[StanD]

   
 
This is getting old, but it seems there is a misunderstanding and assumptions about what I've said that are not true.
 
Heck yes, the tubes present a h*ll of lot more noise than the rest of the circuit - especially in the Vali. I was talking specifically about the hybrid buffer, not the tubes.  Even a CMoy sometimes uses resistors on the output to control hiss.
That statement is absolutely false.  Every amp that does not have transformers on the output will produce different ratios of voltage and current, depending on the impedance of the load.  No offense, but I've probably built and sold more tube hybrids than many people.  In the real world, load impedance will change both output and noise, even with transformer output (admittedly the change is a lot less, but transformers are not perfect devices, either).

Regarding the CMoy, that 38 Ohm resistor is probably more for preventing the load from causing instability by disrupting the negative feedback loop. I really don't see how that is going to be a factor in the perceptable noise level. By the way that 38 Ohm resistor is not a good thing.

 
 
Sorry but the as I said the output signal voltage is not going to change by much when you vary the load impedance. The reason being is that the the amp is a voltage gain device which will produce a predictible output despite the load. Whatever the gain, the signal level will remain the same with some differences in load impedance. In truth there is a small loss of signal due to the attenuation formed by the output impedance of the amp and the load impedance of the headphones, however, if you do the math it is very small and affects the signal just as much as the noise hence the SNR is the same. Changing the ratio of current and voltage is a consequence of Ohms law, however, by changing the load impedance the current is what will change as the voltage remains pretty much the same. As I've said previously, it is the sensitivity of the headphones that come into play. If you take a careful look it is the sensitivity in dB/V one should consider, if you would like further clarification on this last point I'll be glad to explain why this is more important than dB/mW. Thus to the original point it is the sensitivity not the impedance of the headphones that determines what ambient noise levels we can hear in the Vali or for the matter most amps/chains that have a bit of noise.

 

[tomb]

No one ever said sensitivity didn't matter or that it wasn't the most important factor.  I simply stated that load impedance plays a part and that it, too, has an effect on noise.
 
I don't know who Mark Houston is, but Tangent is probably the best source on the CMoy besides Chu Moy himself.  R5, as its known in his schematic, is not recommended but sometimes necessary to control noise.  He recommends anything from 10 to 100R, but points out that it's a last resort:
http://tangentsoft.net/audio/cmoy/tweaks.html#r5
 
Does the resistor reduce gain (and thus, noise) by introducing a voltage drop ahead of the headphone or does it increase the overall impedance seen by the amp, enhancing the stability of the circuit and reducing distortion?  It depends.  I've seen both at play.  It's why I state that impedance plays a part.  Again, I never said it was the most important part.
 
You chose to not include your previous statement, but I'll quote it again here:

"If you put a meter on the output and change the load impedance neither the voltage of the signal nor that of the noise will change."

 
That is still incorrect.  The voltage will definitely change when the load impedance changes and so will the current.  You admit that it's simply Ohm's Law, but you somehow maintain that doesn't really happen, and I quote again:

... however, by changing the load impedance the current is what will change as the voltage remains pretty much the same. ... Whatever the gain, the signal level will remain the same with some differences in load impedance.

That's wrong, too.
 
The actual voltage gain will change with the load.  It will be higher for a high-impedance load than a low-impedance load.  This is especially true in a tube hybrid.  You attempt to recognize that we're not dealing with a perfect voltage drive amplifier, but then you minimize it:
 

In truth there is a small loss of signal due to the attenuation formed by the output impedance of the amp and the load impedance of the headphones, however, if you do the math it is very small and affects the signal just as much as the noise hence the SNR is the same.

 
The truth is, that a tube hybrid (and almost every two-stage headphone amplifier) acts with both voltage drive and current drive amplification.  The tube signal stage attempts to act as a perfect voltage drive.  However, the output buffer is set up with zero gain and attempts to act as a current drive.  The result is that you get some mixing from both and these effects are measurable and audible.
 
Here are some measurements from a typical and well-known tube hybrid:
Voltage gain
330R load: 12.4 (21.9dB), 33R load: 7.2 (17.2dB) (same tube for both)
Noise
330R load: -86.1dB, 33R load: -90.6dB
%THD
330R load: 0.287, 33R load: 0.457
 
All other things being equal, the simple change in static impedance resulted in gain changes, noise changes, and %THD.  Notably, as impedance went up, noise went up but distortion went down.  As impedance went down, the reverse was measured.  One might argue whether a 4.5dB difference is audible, but certainly a 2X distortion is.  I have measured similar numbers even with output transformers on a hybrid.  Transformers are not perfect devices, either.
 
In the real world, changing load impedance can have a significant effect and can be audible.  Whether one is noise, distortion or hiss can be a terminology debate.  Somehow, we got off on the subject about whether load impedance had any effect at all.  It most certainly does.
 
Anyway, I appreciate the attention but will bow out for now.  This has gone way off topic and I'm afraid we're confusing a lot of folks.  Just realize that the headphone impedance will cause the amplifier to perform differently regardless of sensitivity.  Not saying that's as significant as sensitivity, but the effects are there.

 

[StanD]

@tomb Using a value of output series resistor that compromises impedance is indicative of a poor design. As I said previously, this is done to promote stability at a performance cost. As this resistor gets close to or exceeds that value of the headphones impedance you will attenuate the output of the amp whihc seems to be a bad move. Using headphones that have a impedance that varies with frequency (as some popular Sennheisers and others) will result in a messed up frequency response. Not to mention that in many cases the damping factor (mostly for bass) will be reduced, not good. I wouldn't buy such an amp, one is better off with a Vali, as long as its 6.5 Ohm output impedance isn't a problem. If that or noise is an issue, there's always the Magni if one is to keep to the same price range.
Looking for such fringe cases to establish your point seems off base to me. I'm an EE and worked in the field as such not to be swayed by this. Nuff said.

 

[jimmers]

I guess this is all OT, however ...
 
I don't think zero gain will give you anything but zero out (unless you mean something else like 0dB )

...  However, the output buffer is set up with zero gain and attempts to act as a current drive. ...

...
Here are some measurements from a typical and well-known tube hybrid:
Voltage gain
330R load: 12.4 (21.9dB), 33R load: 7.2 (17.2dB) (same tube for both)
Noise
330R load: -86.1dB, 33R load: -90.6dB
%THD
330R load: 0.287, 33R load: 0.457
...

I would like to know the circuit and testing conditions for these, what I consider strange, results.

 

[tomb]

...  However, the output buffer is set up with zero gain and attempts to act as a current drive. ...

...
Here are some measurements from a typical and well-known tube hybrid:
Voltage gain
330R load: 12.4 (21.9dB), 33R load: 7.2 (17.2dB) (same tube for both)
Noise
330R load: -86.1dB, 33R load: -90.6dB
%THD
330R load: 0.287, 33R load: 0.457
...

I would like to know the circuit and testing conditions for these, what I consider strange, results.

The definition of a "buffer" is something that has unity gain. I admit mixing terms up there in that I meant zero amplification. Zero amplification = gain of 1 (unity). Another way of stating it is that the buffer has zero voltage gain.

Perhaps you can indicate why you think the cited measurements are strange?

 

[StanD]

The definition of a "buffer" is something that has unity gain. I admit mixing terms up there in that I meant zero amplification. Zero amplification = gain of 1 (unity). Another way of stating it is that the buffer has zero voltage gain.

Perhaps you can indicate why you think the cited measurements are strange?

Please show us a schematic that places this resistor in the circuit as well as the headphone so we can make sense of it. Otherwise there is no context for the measurements. At one point this is best resolved with PM's as this is OT.

 

[tomb]

 

The definition of a "buffer" is something that has unity gain. I admit mixing terms up there in that I meant zero amplification. Zero amplification = gain of 1 (unity). Another way of stating it is that the buffer has zero voltage gain.

Perhaps you can indicate why you think the cited measurements are strange?

Please show us a schematic that places this resistor in the circuit as well as the headphone so we can make sense of it. Otherwise there is no context for the measurements. At one point this is best resolved with PM's as this is OT.

 
There is no headphone, nor should there be.  The resistances were clearly specified.  Honestly, I don't have the time to draw a picture showing how to connect a resistor to the output of an amp.
 
Read this about current noise in a resistor: http://www.vishaypg.com/docs/49997/49997.pdf  That's how this argument started.  All I did was suggest to you guys that impedance also plays an effect in the noise of an amp.  Somehow, that progressed to disbelief in whether gain can change at different impedances and whether it was a good idea to add a resistor to the output of an amp.
 
As for resistors on the output (something else you also questioned), maybe you should read up on Walt Jung, Pete Millett, and Alex Cavalli.  The last two have designed some of the most notable tube hybrids in existence.  All of them used resistors on the output at one time or another.  That includes Walt Jung on his diamond buffer, too: http://waltjung.org/PDFs/WTnT_Op_Amp_Audio_2.pdf  He refers to it as a "load isolator" that "can be increased if necessary."
 
I have personally built, tweaked and measured dozens of diamond buffers.  A resistor on the output is not crazy, neither is it a reason to reject an amplifier.  I don't subscribe to the religion of super-low-output impedance.  There are instances where that can be a detriment, not an advantage.
 
I agree that this should be in a PM, but honestly, I don't have time for this.  I already spent too much time in this thread a year or so ago.  I am sorry for the digression. You guys go on to something else ... please.
 
 
P.S. The measurements were done on a revMH Millett Hybrid built by AMB a few years after Pete's original article.  I have made similar measurements and noted gain and noise differences on many amps, depending on the load impedance.  There's an explanation for it, but it doesn't jive with perfect-world theory.

 

[StanD]

@tomb Lets, see. The first article about resistors, now calculate the level of noise that a cheap carbon film resistor will contribute to a headphone amp. Yes, go do the math in the article and discover the level is nowhere near what you will be able to hear at the gain levels of a headphone amp. One can use Google to find articles, but that doesn't mean the information is relevant to the discussion as this is not a high gain instrumentation amplifier for strain gauges.
The second article has nothing to do with the discussion at hand. Are you trying to confuse the discussion? As far as the resistor goes you didn't state if it was a simple load resistor or a resistor in series with a headphone. In fact the R8 resistor in the article referred to as a load isolator is in series with the actual load, on the outside of the feedback loop, it is not the load. It is probably there to prevent instability due to capacitive loading. You should read the article more carefully and not try to confuse readers.
The Millet amp does not have short circuit protection so it was recommended to use a series resistor similar to the R8 load isolator to limit the output current. So if you vary the final load impedance it will form an attenuator (voltage divider) and thus appear to affect gain. I wouldn't use such an amp, get a Vali instead. IMO you made an argument that has no value to the case.
 
Now for people with a Vali that have sensitive IEMs or headphones that have the volume turned low and still hear noise, one can use a resistor L-Pad to cut down the output level while controlling the load impedance presented to the Val while keeping the series resistance to the headphones low enough not to cause side effects. Unfortunately this requires some calculations if one wants to get a good result.

 

[Astral Abyss]

Can't you guys just let it go?  There's no winners here...
 
 
Anyway, I'm thinking of replacing my Vali with an Ember II, mostly for more control and to roll some tubes.  The Vali was kind of an experiment to see if I liked the tube sound, and I do.  I've since bought a Bifrost MB and Lyr 2 that I use in my office.  The Modi 2U / Vali combo is connected to my gaming PC (although I just use it for music) to a pair of HD650 (and I have a Fostex TH-X00 on the way to try with it as well).  Think it'll be noisy with the Fostex?  They're pretty low impedence.
 
What do you guys think about all this?

 

[jimmers]

  Can't you guys just let it go?  There's no winners here...
 
...  Think it'll be noisy with the Fostex?  They're pretty low impedence.

 

[StanD]

  Can't you guys just let it go?  There's no winners here...
 
 
Anyway, I'm thinking of replacing my Vali with an Ember II, mostly for more control and to roll some tubes.  The Vali was kind of an experiment to see if I liked the tube sound, and I do.  I've since bought a Bifrost MB and Lyr 2 that I use in my office.  The Modi 2U / Vali combo is connected to my gaming PC (although I just use it for music) to a pair of HD650 (and I have a Fostex TH-X00 on the way to try with it as well).  Think it'll be noisy with the Fostex?  They're pretty low impedence.
 
What do you guys think about all this?

It's not the impedance it's the sensitivity. My HE-500's are low impedance and haven't any issues with noise with the Vali or any other amp.

 

[senorx12562]

It's not the impedance it's the sensitivity. My HE-500's are low impedance and haven't any issues with noise with the Vali or any other amp.

These factors, along with (electrical) efficiency are often conflated, ime. Though they are related, i agree that it is the sensitivity that makes 'phones sensitive (pun intended) to noise.

 

[StanD]

These factors, along with (electrical) efficiency are often conflated, ime. Though they are related, i agree that it is the sensitivity that makes 'phones sensitive (pun intended) to noise.

Yes efficiency and sensitivity are confusing terms. I think that sensitiviy in mW/dB to be tricky when headohones have varying impedance curves since the output signal voltage of an amp doesn't really change with impedance (unless there are driving issues). The sensitivity is usually given at a particular frequency, typically 1 kHz. Yet if the impedance changes at another frequency then the power (mW) will be different, yet the volume stays the same. Of course if one inserts a series resistor of a significant value then the FR will change. This is why I prefer sensitivity specs as dB/V.
 
One day when I have some spare time, I'll throw together a resistor L-Pad that might work well with a Vali to reduce the output level which should help with noise. If I do, I'll post the schematics.
Something like this in stereo and perhaps 32 Ohms might be nice. We could live with less than 100W as well, but that shouldn't be a problem.
http://www.amazon.com/Parts-Express-Speaker-L-Pad-Attenuator/dp/B0002KR1K0

 

[MadsSK]

Hi Head-fi members,
 
I was wondering if you think that the Schiit Vali would go well with my Shure SRH1840 ? They are 65 ohms so I am dont know if the impedance is too low to use with this amp. I am interested in an amp which will add some warmth to the sound without removing from the detailed highs or mids. I am especially intersted in more punchy bass (not out of control).
Do you have any recommendations?
 
/Mads

 

[Mr Rick]

  Hi Head-fi members,
 
I was wondering if you think that the Schiit Vali would go well with my Shure SRH1840 ? They are 65 ohms so I am dont know if the impedance is too low to use with this amp. I am interested in an amp which will add some warmth to the sound without removing from the detailed highs or mids. I am especially intersted in more punchy bass (not out of control).
Do you have any recommendations?
 
/Mads

It would appear to drive them fine.
 
http://www.audiobot9000.com/match/shure/srh1840/with/schiit/vali
 
I can't speak to your sonic preferences.