[brokentofu]

No.  You can select an output coupling capacitor to push the bass cutoff well below audibility, but an electrolytic that size with low-impedance planars will cost more than an output transformer.  At the same time, it does nothing to increase the current.  You get what comes off of the tube, period.

Contrary to the post above, output transformers (good ones) only need two impedance outputs to match well with just about any headphone impedance on the market.  Things are not that precise in matching impedances.  32 or 300 will work quite well for Grados, THX00's, AKG 701's, and HD800s.  All that requires is a DPDT switch - a single one for both output transformers.  Edcors run about $10-$15 each, that way.  You'll spend a lot more on the tubes.

BTW, an output transformer is much more transparent than an output electrolytic capacitor.  OTL gives up a lot, IMHO.

Is there a build you you suggest for using the cheap output transformers? Something that has a PCB or well documented enough to be someone's first point to point?

 

[Sodacose]

To calculate the needed output cap we use:
 
c = 1 / (2 * pi * Z * f)
 
If headphones are 32 ohm and we want the -3db roll off at 5hz, the capacitor is about 1,000uf.  For 10hz, we'd need just 470uf.  These are not expensive electrolytic values, but they would be crazy pricey in film caps and electrolytics are generally considered bad for SQ.  Bypassing with film might help (in theory it kind of makes sense) but I'm not a cap head so don't ask me.
 
Coupling caps and transformers block the DC current from the tube from reaching the headphones (that's why we use them).  What's important is the output impedance of the tube (determines power transfer and damping factor).  Coupling caps in OTL amps do not offer any assistance in lowering this.  Something like a 6AS7 cathode follower (common in OTL hp amps) has a textbook Zout of about 140ohms; it works well with high impedance but less well with low impedance.  Push pull configurations in OTL can push the Zout lower, but not nearly as low as can be achieved with hybrid SS output designs (Starving Student for example).  All of these OTL approaches would still need a big coupling cap because the high pass is determined by headphone impedance.  
 
It's true that an output transformer with 32 or 300 ohm taps would cover the majority of headphones, but I haven't seen many available with an air gap.  Edcor makes the WSM, XSM 10k:150 (center tapped primaries and secondaries) that would be about right for a single ended tube with approximately 1000ohms Zout (12B4A, 6AH4, 6CK4, 6C45P).  The relevant ratios would be with the 5k:150 and 10k:75 configurations.  Edcor also makes the 10k-7k:300-32 model (used in the Torpedo IIRC) which would allow for a single ended tube with a slightly higher Rp.  These transformers are all pretty affordable but for single ended use they all will require a coupling cap on the order of a few uF (AKA parafeed).  Film caps of a few uF can get pricey.
 
Sowter makes an air-gapped tube headphone output, but the price is way up there.  I've also seen a pair for sale out of Taiwan on eBay for a while now (2W, air-gapped, lots of taps), but I don't have any experience with them as they are also a little too much money just for experimenting.  The air-gap means we can run DC through the transformer and forget a coupling cap.  This is how most single-ended tube speaker amplifiers work as well.  In my opinion it's the ideal situation from a design perspective (good performance and low complication), but it's also the most expensive because no one is making small affordable air-gapped iron with multiple headphone-minded output taps.  
 
Pick your poison: SS outputs, electrolytic coupled OTL output, film cap + transformer coupled output, big bucks for air-gapped transformer, push-pull output.  The 6DJ8 design I linked earlier chooses the last (self inverting push pull).  Push pull means we probably lose some of the single-ended magic and is slightly more complicated than a SPUD, but we don't have to use a coupling cap on the output and the iron is fairly cheap.  Going electrolytic coupled OTL (Bottlehead Crack and similar) is very simple and cathode followers are inherently linear, but it means the output impedance is pretty high for low impedance headphones (low damping factor).  Film cap + transformer (AKA parafeed like the Torpedo from Beezar) addresses some of the short-comings of OTL, but it still leaves a largish film cap in the output (expense in addition to the transformer) and requires some more considerations in how we load the tube.  True transformer coupling with an air-gapped transformer means no cap and simple topology, but it is pretty expensive.  

 

[DutchGFX]

No.  You can select an output coupling capacitor to push the bass cutoff well below audibility, but an electrolytic that size with low-impedance planars will cost more than an output transformer.  At the same time, it does nothing to increase the current.  You get what comes off of the tube, period.

Contrary to the post above, output transformers (good ones) only need two impedance outputs to match well with just about any headphone impedance on the market.  Things are not that precise in matching impedances.  32 or 300 will work quite well for Grados, THX00's, AKG 701's, and HD800s.  All that requires is a DPDT switch - a single one for both output transformers.  Edcors run about $10-$15 each, that way.  You'll spend a lot more on the tubes.

BTW, an output transformer is much more transparent than an output electrolytic capacitor.  OTL gives up a lot, IMHO.

I didn't know escor made cheap OPT's, so I stand corrected on that front. But as mentioned, the coupling cap only serves to block DC from the headphones. You get what you get from the tube, the cap doesn't help lower the output impedance.

I would go hybrid with an EHHA probably.

 

[DutchGFX]

After re reading the above post, which I must say is excellent, I think it should be reiterated that the issue with an OTL is not power (generally), but impedance. Cathode followers have much lower output impedance than typical cathode biased gain stages, but still not low enough for low impedance cans. They can push enough current, but the damping factor is generally poor unless you use multiple tubes.

 

[tomb]

I think everyone focuses waaaay too much on output impedance.  With regard to power, it's probably more appropriate, actually - it's an indication of where the power vector is aligned with regard to load impedance.  I don't think it translates to better control over a headphone diaphragm with some unknown damping, anyway.  This all goes out the window with output transformers, of course.  The windings dictate the impedance.  An output transformer coupled tube amp or hybrid tube amp - will produce the same current and voltage at 32 ohms as at 300 ohms.  There are some slight differences in testing, because the transformer windings themselves present a DC load of a higher value at the higher impedance tap (more wire), but the differences are small and sometimes within the data scatter.
 
The point made about air gaps is true, although Edcor will tell you that DC is blocked, anyway.  Still, I wouldn't want to trust that.
 
What really surprises me is the suggestion that film caps of a few uf get pricey.  Yes, if you want gold-oil Mundorf or Duelund copper, etc., but there are many, many film caps down around 5uf and below that can be had for $5 or less - Solens, Wimas, Epcos, for a few dollars more - Mundorf, Clarity, Obliggato, etc.  You're still light years ahead in transparency combining those with OPTs compared to a giant electrolytic on the output of an OTL amp.
 
Besides, you could apply a servo to control DC for not much money - about as much as the caps.  The SOHA II and the EHHA use one (although without output transformers) and a number of ECP Audio designs (DSHA).
 
To answer an earlier question, as a MOT I'm supposed to shy away from direct answers of products I sell.  Yet the ECP/Beezar Torpedo I is a great example of a true, transformer-coupled tube amp that comes in DIY kits (all on a single PCB).  The ECP/Beezar Torpedo III is truly summit-fi and adds a differential solid-state buffer prior to the output transformers (the tubes already split the signal into differential) to bring power up to about 1 W at both 32 and 300 ohms. 

 

[Sodacose]

The cost of film caps statement is relative.  If the output transformer costs $15, a $5 cap raises the overall cost by 33%.  Add a few bucks for a CCS, and you approach a cost per channel that makes the air-gapped options look not quite so crazy.  You also might start to wonder if you can live with OTL and a big electrolytic to save some costs.
 
Your points regarding output impedance are well made.  With speakers it becomes a real issue, but I too wonder how much we have to wonder about the amp's control over the transducer at such low power levels and when it moves so little.  Still, if you can optimize it without major sacrifices in other design goals, I don't see why it shouldn't be considered.  It's all trade offs.  
 
Regarding transformers and DC blocking, it's necessary with single-ended tubes and un-gapped transformers to block DC (AKA cap couple) in order to preserve the primary inductance.  With push pull or air-gapped transformers, the cap isn't necessary as far as inductance and there's no danger of DC not being blocked as far as the secondary is concerned under any normal operating conditions (not sure if that's what you meant).

 

[brokentofu]

Is there any tube amps that can take balanced input and balanced output on a budget. I'm sure hybrid would be the cheapest. Maybe I could build 2x 12au7 millet hybrid and have one knob control 4 channels. Do you think it would work?

 

[tomb]

Is there any tube amps that can take balanced input and balanced output on a budget. I'm sure hybrid would be the cheapest. Maybe I could build 2x 12au7 millet hybrid and have one knob control 4 channels. Do you think it would work?

Just to be clear, the Millett Hybrid is a totally different amplifier than the Starving Student.  it predates the Starving Student by a number of years and its first version was well documented in November, 2002 by Pete Millett in AudioXpress magazine: http://diyforums.org/MAX/history/ax_hybrid.pdf.
 
There were many other versions of the Millett Hybrid designed and built, beginning with Drew Dunn/Nate Maher's version commonly referred to as the revMH Millett Hybrid.  A couple of years later, Colin Toole (Head-Fi user cetoole) designed the Millett Hybrid MAX.  Later, a refined version with an improved power supply and more compact casework was produced and sold for many years as the Millett Hybrid MiniMAX.  Finally, Colin included the low-noise power supply design into the Millett Hybrid MAX and that became the Millet Hybrid MOSFET-MAX, still sold as a PCB at beezar.com and as a kit available from Jeff Rossel at GlassJarAudio.com.
 
Just an FYI, but the Millett Hybrid MOSFET-MAX is an extremely powerful tube hybrid amplifier and is normally Class-A biased at 100ma-125ma or higher, about the same amount you'll find in the B22.
 
Anyway, the Starving Student came about a few years after that, as more or less a completely independent amplifier design by Pete Millett.  It uses different tubes, a different style power supply, and a different buffer than any other amplifier in the Millett Hybrid family.  Yes, it's a tube hybrid and it's designed by Pete Millett, but beyond that, it has little to no similarity.

 

[Sodacose]

Is there any tube amps that can take balanced input and balanced output on a budget. I'm sure hybrid would be the cheapest. Maybe I could build 2x 12au7 millet hybrid and have one knob control 4 channels. Do you think it would work?

 
Hey Tofu, you could try doing some reading on Tube CAD about balanced tube front ends.  
 
The Balanced Aikido is something to look at - gonna be a lot of tubes, but I've played a bit with this style circuit (single ended) and I like it.  You'd need to decide on an output style, but there are some simple solid state solutions that would probably work fine (a IRF510 totem on a 10M45S, for example).
 
Here are a couple of good articles to read through:
http://www.tubecad.com/2008/03/blog0136.htm
http://www.tubecad.com/2010/03/blog0182.htm

 

[brokentofu]

Those links pointed me in a good direction. I need to learn a lot more before I can start building it seems.

 

[Sodacose]

Update:
 
Brokentofu and I worked together to come up with an affordable (and simple) DIY tube design with balanced input/output.  Here it is for all to enjoy.
 
Lower power no-caps transformer coupled (100mW ballpark with low impedance headphones):

 
 
Higher power (~600mW) because why not (also tested with LM317 CCS in first stage):

 
Power supply (choke DCR 100-150 ohms, whatever henry value is cheap and available):

 
You can find a much more detailed write up on my website in my sig.  No commercial interests, just a free Wordpress blog.
 
Brokentofu built the high power version and by all accounts it's a huge success.

 

[brokentofu]

I made a build album. It has 60+ pictures in it so I uploaded it to imgur.

Build log

 

[Whitigir]

I made a build album. It has 60+ pictures in it so I uploaded it to imgur.

Build log

Where can I find more info about this amp ? Thank you

 

[Sodacose]

You can find a write up here:

https://wtfamps.wordpress.com/el-bad-hombre-headphone-amplifier/