What tubes for a tube amp?

[atmasphere]

Originally Posted by markusG

Personally It doesn't matter to me what the numbers are as long as it sounds good.
However I've always been told that low source and high load impedance is important?
My speakers have 5ohm at the lowest and 16ohm at the highest (I think).
Normally a 8ohm output impedance would put me off but if you say I'll be ok, I'll believe you.

Doesn't this however result in very low efficency?

Correct on both counts. The issue revolves around how your speakers were designed. Part of the reason I produced that article on the link I posted is simply because so many people waste money on equipment matching- with our stuff or anyone elses' for that matter. So, especially in the case of 4 6AS7s, the amp may or may not drive your speaker; IOW some speakers with an impedance curve like that will be a piece of cake to drive and others won't. Its all on the intention of the speaker designer. In a nutshell: if the speaker cannot be driven by an amp that otherwise sounds like real music, then there is no way the speaker will sound like real music. Its that simple.

[markusA]

The speakers should be easy enough to drive since they are full range with no filters at all.
(Jordan fx92s)

[atmasphere]

A full range driver should not be much of a problem! I would certainly give it a try!

[bacon665]

The measured dc resistance of your speakers as found in this thread is 5.3 ohms

Now as anyone can point out im by no means an expert on any of this, but I have seen some people do this in a crossover and it may not be such a great thing.
I found this resistor.
Its 30 Watt with a 2.2Ohm resistance.
in my eyes you should be able to put this in series BETWEEN the speaker and ground. 5.3 + 2.2 = 7.5 Ohm this raises you just below 8 ohms so you shouldnt have to worry about sensitivity loss. Then to keep this from being added to that massive 16 ohm bass response you can calculate an inductor for use as a low pass it will need to be placed in parralel with the resistor to bypass the extra resistance for signals in its pass band.
Judging from your impedance chart i would design the cut off for about 90hz.
With a little bit more work im sure you could work on the lower and upper sides of that hump to flatten the impedance to an average 16 ohm load.

Now before you get too excited wait for someone to post about the cons/pros of this. As i said im not expert but i know if you do that to a dc motor is gets hot and this idea could be damaging to your speakers or the frequency response.

[markusA]

I suppose one could take it even further with a series resistor and a zobel network.
It wouldn't be too hard to tailormake a "perfect" impedance. Othoh it would result in powerlosses in the newly added circuit. I'd rather not take that rout unless I had to.
But we're straying way off target here. This thread is about amplifiers!

Like atmasphere said, it's worth a try.
And adding tubes will lower the source impedance as well as increasing ouput. This should keep one busy for a while I think?

Now for some amp questions...
+/- 140V (B1) and +/- 300V (B2) seams like standard values for tube psu's.
How much current would they be able to deliver?
I think I read each output tube dissipates up to 2x13W and designing for maximum load (tubes per channel) we get 8*2*13=208W ( 0.7A@300V)
Lets say 20% losses and make it 250W. (0.83A@300V)
And choosing components for 1A should put us in the safe area?

15W (maybe 20W with some to spare?) for the 4 sn7gt tubes?

And then the heaters... what voltages? And how much current? (wattage?)

(It's more stupid not to ask the questions than to ask stupid questions.)

Am I doing this right? Or have I missed some vital stuff? I'm a n0ob and learning as I go along.

[SY]

Understand that adding a resistor will not help the sensitivity- resistors don't make sound, they just convert watts to heat. A series resistance will certainly change the frequency response; sometimes that's for the good, sometimes not. See, for example, the "Arpeggio" loudspeaker project, where the inherently high source impedance of an SET amp is used to advantageously increase the bass extension of a closed-box system. That's one reason why OTLs can be controversial- they DO sound different, and much of the difference is due to that high source impedance. For some speakers, the resulting EQ is a positive thing, and those OTL owners are happy. For others... not so much.

[bacon665]

Originally Posted by markusG

I suppose one could take it even further with a series resistor and a zobel network.
It wouldn't be too hard to tailormake a "perfect" impedance. Othoh it would result in powerlosses in the newly added circuit. I'd rather not take that rout unless I had to.
But we're straying way off target here. This thread is about amplifiers!

Like atmasphere said, it's worth a try.
And adding tubes will lower the source impedance as well as increasing ouput. This should keep one busy for a while I think?

Now for some amp questions...
+/- 140V (B1) and +/- 300V (B2) seams like standard values for tube psu's.
How much current would they be able to deliver?
I think I read each output tube dissipates up to 2x13W and designing for maximum load (tubes per channel) we get 8*2*13=208W ( 0.7A@300V)
Lets say 20% losses and make it 250W. (0.83A@300V)
And choosing components for 1A should put us in the safe area?

15W (maybe 20W with some to spare?) for the 4 sn7gt tubes?

And then the heaters... what voltages? And how much current? (wattage?)

(It's more stupid not to ask the questions than to ask stupid questions.)

Am I doing this right? Or have I missed some vital stuff? I'm a n0ob and learning as I go along.

ah my friend i finally do have some truely usefull info for you.
Tubes have datasheets just like transistors one of the largest collections can be found here.
In the american naming scheme models start with a number followed by a product designation that starts with a letter the first number 6,12 denotes the heater voltage but there are exception 300b 845 80 ect.

http://scottbecker.net/tube/sheets/135/1/12JN8.pdf
12.6 volts per tube heater, .3 amps.
Typically people just wire the heaters in parallel
so for 4 tubes your going to need 1.2 amps allow for a 5% deviation from that current draw gives you 1.26 amps so id just as well design the heater part of the psu for 1.3 or 1.5.

If you want more regulation theres LM78 series just pick your voltage and tack it on to the end so you need an LM7812 to regulate positive 12 volts these things have a limit of 1 amp so you will need two for four tubes

this page has a nice set of regulator circuits with the formulas to calculate them for whatever your needs may be Regulators and Power Supplies

[Miles Prower]

Originally Posted by markusG

If I wanted to build a tube amp without output transformer, what tubes would be a good choice? Lets aim for 20w or something along those lines?

Trust me on this: you do not want a hollow state OTL. If it's an OTL you want, your best bet is to put the hollow state electronics up front driving power MOSFETs on the back end. It'll sound better, and it won't be a room heater.

The vacuum tube is fundamentally a high voltage, low current (Hi-Z) device. Nothing you do will change that. Any OTL will be running into a nearly vertical loadline, and that means just essssssssssss-loads of distortion. You can, of course, correct for that nonlinear operation with gNFB. Why bother? Yes, OPTs are expensive and nonlinear, but they are one helluvalot more linear with a reasonable loadline, and will require much less in the way of correction.

Then there is the efficiency problem. In order to get reasonable power levels into the usual 8R speeks, you will be paralleling a great many VTs. The heater power alone is really gonna add up. I worked up an OTL design that could do some 40W into 8R, using 6BQ6GAs. That required eight 6BQ6GAs per phase (Heater: 6.3V @ 1.2A). I can get that much from a single pair running into a much more reasonable loadline of 1K1 / phase, and matching impedances with a decent OPT. Which would you prefer? I already have my answer.

I like the thought of a single ended amp so I guess that would be on my wishlist as well. But I don't know enough about designing amplifiers to really kno what I'm talking about.

I just want a place to start my reading.
So, a recomendation on suitable tubes and perhaps a topology. That'll keep me busy for a while.

Yes, it helps to know what one is talking about. Take a good look at Technical Books Online -- lots of freebies to download. Depending on where you're at, you might want to check out the Radiotron Designers Handbook.

[markusA]

I think I should point out that I really don't need this amp. I have Musical Fidelities X-A200 monoblocks and I haven't seen a speaker they can't drive.
My newfound interest in tubes are purly for fun.
If in the end it actually sound better than my current set-up, It's a great wonderful bonus.

Of course I want a Kick a$$ sound in the end, but I have time on my side.

[mullardel34]

Originally Posted by moonbird

I have looked hard at doing an OTL design with the mentioned tubes, but must admit that I usually end up agreeing with Leadbelly. Seems like a pretty (and $$) heater in the end. (Not taking a side here just struck by how close this thread is to an on-going internal dialog (its sick, I know ).

And yet these designs are everywhere. And people swear by them. Could some of the OTL supporters answer the question:

"Why bother?" There MUST be a reason. The tubes are impressive -- how much does that enter in? How much heat do these things produce? Are they recommended below the 40th parallel ? How much do they weigh??

Regardless of the answer to the why question - I hope a design gets discussed. Hopefully, lower output options might be discussed too -- 80 watts of tube amp seems like a awful lot to me in my small listening room.

I'll take a run at the "Why OTL?" question. But first, let me preface my comments by stating that I'm not trying to instigate a flame-fest over subjective-versus-objective sonic attributes. I firmly believe (and have proven to myself over the years) that good engineering has to be the underlying basis for audio design. Once the engineering fundamentals have been addressed, the subsequent "voicing" of audio components in order to maximum the perceived level of music reproduction is simply the use of the ear as final arbiter in finalizing the audio design. So hold the flames, please...

I built a pair of Rosenblit-based 25W OTL monoblocks and I just love the way they reproduce music, regardless of musical genre being played. A properly-developed tube circuit just audibly preserves more of the music. I haven't seen any objective basis for this stance presented, but once you've heard a really well-developed vacuum-tube audio component, there's no going back... The bottom-line is, OTL's are all about the music.

Really well-designed vacuum-tube preamplifiers (low-feedback, locally-degenerated, balanced, fully-symmetric circuits) can inherently convey the musical essence in our recordings at a performance level that is truly astounding. And we're talking about circuit topologies that present excellent measured performance --- no "magical" circuits that "sound great", but "measure bad". But that brings us to the subject of vacuum-tube power-amplifiers. One of the key limitations of most vacuum-tube power-amplifiers is the set of sonic constraints imposed by the output transformers used to impedance-match the amp and speakers; OTL's eliminate the output transformers, bypassing all of the negative aspects of output transformers. With OTL's, all of the musical nuances are directly coupled to the loudspeakers with none of the sonic artifacts and degradation incurred when employing output transformers.

Most contemporary loudspeaker designs need a fair bit of current to be delivered by the power-amplifier in order to achieve reasonable sound-pressure levels in the listening room. Since vacuum-tubes are (for the most part) current-challenged, a virtual phalanx of vacuum-tubes have to be paralleled to provision the necessary current-delivery capacity. Vacuum-tubes are, relatively speaking, high-voltage power-amplifier output devices. Most OTL's run with a +/- 140VDC dual-rail power-supply, contrasted with a solid-state design's +/- 20VDC dual-rail power-supply, to power a 25W/8-ohms monoblock design. Since power-dissipation equals (current)x(voltage), a comparable standing idle current in the output stage will result in substantially greater heat load for the higher-voltage vacuum-tube OTL as compared with a solid-state design. Due to operational constraints for the vacuum tubes, we can't drop the OTL power-supply rails to the voltage levels of a comparable solid-state design, so the OTL must pay a fairly substantial energy-efficiency penalty to deliver a given level of power output. But, remember, OTL's are all about the quality of music reproduction...

I'm really interested in investigating "atmasphere's" low-feedback, balanced/bridged OTL topology. My current Rosenblit-based OTL's are comprised of cascaded, single-ended gain stages feeding into a "totem-pole" quasi-complementary output stage, all within a relatively high global negative-feedback loop --- while they sound great and measure really well, there's always room for improvement. Through my own series of experimental line-level prototype designs, I've become convinced by what my ears tell me that a low-feedback, fully-symmetric, balanced OTL design has great sonic potential. So I'll be following this thread with great interest and anticipation.