|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
212.136.56.20
In pentode mode?I tried it last night and liked it a lot. Very fast sound. But I need to do some tweaking...there was a little sibilance I think. I used an SV83 like Thorsten used to do in one of his "Legacy" amps.
I'd like to hear experiences of those who liked it and those who did not like it.
Follow Ups:
Usually better than a triode in my opinion, if you want a 1-stage driver:D3a -> KT88 http://www.pmillett.com/elinear.htm
12HG7 -> 813 http://www.pmillett.com/813_se_triode_amps.htmFWIW, the statement that "pentodes are less linear than triodes" is an untrue generality. As output tubes, it is usually true; for small signal and driver applications, it is almost always UNTRUE. Look at the numbers...
That said, there are still cases where something else works better. Pentode into a cathode follower or SRPP will push a bit more grid current to soften clipping, for example.
Pete
pentode for the linestage, pentode/cascode for amp's input stage. I don't think I will do anything else with triodes any time soon!
I have to agree it does some things very very well. But other things less well...IMO.So far I think the advantages are.
Fast, nice tone and detailed.
Disadvantages:
I just could not listen too long..something bothering me. A little forced is how I'd describe it.
To infinity and beyond!!!
I have not run into that effect you describe. Every time I put a triode pre back in, I find it missing something. The pentode seems to be delivering just as its lack of Miller suggests. I have used one with a low triode mu, and plan one with still lower mu( less than 4 ).What sort of plate load? I run less than 1k and more than 50 mA per tube on mine...:) takes a bit ov power supply for a simple pre though.
I'll try a lower plate load...see how that goes.
that stage must have some serious gain. I have no need for that much in my linestages. My two stage amps need just a bit more than my CD player will put out. A 1k should give you a a gain of ~10 with an EL84. That is also where the output Z winds up, and I can easily deal with that number.
and at the moment the pre is passive. I'll halve the load ... and have a listen.
Bas, I run my el84 driver with 12K plate load, around 12ma. Screen voltage is set by a simple resistor network to around 135 volts (from memory), plate voltage is around 225 (also from memory). The amp displays no tendancy toward sibilance, is easy to listen to for hours on end. It isn't my main system, they are pp300b, but this amp is no slouch. I made a big change in the sound of the amp by changing the coupling cap - I started with pio, ended up with a musicap.
this looks like a real keeper!
To infinity and beyond!!!
A true pentode takes special attention for its load line placement; the g2 voltage is one of the critical inputs.
Yes, you are correct. In this application I wanted to make sure the plate voltage at full drive would never be lower than the screen voltage. I checked my notes last night, I actually have 250 on the plate, 140 on the screen, giving 110 volts of swing room, far more than the 300b requires.
"In this application I wanted to make sure the plate voltage at full drive would never be lower than the screen voltage."
When I first started researching the pentode/300B scheme, I read some of Thorsten's background material, including an article from the French publication that originally proposed using the EL84. TL credits the design as being a Verdier design. Somewhere in all that stuff I had remembered the comments that the tube was much more linear if the screen voltage was lower than the plate at full drive. The original model 91 had somewhere around 100 volts on the screen, while the anode was much higher.
> > "In this application I wanted to make sure the plate voltage
> > at full drive would never be lower than the screen voltage."
>
> why?I would have thought a look at the curves, including screengrid current vs. anode voltage would provide a BFBFOTO....
Think happy thoughts, use fairy dust - Peter Pan
BFBFOTO = Big Footballing Blinding Flash Of The Obvious
Or maybe it's the lack of fairy dust.I don't see anything footballing obvious happening around Va = Vg2 (250V in this graph.)
If however you look at the way in which the curves differ when going from G2 voltage of 200 volts to 300 volts the picture begins to get a bit clearer. Even moreso if you do a transfer function. You won't have near the useable range but what you do get is more linear. So "if" you don't need the gain it is possible to improve some with this technique. It might be easier to see if you look at small signal pentodes.
HiI've tried so many drivers I can't count them. I've tried pentodes on several occasions, but they never pushed my button. I always found them lean and fatiguing. I also don't like high gm triodes (same steely sound).
Believe it or not, my favourite driver stage is now a 6SL7 in SRPP.
> > I also don't like high gm triodes (same steely sound). < <Perhaps you took inadequate precautions against parasitic oscillation. Never mind "in cardboard box oscillators", like the 5842 and 6s45p, something as commonplace as a 6922 can easily go "beserk". A 1 KOhm Carbon comp. grid stopper may be insufficient. A 100 Ohm Riken Carbon plate stopper may be needed too.
HIGH gm types are intended for RF service of some kind. We've adapted the tubes to audio service. Failure to fully account for the RF capability of the HIGH gm tubes is (IMO) an error.
Eli D.
Hi Doug, to each his own, so I'd disagree with high gm triodes sounding "steely" as I have both a 10Y/50/300B amp and 45 amp both driven by 417A and not sounding steely.I'm with you on the 6SL7 SRPP, I like this better than a two stage 6SN7 driving a 2A3 or 300B.
the reason they reportedly work so well as drivers in SE amps is actually due to their high distortion (relative to triodes in general) which cancels with some of the distortion produced by the load on the OP tube. I say this because I've observed (measured and heard) a similar improvement by biasing a triode driver to a less linear region through reduced current. One day I'll try it and take some measurements to prove or discount the theory.
hello, you are wrong with the high distortion of pentodes. Pentodes are very linear, compared at equal output voltage.But you are right with your experience. I have made many experiments with pentodes and triodes. The best results I have with combining one of both, with local feedback between both tubes.
In my Darling project (schematic is here: http://www.roehrentechnik.de/html/darling.html )I use your configuration pentode-triode, and it sounds much better than triode driver, compared with the same output tube, OPT and speaker.Combination of triode with pentode allows to work with only 2 tubes, married to one supertube because the local feedback, with high gain and low output resistance. And the distortions are similar to a triode, also if I use a triode driver with output pentode.
My theory says, there is 1 stage with high gain and no interesting sound and 1 stage with beautiful triode sound, but low gain. And combining this both tubes gives the characteristic of the tube with low gain. And this is the triode with dominant 2nd harmonics.
My last project was this combination in push-pull-configuration with cancelling 2nd harmonics in the OPT. (schematic: http://www.roehrentechnik.de/html/2a3-pp.html).
Hi Gerd,Interesting tube choice. I can't find curves for the pentode you use but generally speaking pentodes ARE less linear than triodes and that was the basis of my observation. Thorsten has apparently done some more research which elaborates on the subject. I'm the last one to argue that your combination doesn't sound good not having tried it; in fact there is plenty of evidence to support that indeed it does. However I haven't heard another theory that better explains the phenomena technically ... but I'm always willing to listen and to be educated.
Hello,with no feedback you are right with nonlinear pentodes, but comparing triodes and pentodes for gain stages needs similar conditions for output voltage and gain.
You can give a negative feedback with an voltage divider from plate to grid (Rk must be increased for making the correct Ug1) to reach the same gain. In this case pentodes are more linear than triodes.
For Output stages it is very simple. Power Triodes are more linear but needs higher grid swing than pentodes, and this gives more distortion in the driver stage. Thorstens experiments with cancelling 2nd harmonious are interesting, but it is very complex to find perfect matching driver tube.
Pentode with feedback can reach the linearity of a triode and need less voltage swing on the grid to drive, and the driver produces less distortion.
It is easier to use a pentode and feedback between both tubes.
The relationship 2nd and 3rd harmonious for pentodes can be changed through increased or made smaller the anode resistor. In this way it may be possible to cancelling 2nd harmonious.But there is another fact for using pentodes. Fullrange speakers have a nonlinear impedance curve, impedance rises toward both sides of the frequency range. Triodes with low output impedance (damping factor > 1) give less power to the speaker if the impedace is higher, pentodes with higher output impedance (damping factor < 1) give more power on higher impedances. It is very complex to fit Amp and speaker for best results.
Hi Gerd,Good discussion here … a few comments for you to ponder.
> > You can give a negative feedback with an voltage divider from plate to grid (Rk must be increased for making the correct Ug1) to reach the same gain. In this case pentodes are more linear than triodes. < <
Partly agree but the use of NFB to lower distortion can give rise to an undesirable harmonic distortion structure and a correspondingly detrimental effect on the sound. 2ndH may end up better than triode but not high order distortion, particularly odd order.
> > For Output stages it is very simple. Power Triodes are more linear but needs higher grid swing than pentodes, and this gives more distortion in the driver stage. < <
True but most of the distortion is generated by the load imposed on the
OP tube and it's easy to design a low distortion driver anyway.> > Thorstens experiments with cancelling 2nd harmonious are interesting, but it is very complex to find perfect matching driver tube. < <
Agree in principle but I haven't tried it with a pentode driver. I was surprised how well it works with a triode though and in theory a higher distortion driver would work way better.
> > But there is another fact for using pentodes. Fullrange speakers have a nonlinear impedance curve, impedance rises toward both sides of the frequency range. Triodes with low output impedance (damping factor > 1) give less power to the speaker if the impedace is higher, pentodes with higher output impedance (damping factor < 1) give more power on higher impedances. < <
There are many issues here. 1) Actually, a high OP impedance only helps smooth frequency response through its inefficiency, ie higher attenuation of the mid band rather than by delivering more power into the high impedance parts of the spectrum. 2) The poor damping factor can cause bass problems 3) A high OP impedance (poor DF) implies a poor load line and therefore high distortion. 4) Even if high OP impedance was desirable then it can just as easily be achieved by loading a triode OP stage more heavily, the bonus being greater power and reduced turns ratio in the OP tranny.
To summarise, I have to agree that low distortion is possible with Pentodes but not without compromises, especially when used in OP stages.
Rgds,
Naz
Hello> > Partly agree but the use of NFB to lower distortion can give rise to an undesirable harmonic distortion structure and a correspondingly detrimental effect on the sound. 2ndH may end up better than triode but not high order distortion, particularly odd order.
yes, this is possible, but not probably. I think, NFB will reduce all harmonics because linearization of the entrance characteristic
> > .. and it's easy to design a low distortion driver anyway
and it is much easier to design this for less needed output voltage to drive a pentode with an 1/5 or 1/10 needed grid swing ;-)
> > but I haven't tried it with a pentode driver
than do it, I heard the difference, Darling with triode driver without NFB and Pentode driver with NFB.
> > There are many issues here. 1) Actually, a high OP impedance only helps smooth frequency response through its inefficiency, ie higher attenuation of the mid band rather than by delivering more power into the high impedance parts of the spectrum. 2) The poor damping factor can cause bass problems 3) A high OP impedance (poor DF) implies a poor load line and therefore high distortion. 4) Even if high OP impedance was desirable then it can just as easily be achieved by loading a triode OP stage more heavily, the bonus being greater power and reduced turns ratio in the OP tranny.
to 1: no, pentodes with high impedance are more efficient than triodes with low impedance, because of the small not controlable residual voltage. And you are wrong with higher attenuation of the mid band. Higher OP impedance behaves like current source, and higher load resistance have a higher power consumtion. For nominal load impedance in the mid band you get nominal output power, and the rising impedance at the ends of the frequency range makes another distribution of output power.
Triodes with low impedance are like voltage sources, at rising load impedance the current will be smaller and power is reduced.to 2: yes, this can be a problem ;-)
to 3: not necessarily, a loudspeaker is a magnetic transducer, and magnetic forces result from the current, which flows by the voice coil. And pentodes with current source characteristic delivers current..
to 4: no, it is not so easy. Example 300B with Ri 700 Ohms. To get low distortions ra should be around 3,5k. transformed to 8 ohm load, without any losses this gives a output impedance of 700/(3500/8), around 1,6 Ohms and DF=5. You can not wound an transformer for DF> 1. For DF=1 you get only ra 700 Ohms, this gives much power with much distortion, but it is still not a current source.
A pentode like EL12 works also on 3,5k Ra, I can use a transformer with the same impedance ratio, but plate resistance of the EL12 is 25kohms, I have an output impedance of 25000/(3500/8), around 57 Ohms, DF = 0,14 (without NFB). NFB reduces output impedance, but not so much, DF will be ever smaller 1.I think, this is a reason because EL84 sounds very good on fullrange speakers.
regards, gerd
( sorry for my english, it is very difficult to interpolate between my memory and babelfish ;-)
Hi,See link.
Refering to a Komuro design Thorsten concluded: "Of some interest is here the very distinctive distortion of an "unloaded" triode wired pentode, which has a very strong 2nd Harmonic cancelling some of the distortion of the output triode if designed correctly."
So, you're not alone with this line of thought...
Regards,
NC
Thanks NC, pity I seemed to get on the wrong side of Thorsten early on in the piece (my fault I’m sure) because I share many (but not all) of his views and have had some very similar experiences. I too tend to run drivers at lower currents these days, although I still like higher currents for OP tubes. I also like grid chokes and prefer mu follower to textbook SRPP.On the subject of mu followers, I've done a lot of testing of different circuits recently and I'm surprised by the very different sound I hear when taking drive from the cathode of the top triode (mu follower) vs the plate of the bottom triode (CCS). And the cause can't be the difference in OP impedance because a high impedance load was used in all cases. Mu follower is subjectively much stronger at the frequency extremes particularly bass, whereas CCS tends to favour clarity in the mids. I can see why so many don't like SRPP in general because of the apparent loss of mid range detail causing a kind of muddiness, loss of precision and depth, or as Thorsten describes it "opaqueness". Perhaps it’s simply a by-product of their strong bass/mid bass. Paradoxically, they can also suffer from sibilance. But all that said I can also see that their strength might be of great benefit to restoring tonal balance in systems that need what they have to offer.
Hi.From the link, I can see 6SL7 SRPP driver topology exhibits the smoothest overall HD profile from very lo to hi O/P power.
He also said about the operation conditions of SRPP to deliver good sound:
"Most circuit benefit from a stiff regulated PS, the SRPP & most
other half assed PP circuits really NEED them.""It (SRPP) is a very linear & good sounding topology for as long as you have it working with a FIXED load".
So constant O/P load is crucial for such topology.
I would agree with Mr. Konichiwa (Thorsten's moniker here) that smooth & even decending profile of least orders of HDs would deliver good sound to our ears.
c-J
Hi c-J,Before you 'jump into conclusions' as I did when I first saw these graphs, maybe it is good to read part of a private e-mail conversation I had with TL about this (which he told me I could share).
Thorsten wrote:
First the graphs from the above, the perspective makes it difficult to see, but the peak at the 5th Hamrmonic at lower power levels on 417A and EL84) is from output valve (not sure why it did that, but it sure did), so if you do not see a distinct peak-up to the 5th it means the driver produces more 5th than the output did.
The 6SN7 cascade always does more distortion even at high orders than the 300B, in effecting swamping out the 300B's distortion, but it adds some more of it's own at the 5th causing that peak in the 2.5W graph.
When comparing the spectra I felt that the EL84 (and C3m) provided the most evenhanded (and low high order) distortion overall, arguably the 6SL7 SRPP is second and IMHO better than the 417A (same trend with 437A) purely from a harmonic profile.
That said I liked the various pentode driven 300B Amp's I have build about equally well as the 417A/437A/6S45/E180F/E810F/D3a TWGC Amplifiers (TWGC - Triode With Gridchoke), all BTW using rather low anodic current, usually around and below 10mA in the driver, feeling they offered a similar overall performance levels.
No, they do not sound alike, either between Pentode or Triode nor among the different types, but they share some overall charateristics, with the Pentodes offering more "Tone" and "Delicacy" while the Triodes tend towards "Detail" but also directness and notably greater macrodynamic robustness (maybe gridchoke plays a role here, they are not easily used with pentodes).
Compared to that the 6SL7 SRPP (with or without gridchoke, also when connected as mu-follower which I thought best) seems a little "further away from the music" sorta like "through a glass, darkly", giving a certain sameness to varied music, more so than the others).
The 6SN7 cascade seemed to combine the worst of all worlds, managing to sound opaque, without variation in the music (making all sound the same) while giving me headaches with edgyness at the same time. Didn't care for that at all.
A cascade with 6072A & 5687WB sounded more balanced and had more immediacy and directness, but not quiet up to the single pentode or triode levels.
Not sure if any of this of use.
I'm currently working (extremely slowly) on a topology provisionally called "Bushido" which aims to combine the best sonic features of the pentode driven amplifiers (the "delicacy", "lightness of touch", "speed" (whatever we want to call it) and the nice "tone" with the macrodynamic cojones of the "TWGC", it might see the light of day at some point.
Ciao T
Hi.Yes, it was TL's email to me yesterday with the same texts.
On reading TL's 3-D HD spectrum charts alone without other supporting papers showing the HD spectrum of the 300B involved, & sonic comparison, I would still find 6SL7 SRPP shows "arguably" better than EL84 pentode in overall HD figures from 0.66W to 11W except its 4th order HD levels at 0.66W & 2.5W
One thing I would agree with Thorsten in general sonic characteristics triode vs pentode. A triode sounds more "detail" & "macrodynamic robustness" & less "tone" than a pentode. Given proper component mating, a triode amp can deliver "lightness of touch" & "speed" too.
Hi.Was it from you or directly from Thorsten in U.K.??
c-J
I ran an EL84 into a UL OPT and cap connected the signal from the primary to various output tubes - 6L6, 7591A, 2A3. Seemed to work pretty well, although the sound was a little aggressive for my taste and maybe a bit sibilant.
Like others here, I have also used an EL84 driving a 300B, a typical Thorsten's Legacy implementation. This driver replaced a standard 6SN7 driver (two triodes cascaded - DC coupled).Way better sound with the pentode: faster, more accurate, better resolution, clearer overall especially in the highs. No screen grid voltage regulation, just Thorsten's "passive shunt regulation", exactly as in the Legacy schematic.
Yamamoto uses the 717A a door knob style pentode as a gain/driver stage in his amps.
I've collected the parts for a 12SJ7 with a 12SN7 as white cathode follower. A implementation of the Pentode based Aikido in Tubecad.com. Anybody else tried this ?
I am using EL84 to drive a 300B, (see posts below on Legacy)it sounds good, no sibilance. I did not use a regulated screen supply.
Bas,Funny I was doing this last week. I needed a larger gain than I could get for a new 20-25W amp. I have three gain/driver stages I am going to try.
I basically setup a EL84 with regulated screen with two 22.1K 12W mills in parallel for the plate. Running the whole thing at like 25ma. So far so good.
The second idea was parallel 12AX7A, running them pretty hard. I think this is a waste of time but still best to try all avenues.
The third is this cascaded 6922 I did years ago for the Cornet integrated 300B amplifier. It's pretty hefty and has about the same gain as the EL84.
I also thought about using the Telefunken EF804. They are suppose to be real nice sounding and much higher drive than say an EF86.
Hi Gordon, not the 804 nor the 86 as these are not intended to deliver large grid currents . c3m or c3g seem to be a better choice , even trioded ! cheers
Hi.Check out ACE 8 watt EL-34 SE monoblock, also available in
DIY kits (www.chimeralabs.com), using only a 12AX7 parallel to drive the EL-34. Response: 16Hz - 33KHz +/- 1dB (1W), & 23Hz - 33KHz +/- 1dB (8W).It earns rave appraisals from critics & end-users alike.
c-J
PS: someone posted down below about its "ugly sister" - ECC82/12AU7.
Are they really that "ugly"?Ask Andrea Ciuffoli (www.audiodesignguide.com). He posted his OTL+OCL project years back using one ECC82 parallel simple cathodyne phase splitter, to drive 2x6C33C-B twin power triodes. What can be worse to drive the current hungry power triodes run in parallel ???
.
To infinity and beyond!!!
I also used an SV83 in a SE 300B ala Thorsten - very snappy detailed sound. It didn't sound 'tubey' - just very neutral, even with the cheap build I did using an outboard regulator, digikey sourced passive parts, and One Electron output transformers.
I have been looking for a "faster" sound. I think everyone should at least try it ....it might just be what you are looking for.
To infinity and beyond!!!
Yes, using a similar "Legacy" implementation of an EL84 driving a 300B. Sounds great to me, no sibilance (I'm using a Mullard EL84, maybe that's why).
for pin 9. So I think I'll sort out the sibilance soon.
To infinity and beyond!!!
.
.
To infinity and beyond!!!
This post is made possible by the generous support of people like you and our sponsors: