 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
85.2.23.243
Gang,
Just ran a series of tests on a SVP-1 line stage, which is very close to the FVP5A line stage linked below.
It uses a single common cathode 6DJ8 gain stage and a SLCF output stage - the gain stage has NO cathode bypass cap (of course) and you can see from the map that there is no loop NFB. Otherwise it looks kinda basic, no CCS anode loads etc.
At 0dBV (1VRMS) out: 2nd H = 0.012% 3rd H = 0.006% 4th H = 0.001% 5th H = 0.002%
At+10dBV (3.16 VRM): 2nd H = 0.033% 3rd H = 0.001% 4th H = 0.000% 5th H = 0.002%
At +20dBV (10 VRMS): 2nd H = 0.110% 3rd h = 0.002% 4th H = 0.001% 5th H = 0.001%
This is using a 16 Bit/44k DAAS spectrum analyser card in a PC - could post the spectrum graphs for the non belivers from home.
The second harmonics naturally rise in relation to the output level - it is after all a single ended circuit - but where's all the horrific odd order harmonics that so many experts claim this tube generates?
And it sounds as good as it looks!
Regards, Allen
Follow Ups:
As noted below much of the "Myth" was probably associated with 6ES8 being relabelled as 6DJ8.
To make two WRONGS make a WRIGHT (Appologies Allen) what you do with the 6ES8 is run it as a Miss Piggy (Cathode Follower) at around Va = 150V and Ia = 10mA as the output of a line stage. It imparts a delicate euphonic balance of 2nd and 3rd harmonic distortions ideal for "warming up" that SS Power Amp you've had collecting dust on the shelf.
No - Its not really suitable for driving a tube power amp but that particular combination of "warts" does impart a surprising satisfying "tubiness" to an SS Amp.
Cheers,
Ian
I remember complaints about 7-9KHz noises not high distortion. The older 6DJ8 stuff I heard, shrieked at you when there was a problem(a huge problem in my mind!).
...but often it's not the tube that's bad, but that it's on the edge (or over the edge) of VHF/UHF oscillation and that, for some reason, either manifests as this upper mid harshness, or 50/60 hz hum.
A 100 ohm R *absolutely* on the tube socket's grid pin normally gets the job done 100%.
regards,Allen.
x
nt
all the best,
mrh
Although not all are created equal the best will outclass most other similar types IME and that includes 6SN7 varieties. Your measurements mirror mine and they get even better with CCS loading but only of course only when lightly loaded as your circuit is, which a lot of people seem to overlook.
> > It uses a single common cathode < <
??? Shouldn't it be grounded cathode? - Just getting even for you picking me up on CSS instead of CCS a long time ago ... Aussies have a long memory since there's not much else goin on in there!
Tre has a point on your operating points but I'd only be concerned about the plate voltage. Tube Cad predicts 185V on the plate but we all know that's not necessarily accurate in real life. Although I personally like more current, these tubes are still surprisingly linear with small signals even when operating in what appears to be a non-linear part of their parameters.
Your point is well made Alan, particularly since the audio reviews back up your measurements.
Regards,
Naz
Interesting, Allen. The only source I have for measurements of 6DJ8 distortion is a Vacuum Tube Valley, Issue 7 article by Eric Barbour. His test conditions were 48 kilohm plate resistor, 250 V B+ and 10 VRMS. He measured 3rd harmonic distortions at between 50% and 100% of the 2nd harmonic distortion. 2nd harmonic measured between 0.037% and 0.235%, 3rd harmonic between 0.027% and 0.215%. I have no reason to doubt either his results or yours, nor do I have an explanation for the apparent discrepancy - different operating conditions perhap?
> His test conditions were 48 kilohm plate resistor, 250 V B+ and 10 VRMS <
OK, but how much current was he running - these numbers don't tell us that. It needs current to become linear, the original curves show it to be extremely linear at 10mA and above - I use only 5mA so I'm not at it's best place, but many try and treat it like a 12AX7 and certainly then it's NFG!
Another MAJOR difference, a 2k49 unbypassed cathode R in mine.
Regards, Allen
Don't have any other info on the operating conditions, other than he used a negative grid bias supply with about 1 volt negative bias (range -0.67 to -1.29 V), so I assume that the cathode was tied to ground. Presuably the bias was set to achieve the same operating current for all tubes tested. He did comment that the plate current was significantly less than 10-12 ma but did not give the value. And I forgot to mention that the 10 VRMS was the output signal level.
> I assume that the cathode was tied to ground.
In that case, there was no degenerative feedback. Measured distortion could be expected to be significantly higher than in Allen's tests.
That's reasonable, although the 2nd HD levels at 10VRS out seem to be similar to Allen's tests. However the 3rd HD is significantly higher. Not sure why lack of degenerative feedback should boost the 3rd but leave the 2nd around the same level. From what I've read, feedback tends to lower the low harmonics but add higher harmonics into the mix, albeit at lower total levels.
> feedback tends to lower the low harmonics but add higher harmonics into the mix, albeit at lower total levels. <
NFB can do this, but cathode degeneration linearises the tube curves and also has no time dispersive bad effects. It makes the tube act like a lower transconduction, but vastly more linear tube
Regards, Allen
I understand that you want to make a distinction between loop FB and what happens when one leaves a cathode resistor un-bypassed. But please don't change the language.
"Degenerative Feedback" or "Local Current Feedback" is what this type of feedback has been called for years and years.
RDH4; index
Feedback, cathode-degenerative amplifiers pg 327-330
Feedback, degenerative cathode impedance pg 330
Feedback, current feedback pg 307, 312-313, 315-316, 327-330, 1236
Feedback, degeneration or regeneration at any frequency pg 342
On page 328, fig 7-20 shows a un-bypassed cathode resistor common cathode circuit. The caption reads, "Resistance coupled amplifier with unbypassed cathode bias resistor, giving cathode degeneration with current feedback."
Thank you, Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Then again, Henry often distorts the big picture in favor of isolated and irrelevant minutia. That allows him to justify different conclusions at different times.
Allen, I use "degenerative feedback" mostly because it's universally recognized. I agree that referring to it as simply "degeneration" would be more accurate (and probably less contentious).
Says H.P: "The use of modest cathode resistor will provide degenerative feedback and help to minimize the impact of tube-to-tube transconductance variations."
Sorry, TK, but the quote is correct.
"Degeneration" is just a vintage term that means exactly the same thing
as "negative feedback."
Allen is right that the resistor linearizes the tube and lowers the total
transconductance. It also makes the transconductance less dependent on
the tube characteristics (can you see why?). And it raises the plate
resistance.
There is a delay in the cathode feedback loop, but it's very, very small
and can pretty much be ignored.
I believe I invited you once before to compile a list of my so-called
lies and distortions, and we can go through them one by one and see
which, if any, of your criticisms are valid. The offer still stands.
By the way, that FAQ came from a usenet posting I wrote many, many
years ago when I was still relatively new to tubes. Many people have
copied it to the Web. I think it's mostly correct, but it's mot my best
work.
-Henry
henry, if you'd like to make any changes or addenda, please let me know.
:)
-ken
ps. long time no see. hope you're doing well!
Still no clue as to why some tubes sound so much different from others. To me, the 6dj8 has a very pronounced 'signature.' Whether or not you like that signature is a personal thing. OTOH, the 6SN7, which is also regarded as low distortion, seems to have less of a signature.
Then there are other tubes that have higher measurable distortion, but sound more realistic (for want of a better word) than some with low distortion. Certainly there are many that sound much smoother than the 6dj8. I don't buy into the theory of Mr. Jeeves that I like the distortion in the same way that I like sugar in my coffee. It's just not that simple.... No doubt there's some truth to it, but it does not explain everything; it's not that simple.
-- Dave
"No doubt there's some truth to it, but it does not explain everything; it's not that simple."
That's why I used the word "necessity" and took great pains to avoid claiming "sufficency."
I don't buy the sugar-and-coffee analogy. It's more like something I see in my profession- if one lets a red wine ferment to (say) 0.25% residual sugar, the taster will not get a sweet sensation, but just perceive the wine as "bigger, broader flavor, smoother." Or in my other avocation besides audio, like adding very small amounts of MSG to a good stir-fry dish.
I love stir-frying! So please tell me what MSG is :-D
To add to the info Bas kindly dug up, it's a source of umami and a cheap chef's trick. I'll use it if the ingredients aren't perfect, but living here in the finest agricultural region in the U.S., that's not often.
Ah, now I understand your first remark :-D
The link Bas gave me was not all too positive about the substance, in fact is was mainly health related.
The health stuff is, predictably, overwrought. As with sulfites or peanuts, a few people have sensitivities, but the vast majority do not. It's not really bad for you (after all, it's a major component of soy sauce and mushrooms), but it unbalances the flavors of food cooked with it, rendering them unnaturally vivid and obvious.
Monosodium Glutaminaat
http://www.aspartaam.nl/artikelen/msg.html (dutch)
http://www.rense.com/general52/msg.htm (english)
To infinity and beyond!!!
Dave, could it just be a matter of synergy? It's not a matter of you "liking it", it's a matter of the rest of your system "liking it".
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
I first recall hearing about distortion in 6DJ8 circuits in the early 90s. IIRC, Roger Modjeski discussed the situation in one of his interviews/mfgrs comments in Stereophile or Absolute Sound.
Roger found high distortion in some of his designs using the 6DJ8 esp at high signal levels. Investigation revealed that what he though were 6DJ8s were really mislabled 6ES8s. The 6ES8 is very nearly the same as the 6DJ8 except that the 6ES8 is a variable u device designed for RF service where gain is to be controlled by AGC voltage. Use of true 6DJ8s solved his distortion problems.
nt
I had good experience with the 6DJ8, but some can be a bit noisy.
Those figures around 0.001% is that reliable or just noise? If you can, would you please post the spectrum graphs? I find them much more intuitive than percentages.
It's down at the inherent noise for sure - especially with the 20dBV measurement range.
. I'll try and post the graphs - have to remember to take the floppy home - Macs haven' thad a floppy drive in 10 years...and that's the only way I can export out of the 386 the SpecA sits on.
Regards, Allen
In Morgan Jones' book (3rd edition), where he measured many samples of many different medium mu triodes and came to the conclusion that the 6SN7 and its variants had the lowest overall distortion, do your results for the 6DJ8 coincide with his? Or did he even test any 6DJ8/6922s? His results are given in db whereas yours are in percent, so it's not easy to compare, but I was just curious.
My circuits, especially the SLCF require high gM tubes to work well, so I have used a 6SN& in 20 years and hence haven't ever measured one.
Regards, Allen
ECC88 was not included in MJ's measurements. That may change for the 4th edition (whenever THAT comes out), and he's aware of several tubes which show even lower distortion than the 6SN7. Don't ask me what they are, I'm not tellin' until I have the market cornered.
But here in tubeland we also have this inherited belief that such distortion measurements have very little/little/not a lot (pick one) to do with the perceived "accurate" reproduction of music. Otherwise, wouldn't we all be listening to solid state devices that give vanishingly low harmonic distortion measurements?
I'm not so sure, all things being equal, that SS devices are lower distortion. In their traditional use and circuits maybe but that's not comparable topologies to valve circuits AFAIK.
A circuit which does not have good linearity will not be accurate. You might LIKE it, but it's a signal processor.
This is VERY different than saying that low static distortion is sufficient. But it is necessary for accuracy.
I forget which one, but one publication had an interview with a "tube expert" that just railed on the 6DJ8. Claiming it was a TV tube not designed for audio and that frame tubes were different than the typical 12XX7 tubes. Most people took this as settled business and decided that they no longer had to listen to them. It gives them something easy to slam because anyone who knows anything about tubes has read, heard about or has been influenced by this artical and therefore you are not suppose to like 6DJ8s. Which is worse, the shepard or the sheep.
_______________________________
Long Live Dr.Gizmo
...the vari Mu 6ES8 tube thatlooks like a 6DJ8 and was remarked by some ripoffs 'cause the 6ES8 was useless once TV sets went solid state, but there were millions around ..
Allen
This was in Glass Audio, early nineties. The remarking seems to have been done by the Sov manufacturers not so much dishonestly but because the chars of the Soviet tube (not exactly 6ES8 as I recall; a close sov equivalent) were a close general match.
It was Modjeski that outed the substitution, along with tests of some nice 6922 and 6DJ8 types to illustrate.
I've never used them, Allen, but I darkly suspect that maybe the calls of 'hard' sound from this device derive from people using them up at 10-15mA, to get the big gm and low rp that they're famous for. This can cause a lot of tubes to sound hard. I've gotten suspicious of the old manufacturers' data sheets for this fact; that they pound the devil out of the devices, operating them at absolute rating max, to get good numbers.
Aloha,
Poinz
... that maybe the calls of 'hard' sound from this device derive from people using them up at 10-15mA, to get the big gm and low rp that they're famous for <
I suspect the other way. When used at 12AX7 current levels it;s a rather non linear tube - and that's how many people have used them, and also at low anode v's. I find they get better sounding the more current you use - but then you start to run into it;s rather low dissapation limits.
BTW, the printed 90V max anode V rating was never follwed by TEK who used then by the bucket load in their 50's and 60's scopes - i've seen them used there with 300V anode/cathode!
Regards, Allen
You may very well be right; I have not looked into the device carefully, just because it never filled a circuit need for me.
I've seen a few of the 'bigs' operate devices outside rating over time, but that's pretty scary. Must be nice to have a test lab extensive enough to do life tests on a thousand components operated at 400% overrating.
Aloha,
Poinz
resistor audio colorations, the universal evil of electrolytic
capacitors BS --etcetera. Sure, there are some subtle harmonic
complementarity issues to be considered between various tube types
in a given amplification chain, but to paint the 6DJ8, or any other
reasonably linear device with the "bad tube" brush is a non-helpful
oversimplification.
Thanks for exposing the 6DJ8 distortion myth to a bit of sunlight.
-T.M.
Allen, I don't understand your operating point for the gain stage.
The operating point is down in the curved part, with the grid lines "bunching" to the right and "spreading" to the left.
I would think the distortion would be a lot lower with more current and a lower plate voltage.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
This looks much better to me. Even spacing to the right and to the left. The need for a 510 volt supply could be solved by using a CCS in the plate circuit instead of the 25K resistor.
I'm I missing something?
Thanks, Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
what is the source of the curves?
dave
The one in TubeCad. I don't know beyond that. John Broskie would know.
Above are the curves generated by the Sofia curve tracer of a National 6dj8 Made in Holland. http://www.mclink.it/com/audiomatica/tubes/6dj8.htm
Here are the Phillips curves.
http://www.mif.pg.gda.pl/homepages/frank/sheets/030/e/ECC88.pdf
To me they all show better linearity at higher current. I'm not saying that either of the OPs I picked would be best, just better than Allen's.
Unless there is something I am missing?
Thanks, Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
The other effect of running higher current not accounted for here is that (necessarily) the bias is reduced. This can run the tube closer to the point where grid current distortion happens (typically starting at -1V). That's a serious issue when the source for the tube is a volume control...
Load lines are a good first approximation, but you have to be aware of what else is happening in the circuit.
Well in theory all tubes show better linearity at high current and with higher value loads. The reason i ask about the data source is there are several programs where the curve data can be plugged into to spit out the mathematical distortion numbers.
all of the sims apply some sort of mathematical model to the tubes and the curves when plotted are perfect. No kinks no dips no errors... hell you can even sim a 2A3 at 1A of plate current :-)
I have often noted odd "kinks" in the single sample curves and the ones you posted had a number of "wiggles" that i would like to know the source of.
The sofia trace shows some definite deviation from the math model whereas i suspect averaging 100 tubes would better approximate a mathematical model. The previous model you presented (broskie's?) had more distinct kinks.
So what is the cause of the wiggles? IS it true measured behavior showing individual tube deviation form the ideal or is it single tube measurement error which can be covered up by averaging multiple samples?
this is where we need the big-time math guy who can crunch the actual numbers of a large sample of individual tubes and then relate the single tube results to the average results and figure out where the errors reside.
the converse of this is actual measured distortion results. this is one case where science tends to lie based on statistical insignificance and 100 samples from 50 random tubes would tell the greater picture on the behavior of the individual vs. the average and how much deviation is involved. (of course a larger sample would be great, but in tubeland samples of 1 or 2 are often considered fact)
dave
"the converse of this is actual measured distortion results. this is one case where science tends to lie based on statistical insignificance and 100 samples from 50 random tubes would tell the greater picture on the behavior of the individual vs. the average and how much deviation is involved. (of course a larger sample would be great, but in tubeland samples of 1 or 2 are often considered fact)"To me, this would be the value of something like the Sofia curve tracer. One could custom tweak the circuit to match the tube being used with less trail and error involved. Assuming the tracer is accurate.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
At low signal levels, operating in what APPEARS to be the non-linear part of the curves is not necessarily such a big issue. The important factor is only the spacing between identical grid voltage steps at the operating point (as predicted by a proper load line which takes into account total loading).
Unfortunately the curvature of the lines gives us the impression that the distance between the steps varies dramatically but expanding them out reveals that it's not the case and in fact they can actually be VERY linear. Of course all of this changes dramatically at high voltage swings but here we are only interested in a few volts P-P.
Regards,
Naz
Now calculate the anode dissipation...
I did, It's at 99% of max. So this one is 80%. What is your point?Just as an aside, Allen's operating point has the plate voltage at 142% of max. The plate dissipation at only 25% and the current at only 10% of max.
Those max numbers, though they need to be adhered to, are not as important to me as the linearity of the tube at the operating point.
What I am questioning is the linearity of the tube stage, at Allen's operating point.
Thanks, Tre'
.
Have Fun and Enjoy the Music
"Still Working the Problem"
I said the circuit is SIMILAR to thatshown in the FVP-1 map on my site - but I never said it was identical.
The actual operating points are 260V supply, 100V anode, and 6.8mA current, with a 22k load.
How I get all these points at the same time I'll leave as the problem of the week...
And what you can't tell from the basic curves is the HUGE positive effect on linearity of the un-bypassed cathode R of 2k49.
Somewhere in the RDH is a graph showing the almost perfectly straight lines (compared to the normal curves) resulting from this design concept.
Regards, Allen
Sorry, I was only speaking to the circuit shown. I was only identifying that circuit as yours in the sense that it was the circuit that you had shown.
"How I get all these points at the same time I'll leave as the problem of the week..."
My guess is there is some (resistive) direct coupling, for the sake of the low frequencies, from the preceding stage?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Or provide an accurate description since it's not the same as the FVP-1 (I also mistook this). I agree that the unbpassed cathode R makes a big difference, even moreso when it's so large but it must affect gain considerably. Anyway, you have my attention and I'd like to know more of the test circuit. I assume you tested it with no load? BTW, if indeed this is your operating point I've also found that it's a VERY linear point for relatively small signals.
Regards,
Naz
Yes, which makes it perfect as aline amp. Bypassed, a 6DJ8 has WAY too much gain!
Allen
... is a term often used to compare things. However it would seem that is not the case here. Topology is the major effect while the device is largely secondary if it is being operated linearly. So lets assume you are operating it linearly. Your circuit would seem to have huge cathode degeneration by using a big resistor. This is a further form of "linearising" and so lowering distortion. To then claim that the 6DJ8 is a low distortion device with this arrangement is pushing things a bit far and could even be misleading :-)
Unless you compare it in the same fashion to other valves, then the figures provided are largely meaningless. AFAIK, and this is by no means a speciality area of mine, valve distortion figures wouldn't be normally presented in this fashion i.e. in that type of circuit.
cheers,
Stephen
Operating at full anode dissipation raises temperature and shortens life. You can expect anodes and micas to outgas and poison the cathode. You can expect higher cathode temperature and consequent evaporation of cathode emissive material onto the (nearby) grid, causing grid emission and nonlinearity. There are usually better ways of obtaining linearity than running the valve right up to the brink. 66% of Pa(max.)is better, and at 33% you can expect a valve to last longer than the associated electrolytics.
By the way, it's well worth looking up the fuller data sheet for the Mullard E88CC; Va(max.) = 250V for Pa < 800mW.
With a CCS load and LED biasing in the cathode, the 2nd drops quite a bit, too. I scratch my head in wonder about where these legends come from. "A friend of mine told me that..." is an almost certain indicator of bullshit.
There's a lot of variation in distortion between different manufacturers. What were the particular ones you were using?
Current manufactureSOVTEK 6922 which of course are actually 6H23...
But tests with real 6DJ8/ECC88/E988CC come outquite similar.
Regards, Allen
I remember a Glass Audio article about 6DJ8's in audio, which reported that the 6DJ8 had gotten an undeserved reputation because of a large lot of 6ES8's that had been relabeled 6DJ8. As discussed in the other forum, the 6ES8 has a remote-cutoff grid, which would not be appropriate for a large-signal audio application.
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
