half is running 50Vp-p, or about 17VRMS. That's still fairly high for that tube, high enough that I suspect your distortion number is after feedback- the "irreducible" 3rd order distortion for the best ECC88 type I've measured (CCa) at that level is about twice that."

Some corrections. First each section is swinging 100p-p reference to ground. Each 6550 has -50, so 100 p-p. Secondly, No feedback. Thirdly, I am using a JJ, with about 1/8 to 1/10 the distortion of any other 6dj8. (And I could do better with a higher cathode Z.)

At that signal level, a diff pair of 6SN7 will have about 0.02% open loop third harmonic. Morgan Jones reported that a diff pair of 6SN7 was able to swing 177VRMS (250VRMS) at 0.11%."

I could do some special things as well but I am not. For Jones' certainly does require special treatment cause it isn't a normal design. 250 vrms is about 700 volts peak to peak. But let's play the game.

He isn't within specs, so I hope anatech gets on Jone's case. Let's find out by his response, or lack thereof. And both designs have serious problems.

First off, either a very high B+ and large plate resistors, or using constant current ss devices. SS, both as constant plate current devices and high Z cathode devices just flat out ruins the sound, so we can dispense with that right now without further discussion.

If large plate resistors are used, then a very high voltage supply, with either series filter caps, or some exotic high voltage caps are used. So we aint talking your normal design. Want to try a normal one? But there are still several other problems.

First, the RL (plate load resistor/device) raises the plate circuit impedance. Secondly, the Rp of the tube is around double the value of a typical common cathode circuit. So the effective plate Z is at least twice or more higher than even a simple design, such as I mention used in a preamp.

This means the highs will be even worse than in a conventional circuit. Of course I guess one could use negative feedback to reduce distortion.

"Now, let's look at the argument about output stage drive. For universality, we consider the Williamson amp as our example. The driver stage has a source impedance of about 8k7, the output stage has an input capacitance of 55pF, so we're facing an f3 over 300kHz. That might be too constrictive for you, but around here, output transformers have cutoffs 5-10 times lower."

So? You obviously have never bothered to check it out.
That is what this string is actually about; making wild claims as if fact, but not even tested.

As far as changing subjects, why change subjects in the first place?

My first post is in response to Mile's comment:
"I have some older preamplifiers that use 12AX7's and 12AT7's. They sound better than many newer 6DJ8 type units. What can I say??"

So I respond: "I agree, many newer components using 6dj8s are poor sounding." I then state later.
"Let's take a look at the 6sn7." and the rest of my post has to do with preamp use, responding to Mile's. Very simple.

Next Sy replies: "Steve, it's a matter of application. If I had to swing 70V, the 6SN7 would be one of the first tubes I'd reach for (large signal linearity is quite good) and the ECC88 one of the last. If I want a cathode follower for a preamp where the swing is 2-4V, the 6SN7 would not be optimum, but an ECC88 would be close."

First of all, we were talking preamps. Secondly, you fail to mention the E88cc as a gainstage. Just forget?

Thirdly, you were the one who upscaled the discussion to large swings, not me.

I then simply responded that the high frequency problems also occur when both used as a converntional common cathode preamp or as a common cathode driver.

Then Miles mentions about HF response not being needed. I defended my position that above 20khz is important.

Then Anatech makes this statement: "Hey Steve,
You keep dragging your comparison back to a line stage output.
You are arguing a point while focusing in one direction only. Open your eyes man!"

Well, DAH. Preamps is the subject. That is what I was responding to anatech, Miles comment about preamp stages.

And Sys is the one who expanded it to 70 volt signals etc.

And anatech came off with some stupid comment about Rp that demonstrated he is playing a charade game. Anatech doesn't really have a clue what he is talking about.

Now the real question is: 'what is the real reason you guys are trying to refocus the issue to another subject, so as to improve the image of the 6sn7? Anatech even went so far as to make a complete fool of himself.
Why not use 6H30s, 12gn7s, 5842s, El34 or 6550?

Why do you need such high swing voltages? Who cares?

"I'm not trying to sell this tube or make audio lore, I use both in my system, they are just very different tubes with very different strengths."

Fine, no problem Sy. Then don't change the subject at hand, the use of the 6sn7 in preamps.

"And I caveat all this by assuming GOOD examples of both tubes, not things like relabeled 6ES8 or 7A4."

Well, of course.

"edit: I hadn't noticed until Chris pointed it out that you're in the business of making and selling amps. I do not intend to step on your commercial toes, but I won't avoid them, either. If you're using ECC88 in your preamps, I can't object to that since I do the same thing."

Do I need your permission to use a tube?

[rdf]

Hi Chris. Without drifting too far off topic, I don't find issue with anything you wrote. To be completely fair, with the output swap, some power supply work and replacement of the 101 ceramic caps with poly* or silver mica it's a tremendous tuner and a testamant to their engineering prowess the design was stable enough for kit form. The Scott wasn't intended to demonstrate a universal but as an example from a name manufacturer that fashion, marketing, stock convenience and accountants were operating considerations in the Sixties too. The output was clever in one sense, the feedback ratio rose with increased static output level, a kind of 'on demand' degeneration. The Scott isn't however an argument for using 12AT7s.

I saw the same sort of thing in a Rowe jukebox amp, a "good 'nuf" circuit designed mid-Fifites carried through for fifteen years to the end of the glass era. Here a uamp biased 12ax7 was used as front end and splitter into near-Class B 7868 outputs. It was suitably engineered for its purpose - very long MTBF - but the presence of a 12AX7 is more a testament to its high gain than sonic superiority. I don't know enough detail about the times but suspect the late Fifties and Sixties was also when engineers became completely comfortable with GNFB and gain rose in importance against base linearity.

And never discount simple error, even acknowledged classics like the Dyna FM-3 are recongnized today to have incorrect NRSC de-emphasis (Dick Sequerra designed a correction.)

Without knowing why I would never choose a tube based on its popularity with 'the ancient ones', no matter what their skills or how ubiquitous. We also have commodity measurement hardware they didn't dare dream could exist.

[SY]

The 12at7 is a very high distortion producer, as well as limited bandwidth. Take a look at the curves from the links below. And actual measurements have been performed.

I missed that one before. Have you actually set up a good 12AT7/ECC81 in a grounded cathode circuit and measured linearity? I have done so with about 30-40 samples at varying currents and bias. My results have been posted here several times; to summarize:

ECC88 (Amperex Bugle Boy, average of four sections) 1.7V bias, red LED
14VRMS output, 1kHz
-44dB 2nd
-71dB 3rd
-87dB 4th
-90dB 5th

ECC81 (JJ, average of ten sections) 1.7V bias, red LED
15VRMS output, 1kHz
-66dB 2nd
-91dB 3rd
-99dB 4th
<-100dB 5th

Sorry that these weren't at exactly the same output voltage, but they weren't done for comparing one tube type to the other and were run a year apart. Same test setup, though.

edit: corrected test level units

[SY]

Thirdly, I am using a JJ, with about 1/8 to 1/10 the distortion of any other 6dj8.

That's good to know- I haven't used them, but I saw the same thing with their ECC81- its distortion was 6dB better than any other ECC81 I've measured.

Regarding Jones's design, you might want to actually look at it before throwing stones. It is a VERY carefully engineered circuit. It's described in great detail in "Valve Amplifiers" 3rd edition, the chapter on the Crystal Palace amp. I've spent a few evenings listening to that amp and if the cathode CCS ruins the sound, give me more ruined sound.

I don't understand your comments about the Williamson and the circuit analysis. Could you be a bit less elliptical? I'm rather slow.