Tube amp for both AKG K340 and Grado (µ-follower)

[00940]

Ever since I sold my RKV a few years back (don't ask me why), my K340 have been longing for a tube amp. I'd like to remedy to that situation. But I'd also like to build an amp that would be truly useful with a wide range of cans. For once, I'd also like to avoid OTL and big output capacitors.
 
I've got a few parts lying around that I'd like to use for this project. Among those parts:

- A xformer with 330-0-330 @85ma and 6.3@4A secondaries,
- A pair of 25W 100V line transformers with 4, 8 and 16 ohms taps on the secondary and with 25, 20, 12, 6W taps on the primary (advertised bandwith of 30hz-20khz).

You'll find below what I'm considering for now... but if you have better ideas or improvements to suggest, I'm all ears.

 
 
 
 
 
Obviously, the inspiration is coming from Alan Kimmel's article on mu-follower stages with pentodes and from the use of line transformers in the Bottlehead S.E.X.

If my calculations are ok, using the 25W and 16ohms taps of the line transformers, headphones with a 320ohms impedance should create a 8K load on the primary, with a 5/1 ratio. Using the 6W and 8ohms taps would convert a 32ohms pair of headphones into a 6.6K load, with a 14/1 ratio. Such loads should be quite ok for a beefy mu-follower stage. Hopefully, as I'm not loading the xformers too much, I should also get a decent frequency response.

The xformer will give me 350VDC to play with (using an EZ81 for rectification), with current to burn. By running the pentodes at 30-35ma (the ECL805 can provide 75ma), I'll have plenty of current for low impedance cans. The concern is more the limited voltage swing on a 350V B+ (340V most probably after a choke). Sims give me "only" 8.5Vrms at the output for 300ohms cans before thd figures start hitting 1%.

I've got a question still about the pentode screen biasing.

  Kimmel explains well how to set all the resistors values... but this part is more obscure. I've used what gave best results in sims but I've got a limited trust in the tubes models for such fine adjustments. If anyone can comment on this, thanks a lot.

 

[regal]

You may want to consider a SET with the gapped transformer on the anode,  as what you are proposing isn't a huge difference from an OTL IMO (still has a coupling cap.)  Just a consideration.

 

[FrankCooter]

I agree with regal's  OTL opinion. You're taking the output from a transformer connected to a capacitor connected to a cathode follower. I admire your creativity but I think you are mixing apples and oranges here. If you're going to use a transformer, take the output from the plate of the lower triode. If you want to go OTL, eliminate the transformer and increase the output cap to an appropriate value. As it stands, you've sort of got the worst of both topologies without the benifits of either. I'd also substantially increase the current through the triode and increase the voltage drop across R4. Ideally, it would be nice to drop about 25% of the total B+ across the plate resistor of the triode. I'd use an LED for bias. I've built this circuit a number of times, like all SRPP variations, output taken from the plate of the lower tube almost always sounds better than output taken from the cathode of the upper tube.

 

[regal]

Quote:

, like all SRPP variations, output taken from the plate of the lower tube almost always sounds better than output taken from the cathode of the upper tube.

Yep when taken from the plate of the lower tube it is SET (Broskie calls this a totem pole),   taken from the cathode of the upper it is a push-pull IMO becomes unbalances without a specific load.  Always found better sound with the totem pole config,  may sound good with a parafeed OPT,  basically this is what the Stacker II is but it uses a SS buffer in lieu  of a transformer.

 

[00940]

First of all, thanks for the comments. It's always good to receive some food for thought.
 
Quote:

I agree with regal's  OTL opinion. You're taking the output from a transformer connected to a capacitor connected to a cathode follower. I admire your creativity but I think you are mixing apples and oranges here. If you're going to use a transformer, take the output from the plate of the lower triode. If you want to go OTL, eliminate the transformer and increase the output cap to an appropriate value. As it stands, you've sort of got the worst of both topologies without the benifits of either. I'd also substantially increase the current through the triode and increase the voltage drop across R4. Ideally, it would be nice to drop about 25% of the total B+ across the plate resistor of the triode. I'd use an LED for bias. I've built this circuit a number of times, like all SRPP variations, output taken from the plate of the lower tube almost always sounds better than output taken from the cathode of the upper tube.

 
I've had OTL before and the output impedance and current handling of those amps are not good enough in my view, especially SE and without global feedback. The big caps bother me a bit too. So OTL are out (at least this time). The problem is that those line xformers won't stand DC through so I need a cap whatever way I go (unless going push pull maybe, but I'm not quite certain it'd be better). The capacitor (that should be bumped to at least 10 uf by the way) can be high quality mkp; such a cap doesn't bother me much.
 
So, unless I buy more stuff, I'm limited to cap+xformer (parafeed). The question then becomes, how to connect the xformer ? Two main options here, take the output from a cathode follower or from a grounded cathode amplifier. As my B+ can be quite high, using a gain stage + cathode follower output had the advantage of further reducing the output impedance. Hence, I went the mu-follower way, which nicely integrates the gain stage and a beefy output stage. However, the ecl85 isn't that good a tube in this application, so I'm currently back to the drawing board...

 

[nikongod]

I like the presence of R8 in the schematic in the OP

too many people leave this out and try to figure out how to get high currents through small signal tubes.
 
Anyways, I would echo the comments of others, and say build a simpler parafeed. Drive the load off of the plate of the output tube rather than the mu-follower output.
 
The original Espressivo (link) and Dsavistik's Less Espressivo (link) have always been quite helpful to me.

 

[00940]

I'm going to have a serious look to the less-pressivo.
 
In the meantime, I've been fooling around with designs in ltspice:
 
This one is probably the simplest.. and still the apparently the best one. Sometimes less is more.
 

 
PS: there are btw designs around from serious designers using cathode followers+ caps+ xformers. See for example:
 
http://www.decware.com/headphone/MLBschematic.pdf
http://www.pmillett.com/images/ECC99_SRPP_2.PDF
http://hollowstate.netfirms.com/Super2manual.pdf (mapletree ear+)

 

[digger945]

I thought your last post had 2 possible circuits.
It can be tough sometimes to build something sometimes with just what you have in the parts box.
Did you use the word "ain't?"

Maybe a couple of other sites that may(or may not) help that I have used some...
 
http://livinginthepast-audioweb.co.uk/mainindex.php
 
http://mysite.du.edu/~etuttle/electron/elect27.htm
 
...and one from a friend on another forum that I found pretty interesting....
 
http://syclotron.com/

 

[00940]

Thanks for the links, especially the second one.
 
Quote:

/img/forum/go_quote.gif

I thought your last post had 2 possible circuits.
It can be tough sometimes to build something sometimes with just what you have in the parts box.
Did you use the word "ain't?"

 
You're right, I posted a second circuit earlier on. But the second one was a typical simulation error, when you follow an idea in the ideal world of spice, without reality checks. To make things short, it was a grounded cathode pentode amplifier, with a ccs on top and a xformer from the plate in parafeed... which meant that the output impedance was very high, even with an healthy dose of ultralinear thrown in. It was even more stupid that I had read this page http://www.ecp.cc/CagedFrog.html before. It would have made a funny gain stage though, as varying the UL ratio changed drastically the thd pattern.
 
PS: I couldn't say anymore if I used ain't or not... why the question ?

 

[digger945]

Quote:

Thanks for the links, especially the second one.
 
 
 
PS: I couldn't say anymore if I used ain't or not... why the question ?

I guess you already know that Pete Millett has another website devoted just to books and the people who help to scan them into the computer...
http://www.tubebooks.org/
...found when trying to learn a little more about Child's Langmuir 3/2 Law (Child's Law).
Lots of very interesting reading.
 
I'm not sure it was you I saw use the word. I was reading in the drive thru at Taco Bell today and it just struck me as odd that someone from your part of the world would use the word. I use it all the time. No harm, just curious.

 

[00940]

Yes, I know of Pete's depository. Pretty amazing work.
 
As for "ain't", it is indeed quite possible that I would use it; I've spent one year studying in your part of the world (Colorado University at Boulder). Lovely place and great memories... That's actually where I discovered head-fi too (I needed cans to avoid bothering my neighbours in the dorm).
 

 

[nikongod]

 
Quote:

You're right, I posted a second circuit earlier on. But the second one was a typical simulation error, when you follow an idea in the ideal world of spice, without reality checks. To make things short, it was a grounded cathode pentode amplifier, with a ccs on top and a xformer from the plate in parafeed... which meant that the output impedance was very high, even with an healthy dose of ultralinear thrown in. It was even more stupid that I had read this page http://www.ecp.cc/CagedFrog.html before. It would have made a funny gain stage though, as varying the UL ratio changed drastically the thd pattern.
 

I heard that amp at canjam. It was quite nice.
 
I dont think its output impedance that screws up headphones. I think its how the amp distorts. When people think of high output impedance, they think of an OTL amp. measure one... they suck with low impedance headphones! Even people who dont particularly care for measurements will agree that most open loop OTL amps suck on paper with low impedance headphones. I think a poor parallel was drawn between damping factor in headphones due to this. Try grados from (insert any SS amp with more than 20db of global feedback) BUT with a 100 ohm source resistor to reduce the damping factor. not too bad. quite nice in fact.
 
To expand on the other amp check out Garry Pimm's article on a CCS loaded pentode. http://www.pimmlabs.com/web/drivers.htm
It could be cool. dunno. Its still a mu-follower, but different. 

 

[FrankCooter]

Have you thought about using some sort of regulator as your output tube? The 5998 has a mu of about 5.5 and a plate resistance of about 350 ohms. Using parafeed, something like this might drive your transformers off the plates.

 

[00940]

@FrankCooter: nice tube that 5998.

With such an rp, the output impedance of the amp would be similar to a cathode follower. A bit hard to come by around here though.  The EL36 could be a nice alternative. Triode strapped, it has a rp of 800ohms and a µ of 5. Not as nice but much cheaper.
 
@nikongod: just for kicks, here is the schematic I pulled out. It is based on the ecl84. The pentode section is good enough to go without driver. Triode strapped, it has a gain of 31, a rp of 3K and can dissipate 4W. In UL mode, it's difficult to know accurately but, according to the model I've got, the output impedance would be around 10K, as shown. The triode is quite powerfull (1w dissipation, 12ma of Ik) and will do nicely as screen driver. R6/R7 can be adjusted to suit your sonic preferences (the lower R6 is, the more the pentode act as a triode, with rp going down and H2 up). Iirc, the idea comes from a tubecad article I read some months ago (found it: http://www.tubecad.com/2010/02/blog0180.htm ). The top ccs could be improved by cascoding dn2540s.
 

 

[regal]

Quote:

 

I heard that amp at canjam. It was quite nice.
 
I dont think its output impedance that screws up headphones. I think its how the amp distorts. When people think of high output impedance, they think of an OTL amp. measure one... they suck with low impedance headphones! Even people who dont particularly care for measurements will agree that most open loop OTL amps suck on paper with low impedance headphones. I think a poor parallel was drawn between damping factor in headphones due to this. Try grados from (insert any SS amp with more than 20db of global feedback) BUT with a 100 ohm source resistor to reduce the damping factor. not too bad. quite nice in fact.
 

This is a mystery to me as well,   I the 100 ohm resistor is incorporated with the voice coil electrically so in essence you are increasing the damping factor.  Its sort of what came first the chicken or the egg argument.   Of course I have been wrong before.