 
|
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 |
82.44.160.191
In Reply to: RE: 12AX7-EL84 magic posted by Jon L on May 24, 2009 at 20:10:37
It's not going to measure very well and from my own experiences 12AX7 should not be used as a triode driver , the sound seems to lack in sparkle and life in comparison with a lower Rp driver . EL84 strapped as a triode has fairly high miller capacitance and the 12AX7 high Rp :- a recipe for poor bandwidth , the 12AX7 will also have excessive gain . There are much better valves out there to drive an EL84 such as the 12AT7 , 5965 and E80CC
Pentodes on the other hand do not have the same capacitance issues as triodes . If you have a 12AX7 driving a pentode , leave it as it is . Re-strapping to a triode will incurr the issues detailed above , some may like this sort of sound . Unfortunately I don't ;)
Al
Follow Ups:
If you have a closer look, you can see why there is no problem:(A) The EL84, cathode-biased can safely use a 1-megohm grid resistor which loads the driver stage very lightly, and so no strain for the 12AX7 driver.
(B)Because the EL84s are operated in class-A, and thus no grid current is drawn, the input resistance stays high, and presents no problem for the 12AX7 driver.
(C) All triodes do suffer from some degree of miller capacitance, but this is moderate enough that the circuit shown has a 55 kHz -3dB rolloff point, and so no HF problems here.
(D) lower Rp tubes are not necessarily always sonically superior as drivers for class-A stages with high input resistance. Sure, they will drive the input/miller capacitance better, but again: can a measured 55 kHz -3db point in a zero-feedback amplifier (using 12AX7 driver) be considered "poor bandwidth"?
(E) Sonics: Largely due to their inherently high PSRR, the long-tailed pair is much better isolated from LF power supply transients than other more favored ( allegedly technically superior) inverter schemes such as split-load. In fact: I tried several split-load configurations using the tubes: 6SN7, 12AU7, 6DJ8 all of which have much lower Rp than the 12AX7. I presumed, because of the lower output resistance (ability to drive loads) of the above mentioned tubes, and the inherent dynamic balance of the split-load scheme, that it would be clearly sonically superior to the long-tailed pair. Surprisingly, this is simply not the case. The difference is not small either, and indeed the long-tailed pair (with both 12AX7 and 12AT7)sound so radically more "alive" detailed and "fast" as compared to any split-load scheme I tried. I have constructed this amplifier based on extensive listening tests with many many circuit configurations.
BTW: are you certain you have actually heard a T-C P-P EL84 amplifier with a 12AX7/12AT7 long-tailed pair? You might be blown away. The speed and resolution are incredible.
I am presently using a 12AT7 in my personal amplifier (no circuit changes needed, just plug-in the 12AT7) because it sounds quite similar to the 12AX7 in this config, and is cheap and plentiful.
On the gain: I initially planned on NOT having the 12AU7 input stage due to the fact that the additional gain is entirely NOT needed. I put it in place, however, because the harmonic complemetarity of the stage with the rest of the circuit is desireable, and without it, the amplifier sounds decidedly "drier" and somewhat less presence.
This little P-P EL84 amp is a an excellent performer, even as compared to a good SET.
-T.M.
Edits: 05/26/09 05/26/09 05/26/09 05/26/09 05/26/09 05/29/09
Blimey Thom , you sure like editing ;)
I am very surprised you got anywhere near 55kHz , but if you did , good for you . Never got results anywhere near that myself . Driving SE triode connected EL84 with 12AX7 running at 0.8mA I got -3dB around 25kHz at the EL84 grid . Watch out for those 1m grid leaks , that's really pushing it !
Al
1-Meg is absolutely no problem with cathode bias operation of the EL84, this has been widely established in practice.
-T.M.
Edits: 05/29/09
1m is pushing it , I don't mean safety , rather sound . The lower , the better IMO . I can see why your bandwidth is better , you used 100k/91k loads . Lower impedence but much less gain than the 270k I used .
Al
The time constant of the .22uF coupler/1-meg grid resistor can really only become an issue sonically if the amplifier is driven hard enough to venture up into class-B on peaks and draws grid current, otherwise, it's no problem.
This particular amp design was purposely undertaken with the intention of being simple, in that a modest +325V volt supply voltage could be used, therefore not requiring several stacked electrolytics or more expensive types of higher-voltage-rated filter capacitors.The individual cathode bias resistors and bypass caps on the EL84s would seem to only be required with ultralinear operation, and not desireable with triode operation where theoretically a single unbypassed cathode resistor per P-P EL84 pair would seem to be the elegant, no-time-constant producing biasing method. Surprisingly, repeated A/B listening between both bias schemes convinced me quickly of the subjective sonic superiority of the separate, bypassed cathode resistors approach. Why this is, I am not entirely certain. Initially, I fully expected the simple single shared-resistor approach would be found to be the easy winner, and was quite puzzled to find that it was not.
-T.M.
Edits: 05/26/09 05/29/09 05/29/09
8k a-a at 25mA is class AB1 . I used 10k a-a last time I built an amp with EL84 , this gave about 3.5 watts in A1 running at 32mA per side / 300V . Driver stage was an E80CC common cathode stage DC coupled to a 12B4A concertina , the EL84 had the cathodes tied via 4.7 ohm resistors into a CCS set for 64mA . A centre-reading meter was used for bias monitoring . Biasing was variable grid bias and before you mention this is not cathode bias , consider the CCS . Grid leaks 330k , 470 ohm stoppers on the EL84
You probably have way too much gain in your circuit , I seem to remember I could drive the EL84 into clipping from a CD player with the single E80CC input stage/driver .
Al
Sure, if you drive it harder it will eventually transition into AB --this is to be expected. Nonetheless, it will stay within class-A approx 90% or more of the time with normal usage.BTW: I've tried the 12B4A split-load with the EL84. Surprisingly, it was a pretty lack-lustre in sonic results. Bottom line: the split-load inverter just was not found to be sonically comparable to the long-tailed pair fro driving the EL84. In fact, it's much more than a subtle audible improvement with the long-tailed pair.
As far as CCS implementation: I'm not entirely sold on CCS usage. I suspect they can introduce more low-level harmonic artifacts than is generally assumed. And BTW: 32mA is much more current than you really need for clean triode-connection operation at the 5-watt level. It sounds much better with 24/25 mA.
As for the "too much gain" concerns: carefully re-read my previous posts here. There is a sound reason I included the 12AU7 input stage which has nothing to do with the additional (yes, unneeded) gain the stage provides.
You seem to be quite interested in this circuit. I'm curious, Why not suspend your disbelief, try it out, and then judge for yourself? Then you can get back with me and we can discuss your impressions.
Cheers,
-T.M.
Edits: 05/29/09 05/29/09 05/29/09
'You seem quite interested in this circuit. Why not suspend your disbelief, honestly try it out and (listen) for yourself, then get back with me and we can discuss your impressions?'
...don't flatter yourself too much Thom , I stopped playing around with the Mullard circuit over ten years ago ! Since then onwards and upwards , PP 2A3 is what I'm currently using . Whatever floats your boat I suppose
Al
edit
Edits: 05/29/09 05/29/09 05/29/09
Thom , I'll stay away from you in future I can see where this is going . I'm no beginner to this game and I'm not interested in pissing contests regarding what you consider to be good . I'd rather build for my own tastes , thanks !
Al
edit
Edits: 05/29/09 05/29/09 05/29/09 05/29/09 05/29/09 05/29/09
Well you're flat out wrong in that respect , read my original post . I have no requirement for a 3.5 watt amp and no longer have suitable output transformers so sorry I'm not going to build anything with EL84's . I have over a hundred EL84 in my stash , mainly Mullards , so saying I have prejudices against this type is laughable . Besides , like I stated before I use DHTs , for both driver and output stages so there is no place for your MSG seasoned 12AU7 input stage either
Don't expect further replies .
Al
edit
Edits: 05/29/09 05/29/09 05/29/09
They're a fine VA tube, or as LTP a fine PI stage too; nice g2/anode current ratio, and with gm to deliver gain from *VERY* reasonable plate loads( 5-7kOHm ). A pair of EL84 can only deliver a bit more percieved volume than a pair of 2A3, and with efficient speakers, it is a hard arguement for more power. Why rail against DHT's? Is there a single Way to good sound? ( it is a rhetorikal question, if you answer anything but 'no', there are bigger issues than power tube choice with your PoV ). Very similar to the One Way to build power supplies put forth by idealogical fanatics, yes?
I can't say I've ever had better results with IDH tubes for some reason...DH pose a whole bunch of troubles not seen with IDH...and yet for lots of us it is indeed worth the trouble( I've got two sets of DH amps, and two sets of IDH, right now I run one of each, bi-amped ). Attempting to convince somebody they're doing it wrong can be a lot of effort, yes?
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
edit
Edits: 05/28/09 05/29/09 05/29/09 05/29/09 05/29/09 05/29/09 05/29/09
see the link. At least that looks rather like slander. You also claim to have presented a 'reasonable technical arguement', and I see that nowhere...as well as this distasteful accusasation of having some 'veiled agendas'. Give me a break...there is nothing but you doing just what you accuse Al of...it's personal, why not just admit it, and be up front about it? ( another rhetorical question; no response needed)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
edit
Edits: 05/29/09 05/29/09 05/29/09 05/29/09 05/29/09 05/29/09 05/29/09
your answer sums you, and yr thought process up quite well.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
The best policy with false prophets is to back off and let them have their fun with all the newbies .
Don't worry about me I can fight my own battles
Thom is just another cheap-jack , blinkered and full of himself
Enough said , time to crack on with Friday afternoon :)
Al
ps the WE388A's are in the post , I'll drop you a mail once I have the fils fired up ;)
I tried to trace 6 samples but alas 2 were dead so here are the 4.
now toss out the extremes and look at the two in the middle.
edit oops Grid is 0 to -50V in 5V steps!
and those were the two I listened to. I cranked them up to a juicy operation point and listened and that carbon wedge was nice! Here is a crappy pic of the basic construction.
Now lets diverge from the science. The sound was awesome but a few days later it quickly lost its charm. When I checked the operating points of the two tubes they were far off the curves I plotted. So I had a $8 tube that was dead and I thought why not bump the filaments a bit to get emission and 2.2V brought me back to the curves. The sound was back and I listened for a few days and things were nice but I needed to move on since one can only be so anal. That was a year ago but interestingly enough a similar situation that parallels this situation has come up hence my post.
dave
Edits: 05/30/09
Thanks for that ! This current batch of sillyness is Douglas' fault , after all those 388A curves look a bit 316A-ish . I've got a basic idea of how I'm going to use these but not what I'm going to use them in ;) Luckily I'm sitting on some 0.2mH 20A chokes and have a few other 8mH @ 10A and 2mH 20A so the nasty part is more or less catered for . Filament bias will be a bitch , putting more power in the cathode resistor than into the filament is not exactly going to help global warming LOL .
Al
the obvious choice for your filament bias is 3 12V 110W xenon bulbs in series.
A better choice is to build a small toaster oven to operate off your bias voltage. 650W will do some serious cooking :-)
dave
When I was running the 388's i actually put the filaments in series and grounded the center connection :-)
dave
Like the 316, if you run them hard they lose emission quickly. About 4 days ago I started listening to some mesh 2A3's and they were tired. Just for kicks I set the B+ to 250 and the bias to -40V and was only getting about 25ma of current and a Gm of 1500. I remembered the 388A experience and up went the filament voltage and at 3.8V I had 60ma and the Gm was up to 3000 (which is what the tubes measured as new). after a day of use I was running 80ma and I dialed the filament voltage down to about 3.5V. another day and it was 3.2V until eventually one of the tube was pulling proper current at 2.5V and out it went and another tired tube went in. Thus far 2 tubes are much closer to normal and a third is a good way there. The 4th one is putting up a good fight and we may justfind out how long the filament will last at 4V :-)
dave
oh yeah...blame it on me...Al, here are some neat tubes. Now if I didn't tell you, it'd go the other way...'why didn't you tell me?' Can't win no how.
The experience did set a record; out of 8 4E27 he had, 7 were good. In order to get that many, even if they're NIB/NOS I'd expect a third of that between filament damage and gas.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
hey-Hey!!!,
I am considering some for myself; I've been looking for a reason to visit him again...:) Maybe get a ride in his new Shelby Cobra whilst I am there...:)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
I'd get my mits on them ASAP if I was you . $12 a pop is super value for a WE . All you need are some meaty fil chokes/schottkys and some high ripple caps and you're away . A few turns on top of a core with some beefy ECW is all thats needed for the fil tx . Filament bias may be problematic though LOL !
I'd have a field day if I was your side of the pond , could easily see a lot more of my hard earned getting pumped into 'investments' like 388A's :)
Al
heh-heh-heh...you should see this Cat's basement...:)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Separate bypassed cathode bias resistors will sound better if:(a) your OP tubes are not well matched.
(b) you are straying into Class AB1 (i.e. cutoff on negative-going peaks).
BTW, Class A is not guaranteed unless you use a shared CCS cathode bias. This will ensure perfect AC balance but NOT DC balance unless your tubes are matched.
Also, UL and Class A are not mutually exclusive, as you seem to think. There is no reason whcy a UL amp can't work in Class A.
Edits: 05/26/09
Ray, you wrote:
"Also, UL and Class A are not mutually exclusive, as you seem to think. There is no reason whcy a UL amp can't work in Class A"
Yes indeed Ray, this is known, and I'm sorry you interpreted that I stated otherwise.
Also: your conjectures regarding tube matching and the sonic differences between common and seperate bypassing, I am afraid, are not entirely so cut-and-dried.
Take a look at the Williamson circuit: Williamson used the shared-unbypassed cathode resistance route, but still installed a potentiometer in the cathode circuit to enable one to adjust the grid-bias of the output tubes to balance the current flow between the output tubes.
Sure, if you use just a single shared cathode resistor, with no provisions for adjusting each tube's individual DC bias (a privision like Williamson's scheme), then you'd better have a pretty closely matched pair --true.
Bottom line: I've tried both bypassed/individual and unbypassed/shared configurations with closely matched and current-balanced pairs of EL84s, and the sonic differences between the two configurations I mentioned still held. It's a curious result, I know.
-T.M.
Awhile back I asked the same question. Not only does it sometimes sound better, but bypassed topo seemed to be a little less sensitive to the operating points of the specific make of power tubes. Anyhow, I was directed to this site. Quote:
"Since cathode biasing eliminates the need for a special negative DC bias supply, why don't all amplifiers use cathode biasing? Well, cathode biasing is not without its faults. It turns out that in order to keep the DC bias supply voltage at the cathode constant while the input signal is changing, the cathode resistor must be bypassed with a large capacitor. This capacitor effectively "shorts" the AC signal component to ground, while allowing the DC voltage to remain relatively constant. If the capacitor is removed, the cathode DC voltage will have a signal voltage superimposed on it, which will subtract from the grid-to-cathode signal voltage, and reduce the gain of the stage.
The problem comes in when there are large signal level changes, and the average DC level of the cathode voltage changes. This causes a bias shift, usually in the direction of a colder bias point. This bias shift can be audible, but is sometimes desirable for guitar amp use, as it adds varying harmonic overtones to the sound. If the bias shift is severe, the tube will go into cutoff, and large amounts of "crossover" distortion will occur. In addition, the current flow through the cathode resistor generates a necessarily large bias voltage on the cathode for proper tube operation (typically 30-50 volts for most higher power output tubes). This voltage subtracts from the total plate voltage, which decreases the available output power. Between this voltage decrease and the bias shift, the output power in cathode biased operation is reduced when compared to fixed bias operation. Therefore, fixed bias is usually used for higher power amplifiers (50W and higher), and cathode bias is usually used for lower power amplifiers.
Preamp tubes are almost universally cathode biased, because they are used for signal amplification, not power amplification, and the side effects of cathode biasing are not as important. Also, cathode biasing makes the circuit less dependent on tube parameters, and more forgiving with respect to bias point. This allows replacement of tubes without having to rebias the amplifier stage. Output tubes that are cathode-biased should always be checked when replacing tubes, however, because they vary widely in terms of idle current for a given cathode resistor value, and it may be necessary to change the cathode resistor value to return the output stage to it's proper bias current."
Post a Followup:
| 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 ]
