Sauermann Amplifier

[xp9433]

Mani

Le Monster - CaseLabs W8, Asus x79, Intel i7 3960X is certainly a huge machine.
No wonder you can't have it in the lounge!
Nevertheless it must look very impressive. Any pictures for us?

Cheers
Frank

[manisandher]

Any pictures for us?

Hi Frank, well there's really not a lot to see, but here's a low-res pic taken with my phone. In the background you can see one of my totally silent Zalman TNN300 PCs. This itself is not a small machine, but is dwarfed by Le Monster.

To keep some semblance of being on-topic, notice the gorgeous Sauermann amps on the racks. I asked Gerd to finish them off in white to match my NOS1 ('The White Sheep'). Not even a high-res pic would do them justice - they are simply the best finished hifi components I've ever had.

Cheers, Mani.

[xp9433]

Mani
Yes, it is a bit of a monster (and I suspect at a monster price), and hard to hide, but not as big as I had imagined! I understand there are even bigger Caselabs cases?

Your system looks great. I see one of the advantages of having an open baffle design is that their radiation pattern allows you to position closer to side walls without degrading the sound. Less imposing on the living environment as well.

The Suaermanns are somewhat bigger than I imagined for what is a relatively low powered mono amp - which is good of course.

Cheers
Frank

[manisandher]

Hi Frank.

Yes, it is a bit of a monster (and I suspect at a monster price), and hard to hide, but not as big as I had imagined! I understand there are even bigger Caselabs cases?

The biggest cost was the CPU and of course the RAM (64GB is of course totally crazy, but I just felt anything less would not be worthy of the 'Le Monster' name).

Your system looks great. I see one of the advantages of having an open baffle design is that their radiation pattern allows you to position closer to side walls without degrading the sound. Less imposing on the living environment as well.

Thanks. The Quad 2905s use a 'concentric ring, time delay' system to create a point source effect. And as you say, they tend to 'beam' quite which means that they can indeed be placed fairly close to the side walls without any ill effect. But you need to have enough space behind them so that the sound reflected from the back wall remains benign. I like the big Quads and in a large enough room, with them placed near the side walls, they're not as imposing as their 'on paper' size might suggest.

The Suaermanns are somewhat bigger than I imagined for what is a relatively low powered mono amp - which is good of course.

Yep, even 40-50W of pure class A amplification requires reasonably large heat sinks. Fortunately they don't get stupidly hot - even after being on for a while, I can easily keep my hands on their heatsinks.

Mani.

[GerdS]

Hey Gerd,

So you could even hook one output pin to earth ground without problems (what some messurement equipment do when testing on complex load).

I would never do this !
This assumes earth ground to be equal to the (created) potential of the amplifier ... which it will not be.
And there you will have DC offset ...

Or ?


Regards,
Peter

Hello Peter,

no!
Since no part of the circuit has any relation to ground you can connect any SINGLE point of the circuit to ground without changing anything. Except that now any other point of the circuit shows a voltage to ground what does not matter, doesn't it?

Regards,
Gerd   

[PeterSt]

Two things IMHO :

1. The ground you refer to is absolute "zero" with SE all right. ... Within your device.

2. It will influence that same ground reference. Something like the average (RMS) positive voltage will tear it down. Now PE goes down = neutral will go down.

Now, what happens with other devices on that same neutral ? They fight with it. An inherently lower voltage device can't fight back. It will have an offset compared to the stronger device. But merely, your mains will not be a sine anymore.

Gerd, I don't say I am completely correct on this reasoning, but I just see it happening. The max offset I saw was 109V. All it needs is this independent other "PE" reference. I have that here. The more devices you tear out (all SE of course) the lower the offset gets. Mighty difficult to prove in an always consistent setup, because groundloops may undo it and next create problems else where.

So what I think is that the only way to avoid this for 100% sure, is create your own ground reference throughout.
This is what you do with your amplifiers, and this is what I do with the NOS1. Nothing to destroy it, if only PE is left out.
Have one device in there (like my bass amps in "our setup") and the offset is there, even when that bass amp is not connected to PE. The neutral does it, because it balances to PE. The net output will be lower (cancels out) and the noise rejection less.

Again, my reasoning can be wrong (because most of it is just that only), but just too many things add up; Saw Mani's transformer supply ? looks good eh ? well, at the moment he put that in, the DC offset went from near 0mv to 120mV everywhere. And no, I don't know how. And yes, 3x 120 = 360. These kind of relations are in there too (I'm, obsessive now).
These things you see happening when you try to let nature do its work, while nature is different per kind of "mains" (like at some stage I had it working for all, but 110V worked nowhere).
Today I have it working everywhere, but don't ask me what it took.

Kind regards and call me crazy where it's needed !
Peter

[manisandher]

A quick update on how I'm getting on with my amps...

They're great, and I love 'em! Every change in the music (different recordings, XX settings, etc) comes through loud and clear. It's a cliché for sure, but I really don't think these amps have much of a sonic character... and yet they sound like no other amp I've ever heard. Ergo, all other amps I've heard do have a sonic character...

Attached are some measurements of the amps' THD performance (taken from Gerd's white paper):

1. Be careful here because I'm pretty sure K1 denotes the fundamental and K2 the 1st order hamonic, etc. This being the case, it's clear that the amp is dominated by even-order harmonics (K3 and K5), with the 2nd order harmonic rising linearly with power.

2. I'm not sure how common this is, but the THD just doesn't change much with output impedance.

3. Finally, THD remains pretty constant throughout a large bandwidth.

I'm no electronics engineer, but I think these results are impressive considering the design of the amp.

Mani.

[Flecko]

Hi Mani,
1.
K2 is even order Harmonics and k3 and k5 is odd order. That means K2 is two times the basic frequency, k3 three times the basic frequency and so on.

Here is a site where you can calcualte THD from Klirrdämpfung
http://www.sengpielaudio.com/Rechner-klirr.htm

and here is a review of the amp I use which is a similar design (class a, no negative feedback, third order dominated) and delivers similar distortion values.
http://www.jungson.com/files/reviews/ja99c_uk_review.pdf

In the box on the bottom there is a technical interpretation of the measurements, which could be interesting for you.

In comparison to a design which uses negative feedback, k3 is high. But I compared my "old" class a amp to negative feedback designs with lower distortion and finally I kept my old amp. It shows more details and more of the original colors of the instruments. But it can also sound a little hard on high listening levels. I like it.

Greetings
Flecko

[manisandher]

K2 is even order Harmonics and k3 and k5 is odd order. That means K2 is two times the basic frequency, k3 three times the basic frequency and so on.

Hi Flecko thanks for this. You're right of course. I don't know what I was on when I wrote what I did... I suppose I just thought that the 2nd harmonic couldn't be so low for the type of amplifier the Sauermann is.

In the box on the bottom there is a technical interpretation of the measurements, which could be interesting for you.

Yes, they suggest the Jungson gets a little 'sharp' at higher output levels due to that 3rd harmonic. I'm not sure if I listen to the Sauermann at high enough volumes, but I'll explore this a little more. It makes me think that an amp like the Sauermann should really only be used with high sensitivity speakers. Mine certainly aren't. However, because they 'beam', you can listen to them at pretty low volumes.

Mani.

[Flecko]

Yes, they suggest the Jungson gets a little 'sharp' at higher output levels due to that 3rd harmonic. I'm not sure if I listen to the Sauermann at high enough volumes, but I'll explore this a little more. It makes me think that an amp like the Sauermann should really only be used with high sensitivity speakers. Mine certainly aren't. However, because they 'beam', you can listen to them at pretty low volumes.

The question of the "right" sensitivity of a speaker to fit a certain amplifier is a question I ask myself at the moment. Once I wanted to have very high sensitivity speakers (~100db/w) and planed to use very low powered amps (10w). But the problem with this is, you also amplify the noise of the electronics. Also, the distortions from 1w watt to 1mW normaly increase in a normal amplifier design. So with a very high efficiency speakers you will listen in this "bad" region of your amp. From this point of view, it could be better to have a very high powered amp (500w) and a very low efficiency speakers (80db/w). And all should be matched ideally in a way, that at 0 db at the DACs out, the amp should be short before reaching the region, where it starts to produce an higher level of distiortion. That way you would use the digital volume control at its best. Noise should be very low and distortions at normal listening levels too.

[Flecko]

PS.: Just after i read my post I looked again at the sauermanns distortion figures. The distortion does not increase from 1W to 1mW as far as you can see the figures. It would be nice to see the distortion for lower watt and distortion levels. As far as I know this is unusual and could be the reason for its good sound. This amp should match very well with high efficiency speaker.

[gsbrva]

The question of the "right" sensitivity of a speaker to fit a certain amplifier is a question I ask myself at the moment. Once I wanted to have very high sensitivity speakers (~100db/w) and planed to use very low powered amps (10w). But the problem with this is, you also amplify the noise of the electronics. Also, the distortions from 1w watt to 1mW normaly increase in a normal amplifier design. So with a very high efficiency speakers you will listen in this "bad" region of your amp. From this point of view, it could be better to have a very high powered amp (500w) and a very low efficiency speakers (80db/w). And all should be matched ideally in a way, that at 0 db at the DACs out, the amp should be short before reaching the region, where it starts to produce an higher level of distiortion. That way you would use the digital volume control at its best. Noise should be very low and distortions at normal listening levels too.

I like this question of matching amplifiers.  I'll argue the opposite side for fun :-) 

I don't think using a low efficiency speaker just to suppress electronics noise and distortion is a good choice.  Higher efficiency speakers generally have better coupling to the air load and are intrinsically better damped.  Also, low powered amplifiers are much easier to design and build to high quality. 

With regard to amplifier power and matching a speaker, it becomes nearly impossible to achieve the performance of a low power/high efficiency system by increasing both power handling and power generation.  The laws of physics are against it all the way.

There is a big problem with introducing more power at lower efficiency.  The low efficiency/high power system is producing much less sound per watt.  That lost power doesn't just disappear (conservation of energy and all that stuff).  Instead it must eventually get turned into heat.  But in the meantime, that energy causes all sorts of trouble, like overshoot, power compression and re-radiation through enclosure walls and the speaker diaphragm.

If you hear a lot of noise coming from high efficiency speakers then the electronics are not properly designed.  Even with tubes, you can run directly into 115db 1w/1m drivers if it's done right.  Usually electronics simply have too much voltage gain. 

One disadvantage of the digital volume control is that you must listen to all the noise all the time.  I think it's nice to have an analog gain control located as close as possible to the  final amplifier stage to manage the noise.  That way you can run your digital volume control up near full all the time.  This preserves the bit depth of the converter and still allows reserve gain for quiet recordings without the disadvantage of a high noise floor for normal recordings.

There are a lot of reasons to choose a lower efficiency speaker system, but amplifier matching is not one of them IMHO.

Greg

[PeterSt]

If you hear a lot of noise coming from high efficiency speakers then the electronics are not properly designed.  Even with tubes, you can run directly into 115db 1w/1m drivers if it's done right.

Although true, easier said than done ! And careful now, because a 1W tube amplifier doesn't amplify all that much. I mean, the gain is relatively (very) low, so the possible noise will be just the same (very low). But ... the gain will be sufficient for the 115dB sensitive speaker of course (even for larger rooms).

The real sh*t happens when you have a higher gain amplifier and run *that* into this 115dB speaker. Now, funnily enough I just have that. And I can tell you, it won't be a much normal thing when no noise is audible at full gain of, for example, my GainClones (20dB gain or so). But since this is topic is about the Sauermann amplifier (with somewhat more gain IIRC) ... they too are dead silent through these same speakers.

But is it about the amplifiers ? ...

Partly yes of course, but for the largest part it is about the environment; the remainder of the chain. How grounds are connected. Ehm, how much noise the DAC produces ...
... whether there's an analogue volume control in order ... Just saying.

Let me put it differently, and remember, all is about the (very !) high efficiency speaker :

5-6 years ago I obtained these same speakers and same gainclones. I used "a DAC" and a passive preamp (TVC). Played with that for years.
I better did not turn up the volume fully (no playback going on) or otherwise a blast of noise (and mains rattle) would be my part. Of course I did all to get rid of it as good as possible.

I could use a FireFace800 in my chain, and I found out I had to shut off all the unused channels (there are 56 in there) because they all produced their own share of noise. Took me maybe a year to find that out.

If I measured the noise through microphone it was around -70dB. Remember, at full gain. But who cares, because who plays at full gain.
Well, all of us today, because the first thing to get rid of is the preamp. Aha.

Long story short, when the environment is made right, only *then* there's only noise from the main amps and the DAC. And the DAC is the first thing to look at. Not only because it may produce nosie herself, but merely because of the incurred for groundloops, EMI and that sort of problems. Pickup from the mains.

You want noise ? just put in that one good resistor (and think about it as attenuaton). It is totally impossible with 115dB spakers and a gain of 20dB or so. Only that resistor ...
(this is how all analogue attenuations on the NOS1 failed)

At this moment it is fair to say I have "no noise"; The NOS1 produces a tad better than -140dB and measured at the output of the amp that becomes -120dB. It can't be much better with a gain of 20dB, right ?
Well, this is not 100% practice, because my GainClones are single ended (RCA) and it very much depends on how the interlinks go to the amps, and in my situation this varies per day (testing stuff). So, as a result I can hear some noise sometimes when I'm with my ear very much into the horn.

But there you have it : the environment again. When I hear noise (ear in the horn) you can bet it will be at -90dB or worse. Just pickup from the interlinks, the connectors or poor soldering of them. This is why balanced interlinks are a must. This is why an "all the way" balanced (differential is how it's called) setup is a must. By a bit of coincidence for a chain with the NOS1 and the Sauermann this is the case. Nothing faked and all based upon a self created ground (which is what happens in a true differential setup). Have the DC offset under control at both ends (DAC/Amp) and there you have the ultimate (for noise rejection, also "in the box").

So let me conclude that when someone perceives noise from an amplifier (outside a "lab" ituation), chances are close to 100% it is not that amplifier alone, or maybe even not at all.

Peter

[PeterSt]

One disadvantage of the digital volume control is that you must listen to all the noise all the time.  I think it's nice to have an analog gain control located as close as possible to the  final amplifier stage to manage the noise.  That way you can run your digital volume control up near full all the time.  This preserves the bit depth of the converter [...]

No, no, triple No. And Greg, your first sentence here seems to show that you don't believe in it much yourself. I mean, who talked about poor electronics ? you. So, you use those poor electronics ? I guess so. Thus, get rid of them (you implied that yourself too, haha).

But following my previous post, it appears easy to turn things upside down. So, we put in an attanuator to surpress the noise from ... itself. Now *that*'s nice.

[...] and still allows reserve gain for quiet recordings without the disadvantage of a high noise floor for normal recordings.

How is that related ? IMHO nowhere. Well, if you are saying that a higher level recording is allowed to mask the noise which would be audible otherwise, then I understand what you are saying. But never think I will agree with the "solution";

What people often seem to think is that as long as noise is not audible, all is right. But it is not, because noise, at audible levels or not, is in all of the signal, and it creates the signature of it. Take that attenuator again;
It's noise signature already shows that it's totally uneven in the frequency band to begin with. Oh, what ? it's at -120dB and thus inaudible ? NO, because and again, it is in the signal everywhere. This is how any means of analogue attenuation immediately kills the sound. Makes it dead. Dead bird I call it.
But of course you first need the reference to compare with.

So Greg, go hunting for the dead silent system and don't start with replacing your amplifiers.
(for fun : Ask Scroobius (Paul) whom I "guided" to a noisless system, only because I told him it could be done -> took 30 emails, so don't ask me )

When we play 16/44.1 Redbook material (onto a 24 bit DAC) we can attenuate to 48dB and nothing is lost. It needs decent digital (losless) attenuation, and XXHighEnd just contains that (ehm, for the explicit purpose).

There is much more going on, but this can't be explained in here, and some of the things I might have difficulties with myself. But for example, when I use the NOS1 and a test signal at -3dBFS the THD+N is around 0.0018% (never mind this figure in absolute sense). Now, when I digitally attenuate with -21dBFS, THD+N is around 0.0035%. Now think : Attenuation 21dB while THD+N increased only 6dB (2 times). This, while the signal came 21dB closer to the noise (THD+N) - is what you think.
But it ain't so, because thermal noise in the gain stage got down too (less output from the D/A chips). Actually with that same 21dB.

You could of course say that THD+N got 6dB worse anyway, and thus it is not a good thing to digitally attenuate, net. But for starters it is 6dB and not the 21dB you'd expect.
Now stuff in that analogue attenuator and see the noise floor RISE (instead of drop). With what ? depends on the means. But 20dB is the most easy ... Now 0.0018% becomes ~ 0.0054% only because of the N part. And if that were all ... But because of the uneven characteristic of the attenuator (thermal stuff again) it will be worse, because harmonics will appear all over the place too.

The story is infinitely longer, but this is to get the idea. Or something to think about at least.

Regards,
Peter

[gsbrva]

Ha,  thanks for the replies.  Maybe I was trolling a bit there :-)

[...] How is that related ? IMHO nowhere. Well, if you are saying that a higher level recording is allowed to mask the noise which would be audible otherwise, then I understand what you are saying. But never think I will agree with the "solution";

Ah, I see where we are misunderstanding.  No, that is not what I am saying.  Not sound masking, that is not at all a good thing.  I'm just talking about electronic noise generated after the DAC.  If there is excess gain, then there is excess noise.  The only solution for such noise is to reduce the gain by removing (or improving) devices or through attenuation.  A volume control after the noise source reduces both signal and noise.  A volume control before the noise source just reduces signal and reduces the signal to noise ratio.   

[...]The story is infinitely longer, but this is to get the idea. Or something to think about at least.

So very painfully true.  Just when I think I finally understand the big picture, there is always more.   

One thing I'm not understanding is the dislike for analog attenuation.  I'll bet every user on this site uses analog attenuation.  Even if you have a Gainclone amplifier driven directly off a NOS1 you are using a great deal of analog attenuation in the chip feedback loop.

I could even go so far as to claim that "digital" volume is no different than an analog voltage divider.  It is simply selecting different resistors per word in the PCM1704.  After all, the DAC is mostly a bunch of switched nichrome resistors as a current divider.  Music signal currents still must flow through these resistors and all the effects of analog theory apply.  It's just that by careful design BB keeps any filtering to a higher frequency than we will ever worry about.

For the record, my "analog" volume control is a constant impedance network of switched nichrome resistors.     

Analog attenuation is the lowest distortion building block available to the designer.  I agree that using it stupidly, such as driving an interconnect cable through a pot sounds bad.  But, that is just an accidental analog filter.

Often, there is no good spot in the signal chain to place a volume control.  I will certainly agree that if using a chip amp, there is no benefit to adding anything in the chain.  Putting a volume on that amp front end makes no sense because it would effectively be at the same spot in the signal gain stages as the digital volume. 

However, I must still stick by my statement that an analog attenuator placed as close to the last device as possible has benefits over digital volume control that is upstream of noise generation.

Many amplifiers do generate internal noise that is easily audible if you are direct driving horn compression drivers in a multiamped system.

Cheers,
Greg