The 12AV7 couldn't be too bad an audio tube if Graaf use it in their flagship GM200 OTL amplifier... without any doubt one of the finest sounding amplifier out there...
Ok, you have me fascinated... why would this be flawed reasoning? If the tube was non-linear or resulted in poor sound at the beginning of the chain, all that could be done later would be to amplify this or make it worse...
Not necessarily. The Graaf is a high feedback amp- this means that nonlinearities in the circuit path are reduced by the feedback factor. The distortion is still pretty high and has a lot of high order harmonics, so that's a clue that the open loop performance is nothing great (the closed loop performance is nothing special, either, especially considering the price, the heat, and the power consumption). The tube could have been chosen for many reasons having nothing to do with linearity.
I've never had a 12AV7 on the bench, so I have no idea if it's actually nonlinear or whether that's another Audiophile Legend.
I've never had a 12AV7 on the bench, so I have no idea if it's actually nonlinear or whether that's another Audiophile Legend.
So, I'm looking at http://tubedata.milbert.com/sheets/093/1/12AV7.pdf . When looking at the graphs, what would flag that this is a poor triode?
Thanks,
Ray
Thanks,
Ray
There was some careful weasel wording. Marketing. Don't take claims from high end manufacturers at face value.
"When looking at the graphs, what would flag that this is a poor triode?"
When a tilted load line is plotted on the 12AV7 curves, the spacing of grid1 voltage steps to plate voltage steps varies significantly from one end to the other of the load line. Indicating varying gain, hence distortion (largely 2nd harmonic from the monotonic deviation). The question in my mind arises when these are compared to 12AT7 curves, which show even worse variation. Clearly neither of these would be suitable for SE operation unless one wanted some distortion cancellation effects with the subsequent stage.
The 12AT7 is known for working well in a differential class A stage though, where this largely 2nd harmonic distortion cancels out, and may even produce lower 3rd harmonic dist than two constant Mu tubes (the flattened and ramping up gm curves add together better to a constant total gm). One would expect similar results for the 12AV7, but I haven't seen or heard any results.
Another artifact is apparent in the 12AV7 curves (looking at the GE datasheet) which is likely produced by the curve tracer used and not the tube. Every other curve has alternating slightly large or small spacing from the next. This occurs frequently on curve tracers when both sides of the 60 Hz waveform are used for the tube HV B+, since alternating traces are from separate sides of the sinewave power waveform and are traced in opposite direction on the screen. Capacitance induced phase shift, and power supply hum then cause offsets in the alternating traces. Very common in older tracers which need new caps installed in the power supply.
When a tilted load line is plotted on the 12AV7 curves, the spacing of grid1 voltage steps to plate voltage steps varies significantly from one end to the other of the load line. Indicating varying gain, hence distortion (largely 2nd harmonic from the monotonic deviation). The question in my mind arises when these are compared to 12AT7 curves, which show even worse variation. Clearly neither of these would be suitable for SE operation unless one wanted some distortion cancellation effects with the subsequent stage.
The 12AT7 is known for working well in a differential class A stage though, where this largely 2nd harmonic distortion cancels out, and may even produce lower 3rd harmonic dist than two constant Mu tubes (the flattened and ramping up gm curves add together better to a constant total gm). One would expect similar results for the 12AV7, but I haven't seen or heard any results.
Another artifact is apparent in the 12AV7 curves (looking at the GE datasheet) which is likely produced by the curve tracer used and not the tube. Every other curve has alternating slightly large or small spacing from the next. This occurs frequently on curve tracers when both sides of the 60 Hz waveform are used for the tube HV B+, since alternating traces are from separate sides of the sinewave power waveform and are traced in opposite direction on the screen. Capacitance induced phase shift, and power supply hum then cause offsets in the alternating traces. Very common in older tracers which need new caps installed in the power supply.
Test for Broskie Idea?
Hi.
John Broskie has an article on using a grounded cathode stage cascading into a cathode follower at the same bias point (same tube type) canceling the distortion. It occurs to me that this could be a good tube to try this with.
Maybe someone out who there has some of these tubes and is set up to do distortion can try it out and let us know? It would be interesting to know if higher order harmonics are created as low order ones are diminished.
Just throwing it out there.
Rolf.
Hi.
John Broskie has an article on using a grounded cathode stage cascading into a cathode follower at the same bias point (same tube type) canceling the distortion. It occurs to me that this could be a good tube to try this with.
Maybe someone out who there has some of these tubes and is set up to do distortion can try it out and let us know? It would be interesting to know if higher order harmonics are created as low order ones are diminished.
Just throwing it out there.
Rolf.
Well, SY, I can only talk from first-hand experience as to how the amp sounds, and it is head and shoulders above a wide range of amps such as the Nuforce, Wolcotts, Musical Fidelitys, etc. that I have tried. Associated equipment is Wilson Benesch Bishop speakers, Lyngdorf DPA-1 and Sonic Frontiers Line 3 SE preamps, and Esoteric DV50s SACD player.
Admittedly, the Graaf is an OTL amp, which may have something to do with it, but it the first tube in the signal path is the 12AV7 and it seems to have very little effect on the superb sound of this amp.
Which brings into question the wisdom of looking at distortion figures on the sound of amplifiers. I personally think that extended frequency responses and rise times have far more of an effect on the perceived quality of sound than distortion does.
Maybe it's the distortion that you like. Maybe it's the gross frequency response errors. It can't be the "extended frequency response"- the high end rolloff is significant and likely audible if you have good hearing in the top octave.
I would be the last one to tell you what you should and shouldn't like, but I have no hesitation in saying that this is an effects box, not an accurate amplifier. You may like the effect on your particular speakers, in which case, more power to you (which you'll need to run those monsters and the air conditioning to keep the room from baking). The 12AV7 may be deliberately used for achieving the "desired" distortion, if indeed it's as nonlinear as several of the experienced guys on this thread claim. Or maybe the swing in the first stage is limited, and the manufacturer had a cheap source of 12AV7...
I would be the last one to tell you what you should and shouldn't like, but I have no hesitation in saying that this is an effects box, not an accurate amplifier. You may like the effect on your particular speakers, in which case, more power to you (which you'll need to run those monsters and the air conditioning to keep the room from baking). The 12AV7 may be deliberately used for achieving the "desired" distortion, if indeed it's as nonlinear as several of the experienced guys on this thread claim. Or maybe the swing in the first stage is limited, and the manufacturer had a cheap source of 12AV7...
12AV7 could almost be described as semi-remote cutoff which, given its RF intention, is not too surprising. The mu curve is not too far off linear, so I guess in the right circuit it would give mainly 2nd order distortion. So possible use as an input stage, with low signal levels, or a balanced phase splitter (where 2nd will cancel). High gm would give low noise at RF, but not necessarily for audio where 1/f dominates. Something to watch would be the limited heater-cathode voltage rating.
It does have a higher-than-usual heater power, and this seems to be a feature of valves popular in certain circles. My guess is that for most purposes you could find another valve which would do a better job, unless you wanted to add 'warmth' to the sound.
It does have a higher-than-usual heater power, and this seems to be a feature of valves popular in certain circles. My guess is that for most purposes you could find another valve which would do a better job, unless you wanted to add 'warmth' to the sound.
I suppose you get all kinds on these boards, but this is just an ignorant reply...
Maybe, maybe, likely, if indeed non-linear... no hesitation...large amounts of feedback...
I'd say that this was an engineer's usual opinion (i.e. forget what you are hearing, it must sound bad if it measures poorly) if it wasn't for the fact that the engineering statements are so poor! 18dB of overall feedback being described as a "large amount of feedback"? The "high-end roll-off is significant" when the frequency response is 1dB down at 50kHz or 1dB down at 20kHz at full load? "audible if you have good hearing in the top octave"... well, maybe if you are a bat! "Power to run those monsters" and "airconditioning" when the normal power usage is around 800W and temperature measurements on the amp show a maximum of 40C except on the tube cages directly above the tubes, where it is 60C? Compare that to a Pass XA30.5 where "I therefore preconditioned the Pass amplifier by running it at 40Wpc into 8 ohms for an hour. The heatsinks were just over 60°C (140°F) at the end of that period;"
What is even scarier is that this rubbish is coming from a forum moderator! Get your facts right and your very obvious biases out of the way before replying please. And please also focus on getting your basic engineering right too.
Yeah, I'll have to work harder on understanding that engineering stuff. In the words of Barbie, "Math is hard!"
Take a look at the measurements published by Stereophile.
Take a look at the measurements published by Stereophile.
SY, I'm ashamed that you would allow measurements to get in the way of a good marketing exercise.
I did, and they conclude that "the amp was capable of conjuring up stunningly beautiful music that inspired and inflamed my imagination", "the GM 200 is entirely recommended" and "The Graaf's test-bench measurements were respectable".
A somewhat different opinion to your "I have no hesitation in saying that this is an effects box, not an accurate amplifier"...
But please don't let me stop you from giving yet another ignorant reply... you obviously know a great deal about the 12AV7 having previously said that “I've never had a 12AV7 on the bench, so I have no idea if it's actually nonlinear or whether that's another Audiophile Legend”
Interesting discussion. The mention of semi-remote cut off behavior induced me to pull up the data sheet and have a look. It does indeed look to this rookie like over a wide input voltage range the gain can be varied enough to make a kind of AGC. However by looking at the output voltage swing at infinite load at 7.5mA (like a CCS) my admittedly quick look seemed to indicate that it would be pretty symmetrical as long as the input voltage was no more than about 2V p-p. So for my purposes it might not be non-linear enough for a MI effects box.
Your observations based on inspecting the data sheet match my limited experience with this tube quite well. I've experimented with them in the second stage of a phono stage where the signal levels were low, and in a few other applications - they do work well as phase splitters. I was just curious about their overall performance, given their lack of availability I was not thinking about using them in a commercial design at the time. I also found other types I preferred. My samples exhibited reasonable linearity in the phono stage, but had insufficient mu for the application in question to be seriously considered.
For best linearity you can operate them at high current and relatively modest plate voltage in applications with limited swing requirements - they'd do well enough IMHO. (I mean right near the limits of their plate dissipation..)
In the world of IDHT/P I much prefer the 5842, or triode connected D3A, 7788 or C3G to any of the 12xx7 series. Much less noise, much lower rp, better linearity..
To the OP I'm not a big fan of distortion cancellation and you can't do it at all with a cascade common cathode - cathode follower as they don't have similar distortion characteristics. You can get some partial canceling with careful design choices, but not necessarily by cascading two identical stages unless you assure that they closely matched and the input levels are identical in both (essentially you would attenuate the output of the first stage so that the input level to the second stage matches the first - what's the point, why not choose a more linear tube if large signal swings are needed??) One area where this might be interesting would be to provide some distortion canceling in an SE amp based around a tube like the 6CK4..
I've heard just one Graaf OTL quite a few years ago and I thought it sounded quite good relative to other OTLs I had heard up to then, but it did sound quite warm and romantic compared to what I had expected. (Big Atmosphere OTL amps and high power ARC and other UL PP amps) These days I listen to the most neutral 300B SE amps I know how to make so I would hardly call any of the above an indictment.. 😀
Last edited:
- Status
- Not open for further replies.