[Kitarist]

Wow this is getting huge

 

[regal]

Quote:

Originally Posted by Kitarist /img/forum/go_quote.gif Wow this is getting huge

What headphones are you planning on using with this amp ?

 

[Kitarist]

Grado 225 and Sennheiser 650

 

[nikongod]

Quote:

Originally Posted by regal /img/forum/go_quote.gif I'm not writing it off, the topic of the thread is "Best Tube Amp to DIY". This amp has expensive iron a lot of tubes, not the greatest output impedance. And I still question with the single input transformer and the two separate phases connected at the input if it is the Best AMP to DIY. Its surely a fine a fine amp, I just always point out the negatives I see as discussion points, Parafeed takes offense to this but this is how engineering design teams work. The idea is to prove the questioner wrong, defend the design.

Im going to skip output impedance. I think everyone else here agrees its been beaten to death. If you think its that important, outdo it. A little better than 20ohms Zo with all the secondaries paralleled.

Regarding your stated grievances:

Expensive iron: Part of that is that the designer of this amp is the USA importer of lundahl. Id cut him some slack, he offers the schematic freely, let him use it as an add. Some of the transformers can EASILY be exchanged for others (the power transformer & choke could be replaced with a hammond) and the input transformers could be switched for something else if the step up option was not required. Sadly I cant think of much (anything? I welcome alternatives, although its too late for me) that does what the LL1689pp does at a similar price point. The amorphous core is very expensive, but there is the normal core OPT, like what any alternative would have.

some prices at the time of posting:

The input transformers:
The lundahls in the amp are $80 each.
The lundahls closest to the other stuff here are $68
Jensen 10K:10K transformers cost between $70 and $100
Sowter 10K:10Kct are $43british pounds=$65 each
Cinemag 10K:10K are $50 each

Output transformers:
The (non amorphous) lundahl LL1689 are $130 each
Jensen and cinemag make transformers that can take push-pull DC for line out duty but only provide 600 and/or 150ohm outputs. bummer.
Sowter doesn't make a Push-pull headphone transformer. Their SE transformers do have configurable secondaries like the lundahl. They start at 70british pounds=105usd for one which makes no claim to run 32 ohm headphones and jump to 100pounds/150usd for the one that does. Then they go up

Electraprint and magnequest will make you whatever you want, for sure. What is a 10Kct:4*150 going to cost? Whats the secondary DCR? They make nice transformers, but do you want to own the first 10Kct:4*150 transformer they made? output taps are not the same as multiple secondaries on a transformer. The space you waste to get windings for a 600ohm tap is bandwidth you dont get on the 32 ohm one. Parallel those suckers. Use every single winding every time you hook something up.

Yea lundahl transformers are expensive, everyone agrees, but their prices are not out of line when viewed in contrast to other MFR's of similar or even unknown quality.

A lot of tubes:
It has 2 double triodes/ch and 1 rectifier tube. 5 bottles in the whole amp. The rectifier tubes it uses are like $6. You can buy the audio tubes for about $100 or perhaps a bit less if you use russian input tubes. A few beans for shipping and its $110 for all the tubes. Could he have used uncommon tubes that were $5 for input and $7 for output, sure! However 6h30 and 6dj8 (and the russian variant) are still in production and many people already have them in their parts bins & tube collections. When viewed in that light, you may only need to buy the rectifier tube for $6. The fact that people who might build the circuit have the tubes (and have no doubts as to future availability) is key if you are going to draw up a PCB. If you like the design but not the tubes he selected, take the basic idea for free like he gave it and modify it to use whatever.

The input transformers. HUNH?
Quote:

the single input transformer and the two separate phases connected at the input

What do you mean?
The way this is shown in the schematic is exactly how 50% of anyone using an input transformer into a differential amp would set it up. The other 50% would set it up with the CT floating and the balance set by matched resistors to ground. Regardless of which camp people prefer they all agree that there is really nothing unique or unusual here at all. The "resistors to ground camp" will no doubt have something to say about the center tap being used like it is like they always do. The center tap camp will have a thing or 2 of their own to say about the resistor people. I vote for a cage match to the death regarding the superiority of center tapping with windings VS resistors to ground. Its been raging for as long as there has been hi-fi audio, were not going to finish it here, its another thing just not to get into.

As they sit the input transformers accept a single ended or balanced signal without change to the actual active circuitry. They isolate the source from the inputs of the amp which is quite nice. (Is the raven ground loop proof? Indeed!) They output a balanced signal regardless of what is fed in. In the event that you manage to drive the input tubes into grid current (ok thats insanity, but maybe...) the transformer sucks it quickly to ground without the blocking effects of a cap. Considering that not everyone who would build this amp has a balanced source (for better or worse, we are not talking about sources here), and that even if they do have a balanced source other advantages apply its an extremely elegant way to do many things with 1 part. .

The raven is a very solid well thought out design.
Quote:

Originally Posted by regal /img/forum/go_quote.gif One thing I wanted to throw out there for this thread is I see a lot of respected members like Zanth, HeadphoneAddict, etc say that the old Melos SHA-Gold was one of the finest amps, some even say better than a ZD for Grados. The company is defunct, the schematics out there. Do you think this would be a good Best Tube amp to DIY candidate ?

I have a modified Melos SHA-gold/maestro

I like the amp QUITE a lot. In many ways I think it is a runner up for "best" grado amp, in others not as much and its really all subjective.

I really like the overall tone of the Melos but it clearly has tone which to some is an issue. My best attempt at nailing down the tone is that the bass is a little on the loose side, BUT the top end has good detail and separation. A strength of the melos (how do "better" amps fail at this? I consider this a basic test of quality. whats the point in anything else if this dosnt work...) is that the image placement and separation is excellent. With the melos you can "walk around" in a well recorded performance and hear precisely where everything is coming from with all they layers shining through very well.

Compared to the ZD the Melos does give up bass control and detail, but there is something about the melos that draws me in. The impact with HP-1000 on the melos is really well balanced with the rest of the music. The ZD got more detail, and a flat dead "thump". I certainly side with them in saying that the Melos is a better grado amp than the ZD, although I think its all about the faults of the amp working better with the faults of the Grado.

With the HP2 the ZD had a more controlled bottom end but I think almost a "too controlled" bordering on etched/brittle top end. The ZD got more of the minute details out of the HP2 than the Melos but I really just wanted to smash the ZD to little bits after a few minutes. It was a technically excellent and simultaneously totally unsatisfying.

Not all headphones sound good to me on the melos, and Id say that more sound good on the ZD although the thought of something a little thin like an AKG K701 on the ZD flat out scares me. Either way listening to an HP-1000 on a melos is something I think everyone should do. An RS-1 on a melos would be a second pick. If you ever get a Sony R10 or qualia 010 in the same room as a melos try that too, its a VERY good pairing. The sonys sound good on ZD too, Im just in fanboy mode now and very few people think to try the kings of dynamics with that old clunker...

I guess now that my opinion of the amps is out there there are some issues with the Melos. It is a noisy amp. I think some of the noise could be reduced with a better board layout than melos gave us but dsavistik's work on his melos SHA-1 (SHA-1 actually has a better board layout than the gold in many ways, although the HV regulator is better on the gold.) never got his to the point of a totally silent amp. My own (less scientific attempts) on my gold have fluctuated between dead quiet and still some hiss. I dont know where the hiss comes from, and to make matters worse its intermittent and happens with tubes that are quiet in my phono stage! I could pull my hair out or just listen, so I just listen. It is quiet enough that it dosnt bother me but it is something to look out for.

As a VERY far fetched idea of a possible source for an alternative:
In many ways the Cavalli SOHA 2 looks close to a Melos on paper, perhaps with some upgrades hidden in there that take it a step or 2 beyond the Melos as far as design is concerned although I would immagine that they will change the sound even if some other changes are undone.

In particular, the gain stage is similar to the Melos gain stage, LTP with CCS tail. They both idle the tubes at a similarly low plate current (2ma SOHA2 VS 1.6ma Melos) and operate with similarly low voltages across the tubes (melos runs with about 30V between plate and cathode, SOHA2 looks like about 38V from what I could get off the cavalli site). The SOHA uses current mirrors on the plate where the Melos uses plate resistors. The other KEY difference in the gain stage is the way that the tail CCS & bias of the tube is done. The Melos has the CCS to ground, and the grids of the triodes are actually ~20V above ground. The input coupling caps on a Melos are NOT optional

The Soha2 has global feedback, Melos does not. Yea, this is going to be the big one in the difference between the 2. I think thats just the sound Alex likes, but he usually makes it easily removable. I guess there is something to say for him because many people like the way his designs sound. On that note note he dosnt mention the option of removing the feedback loop in the write ups on his site like he does for some of his other amps, I wonder if the amp goes unstable without it or if there was another reason. Perhaps he wanted to keep options to a minimum to reduce build errors? Maybe this has been discussed in a different thread.

The output stages are different, although simultaneously sort of similar. The are both single ended, they both use a small transistor to drive a large output transistor, and both have the output transistor DC coupled to the load. The SOHA is all BJT, where the Melos is all FET. The SOHA has the CCS to set the current of the output follower, where the Melos uses the tube heater.

Maybe they work out similarly, maybe they dont. Maybe disabling the feedback loop throws the amp unstable and blows the **** out of it. I dont know.

 

[sachu]

Quote:

Originally Posted by nikongod /img/forum/go_quote.gif The first stage of the exstata is a differential pair phase splitter. The amp is balanced regardless of what signal you put in. Unless the designer screwed something up (unlikely, the last versions of the schematic I saw looked quite nice) there should be little difference between the 2 inputs. FWIW, from my previous experiments with it, I preferred the balanced inputs on my STAX gear, using a transformer to convert a SE signal to balanced rather than the input differntial pair. I dont know what changed so it may not be worth much. Welcome back

Thanks Ari.

Now, i brought up the exstata cause i believe a few people who listened to the prototype amp at RMAF swore that there was huge difference between differential and SE. But the new amp I honestly couldn't pick out any differences in the type of input being used. Could be my hearing is shot, could be the other folks were imagining things or possible that the Meridian G08 is a POS source.

 

[regal]

Thanks for the info on the Melos, very informative. I have a lot of experience with the SOHAII and its big brother the StackerII. The SOHAII has a lot of potential, you can remove the negative feedback, he just put that in so you could roll a lot of tubes and have similiar gain. It is a dead silent background amp. I found that the tube was biased too low and posted a mod on how to raise the bias to 2.75ma per section. This is still too low to compete with the best as a 6DJ8 needs at least 5 mA's too sound real good.

I guess a am very enthuastic about hybrids and do tend to get overly negative with tubes +iron, I think in 10 years we will be laughing at ourselves for using iron in tube headphone amps, thats just an opinion. The Melos sounds unique in the hybrid realm in that the bass isn't too tight (SS-like) with grados, that's the only issue I have with the hybrids I've built , it must have to due with the simple mofset output? I would like to try a HV-better powersupplied SOHAII with a Melos buffer.

 

[nikongod]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Thanks for the info on the Melos, very informative. I have a lot of experience with the SOHAII and its big brother the StackerII. The SOHAII has a lot of potential, you can remove the negative feedback, he just put that in so you could roll a lot of tubes and have similiar gain. It is a dead silent background amp. I found that the tube was biased too low and posted a mod on how to raise the bias to 2.75ma per section. This is still too low to compete with the best as a 6DJ8 needs at least 5 mA's too sound real good.

Np
Thanks for confirming my guess at the feedback thing.

Quote:

I guess a am very enthuastic about hybrids and do tend to get overly negative with tubes +iron, I think in 10 years we will be laughing at ourselves for using iron in tube headphone amps, thats just an opinion.

I think well be in about a similar place. 30 years ago, they were saying exactly the same thing about power amps.

Tubes work very well with transformers. Its quite possible to make a headphone amp that sounds great and measures well with 1 tube and a transformer. Depending on what you want for speakers maybe them too. but very good sounding simple amps can be made with tubes. Transistors, eeeeh. A few single transistor amps have been tried, but do any really catch on? They are simple voltage followers or very simple gain stages both of which inevitably have high input capacitances/currents for BJT's. Neither has good bandwidth: A single SS power transistor without something to drive it is apt to roll off in the audio band or do uglier things. What the audio industry did in the 70's was say we are trying to outdo a transformer coupled tube amp, and look at that expensive output transformer that we no longer need and how we can buy power transformers that dont have heater windings. That buys a LOT of transistors and STILL comes out to a lower part cost. So rather than look for a more linear device, just wrap 3 or 4 or more of whatever you can get in more feedback. The problem is EVERY device adds its own distortion (and distorts the distortion from the previous stage). you cant get enough negative feedback with version 6? Add another transistor to get more feedback to get the THD you want. its easy! People have been trying it for years, and it simply dosnt work. The single transistor amps dont hold up when you scrutinize them closely, nor do the more complicated amps with lots of feedback around lots of transistors to get a good THD number. There is probably a happy medium with transistors and it may be a hybrid, or an amp built exclusively with transistors designed for audio use

(I think nelson pass is on the right track here but he is stuck using the transistors he can get like everyone else who uses SS), but there is certainly a place for transformer coupled tubes.

Transistor MFR's have never really made any honest effort to improve their devices for audio. and why should they? If they make the next generation of transistors with much more gain, the audio guys can apply more feedback and get better THD! For sure they make enough that work out, but at best there are VERY few audio specific transistors and that borders on none.

OOH! interesting tangent: Some crazy folks have made transformer coupled SS amps, using the SS transistors under high voltages/low currents like you would expect a tube to run. neat stuff.
Quote:

The Melos sounds unique in the hybrid realm in that the bass isn't too tight (SS-like) with grados, that's the only issue I have with the hybrids I've built , it must have to due with the simple mofset output? I would like to try a HV-better powersupplied SOHAII with a Melos buffer.

I have also guessed that the output stage contributes to the way the melos sounds, but never really looked into it much.

I'd say build just the output stage as a buffer and try it out with either source resistors or CCS on the output mosfet.

as another guess, have you tried an output stage with mosfets, or just BJT? I have not followed the hybrid scene for some time but know everyone was leaning towards BJT :facepalm: Even just switching to mosfets in the same basic topology can change the sound a bunch.

 

[dsavitsk]

Quote:

Originally Posted by regal /img/forum/go_quote.gif tube was biased too low and posted a mod on how to raise the bias to 2.75ma per section. This is still too low to compete with the best as a 6DJ8 needs at least 5 mA's too sound real good.

Quote:

Originally Posted by regal /img/forum/go_quote.gif I guess a am very enthuastic about hybrids and do tend to get overly negative with tubes +iron, ... The Melos sounds unique in the hybrid realm in that the bass isn't too tight (SS-like) with grados, that's the only issue I have with the hybrids I've built , it must have to due with the simple mofset output?

a) the Melos runs something like 1mA through the 6dj8

b) the "too tight" bass from many solid state amps is due in large part to the too low Z out and too high damping factor.

Quote:

Originally Posted by nikongod /img/forum/go_quote.gif Tubes work very well with transformers. ... Transistors, eeeeh.

Best Grado amp I've built/heard is all transformer coupled, all solid state. Runs at ~32V

 

[regal]

Quote:

Originally Posted by dsavitsk /img/forum/go_quote.gif a) the Melos runs something like 1mA through the 6dj8 :

I'll quote Jack from ElectraPrint: "Triodes all have one thing in common and thats, the least amount of distortion is when the plate is near full dissipation or highest current/ voltage. its in the books"

Quote:

Originally Posted by dsavitsk /img/forum/go_quote.gif b) the "too tight" bass from many solid state amps is due in large part to the too low Z out and too high damping factor.:

This would explain why the Melos performs well with Grados as it has a relatively low damping factor for a hybrid. I just have a hard time grasping that concept as when you design a sub amp you want a low damping factor because of the stiff drivers (physical damping), a Grado driver is about as opposite of a subwoofer as I can think of.

 

[adamus]

Quote:

Originally Posted by dsavitsk /img/forum/go_quote.gif Best Grado amp I've built/heard is all transformer coupled, all solid state. Runs at ~32V

interesting. best amp i have built for grados is the starving student

no joke, lovely synergy.

bijou + grados .... hmmm not great
soha II / soha 1: ok
darkvoice 336 - horrible
ZD: not bad

I must book a slot in the crazy clinic...

 

[regal]

Quote:

Originally Posted by adamus /img/forum/go_quote.gif interesting. best amp i have built for grados is the starving student no joke, lovely synergy. bijou + grados .... hmmm not great soha II / soha 1: ok darkvoice 336 - horrible ZD: not bad I must book a slot in the crazy clinic...

Not crazy that amp has as much in common with the Melos that Mr Grado tunes his designs with than most any other, thanks for the input.

Agree 100% on the P-P amps with Grados, have yet to hear one that sounds good with them.

 

[Parafeed]

Quote:

Originally Posted by regal /img/forum/go_quote.gif I'll quote Jack from ElectraPrint: "Triodes all have one thing in common and thats, the least amount of distortion is when the plate is near full dissipation or highest current/ voltage. its in the books"

That'll be another way of saying - stay away from the knee of the curves. Good advice, except a lot of vintage stuff doesn't take too kindly to be run at max dissipation. I've learnt that the hard way, with burnt out screen grids on Mullard EL84's and great sounding vintage 6SN7's that have gone noisy when run hard. For drivers, use a CCS or choke, make sure they are not driving low impedances and run 'em sensibly. Low distortion and long life.

 

[adamus]

or how about solid state choke (gyrator?), people seem to like them in the 'anti-triode' thread over on diyaudio. they sim like a choke.

I am playing with a dac otuput stage at the moment. It has responded very well (and not a subtle change) to led bias. I am going to knock up some dn2540n5 cascode regs for a CSS a some point soon, it will be interesting to hear the change as i am now very familiar with the current sound.

 

[Parafeed]

Quote:

Originally Posted by adamus /img/forum/go_quote.gif or how about solid state choke (gyrator?), people seem to like them in the 'anti-triode' thread over on diyaudio. they sim like a choke.

Anti-triode? Got a link to the thread over on diyAudio?

My only experience with an electronic choke (gyrator) and capacitance multipliers comes with having used the DIYHiFiSupply Universal Power Supply modules. The thing is, at least with that implementation, although the ripple is very low and by rights it should sound excellent due to the lack of electrolytics, I found that it "strangled" the sound - compressed dynamics.

 

[adamus]

starts here...

Anti-Triode SEPP, how to do best? - diyAudio

Read from page 7 onwards if you want to cut to the chase.