[SonicTrance]

Superglue is my most important material after solder and wire. Look at Sonics great quality pictures, he has adopted the superglue approach as well. What superglue cannot fix cannot be fixed.

Haha! Yes! Though I didn't use any superglue in the MK6, only cable ties. I would not recommend superglueing something to the PCB. That stuff is so strong and could possibly rip some copper trace if you have to remove the glued part.

 

[baronbeehive]

Transparency and realism are not mutually exclusive in any way. In fact if the recording sounds 'real', then a transparent amp will convey a most convincing amount of realism.

Just to bring an example, back when I was still in possession of my flagship amp, I had a cello player friend of mine listen it. He brought a recording of his cello master. The recording was very well made, and very lightly produced.

When we played it, he was stunned at how much it sounded like he was playing in the same room with us. He even recognized the cello his master uses, by sound. He said he hasn't been able to hear that specific signature on any other system.

Now that I call transparency. The realism was there in the recording, and the system allowed it to be played back, to be reproduced without destroying it somehow.

The easiest way to destroy a delicate signal with complex spatial and harmonic signatures (i.e. realism) is to splash some extra harmonics (no matter how benign or natural) on top of it. But more on that later.

Yes, one of the ways I test out an amp is to see what a violin sounds like, and if you can almost feel the sound of the bow going across the strings you know its not bad! As a violinist used to practice at our house I'm familiar with the sound! My other amp has all this information in the leading edge of the music and lends a lot to the transparency. The LD is good but mellower without that incisive edge, some of that if not all I put down to it being a speaker amp and the sound is much more visceral. But with more information in that leading edge it all goes to build up the texture more realistically.

This message could've been written by past me, I've been in this place. I listen to a lot of old stuff that is recorded quite poorly. One of the best examples is the catalogue of Robert Johnson. While I thoroughly enjoy his music, the sound quality is so rough it can be quite taxing to listen to.

Fortunately I found a way around this.

What the last sentence of point 3 tells me is that your system produces some amount of harmonic content, more than it should according to your ears. Let's go back to this:

The easiest way to destroy a delicate signal with complex spatial and harmonic signatures (i.e. realism) is to splash some extra harmonics (no matter how benign or natural) on top of it.

Distortion measurements are usually done using as pure a sine wave as possible, then looking at the output and it's harmonic content. Let's say the output at 1kHz has some 2H and less 3H, no higher harmonics.

Ok then, what happens when you input into this hypothetical amp a signal with equal amounts of 1kHz and 3kHz. What comes out?

Both of these signals produce their 2H and 3H. Then these 4 newly created harmonics also SUM TOGETHER, creating 4H, 5H, 6H and 7H at lower levels. So now we have quite a complex mat of harmonics from these two (1kHz and 3kHz) inputs.

Now imagine you input a complex signal with multiple instruments into this amp.

You are going to have a mess. Granted these sum harmonics are at a low level, but the higher you go, the nastier they are even at low levels.

Incidentally this sum effect is how and why great amounts of gNFB produces a complex mat of very low level harmonics. It's an "infinite sum loop" to simplify a bit.

Wave physics 101. I remember this part from my days in EE university (I changed majors since). Think about a still surface of water. You drop a perfect round object into water; it creates a ripple. The ripple waves are bigger the closer they are to the point of impact (W1 and W2 are biggest, rest are quite small in comparison).

Now if you drop two round objects of about same size simultaneously, they will exhibit the same ripple behavior close to the impact site, but where the waves meet each other, they produce a lot of very small erratic ripples (higher harmonics of the wave), because the waves sum.

This metaphor is not 100% applicable to electric waves, but in essence the same thing happens with them.

For this reason alone (to avoid complex IMD products) I personally would like to minimize the amount of harmonics my amp adds to the signal.

My SE amps have a substantial amount (depends on what's the comparison I guess?) of 2H and some 3H. While they sound very nice, there is a 'something' on top of the sound at all times. The dynamics ("the kick") largely compensates for this when using speakers, and I wouldn't make a SE headphone amp anymore.

This is also the 'dirt' / 'veil' that Sonic described in comparing amps in another thread.

Distortion (however 'benign') always masks detail. Detail conveys realism; the subtle cues that come from the room, or from the instrument. One consistent comment I get from people who listen to open loop low THD amp for the first time is that they become suddenly hyperaware of the fact that the instrument they are listening to has a physical form. Like you cannot stop imagining the acoustic guitar you are listening, because you hear these subtle audio signatures that come from the body of the guitar.

This effect is most pronounced on human voices for understandable reasons.

These audio cues are transmitted in the higher register. You can rough test this by cutting frequencies above 12kHz and see what happens to spatial information. Now if your amp has even a tiny amount of higher harmonics, you cannot hear these, they are drowned in these harmonics.

Yes to all that! My grandmother was a piano teacher and I'm familiar with the sound of a piano being moved around and all the complex harmonics being produced whenever it was moved on the floor! Also what you said reminds me of the use of the sustain pedal which is OK to use for single passages of music to produce more harmonics but must be lifted after that passage of music otherwise the whole thing rapidly becomes a mess.

Its a shame there is so much bad recording around, even now. I have some old jazz recordings, for example "Milt Jackson" the vibes player from the 1950's and the sound quality is superb although the technique of instrument placing is a little dated.

Now to bring it back to your quote;

A rough recording with quite high distortion is going to sound bad when EVEN MORE distortion is added, because it will further mask the "realism information" and it will produce nasty IMD products. Get a cleaner OPEN LOOP amp and these problems are greatly alleviated. At least this has been my finding.


My experiences with shaping distortion back in the instrument amp world are about 60 to 70% on CLEAN guitar. Plug a guitar to a HIFI amp and play clean. It's horrible. The 'clean' guitar sound, whenever nice sounding, is in fact quite distorted (in HIFI terms). There is no reason to add any distortion during reproduction, if a nice guitar amp was used; a transparent amp will make it sound every bit as nice as it was in the studio. Unless the engineer was an idiot.

Yes, right. The main way I test out the sound of the amp, for example when changing tubes is if I get the incredible sound of the guitar right. The Fender sound twangs, and the Gibson sound sustains, and both these sounds are what gives me most pleasure in music.

 

[baronbeehive]

Hello

Thx. The problem with glue is unmount sruff.... Well fist i will try patafix ...

The other point : don't what bias driver mod to use for 6SL7WGT the one with 33kr/400r or 77k/1k and which method (// soldering or dirty replace with cutting the legs) ....

Yes glue pads are quite useful, to fix together adjacent caps for example.

I don't know what the others think but really I would always desolder, and solder properly, unless you are attaching a parallel resistor. Some large caps, or ones that need longer legs might be done with wire such as I mentioned but you still want to solder properly, its good practice and I don't like the sound of too many cut legs and reattached parts etc.

Desolder, get rid of as much old solder as possible to avoid cold solder joints. I always used extra flux for resoldering but if you use Cardas solder it has plenty of flux and is easy to work with.

Desoldering is what I hated most lol !!

 

[Maxx134]

Hello

Thx. The problem with glue is unmount sruff.... Well fist i will try patafix ...

The other point : don't what bias driver mod to use for 6SL7WGT the one with 33kr/400r or 77k/1k and which method (// soldering or dirty replace with cutting the legs) ....

Easier to add resistors to test before buying exact value and remove old resistors..

If you like closer imaging I believe SonicTrance lower settings will do.

My higher settings seemed more balanced to me.

Haha! Yes! Though I didn't use any superglue in the MK6, only cable ties. I would not recommend superglueing something to the PCB. That stuff is so strong and could possibly rip some copper trace if you have to remove the glued part.

That's why I use hot glue as it goes on fast, and you can pull off i need.

The LD is good but mellower without that incisive edge

I am almost certain that is due to your extremely low stock driver tube bias settings...
You really need to try what me and Sonic did to increase bias.
Just add some parallel resistors to test before settling on a value.
It is a noticable improvement!

 

[baronbeehive]

..........

I am almost certain that is due to your extremely low stock driver tube bias settings...
You really need to try what me and Sonic did to increase bias.
Just add some parallel resistors to test before settling on a value.
It is a noticable improvement!

OK, that's quite possible. Also I'm using RCA 6AS7G's as well, I haven't tried the 421A's yet which will add more dynamics!

I'm not saying the Miniwatt is better, just better at some things, and the LD is better at others, but as you know I've been shocked at the improvement in the little amp after the mods!

 

[baronbeehive]

My other amp, the Miniwatt N3:

Jupiter coupling caps, Kaisei cathode caps and now Audyn True Copper cathode bypass caps. Chassis extended to fit.

 

[MrCurwen]

baronbeehive, that's a pretty little amplifier! What are the tubes you have there? Do you have a schematic?

Haha! Yes! Though I didn't use any superglue in the MK6, only cable ties. I would not recommend superglueing something to the PCB. That stuff is so strong and could possibly rip some copper trace if you have to remove the glued part.

Yeah, no, that's not good. No glue on PCBs.

The safest thing to do would probably be to extend the chassis downwards, put the caps on P2P soldering posts (like the one somebody linked to Sonic in his modern balanced amp build thread), and just connect the caps to the LD chassis with wires. That way you can also easily change the caps for experiments.

 

[coinmaster]

Also I'm using RCA 6AS7G's

Eew, no wonder. At least mixem with the stock 6080s, 6AS7s alone are meh.

 

[baronbeehive]

baronbeehive, that's a pretty little amplifier! What are the tubes you have there? Do you have a schematic?



....

Its similar to Maxx's headphone amp, the APPJ PA1502A. Mine is the APPJ PA0901A, or Miniwatt.
I've just got adapters for Russian 6F6C Kobra's on the recommendation of APPJ thread members. The driver tube is a Tungsol 12SL7. The original tubes I used are Amperex EL84 power tubes and Tungsol 5771 driver tubes.
I don't have a schematic. I have continuity tested every component to check on the traces, so I know what's connected to what, but don't know how to do the schematic. This amp was so much easier than the LD to test parts because the left and right channels are laid out virtually identically.

Eew, no wonder. At least mixem with the stock 6080s, 6AS7s alone are meh.

Alright, alright, I will give that a try!

Edit:
Sonic, can you give me the link to your balanced amp build thread?

 

[coinmaster]

Alright, alright, I will give that a try!

Just remember to try both positions.

Sonic, can you give me the link to your balanced amp build thread?

https://www.head-fi.org/f/threads/modern-balanced-tube-amp-build.852879/

 

[coinmaster]

 

[coinmaster]

Finally got my "ultimate" amp up and running.
Well technically it's only my "ultimate" input stage design, the output stage is a standard sandman but you can't really go wrong with a sandman in a pinch. My real "ultimate" output stage is next on my list.

My efficiency went up like 5000% ever since I bought these solderless breadboards a week or so ago, I shoulda been using these for prototyping from the start I would have never been able to build this on anything else.

It was one seriously noisy beast when I first got it up and running, almost 10v of noise from 0-10Mhz all over. I managed to quell most of it but the solderless breadboards aren't the best thing to use for this.
Part of the reason for all the noise is the distance between the components and the power supply. It can't regulate all 3 boards effectively and my low bandwitdth opamp in the reg isn't helping things.

Even with semi-unregulated to unregulated supplies I still get this beauty

Blue is the input signal to the input stage, yellow is the output signal of the input stage through the other end of my duelund cap shrunk down. Not too bad for a poorly regulated input stage. This is my first time seeing such a good copy of a complex signal since I got my oscilloscope and it can be further improved with better supply regulation and a few other tweaks.
Actually the input stage is only using a C filter of a few thousand uf on the B+, no resistors or regulation which isn't good because the design is susceptible to poor power supply regulation and output impedance (so much for ultimate?)

Although to be fair I've yet to measure an amp with a regulated supply which may be why I often see way more imperfections in the amplified signal (you don't even want to see what the mk6 looks like in SE mode) but considering how this isn't regulated it's pretty darn good.

The sound is clear as a bell, micro detail galore. Nowhere near the euphony of the MK6 though which is unsurprising considering it should be very low distortion, although the clarity, dynamics, and effortless sound separation is in itself mesmerizing.
Either way I'm satisfied that the results meet my expectations and my design goals.
In theory as far as tube input stages go this really is the ultimate theoretical design if the goal is linear amplification. The only way to improve it is to use better matching tubes but even then I built circuits to compensate for unmatched tubes as best as can be done without some sort of digital software curve analysis and algorithmic matching or something crazy like that (which I would love to do).

All I have to use for a DAC right now is my ipod, interestingly it still sounds amazingly good, unlike my MK6 which is in euphony only mode right now since its power supply is a piece of garbage without balanced operation

I'm going for a DHT and a nutube version next before I build my "ultimate" output stage. I built filament regulators that should in theory far outperform the coleman regs so even the filaments will be ultimate.

Then I can put my focus into mastering distortion generation rather then distortion elimination. Which reminds me I should probably connect this new stage to my mk6 output stage.

 

[coinmaster]

So I connected my new input stage to the mk6 and it was the most celestial thing I've ever heard, even better then my modded MK6 in its prime, even though I'm using an ipod "dac" in SE mode.
It was like that for about 5 seconds anyway, until I accidentally ripped the ground cable out of my input tubes which caused some sort of voltage anomaly in the mk6 which was apparent due to the relay kicking in for a few moments. Afterword the sound was garbled and terrible in both channels. I've deduced that my input stage is fine, it's just the mk6 that is busted.

I'm a bit stumped on what I did exactly though, the DC conditions are all normal which makes it unusually difficult to figure out that I did.
Damn it, I was so close to audio nirvana too! God keeps screwing with me.

In any case there is something to take away from the fact that my SE mk6 sounds that good with my new input stage and terrible without balanced operation with the stock input stage.

It seems that the input stage really is the downfall of the mk6. Balanced operation is the only way to regulate the input stage beyond some minimal amount. It seems regulation via matched tubes and balanced operation and/or using an actual regulator is the only way to make it work decently with the stock circuit.

The output stage clearly isn't as bothered by the lack of regulation due to how clear and amazing it sounded by just replacing the input stage.

 

[Maxx134]

My other amp, the Miniwatt N3:



Jupiter coupling caps, Kaisei cathode caps and now Audyn True Copper cathode bypass caps. Chassis extended to fit.

Hey I see you finally have the Russian tubes..
Nice Right?

I've just got adapters for Russian 6F6C Kobra's on the recommendation of APPJ thread members

So what you think?

I built filament regulators that should in theory far outperform the coleman regs so even the filaments will be ultimate

I read also high frequency to the heaters is a good idea..

Afterword the sound was garbled and terrible in both channels.

Sounds like the driver PSU...

 

[coinmaster]

I read also high frequency to the heaters is a good idea..

High frequency solves noise issue but wont sound as good as a current source filament because it doesn't isolate the heater PS from the signal.

Sounds like the driver PSU...

Nah the driver PSU is fine, it seems to be a pretty robust circuit overall.
The issue was some of the mosfets in my new output stage mod were partially but not completely fried. I guess I should start adding zeners or TVS diodes to my mosfets, I usually neglect that.

Now I'm going to wait until next month to do some real listening so my ear gets a chance to heal. I also need to repair my dac and my headphones, my headphones are all kinds of messed up at this point, too many times have I shot DC through them.

I need to use a DC sense relay circuit between them and my prototypes from now on after I put new drivers in them.