How about some feedback to decrease the output impedance?

[xcortes]

I may be saying something stupid but I understand that a small amount of feedback reduced the plate impedance and thus allows for lower impedance headphones to be run a certain amp while maintaining the OTL topology?

I recently acquired a set of Sony MDR-R10 headphones which are 40 ohms. I don't think I'll be listening to a different headphone after them. But from the posts the crack is not ideal for them. Before someone recommeds Sex I do have two Sex amps and I'm also building a WE 437A spud amp with TL404 autoformers. But I'm still intrigued by an OTL amp (I hope Mike LaFevre doesn't see this post!).

Any ideas?

 

[Paul Joppa]

The least damaging form of negative voltage feedback is generally said to be local feedback, i.e. confined to a single tube. (We'll be experimenting to see if we agree, but for now I'm just quoting the accepted view.)

The simplest form of local feedback is a cathode follower, which is used in most OTL amps including the Crack. (It's also used in the Foreplay.) The simplest way to get more feedback is to use a tube with higher mu; to get lower output impedance you need greater transconductance which is just mu divided by rp. The 5998 or WE 421A for example has more than twice the transconductance of a 6080. The 7802 has even more, the 7236 a bit less. All of these have the same pinout.

 

[ironbut]

Hey Xavier,
I have a Tung-sol 5998 winding it's way to me as we speak. My favorite headphone is 40 ohms too (audio-technica ad2000). I'll let you guys know if the 5998 helps. The ad2000's are very efficient so I'm only turning the Crack up to about 9 oclock for a normal listening level. For the same listening level, my Beyers (DT990/880) which are both 250 ohms require about an 11 oclock setting (I've replaced the stock pot with one of those TKD conductive plastic jobbies that has the weird taper though).

The bass does seem to vary quite a bit with different tube sets though. Other than the RCA 6080WA that Doc and Eileen sent me, I've tried a Sylvania 6080WB and a Tung-sol 6080WB with the graphite plates. In the 9 pin position I've tried the Toshiba 5963 that was supplied with the kit, a Sylvania 12au7, RCA 12au7 clear top, a Brimar 13D5, an RCA black plate 5963, a CBS/Hytron 5814a and a Tungsram E80cc.
Obviously I haven't given any of the combos their just due yet but the graphite plate with either the CBS or the Tungsram are leading the pack. Initial impressions are that the Tungsram is more audiophile and the CBS is more musical to my ears and set up.
Of course, that could change tomorrow!

 

[grufti]

Data for some of the tubes mentioned [to the best of my knowledge]. Please respond with corrections, if I messed up.

7802   MU 9 / transconductance 20000
WE 421A   MU 5.9 / transconductance 18000
5998A  MU 5.4 / transconductance 15500
5998   MU 5.5 / transconductance 14000
7236   MU 4.8 / transconductance 12500
6080   MU 2.0 / transconductance 7000
ECC230  MU 2 / transconductance 7000
6AS7G    MU 2 / transconductance 7000

6N13S   MU ~2 / transconductance ?

Please add the values for the Russian 6N13S tube if you know them.

The least damaging form of negative voltage feedback is generally said to be local feedback, i.e. confined to a single tube. (We'll be experimenting to see if we agree, but for now I'm just quoting the accepted view.)

The simplest form of local feedback is a cathode follower, which is used in most OTL amps including the Crack. (It's also used in the Foreplay.) The simplest way to get more feedback is to use a tube with higher mu; to get lower output impedance you need greater transconductance which is just mu divided by rp. The 5998 or WE 421A for example has more than twice the transconductance of a 6080. The 7802 has even more, the 7236 a bit less. All of these have the same pinout.

 

[grufti]

I'm continuing the monolog here, but I hope you don't mind. The 6N13S seemed like an interesting tube until I discovered that its load lines may well be reasonably linear, but they are not parallel. That tends to cause higher levels of THD than some of the other compatible tubes will produce.

 

[Dr. Toobz]

I'm currently running a 6N13S in my Crack and I like it. It seems to have more bass and a richer sound than the 6080 in there before. I do wonder, however, how much of that is placebo. At relatively low power levels, I wonder if one could even hear much difference between the different 6AS7 variants, as none of them should be putting out high levels of harmonic distortion (THD) with the amp only running at 1/4 power (even with Beyers, I rarely turn the thing up much more than 1/3 of the way). I would imagine a bigger difference if the tubes were really being driven hard.

 

[xcortes]

Thanks guys.

Interesting info to me. One question: Is the drcrease in output impedance directly proportional to the increase in transconductnce? If so a 5998 would reduce output impedance from 100 ohms to 50 and using a 421A impedance would be around 40 ohms?

Steve, let me know what your experience is with the 5998. I may give crack a chance?

 

[Dr. Toobz]

How sensitive are your 40 ohm phones? My 32 ohm Grados work no less well than my 250 ohm Beyers, as they are pretty sensitive. I wouldn't rule out the stock Crack too quickly - you might not even need the 5998 if your source is loud and your headphones sensitive.

 

[xcortes]

How sensitive are your 40 ohm phones? My 32 ohm Grados work no less well than my 250 ohm Beyers, as they are pretty sensitive. I wouldn't rule out the stock Crack too quickly - you might not even need the 5998 if your source is loud and your headphones sensitive.

The Sony MDR-R10 are 100dB/mW and I can make the source as loud as I want. So volume won't be an issue. But what about distortion, damping and bass response?

More info. There's a commercial headphone amp that also uses the 6080 as a cathode follower and the specified output impedance is also 100 ohm. The amp is the HA-2 by Wheatfield Audio. According to them:" The output impedance and noise can be lowered as well by substituting different type output tube.  Especially if you want to use 32-ohm headphones (like Grado's), you may want to replace the 6AS7/6080 with a 7236 or 6528(A) tube".

Grufty had already said the 7326 had a transconductance of 12,500. The 6528 has a transconductance of 37,000! I'm not sure that thay are 100% compatible though and right now I have to put the kids to sleep!

 

[grufti]

The 6528A consume a juicy 5A of 6.3V heater current. The amplification factor is 9. All of that is probably too much of a good thing.

 

[xcortes]

The 6528A consume a juicy 5A of 6.3V heater current.

That rules it out. The Crack transformer can deliver 3.8A AC to the 6.3V heaters. The 6080 draws 2.5A and the 12AU7 draws 0.3A.

 

[Paul Joppa]

Time for someone to find a review of the Sony MDR-R10 and see how flat its impedance curve actually is. This thread is spending a lot of time lately trying to solve an intractable issue which may not be a problem in the first place. Most headphones are quite flat and resistive, so the impedance of the source is a non-issue. Most good headphones are compatible with the standard 120 ohm source impedance and have a flat impedance curve.

If you really do need a low source impedance then you really need an output transformer, or a solid state amp. You can get the source impedance a bit lower, maybe even down to half, with various heroic measures (rare and expensive tubes, run tons of current, convert to monoblocks and use parallel triodes, implement tons of negative feedback with a high-gain first stage, etc.) but for the most part it won't get you very far, and/or you won't like the resulting sound for reasons other than the impedance.

Just like SETs almost always work best with high efficiency speakers, OTL tube headphone amps nearly always work best with high-impedance headphones.

 

[grufti]

Paul, I couldn't agree more. I've just been supplying one half of the data needed to get on with this subject, because the information on tubes available to replace the 6080 is probably of lasting interest and readily available.

I won't have an opinion of my own until the Speedball update becomes available. My Crack kit is sitting here waiting to be built once the Speedball kit starts to ship. De-soldering components is often a necessary evil, but I'll gladly avoid it, when a little bit of patience is all that's required.

 

[Paul Birkeland]

The speedball requires very, very little in the way of desoldering.  I worked pretty hard to be sure there would be plenty of room, and that the upgrade could nearly be done by just cutting a few resistors out. 

 

[xcortes]

"ust like SETs almost always work best with high efficiency speakers, OTL tube headphone amps nearly always work best with high-impedance headphones"

Fair enough! Can't say I didn't try.

Couldn't find the R10 curves.

I guess I'll stick with good old Magnequest!

 

[xcortes]

Time for someone to find a review of the Sony MDR-R10 and see how flat its impedance curve actually is. This thread is spending a lot of time lately trying to solve an intractable issue which may not be a problem in the first place. Most headphones are quite flat and resistive, so the impedance of the source is a non-issue. Most good headphones are compatible with the standard 120 ohm source impedance and have a flat impedance curve.

I found an impedance measurement so I decided to resurect the thread. Any comments will be appreciated!

There's a lot of measurements here:
https://5991968839426898461-a-headphonehobbylab-org-s-sites.googlegroups.com/a/headphonehobbylab.org/filehost/home/files/CanJam2010HeadphoneMeasurements.pdf?attachauth=ANoY7cpq_A-JQHQ18Q828Jv3hPQnIXXJ0jnEIOmmBkE_E8Sb06Vej5KNQS8RXqmQR0u4BYRoKaO2cYOy8A-IZp9sgXKoe4yFjZ5uXJy6Np4IFdtUooe97qMg43vXjcSEc4r3xS05DH_WCBKmYJa2auRY3c1e_btUiCYvAKAIZQRipparyJ6kwFvI2BXJEK-WxykYBWic7PPtvPMmpSJBaYNbKHchYco2bGKMEQ6673_H2mwSd9qUV9NR3NVx2dyRVdhIVc6NX9sp&attredirects=0

 

[ironbut]

I visited Tyll and his headphone measuring rig in Chicago (good single malt in there!). He measured a pair of Audio Technica AD2000's while I was there. It didn't do so well with a good sized bump in the "presence" area (around 4k IIRC). In the real world, compared with most speakers with their room interactions most headphones are super flat.
That said, I don't think anyone would deny that listening to a great speaker rig is a richer experience.
I think that's one of the things that set the R10's apart. It seems to convey some of that speaker sound in a much more controlled environment.

 

[Doc B]

Well, it's between 35 and 45 ohms. So the Crack isn't going to be the best match. We're working on a design that - if we decide it sounds OK - will be optimized for this kind of headphone load. It's still in the handwaving phase, we don't know if it will be good, no promises, etc.

 

[Mark424]

My apologies if this is a dumb question, as it's been 25 years since I've talked to Messrs Thevenin & Norton, but could you give the amp a friendlier load by putting a series resistor on the headphone output?  Not sure what that would do to the back EMF and ultimate performance.  e.g. I have the Westone UM3x at 56 ohm input impedance.  But since they are 124 db/mW, I could afford to loose 3-6db gain, no?  

One other question, does the speedball have any effect on performance for lower impedance phones?

Thanks

Mark