Number of tubes, or power?

[greiswig]

I've often heard people say that there is a girth associated with the 100W, four-tube version of the ODS that can't be duplicated in a 50W two-holer.

Is it really the number of tubes, or is it just power? IOW, would having two tubes with as much power sound as girthy as four 6L6s? Anybody made the comparison?

[markusw]

Interesting question!
I'd love to know the answer myself
BTW, a friend wants me to to do him a layout for a ~50W power amp using 16 (!!) EL95 tubes push/pull. He's got a box of roughly 200 old Mullard EL95s from his father and wants to use them.
It most likely will take us until next winter but I will report back on the results......

[brewdude]

I am no expert. In fact I have not yet built a 4 tube amp, but I am under the impression that a factor in the difference between the sound of 4 tubes vs. 2 tubes is the fact that you have two tubes in parallel on each side of the push pull arrangement, not simply double the power. I would expect this would sound clearer and less constricted and more open sounding.

I have been wanting to build a 4x6v6 amp, but have not had the funds to allow it... yet.

[Structo]

I've heard it explained before but it was a bit over my head.
Something about the transconductance of the extra tubes, plus the different primary impedance.

I for one can attest to the difference in tone (and volume) going from a 2x 6L6 D'lite to a scratch built 100w ODS (4x 6L6).

With a Dumbleator to tame the volume I am really enjoying the advantages of a four output tube amp.

In reality the added expense is mainly the two extra power tubes, a quad rather than a pair.

The extra sockets, a handful of resistors plus the increased cost of a 100w output transformer. All worth it in my opinion.

Nothing wrong with the 50w amp, it has it's own unique signature.

[talbany]

g
A number of contributing factors to consider.. The three of the big ones I think of are

0ne- More IRON=Less saturation

Second- The increase in power and current generated by the extra tubes promotes greater Flux density..flux is the term regarding the magnetic field around a wire that is conducting current. Those invisible lines of magnetic force, if you will.Obviously this is the fundamental force that excites the secondary.
( Flux Density is, roughly, the combined strength of the flux -- that field of magnetic force radiating out. More power on the primary (even at idle), greater flux density, more induced on the secondary, etc..)
Greater flux density can also promote things like touch sensitivity harmonic content etc..

Third-100 watt amps increase in bandwidth propagated by the extra output tubes, this decreases the load on the tubes themselves so you don’t get the anomalies associated with a tube about to clip as soon..Things like compression greater headroom etc..

Others can add to this..

Tony

[greiswig]

I don't know...I'm still not convinced. If you use the same transformers (no bigger iron) with a dual KT-88 amp, other than some undescribed effect of transconductance (which might actually be the same for a pair of 6L6s versus one KT-88 ) I'm still not hearing a good scientific explanation for why there would be a difference.

(OTOH, 6L6s just sound different from KT-88s...but let's put that aside for now )

[Firestorm]

I hadn't thought about the magnetic effects that Tony posted, but in addition to those, there are a whole bunch of parameters that change when you parallel tubes. The transconductance doubles (and if the tubes are enclosed by a feedback loop, the loop behavior must change, too), plate resistance is halved, the input capacitance doubles (so you likely lose some high frequencies). With the signal on each side of the PI split between two tubes, the point where the tubes begin to cutoff would probably at a higher signal level, so the interaction between output tubes and PI must be different. Factors like these can likely be compensated for, but Tony's magnetic effects may be off the table.

[David Root]

I don't have any problems buying the physics of it, makes sense to me. Same as paralleling preamp tubes, basically.

One thing that does mitigate against many multiple parallel pairs, like 8 pairs of EL95, for instance, is the Miller capacitance goes up with each pair added, rolling off the highs. Also with 8 pairs you would be running the Z primary of the OT at one quarter of what it would be for two pairs, other things being equal, which is not likely I think.

I have seen four and six pairs of 6550s used successfully in hi fi amps, so it can be done, up to a point. These amps always run multiple tubes at VERY low individual power compared to MI amps. At MI amps' plate voltages, three pairs of 6550s will get you 300W, but note that's in a bass amp, eg Dumbleland, so no extended top needed.

[Zippy]

I still have to wonder if the added "complexity" is just a function of the summing tubes with varying frequency response. I guess the "experiment" would be to compare 4 6V6s with 2 6L6s with simlarly sized iron - or perhaps 4 K66s vs 2 KT88s.

[greiswig]

Interesting discussion, for sure. Thanks, guys! Maybe at Portland Dumblefest we can take my preamp into someone else's power amp and vice versa and compare.

[markusw]

One thing that does mitigate against many multiple parallel pairs, like 8 pairs of EL95, for instance, is the Miller capacitance goes up with each pair added, rolling off the highs. Also with 8 pairs you would be running the Z primary of the OT at one quarter of what it would be for two pairs, other things being equal, which is not likely I think.

Thanks for your help!
I didn't even think of the added capacitance when using eight pairs of EL95s.
Just checked: compared to e.g. a EL34 the capacitances of a EL95 are roughly one third. Should be OK? What do you think?
The question is where to find an 60-70W OPT with appropriate prim Z ? Should be something like 1.2k....

Peace,

Markus

[Cliff Schecht]

The big thing when paralleling tubes is that the output impedance is halved and the input capacitance is doubled. Most of the effects were covered above but here's a couple of other things to consider. Halving the output impedance of the tubes will have an effect on the bandwidth of the output stage (increases) as well as the damping factor and how the output stage reacts with the speaker load. On the input side, the increased input capacitance is compensated for by decreasing the grid resistance, as is done with a typical quad output stage (usually goes from 220k to 150k). There are realistic constraints here though, obviously as you drop the grid resistance too much you'll be attenuating the signal more than you can compensate for with drive. Also, being that any active device is essentially a charge controlled device, you have to consider that doubling the gate capacitance means that it takes twice the current to turn these devices fully on. This triples with three tubes in parallel and quadruples with four, etc..

I'm not sure I get the flux density argument though. Are you saying that for the same piece of iron you're taking better advantage of the available flux by using four tubes instead of two? All iron is limited to a certain flux density before it saturates but at the same time tiny output tubes will have a hard time energizing a large primary inductance. In other words, the best way to get a good transfer of energy between the primary and secondary, as well as good bandwidth, is to size your iron properly.

[Ian444]

A pair of mosfet followers would fix any drive requirements for multiple tubes with minimum fuss and minimum components.

Edit: maybe not true. If you run the followers after the coupling cap from the PI, then you need a negative power supply as well.

[llemtt]

@greiswig

you can make a 50w sounds almost like a 100w

OT, power supply and feedback loop are key elements

what you can't easily replicate is the fact that paralleling two tubes act as one tube but with a different transfer function

not only transconductance doubles, which is compensated by halving the p-p load, but being the two tubes unequal a different harmonic footprint will result

but you can try different tube brands or different tube types

cheers

[greiswig]

Okay, I'm starting to understand, but I still have questions about an assumption that seems to be being made by several people: that the tubes in question, whether one pair or two, are the same. "The (insert parameter here) doubles" argument applies to pairing the same tube, but I'm trying to figure out if it would be double of what a different type of tube in a single would be.

So...is it true that the relevant factors (transconductance, miller capacitance, etc.) are so similar for KT-88/6550/KT-90 tubes as for 6L6s?