WAD300BPP - extraordinary amplifier

[Max N]

I think the first thing to try if you want less gain from the first stage is removing the cathode bypass on the 6072 as Richard suggested - an un-bypassed cathode resistor will be the easiest way of dropping some gain.
FWIW, I think most of us with this amp ended up sticking with the 6072 and using fixed bias (NiCad), which of course gives the same gain as the original circuit.
But please do have a go with alternatives and share the results :-)

I have a theory about why this amp sounds different to (better than?) the Amity/Aurora types. With Amity/Aurora, every stage is push-pull, so 2nd harmonic distortion (actually all even harmonics) are being cancelled at every stage. So you end up with predominantly odd order harmonic distortion, which we all know the ear/brain does not like.
With the WAD amp, the first two stages are single ended so will produce a bit of 2nd harmonic. The output stage of course is push-pull, but remember this will only cancel the even harmonics that are actually generated in the 300Bs, it won't cancel the even harmonics generated in the first two stages - the 2nd harmonic that was generated in the first two stages is embedded in the signal by then.
So I think that what you end up with is an attractive blend - all the virtues of a push-pull output stage, with just a little bit of 2nd harmonic to give it just a soupcon of that SE magic.

Andrew, if you have a scope and signal generator, check out the square wave performance of those interstage transformers. You may be surprised :-)

[Max N]

Also probably worth mentioning that when we did the 300B shoot-out at Greg's, we were using the WAD 300B PP. And on that occasion, we all decided that the TJ mesh plates were the best 300B. So if you want to stick some new output valves in, the TJ's would be a safe bet I would say.

[andrew ivimey]

Hello Max, I'm afraid I have posted elsewhere and I am only running to catch up with what is already known. I have tried Ni-cad, resistor, resistor plus cap and gone back to Ni-cad, as I prefer the sound, not the gain.

With 300B cathodes I have tried going to 1.5K, then to those lovely green RS wirewounds (1.32k) and then to Mills 1.32K. - much much better.

Why, I wonder, are Mills none inductive?!?

No, lovely as 6072 is (a great valve) I naturally thought of others ....and Steve's explanation is right.

Also,

Since that 'shoot out' more 300B valves have come on the market so although we would all agree that for the money EHGGs are smashing but I know some of us have WE 300Bs of which I was very impressed and if only I could get some more 32Bs - WOW! but I only have three and the recommended equivalent is KR300BLX. I'd settle on these too, I admit.

I'd be happy with TJs too but I'm loaded up with 2A3s and can't afford to splash. BUT, it isn't the output valve's fault. We all agree that the gain of this partic. design is very high for modern inputs or using a Pre, so something has to be done. I haven't tried 5687 as input valve but it springs to mind as a good'un.

I like your harmonic theory Max, I wonder. I loved the sound of Amity (super fast, like SE but better - ducks!; it was just silly getting something to drive it. Amity mk2 was free of harmonic distortion to an amazingly high degree. Karna probably is the only solution!

What might I find with some square waves and the 'scope.... good or not so good???

The last time I mentioned using a scope on this BB I was nearly burned as a heretic because despite about 10volts of 70KHz oscillation the vociferous wisdom in reply to my finding was a variation on, 'I trust my ears!' Yeah right, ludicrous. It is difficult to be polite to such a comment.

So, Max old chum, what might I find?

[Max N]

Well, its a long time since I did it and I can't find my notes, so I probably should just say nothing until I get a chance to measure them again. Maybe someone else can chip in? From memory, they do not measure particularly well - but the amp sounds great none the less. I think the problem was that they rang quite badly.
Now Kevin at K&K (Lundahl dealer) is firmly of the opinion that a ringing transformer within reason is not a problem at all and certainly sticking a zobel across the secondaries to kill the ringing can sound worse.
Whatever, the measured performance of the SE:PP interstages was one reason why I lashed up the circuit I took to Owston, to see if PP:PP interstages sounded better.
In the end, the only thing that matters is the sound - and this amp is universally liked. Maybe best not to try the squarewave and scope lest you end up less happy with the amp - irrational but then this is always a subjective game isn't it?

[Andrew]

The original article said they had a right job getting a square wave at 1kHz through them.

In the same vein, ever tried a square wave through an inverse RIAA then into an LCR?

-- Andrew

[Ianm2]

ringing in a transformer is caused by the leakage inductance, which is the inductance that doesn't transfer into the secondary winding, effectifely should be called escaped inductance, combining with the capacitance of the windings, it causes a resonance, and overshoot.

thing is, to get a perfectly damped response, requires that you accurately design, and predict the capacitance and leakage inductance.

to reduce leakage means you need to have more layers of primaries and secondaries, but doing this increases capacitance. in the early stages, leakage dominates, and causes a hf roll off, then you start layering, and this reduces the escaped inductance, there comes a time when you extend the bandwidth by this, and then increasing the layers further causes the capacitance to dominate, and thence again the hf rolloff will suffer.

for any design, there will be an optimal amount of layers.

predicting leakage is reasonably straighforward, but capacitance isn't.

once you gain the maximum hf extension, you will have to attempt to try to not let the capacitance interact with that, and keep the resonance as low as possible

I am not quite sure in practise what that means, ie do you have to make the capacitance equal to the inductive reactance, and so it will balance out?

if you make the core size of the transformer bigger, the low frequency will extend down, assuming your turns and air gap are the same, as these 3 factors affect the inductance and lf of the transformer, and also core saturation is dependant upon core size area in the classic transformer equation, ie

B( gauss or teslas = volts x 10 to the power 8 / 4.44 x low frequency x area of core x number of turns.

but a bigger core will make leakage inductance higher, so a bigger transformer will not extend as hi up per se, to improve it you must increase the layers, and then capacitance will start to dominate.

its fascinating how it all interacts.

as we know, the amp sounds great, and many great soudning things do not measure perfectly.

its a nitemare to predict the capacitance, and most transformers probably use a simple primary, secondary, primary, secondary, primary interleave pattern.

5 layers for simplicity, with insulation as there is hi and low volts between the primary and secondary of some tape, as in the dielectric of a capacitor, so it will likely be paper, plastic of some kind, polyester, polyproplene, teflon, any plastic really, all of which have differing capacitance dependant upon material and thickness.

so really, unless I am mistaken, I don't think makers really pay attention to capacitance precisely, they just layer and leave it at that, and the capacitance and obviously resonance and ringing result.

great stuff.

[Ianm2]

a typcial se transformer will have a -3db rolloff of around 30-45 khz, some intersting things result.

even at say 70khz, -3db begins rolling off in the audio band, hence the need to try to maximise bandwith or upper end response, even over 100k which is very very hard to do, you will start to see the treble reducing slightly from 10khz.

but its not just the frequency rolloff we have to worry about, whenever a reactive component, ie one who's resistance varies which ac frequency is present, phase is shifted, so even with a fairly high frequency rolloff, you still and this is at the very edges of my knowledge, get phase being altered down in the audio band. again, necessetating the import of trying to get as higher hi frequency extension as possible.

low down, the lower rollof limit is subject to 3 factors in a transformer.

the inductance, which is due to

1/ air gap
2/ area of core
3/ number of turns
4/ magnetic path length

so really 3 factors, core, turns, and air gap, the 3 components of a se transformer, ie all you have is a core, wire, and an air gap.

this causes 2 things.

a small signal rolloff, which is always ok, its always less than 20 hz, no matter.

and a large signal rolloff, this is where its interesting, as you need.

from XL = 2 x pi x F x inductance equation 1

XL is in ohms,

primary impedance of your transformer, Z is a combination of XL and dc resistance of windings which is dependant on length, ie number of turns, ie longer wire has more resistance
and cross sectional area of the wire.

thicker wire has less resistance, but the fascinating thing is how the wire all magically seems to fit in the core small window area

its pythagoras, ie

Z = root of XL squared + root of R squared.

the reactive inductance is dominant and the impedance of your transformer can roughly approximate to that.

that' s a little obscure, but your

Z is fixed as of your tube choice, in equation 1 above, so you select a target frequency and your inductance will result.


1/ 4 henries per 1000 ohms to get around 50 hz low freq.
a small transformer

2/ 8 henries per 1000 ohms, to give around 25 hz low freq rollof

most medium normal s.e transformers fit in here

and 3/ 12 henries per 1000 ohms, which gives around 12 hz low freq rolloff.

large transformers, hardly seen.

effectivly, its inductance = Z / 126 for 20 hz.

also, the more the primary impedance, the more the primary turns, so these transformers will need more layering to get the escaped inductance down, otherwise hf rollof will result.

you can roughly guess the number or primary turns by an equation I simplified, its

primary turns = 44 x square root of the primary impedance

or for a push pull its about 34 x the same.

the second lf roll off is due to the size of the core, as the gauss will saturate, its typically around 16,000 for silicon steel.

its the equation above

B (16,000) = volts x 10 to the power 8 / 4 x low freq. x core area x number of turns.

my simplified equation is derived from that.

16,000 is fixed,

volts is fixed from your valve output

you pick your desired low freq rollof

and your core area is already selected from another simplified equation, so you can then derive your number of primary turns or if you do it the other way around, secondary turns, and then use the transformer relations to work out the corresponding primary turns.

its easy to design a core not to saturate, you just make it large enough for your desired frequency.

transformer design in a nutshell.

and then don't forget, your cathode caps will cause an lf rolloff, as will your couplers, and also power supply capacitors, which is usually negliglibe.

its all very fascinating stuff, once you get the hang of how it all works together, you've cracked it, here it is.

finally, your loudspeakers, cone size, box size, loading will all affect the lf performance, most speakers, don't really go below 40-50 hz, so is there any need for a massive transformer, hence why the majority tend to be designed for 25-40 hz lf.

and your room won't support bass below 40 hz, too., its all VERY VERY interactive.

[andrew ivimey]

This amplifier is, of course, the best valve amplifer in the world.

Transformers notwithstanding, phew, I am impressed.

I have a list of possible improvements some of which require a total rebuild so I won't go down that alley.

However, all you boys out there who did the mods ....

AC heating on the 6072/5687. Is it really necessay? Doesn't it actually degrade the sound. I would have thought the little bridge of rectifiers and the caps will introduce DC hash rather than anything pure and simple like indirect AC heating - and ........ why not anyway raise the AC voltage a smidge? - not by a divider off the HT (that too can just introduce noise all over the place, but say use a PP3 battery and a couple of resistors??? Surely this is better than DC heating as suggested????....... or AC heating with resistors to ground creating a centre point - why not?

And then ..... ! laters.

[Greg.]

Hi Andrew,

I built mine with the DC conversion from the outset. This because several who built original kits as standard were unsuccessful in getting rid of hum and changing to DC seemed to fix it. If you do experiment with AC, please let us know how you get on.
Best wishes,

Greg

[andrew ivimey]

Well there is no hum at present so maybe I've got a little leeway to experiment. I don't like DC heaters as a principle though using DC current source for heater supply seems elegant!

I also want to put CCS on the input anode - should clean things up a little more.

laters ...