• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Difference bewteen ultralinear & triode mode?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
It's worth noting that triodes are the only fundamentally somewhat-constant-voltage linearizable* amplifying device. All others have a constant current characteristic: pentodes, BJTs, FETs.

*Linearizable, as opposed to devices like SCRs, which arguably are constant voltage (it switches on and stays right about 1.5V, period!), but cannot be operated in a continuous, linear fashion over their rated range (lots of hysteresis).

SCRs actually share another feature in common with vacuum tubes: they are the only two active devices which conduct no current in reverse bias. BJTs and IGBTs can be reverse biased, but (depending on base drive, for the BJT), only by Vebo (Vec for IGT) at most. FETs are either symmetrical (in which case, arguably, whichever electrode is more negative is the source; it can never be "reverse biased" by such a definition!), or have an intrinsic reverse diode characteristic (almost all MOSFETs).

This means you can't build a general-purpose, voltage source, half bridge inverter out of tubes, without adding diodes at least. SCRs require either a resonant load (resonance induces commutation) or a current source power supply (general purpose).

Tim
 
Actually, ultra linear mode gives higher power but *lower* distortion than triode mode. That's why its called ultra linear.

This is not quite the same as ordinary distributed loading which works on the same principle but doesn't tap the screens at the right percentage OT primary to get the lowest possible distortion.

This is not my impression. Ultralinear is a flashier term for the same thing, If it is a different thing, then in all amplifiers being sold as such, the "perfect" balance has been determined and implemented?

I d like to see some links, in any case where the "perfect" percentage of tapping is determined, thus resulting in ultralinear operation, versus ordinary distributed load.
 
Last edited:
This is not my impression. Ultralinear is a flashier term for the same thing, If it is a different thing, then in all amplifiers being sold as such, the "perfect" balance has been determined and implemented?

I suspect in a lot of cases they simply use the term 'ultralinear' to advertise the amp without actually knowing what it truly means,* in which case they're really just distributed-loaded amps. In other cases they will have got the tapping point from the data sheet, so the tube manufacturer should have found the optimum point (e.g., about 40% for KT88s).

*For example, I think a lot of people assume that when Hafler and Keropes patented the UL idea, they were pateneting the general idea of distributed loading. This isn't true- they knew perfectly well that it was an old idea (one of Blumleins). They were only patenting the idea of an optimum tapping- which is why they gave it a special name.
http://www.oestex.com/tubes/ul.html
 
Last edited:
The nice document artosalo posted shows 43% for KT66. I believe this was determined experimentally. If there are no formulas to calculate the "ultra" ratio, this leaves a kind of a gap , don't you think?
Same for ready made transformers that are marketed as made for a variety f different output valves.
On the other hand I only see british manufacturers use the term"distributed load", so I think its just like valve - tube, i.e. british vs american naming convention.
 
IIRC Mullard call it distributed load and give curves for 20% and 43% taps. I also seem to recall there were some arguments across the Atlantic in the pages of audio journals on exactly what had been invented, by whom and when, and what it should be called. It may be that this was provoked by the standard US habit of patenting some minor detail of an accepted practice widely known and used by others, presumably in the hope that strictly non-infringing users will feel bullied into paying for a licence just in case the lawyers get nasty and the judge doesn't understand what he is judging.

Just noticed that Blumlein patented it in the UK, but then died long before it expired - perhaps leaving nobody to defend it?
 
Last edited:
The nice document artosalo posted shows 43% for KT66. I believe this was determined experimentally. If there are no formulas to calculate the "ultra" ratio, this leaves a kind of a gap , don't you think?
A gap for what?

There is no analytical formula for calculating the ultralinear tapping because the whole system is nonlinear. You would need to know the transfer function of the screen grid and control grid, and even if you knew that, the maths would be truly horrendous since you have multiple feedback loops.
An accurate SPICE model would do it for you though, if such models existed.
Moers appears to have attempted a graphical method, but as far as I can tell it is nonsense.
http://www.oestex.com/tubes/Moers UL_2.pdf

On the other hand I only see british manufacturers use the term"distributed load", so I think its just like valve - tube, i.e. british vs american naming convention.

No, the term ultralinear is very specific- as defined by Hafler and Keroes.** That is exactly why the British use the term 'distributed loading' for the general case- to avoid confusion.* (The British tend to be more fastidious about language than the Americans!)
But in the US I don't think they ever settled on a particular term for the general case; they just talk about screen feedback or similar. Over time the issue was clouded, so now a lot of people don't realise that ultralinear is a special term that applies to one particular case of screen feedback. For example, the American website I linked to above appears confused about the distinction.

*Mullard adopted it first, but it caught on.

**Take a look at fig.2 to see how narrow the UL range is!:
http://www.aikenamps.com/UL.pdf
 
Last edited:
I would say that it is much easier for an audio amateur to design a triode amp to sound acceptably good than to design a pentode or UL (or distributed load) amp to sound as acceptably good.

I say this because the triode will need less feedback around it to be acceptably linear than the pentode, and applying feedback is pretty tricky. If I understand the problem correctly, it's that OPT and its resonances, etc. that can cause problems once feedback is applied, and also whatever roll-offs there might be in the power supply and audio coupling circuits.

I am assuming that the operating points and output transformers chosen are reasonably well-specified for the job at hand, whichever that might be.
 
Last edited:
Does anyone know what's the idea behind the triode mode, technically? and according to the sonic difference, is this idea successful?

It's basically to make up for the lack of audio power triodes. Back in "the day", the pentode revolution happened early on. Most power triodes were intended for RF designs: Class *2 operation, frequently "zero bias" types that don't draw very much Q-Point current, and with high Rp's.

For audio, you don't have much choice: 45, 50, 2A3 (or its 6.3V version: 6A3) 300B or 845. There are some TV vertical deflection types like the 6CK4, but these are in 2A3 territory for Pd ratings. So if you need something that falls between 300B and 845 territory, you need to pseudotriode one of the audio pents if you want triode finals.

As for UL v. triode, it's just a matter of degree. UL restores some of the inherent NFB that triodes have by operating the screen(s) at some AC level. With pseudotriode, you restore all the NFB by basically removing the plate from the circuit. UL is still local NFB and no different from cathode feedback methods, or parallel (anode-to-grid) NFB. You get back some of the triode like characteristic as opposed to pure pentode operation where the Vk2 doesn't vary.

Inherent Feedback in Triodes.
 
I read somewhere that triodes have "nice" distortion spectrum with 2nd harmonic dominating, while pentodes have "nasty" spectrum with 3rd harmonic content. How do pseudotriodes behave? Is UL somewhere in between? Is there a linear relation between UL tap and 2nd/3rd harmonic ratio? Assume the same output power in each case.
 
Virtually all active devices have 2nd dominating. It may be true that pentodes have more 3rd than triodes, but 2nd usually dominates in both. P-P reduces even-order, including 2nd. If pentodes had more 3rd than 2nd, as often seems to be said, then pentodes would never be used in P-P because there would be no point.

I thought the reason for using pentodes was higher efficiency. Yes, a pentode will generate more 3rd harmonic distortion, but that -- and the low damping factor resulting from the high plate resistance of pentodes -- can be corrected with negative feedback. The limiting factor is usually the stability of the resulting amplifier. The PP pentode (or UL) amp will get you more power output than the same tubes strapped as triodes, but will require more negative feedback to get the distortion low and damping factor high.

If you triode-wire a beam tube or power pentode, you get something so close to a triode that it acts just like one. Now you have low plate resistance (yielding better damping factor) and lower 3rd harmonic distortion level, so less need for negative feedback. This makes stability much less of an issue. The price to be paid, as you mentioned, is that your 35 watt pentode amp will now only make 10 watts (or something like that).

There are some, like me, who find something more enjoyable in the sound of an amp that uses less (or no) global negative feedback. There's a nice looseness, a sort of ease to the sound. I know the distortion is higher, but I like it anyway. You may not, and that is just fine. But I have yet to like the sound of any PP pentode or beam tube amp I've heard that didn't have quite a bit of negative feedback. (Not to say that there isn't an amp like that out there that I would like -- it's possible. I just haven't heard it.) I have heard quite a few PP triode amps with no global NFB that I've liked quite a lot.

I guess it comes down to how much power you actually need.

--
 
When I had almost finished building my UL EL34 amp I listened to it with feedback removed, just to check that the feedback was going to improve a good amp rather than paper over a bad one. It sounded OK to me, although I didn't leave it in this state for long.

To hear a pentode with no feedback listen to most 1950's radio receivers. Distortion may be low at reasonable volumes, but the speaker resonances are not well controlled because there is almost no damping.
 
Here is an example at 1KHz of a push-pull 6CA7 in UL mode. The 2nd and 3rd are about equal, although pretty low.
Actually that is labeled wrong.
The top plot is without feedback, the bottom plot is with -5dB GNFB applied.
 

Attachments

  • 6CA7thdplot.jpg
    6CA7thdplot.jpg
    466.2 KB · Views: 691
Last edited:
I read somewhere that triodes have "nice" distortion spectrum with 2nd harmonic dominating, while pentodes have "nasty" spectrum with 3rd harmonic content.

Pentodes only show higher levels of odd harmonics on large signals (and only then under certain conditions), that is, when operating right up to the ends of their transfer curve. On small signals, however, they show a decaying series of harmonics much like a triode. In fact, if you operate a pentode with a small-ish load impedance then the distortion is triode-like all the time. You can see this by looking at load lines.

How do pseudotriodes behave?

A triode-strapped pentode is a triode, and so bahaves exactly as a triode. 'Pseudotriode' is just a silly name the audiophiles invented so they can talk loftily about 'real' triodes like they're extra special. Ultimately you still have three chunks of metal in a glass bulb- they're just triodes.

Is UL somewhere in between? Is there a linear relation between UL tap and 2nd/3rd harmonic ratio?

I would guess the relationship is not linear, since you're cancelling out one non-linear transfer characteristic with a different non-linear characteristic. I suspect you could get all manner of harmonic signatures depending on the exact operating conditions. Just guessing though.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.