feed back etc?

[steve s]

looking at it from a purely simplistic nature...

most se triode's give very low watts for the typical 4-7% distortion (at max output(

a p/p pentode will give a fantastic output for a typical 1-5% distortion..
thats not taking feed back into account..

and if your not using all those watts you have probably got more usable low distortion watts than triodes..

so feedback does 2 jobs.. lowers distortion further and corrects the amps output across the frequeny range...
now, so if you turned the volume down( from 30-40 watts) and used a cap you have a lowish distortion pentode with no feed back... would that sound better than todays norm.? I don't know?
now, maybe I'am talking real c**p

steve

[NealG]

Yes, on the face of it the pentode offers better power at lower distortion but that is with an optimum load. Reading Crowhurst he clearly shows how much worse the pentode performance is compared to the 'more toleratant triode' when running into a none optimal load. The pentode stage produces more and worse sounding third harmonic distortion. He then goes on to show how UL improves the performance of the pentode...

[petercom]

Exactly, and Crowhurst has only touched the tip of the iceberg!

The cap across the secondary, mentioned by Steve, is probably to compensate for the rising inductance of a typical speaker driver or speaker transformer. In early moving coil speakers the vc impedance was often quite high. Sometimes this was matched to the amplifier secondary, otherwise it was matched to '16 Ohms' by a transformer.

As I said it could possibly be made to work with a single, full range, driver with a low impedance voice coil. Anything else would probably be a disaster.

But we are all agog to know what James does?

[James D]

Quick note as I'm 'on the road' so not much time. This deserve a full answer and I'll do that in time.

Peter is right that I'm using them with fullrange speakers - you know my preferences by now. However they are generally full range with a lf helper so really they are two way speakers.

John Swenson on AA has detailed his use of the EL84 in this way. He's using it in PP which helps a little but the principle is to use the beastie as a current source amplifier rather than a voltage source and then everything is rather straightforward. Of course a conventional hifi speaker with a conventional crossover isn't at its best when driven like this but if the same crossover is reconfigured for current drive - see Nelson Pass's articles on Firstwatt.com, then they work as intended by the designer.

Crowhurst is only right for the cases he specifies (as he would be the first to admit) - there is more than one way of picking an operating point for a pentode/tetrode - why do you think they make such good drivers for DHTs? Because they can operate in very low distortion modes and with distortion profile that is dominated by even harmonics - predominatly second harmonic - just like a DHT. The knack is to look at where you need to operate them to acheive this and then optimise the operating point, the load on the pentode and the OPT to the speaker in a way that keeps the pentode operating at or close to this point... as it is with a pentode driver for DHT. HInt: I use parafeed OPT connection...

James

[James D]

As an example of what I do I'll work through an EL34 case. I've not used the EL34 so this is theory and not a specially selected valve. My working example uses the 12GN7 - which is a special valve - its an EL84 with three times the transconductance... you might argue special case for that - so lets go with the bulk standard EL34...

My choice of op point would be Va= 150V, Ia=150mA Vsg=360V Vg3=0V Vg=-16V. So miles from standard op point.Load a 1.2K resistor (actually a chain of lightbulbs of equivalent resistance to get round the dissapation problem). Then parafeed trannie of between 5-8K:8R. Load line corresponds to a 1K load and any variations in impedance of the speaker are swamped by the 1K output impedance of the stage (output impoedance is rp in parallel with the resistive load)... all conventional valve theory except for the choice of op point... If you do a load line analysis you will find 2nd harmonic dominates and if you put in a typical ellipse load line for a full range speaker you find second harmonic still dominates... and that is all it takes - the willingness to look at the published data curves and find the linear operating region for the device - it is not the most power efficent so commercial amps and conventional wisdom don't go there but it is a perfectly ordinary operating point as far as the valve is concerned. The only other item is parafeed with a resistive load at least 5 times less than the speaker reflected load impedance and you have the same beneficial conditions that a pentode driver of DHTs has and sounds so good at but applied to a power stage.

ANd it works in practice with the speakers I have (Actually tried it with Ruark Prologues too and it drives them happily too).

Hope that clears things up a bit...

James

[NealG]

You continue to amaze me James

It looks like we all agree about keeping the load optimum though But you have a solution to mask changes in load from the pentode...Neat! However a little voice in me is asking why not just use a DHT in the first place?

[NickG]

not just use a DHT in the first place

I sure James will be along with other better reasons, but you do still avoid the miller on the input side.

I am trying something similar with a phono at the moment, trying to fin if the good sound of the LCR was down to the high current in the first valve, or the LCR module (sadly its looking as if the LCR is the winner in that battle at the moment, but still stuff to try).

It seems to me for a phono, despite the greater noise of a pentode (according to MJ), the big big advantage is that you can set the source resistance into the RIAA stage by the load on the pentode (in the same way as James example), so you avoid the problem of the RIAA shifting as the valve ages. It does mean that the anode load needs to be of the same tolerance as the RIAA resistors though.

And RDH does mention that for small signals, given the correctly chosen load, a pentode a lower distortion device than a triode, though I can't test that at the moment as my sig gen is producing about 0.1% distortion on its own :-(

Hmm, just noticed this is somewhat off the thread about feedback, so I will stop now :-)

[IslandPink]

Before we stop on your train of thought , Nick, I also think it's important that the pentode is less affected by the very non-linear load that the RIAA represents . I suspect that the distortion sprectum of a triode loaded down by an RIAA is rather non non-constant as a function of level and frequency , which does not happen to nearly the same extent with a pentode first valve .
Of course, my next phono upgrade ought to answer one or more of your initial questions . Will have to wait for some other tweaks first, though ..

Mark

[NickG]

Yes, good point that Mark, thats also is a strong argument for the first stage passing a bit more current that the average 1ma ECC83 stage.

I do have a scratch phono that I have buil to explore this, its using E180F's running at about 10ma into a low impedance single stage RIAA (0.1u and 0.033u RIAA caps). Sounds ok, but does need some better coupling caps to remove a bit of hardness before I pass any judgement. I do have to admit the LCR is a hard act for it to follow though.

[James D]

Some more quick comments:

The drawback of the no feedback design is power effeciency as you are burning 5 times as much power in the resistive load as in the speaker load with the EL34 so you get only a little over 1W of output. Fine for me with my speakers or for horns but not for everyone. It might be possible to go to a 3:1 ratio with some pentodes but this is never going to produce a lot of power.

Peter's first remark was about global feedback as being necessary - I hate global feedback and have used partial feedback (a local loop between anode and grid) to enable the pentode to drive awkward loads and at conventional power efficencies. In fact this lower output impedance well below that obtained by global feedback loops - by an order of magnitide or more so it is much more effective than global feedback (which really is the worst of all choices in almost any valve amp). Oh stability with partial feedback is better too...

Nick.

The pentode noise performance is worse than the triode but it's gain structure, lack of Millar cap and output impedance are better - what we ideally need is a combination of the two - enter the cascode. Gain is approx. Gm*Rl and output impedance is of the same order as an equivalent pentode. Noise is as per a triode( more or less) and no Millar cap. Sounds ideal so why is none using them??? I don't know and I am experimenting with the 6C45 in cascode - got some stability probs. to overcome

James