Humble attempt to improve on a Trafomatic inspired amp

[regal]

Not a clone at all but this is an experiement to see if the Salas Shunt regulator can drive a single feed transformer loaded spud amp similiar to the trafomatic.   I heard the Trafomatic and was impresed by it,  but instead of buying a nice well done complete amp I have to experiement with the unknown

,  plus I miss the Headwize forums and hope that more folks try and report new designs on this forum.
 
This amp is basically Dsavitsk's super single feed but with an important spin.
Its a gamble,  the Salas Shunt is reported to have exceptional tone some believe due to the incredible low output impedance (of the regulator.)   This is also fairly unique in that the only capacitor that is barely in the signal path is the 1uf filter on the regulator.  My experiementation with parafeed and hybrid amps have shown to me that the coupling cap sets the tone of the amp so I wanted to try something different.  I chose 5842 tube which is the precursor to the 6c45pi (used in the trafomatic) mainly because it has been reported to sound a little better in side-by side comparisons.  And the 6c45pi is notorius for varying mu (gain) from tube to tube 
 
It took months to get all the parts for the chassis together as it is not usual to have a 30W heat sink on a tube amp chassis.  It is being built with the tube rectified pre-filter in a separate chassis along with the DC heater supply.  I still need to find my 4-40 tap and step bit,  so difficult to get everything together for a build. 
 
Notice that in the design I damped the heck out of the tube with high grid stoppers,  even a cathode stopper.  Also using a 10k pot so that I can use a lowish value grid leak.
 
 
The gamble is the reactive load from the transformer may not be able to be regulated by the shunt., the pay-off should be SET sound without the congestion in complex passages.
 
 

 
 
 

 

[adamus]

NIce!
 
Having now had experience with Jacks transformers i can wholeheartedly say they are very very good (although i went parafeed).
 
I tried the 5842 and the D3a.... if you could try the triode strapped d3a i think you will smile. Word of warning though, buy a few. 2 of the 4 i have are too microphonic for my tastes. They all match very well though. @ 17.6ma CCS anode load, all hit 165v +/- 1V.
 

 

[regal]

Thanks,  Yes if this works,  I'm going to make a PCB so I can easily switch from 5843 to 6c45pi, to D3a, to 7788,  6688, 6n6p...

 

[adamus]

contact Dsavitsk, i remember he did a group buy for 'd3a experimenter boards'... he may have some left.
 
a paralelled 6n6p may work well, its a shame its such a heater current hog though. it will also struggle for gain i guess.

 

[regal]

Quote:

 
a paralelled 6n6p may work well, its a shame its such a heater current hog though. it will also struggle for gain i guess.

 
I measured the AC voltage at twice my normal listening volume with my Grados,  it was only 1V,  with 6n6p's this amp would do 1.6x2v input so I think there would be plenty gain.  My experiments with the 6n6p in line level stages as well as the Bijou were positive,  its a nice tube.
 
I forgot to mention that the heater supply board that comes with the Salas HV PCB is configurable for 6.3 to 12.6V,  so I have it set to run the two 6.3V tubes in series with a healthy 12.6V hammond transformer.

 

 

[adamus]

with grados you are right, it will be fine, with less efficient phones you may start to struggle.
 
slightly off topic, but may be of interest
 
As i built the d3a for a friend that has now paid for and collected it, i am now starting to cobble together a super cheap version for myself. gone are the dn2540 cascodes.... incomes a simple mje350 + bcc558 cascode (super cheap and good performance). Bias will probably be cheap red LED.  Edcors have been bought from mr Dsavitsk as he had them done for a project and the pinouts were returned wrong, $20 for the pair!  bagged myself a couple of 6c45p for practically nothing.  pulled a transformer with 220v and 6.3v secondaries out of an old radio. i also have a couple of 4.7uf caps knocking around (i think i bought them for £4 so super cheap anyway).  £6 for a step pot from ebay. PS will be a super simple CRC solid state rect.
 
depending entirly on the chassis, this could be a sub £50 build that... depending on the edcors.... could compete with some of the best. I admit its very likely that  it wont touch the special build with electraprints but if it gets anywhere near it will still sound special. If it does i will probably document it as a project. the P2p is so simple, this could easily be a beginners foray into tubes (taking all the necessary HV precautions!). Even if one didnt have parts 'in the box' i think this could be done on a really tight budget.

 

[FrankCooter]

Interesting project! I too am curious about how an ultra low impedance shunt regulator would work in this situation. Do you have a source for boards for the Salas regulator?
 
I've built a number of similar "spud" amps with Electra-Print transformers over the years (can't say enough good things about Jack Elliano of Electra-Print!). My best used a triode connected Western Electric 418A and Doug Savitsky's "Electric Avenue" topology. The power supply was a massively overspeced tube rectified LCLC filtered type with motor-run oil capacitors. Very good, but nowhere near the Salas reg in output impedance. High voltage shunt regulators are definitely an area of current interest.
 
I agree with your approach to parafeed and don't think it has any real advantages in this situation.
 
I'm definitely NOT a fan of the Russian 6C45PI.  The 5842 is better to my ears. If you go ahead and build this amp and like it, bite the bullet and spring for a pair of Western Electric 417A, the difference over the Raytheon is easily audible. The 7788 is excellent, but I also think the D3A may be the top tube of this type. The Western Electric 418A is a "sleeper". It's Basically a four element version of the "unobtanium" W.E. 437A. Pops up frequently on Ebay for about $20.00 N.O.S. It's a beefy little thing with a 10w plate and a mu of about 30. Downside is it uses a magnoval socket, easily available, but requires a larger hole than the standard noval socket.
 
I totally agree with you about the lack new designs here and appreciate your sharing your ideas. Keep up the good work!

 

[adamus]

Quote:

 
I agree with your approach to parafeed and don't think it has any real advantages in this situation.
 

 
Frank , one could argue parafeed transformers are less lossy, no air gap. you could also argue it takes the final electrlytic out of the signal path (can-o-worms and a defunct argument if using the sala reg). The counter arguemnt would be the cap can impart sonic signiature, but i find most reasonable film caps to be transparent enough. Essentially swings and roundabouts.
 
Agreed though, if the Salas shunt works here (and it is an if, as people have struggled to get it working into an inductive load) then this could be another special amp.
 
Re the 6c45p.... i had enough trouble keeping the D3a stable, a hf oscillation caused a slight harshness to the sound. increased grid stoppers - no joy, added platee stoppers - joy on left channel, added cathode stopper (5ohm) - JOY. the 6c45p is meant to be even tougher to keep from singing to itself and you see recommendations of smd resistors right on the pins.
 

 

[regal]

Frank,
 
Glad to hear your input here,  your DHT amp design has been an inspiration,  working my way up to that one (just got a set of Colemans DHT filament supply kits.) 

I bought the boards from here: http://quanghao.com/category.php?id_category=8 
 
They are really overkill,  high quality, thick PCBs, also include a nice DC heater circuit.  They include a SS rectified CRC prefilter which I sawed off,  I'm prefiltering with 5AR4-50uf-5H-50uf-R in a separate chassis with a 5uf right at the board,  you have to put the regulator and associated big heatsink in the same chassis as the tube which makes the chassis build pretty tough.  Probably could have saved a little building the reg P to P.   First dual chassis  build for me with a tube amp so I'm hoping a get the grounding right.  I'm using a separate Hammond 369JX for each channel,  anyone doing this project should use EZ81's so they don't have to buy a separate 5V transformer (just poor planning on my part.)
 
If you know where I could find real WE417's let me know,  I've heard a lot of bad reports on people receiving bad tubes when buying these.
 
 
Adamus,  I need to take a closer look at the D3A.  Just found the datasheet that shows the triod curves.
 
 
Quote:

Interesting project! I too am curious about how an ultra low impedance shunt regulator would work in this situation. Do you have a source for boards for the Salas regulator?
 
 

 

[digger945]

Quote:

NIce!
 
Having now had experience with Jacks transformers i can wholeheartedly say they are very very good (although i went parafeed).
 
I tried the 5842 and the D3a.... if you could try the triode strapped d3a i think you will smile. Word of warning though, buy a few. 2 of the 4 i have are too microphonic for my tastes. They all match very well though. @ 17.6ma CCS anode load, all hit 165v +/- 1V.
 

On the microphonic issue, Siemens?
Did you try more than 1 parafeed cap?
Just curious. I should look back to your posting of the pics and description of the amp in the build thread.
 
On the D3a, Ferrari had many good things to say about it in reference to his K&K headphone amp. Lots of $ for good transformers though.

 

[adamus]

siemens indeed. the two 'good tubes' arent microphonic, i can tap the amp and it wont ring. The other two were slightly microphonic, nothing bad and still usable, but when you have the pick of 4 you may aswell pick the best two.
 
I tried 3.3uf, 4.7uf and 6.8uf
 
3.3uf is was fine for my tastes, the high values introduced a bit of bass bloom.
 
the d3A is excellent, other than stability people dont have bad words to say about it. I have bought a big stash

 

[dsavitsk]

 
Quote:

 
I agree with your approach to parafeed and don't think it has any real advantages in this situation.
 
I'm definitely NOT a fan of the Russian 6C45PI. 

 
Quote:

one could argue parafeed transformers are less lossy  
the 6c45p is meant to be even tougher to keep from singing to itself and you see recommendations of smd resistors right on the pins.
 

I definitely am of the opinion that parafeed is the way to go here.  If nothing else, the ability to use permalloy transformer cores makes a huge difference to me.  Plus, the smaller cores and lower number of turns, and the CCSes PSRR are all advantages.  And, a high quality cap will be virtually transparent.  I've never gotten a single feed amp to sound as good (with the exception of a single feed pentode amp, but that's another story).  As I've said elsewhere, the advantage of single feed is that the transformers can swing higher than B+ which makes them more efficient that a CCS loaded parafeed.  But, when driving headphones that need a volt or so, I think this is unimportant and is swamped by the drawbacks of a lower quality transformer. But, to each his own.
 
As for the 6C45, if you think they sound bad, it is almost assuredly because they were unstable. You really do need SMD stoppers on both grids, the plate, and all 4 cathodes.  Done that way, I've not found a lower distortion triode that can drive headphones directly.

 
Quote:

contact Dsavitsk, i remember he did a group buy for 'd3a experimenter boards'... he may have some left.

 
That was an accidental group buy.  I was having some made, asked around if anyone might be interested in the extras, and it took off a bit.  That said, there aren't any more.  The project can be seen here: http://www.ecp.cc/d3a_exp.html
 
Anyway, I too am interested in whether the sand shunt reg works well here.  The 0D3 was a bit of a letdown for me.

 

[FrankCooter]

I have nothing against parafeed in general, I've built several amps with either parafeed drivers or outputs and appreciate it's utility. I don't think that in all cases, all the time, it betters well executed series feed. The quality, and the position (sounds better on the ground side of the transformer) of the coupling capacitor is critical to the overall performance of the amp. If you're lucky enough to have transformers with exotic core materials and no air gaps then parafeed is the only way to go. On the other hand, if you're using cheap transformers with marginal inductance, it's easier to get better performance using parafeed than traditional series feed.With a well executed air gapped transformer with plenty of inductance,  like the one in discussion here, then I think the issue becomes more a matter of taste in trade offs.
 
The really interesting thing in Dave's proposed design is that it essentially eliminates the capacitor trade-offs inherent in the two topologies. All capacitors  are removed from the audio path. Best of both worlds if it works. Perhaps I'm missing something here, but I really can't see any advantage in adding a parafeed capacitor in this circuit.
 
Doug, I'm glad you can get the 6C45pi to work. It's definitely an interesting and attractive tube for a "spud" application. Somewhere in a drawer I have 6 or 8 of them. No two biased or amplified the same. Admittedly, I was trying to use them in a phono stage, which is a much more critical application than a headphone amp. On the other hand, I had no trouble at all with the 7788, the D3a, or the 417a. All were stoppered in every possible way. I've kind of gotten away from these high gm types in general and now prefer a two stage all DHT amp for dynamic headphones. After a year of building phono stages with this type of tube, I returned to the 6SL7/6N7/6AH4 line up. Maybe it's just because I grew up listening to a different sonic signature than you guys (I'm 57).
 
The other day, I was listening to some of the recent Beatles remasters. I think they played on 7189 based Vox amps. It occurred to me that a cool and perhaps quite realistic way to listen to this music would be through a "spud " based on a vintage Mullard 7189. With a mu of 20 or so, you could certinly power a set of Grados this way.
 
 
 

 

[regal]

For this project unfortunately the enclosure I had planned for the power supply is just too small,  so I'm delayed another week. 
 
Yes I like parafeed too,  but IMO you might as well replace the iron after the parafeed cap with a silicone buffer which arguably does the job of the transformer with even fewer drawbacks.   There are SS buffers out there that I have found are more transparent than the coupling cap, not sure the same can be said about parafeed iron?   Hence my wanting to try this direction for a change-up.  Frank hit the nail on the head,  the Salas Shunt has the potential to level the playing field a bit between the two approaches.
 
One day I do want to try that Pentode current source amp  Dsavisk mentioned,    maybe with a little 3S4 .
 

 

[dsavitsk]

 
Quote:

Originally Posted by FrankCooter /img/forum/go_quote.gif
 
On the other hand, I had no trouble at all with the 7788
 
 
 

Interestingly, my experience with 7788's mirrors yours with 6c45's. I've gotten them to work for short periods, but it never lasts, and my samples are all over the place.  I finally gave up on it.  Must just be the luck of the draw.