OTL Amp and Headphone impedance matching question

Jan 24, 2014 at 10:13 AM Thread Starter Post #1 of 19

gatorjon

New Head-Fier
Joined
Jan 9, 2014
Posts
9
Likes
10
From reading other posts I have learned that it is generally recommended to not use low impedance headphones with Class A otl amps. My question is: Is there a general "rule" about how low is too low in terms of impedance? In other words, is there a lowest recommended impedance for use with those amp types? 
 
Jan 25, 2014 at 8:52 AM Post #4 of 19
Let's take the common 6080/6AS7G OTL cathode follower amps. Output impedance is roughly 1/transconductance for a cathode follower. That puts the 6AS7 with transconductance of 7000 at about 142ohms. Changing the tube to a 5998 with transconductance around 14000 gets you down to 72 ohms. Even better is the 421A at 20000 which gets you close to 50 ohms. For OTLs with two output tubes (like WA2), cut those in half. I find the HD558 sounds good (to me) driven with a WA3 in the 5998 or 7236 configuration and it goes loud enough before distortion....but not to ear bleeding levels.
The HD558 is rated at 50ohms, but the graph shows a pretty big mid bass hump....that with an OTL, firms up the bass quite a bit. http://www.headphone.com/learning-center/build-a-graph.php?graphID%5B0%5D=2861&graphID%5B1%5D=&graphID%5B2%5D=&graphID%5B3%5D=&scale=30&graphType=7&buttonSelection=Update+Graph
With 16 ohm skullcandy earbuds (don't laugh)...it goes plenty loud enough because they are pretty sensitive. I'd say for over the ear headphones, 50 ohms is about as low as you would want to go but you lose power.....125 or more being better with 250 usually the high point on the power curve.
 
Jan 25, 2014 at 10:07 AM Post #5 of 19
The most common OTL output valves are the 6AS7G (as discussed above) (6080 also) and the 6N6P family (Schiit Valhalla, Little Dots 3, 4 and 4SE etc as well as many other lesser know OTL headphone amps). Off the top of my head the transconductance of the 6N6P is about 11,000, (please correct me if someone can look it up). These two dominate the valve 'families' supplied with OTLs. 5998, 7236, 421a etc are at the esoteric end here (as are 6N30P DR), although loved on the forum (and used by the likes of me) they are not that common elsewhere and are rarely supplied as standard with amp (though occasionally as a cost extra).
 
Jan 26, 2014 at 12:16 PM Post #6 of 19
My apologies for the long winded response... I had a long DYI-mod journey to get around impedance matching.
 
The other aspect is the output coupling capacitance value designed into the circuit.  This element of amp designs seems to vary wildly from amp to amp.  Paper-film / low ESR capacitors aren't cheap, and they are very big (think 12 Oz soda can), so there are almost always circuit trade-offs made by the designer.  They make compact low-ESR electrolytic caps, but I am not sure why these are generally unaccepted for this application.  Its important because the output capacitance, combined with the impedance of the tubes and with the low-Z headphone voice coil form a high pass filter that cuts off bass.  So while "common tube" OTL amps are not optimal with low impedance loads, some are far worse than others.
 
My work around has been 25W speaker output transformer coupling with my darkvoice.  It looks crude, but it really does work and more importantly it sounds great too.  I run my Grados and K701 in this setup and it sounds great.  Deep-bass extension is excellent, as well as impact and dynamics.  It does not work with all OTL amps though.  I think the circuit needs enough wattage/power to properly energize (for lack of a better term) the primary coils.  It sounds flat and lifeless with my little EarMax, an amp that has a long history powering 300-600 ohm loads.
 
I racked my brains out over an amp mod a while ago with Nikongod, dsavitsk and I think Tomb too (BIG time DIY credit goes out to those guys)... trying to wrench the necessary paper-film capacitance reservoir into the amp chassis and just could not get all the stars to align.  Space and volume limitations inside were the biggest problem.  I would have had to build up the amp chassis by 2-3 inches to squeeze everything in.
 
These 2x caps are ~55uf and would need to be in the vicinity of ~2x that uf value.  Maintaining the 400V ceiling, that would roughly double the cap overall size.  Whereby mounting and space would be a problem.  The cap reservoir is the last component in the amp circuit, its what the headphone plugs into... and I suspect its a very important component in the amps design.  Theres also a safety concern here as well.  Lowewring capacitor voltage rating allows you to UP the uf rating, within the same volume space.  HOWEVER, if you under spec V too much it will toast your headphones.... and probably do other bad things as well.

 
 

 
 

 
Its the same solution Mapletree employs on their Mad Ear amps that everyone (rightfully so) raves about with Grados.  They are using Hammond 119DA 12W speaker transformers, with the 4 and 8 ohm secondary coils powering the HI/LO-Z headphone jacks.  Its a simple, compact and affordable solution that I'm surprised more amp designers are not using.  Here are some www pics of other peoples amps/builds, not mine.

 

 
Jan 26, 2014 at 6:53 PM Post #8 of 19
I never understood either why many dislike the electrolytics in that role. Electrolytics are at the output of the Woo OTLs....and they sound just fine. At 470uf, the bass response is subsonic. :)
I ended up with a pair of those Hammonds after reading your thread....I still need to get them in a classy enclosure.

They make compact low-ESR electrolytic caps, but I am not sure why these are generally unaccepted for this application.

.......Hammond 119DA 12W speaker transformers, with the 4 and 8 ohm secondary coils powering the HI/LO-Z headphone jacks
 
Jan 27, 2014 at 8:38 PM Post #11 of 19
I never understood either why many dislike the electrolytics in that role. Electrolytics are at the output of the Woo OTLs....and they sound just fine. At 470uf, the bass response is subsonic.
smily_headphones1.gif

I ended up with a pair of those Hammonds after reading your thread....I still need to get them in a classy enclosure.
They make compact low-ESR electrolytic caps, but I am not sure why these are generally unaccepted for this application.

.......Hammond 119DA 12W speaker transformers, with the 4 and 8 ohm secondary coils powering the HI/LO-Z headphone jacks


Kramer speaks truth.  OTL amps are at a disadvantage when it comes to low-impedance phones and the reason is not roiled up into impedance mis-matches.  Bass cutoff is a reality and depends on how big a capacitor you can hang off the output.  Electrolytics in high-value uf's are generally more economical (film caps aren't even possible in some ratings - like the beer can analogy he used).  In an ideal situation, the reason electrolytics are not favored however, is that compared to output transformers or film capacitors, they put a fog over the music, literally.  They are simply magnitudes less transparent and inherently higher in distortion.
 
Dsavitsk favors parafeed output-transformer amps and full-disclosure: I sell one of his designs: the Torpedo.  He sold a parafeed, too - under ECP Audio: the L-2 (see the Summit-Fi forum section).  Tyll mentioned it in his Abyss headphone review as an ideal pairing with the HD800 (he setup a shootout scenario).  A parafeed output-transformer design still uses interstage coupling capacitors, but because of the connecting impedances (output transformers) the requirement for uf's is very low.  The Torpedo uses 4.7uf coupling caps, for instance, which are well within range of film capacitors - maybe half the size of a roll of lifesavers, for instance.  The transparency with film capacitors is monumentally superior to an electrolytic.  Just for a magnitude comparison, note that four hundred seventy uf's are quoted up there, compared to a four-point-seven uf film capacitor in a parafeed, output-transformer design.
 
With OTL amps, the connected load impedance (headphones) makes all the difference.  That's not because of the impedance interaction - it's because the headphone impedance combined with the output coupling capacitors forms a textbook-classic RC circuit.  Such a circuit is frequency-dependent for current flow.  IOW, frequencies too low will not pass (high-pass filter).  In the case of typical headphone impedances, even 470uf means bass cutoff is beginning at about 110Hz.  At 20Hz, it's down by more than 1dB.  That doesn't sound like a lot, but remember that we're used to ruler-flat responses in amplifiers from subsonic to almost 40-50KHz, or more.  There are phase distortions that occur well above that 110Hz, meaning the bass is affected well into the range where 32 ohm headphones would be affected
 
Here's a shot that can illuminate the issue.  Remember what Kramer said - even 470uf electrolytics are huge at 200-300V, and there are none that are considered "audio-quality" (Black Gates, Nichicon Muse, Elna Silmic's, Cerafines, etc.):
 

 
For comparison's sake, many OTL amps use only 220uf electrolytic output capacitors.  They get away with this because they expect people to use high-impedance headphones with it:
 


As you can see, even not counting for any phase distortions, bass is severely affected at almost every impedance except for 300 ohm.
smily_headphones1.gif
 
 
Jan 27, 2014 at 9:07 PM Post #12 of 19
Great clear explanation! I've been using the calculator on the Vcap site to calculate cut offs: http://www.v-cap.com/coupling-capacitor-calculator.php
Ew, at 220uf that's looking pretty poor and most likely noticeable by a non-audiophile.

Woo used Nichicon in the WA3 I have...about the size of my thumb...they are not terribly large (nowhere near as big as film caps).
Blackgate can also be found in 100V 470uf and are 22x42mm
Nichicon and Elna both currently make audio grade caps in 100V/470uf/1000uf
 
Jan 27, 2014 at 10:23 PM Post #13 of 19
SOLID Tomb!!  thanks for the detailed explanation.
 
This is exactly what I was talking about in my post above, when I said "(BIG time DIY credit goes out to those guys)".
 
Jan 27, 2014 at 10:27 PM Post #14 of 19
Great clear explanation! I've been using the calculator on the Vcap site to calculate cut offs: http://www.v-cap.com/coupling-capacitor-calculator.php
Ew, at 220uf that's looking pretty poor and most likely noticeable by a non-audiophile.

Woo used Nichicon in the WA3 I have...about the size of my thumb...they are not terribly large (nowhere near as big as film caps).
Blackgate can also be found in 100V 470uf and are 22x42mm
Nichicon and Elna both currently make audio grade caps in 100V/470uf/1000uf

But don't you need voltages higher than that?... for safety reasons more than anything?  I guess that would depend on the voltage levels supplying the main output tubes right?
 
Jan 27, 2014 at 10:31 PM Post #15 of 19
Ooo, yes, let me be clear. 100V is for a typical 6AS7 cathode follower output design like Woo and Crack and others like that. The cathode bias resistor or current sink never generates more than ~60V. For designs that tap the plate resistor, cap values need to be the same as in the power supply.

But don't you need voltages higher than that?... for safety reasons more than anything?  I guess that would depend on the voltage levels supplying the main output tubes right?
 

Users who are viewing this thread

Back
Top