Give me a reason NOT to do this

[manisandher]

That's the problem with summer. I usually have a few weeks off work and try to spend as much time at home as possible. But on Thursdays and Fridays, my wife goes to work and my 20-month old son goes to nursery. I'm at home... with way too much time on my hands...

My two NOS1s should be back at some point and I'm eager to get everthing ready for them. One of the tasks is to get an ultra-fast amp sorted for my office system, with my small Quad electrostatics. I've been looking at some possible solutions, but the one that I keep on coming back to is to simply buy another Sanders Magtech amp for my office, just like the one I already have in my main room, with my big Quad electrostatics. I've tried a number of amps in my main room now, and none of them sound as good as the Magtech with the Quads. The sound comes alive with the Magtech - it's 3-dimensional, tight and really quite dynamic (for electrostatics at least).

But buying an identical amp just seems too boring. So the search recommenced today. Of course, this is nothing new for me. I've been at this before (http://www.phasure.com/index.php?topic=1274.0). But this time, I'm looking for a real end-game, so that I can focus on other elements of my system. Anyway, to cut to the chase, I have an opportunity to buy a used (but absolutley mint) Soulution 710 amp (http://www.soulution-audio.com/en/serie7/710/index.php). The idea would then be to put the Magtech in the office (with the small Quads) and the 710 in the main room (with the big Quads).

But even used, the 710 will cost me 14K UK pounds! Now, I'm not a rich man and this is a totally absurd amount of money for me. But, I can just about afford to buy it without putting my family in a state of total poverty.

Someone... anyone... please give me a compelling arguement as to why I shouldn't do this. Please...

... but you've got to convince me that the USB-NOS1/710/Quad combination won't be heaven on earth, because I'm totally convinced it might well be.

Mani.

[PeterSt]

Ok, you are obessed. Can't be helped. You *have* to do it. But :

What about the way too high power ?

Also, output seems to be 22V-p with a rise time of 330ns. Roughly this means a slew rate of 66V/us.
Let's say I didn't make too many mistakes here, and let's say the slew rate of the NOS1 of 650V/us is something to take into consideration.
Does this look good to you ?

Then, I don't like the graphs;
I can't see the FFT depth, but from the two pictures -one with noise only, one with signal- there's already a 20dB difference. Besides, something like -165dB for the noise floor ? haha. Well, the specs in the end talk about -110dB, so I guess that's the figure, but that's no good figure at all.

The noise-only figure shows harmonics which I don't like to see. I know, amp manufacturer judge this as very good etc., but I just don't want to see it. From the NOS1 comes exactly silch (the noise floor being at a real better than 140dB), and the only thing my Gainclones add during playing the test signal is a 3rd harmonic which isn't higher than the harmonics coming from the DAC. See ? it can be done.


So Mani, I did my best to trash this amp, which most probably is totally unjustified for the normal situations. Even in your situation it may not be justified at all. But at least you now may have a reason to not spend 14K GPB.

Peter

PS: You can always try to dig up similar figures from the Magtech.

[manisandher]

Hey Peter, thanks for your thoughts. I know you're busy, so appreciate your spending time on my 'obsession'.

What about the way too high power ?

I don't think 130W into 8 ohms is a problem - the Magtech is 500W into 8 ohms.

Also, output seems to be 22V-p with a rise time of 330ns. Roughly this means a slew rate of 66V/us. Let's say I didn't make too many mistakes here, and let's say the slew rate of the NOS1 of 650V/us is something to take into consideration.

Actually, I think the slew rate is 31V/0.33µs = ~90V/µs. Now the Magtech operates at plus/minus 92 volts and is quoted as having a slew rate of 500V/µs. The operating voltage becomes important when considering an amp for electrostatics because these speakers are 'voltage operated devices'.

Then, I don't like the graphs; I can't see the FFT depth, but from the two pictures -one with noise only, one with signal- there's already a 20dB difference. Besides, something like -165dB for the noise floor ? haha. Well, the specs in the end talk about -110dB, so I guess that's the figure, but that's no good figure at all.

Now this is where I'm getting a little confused. In the recent Stereophile review of the 710, John Atkinson ends his measurements by saying, "... the Soulution 710 is definitely one of the best-measuring amplifiers I have encountered." I've attached some of his measurements - any comments about these?

Incidentally, the Magtech's noise is also rated as, "More than 110dB below rated output."

The noise-only figure shows harmonics which I don't like to see. I know, amp manufacturer judge this as very good etc., but I just don't want to see it. From the NOS1 comes exactly silch (the noise floor being at a real better than 140dB), and the only thing my Gainclones add during playing the test signal is a 3rd harmonic which isn't higher than the harmonics coming from the DAC. See ? it can be done.

Yeah, but I need an amp that will drive my Quad 2905s 'properly'. On paper, they're pretty benign. But their impedance dips sharply at the frequency extremes and I'm sure that this is why they sound woolly low down and too smooth up top with most amps. But the Magtech seems to cope very well.

So Mani, I did my best to trash this amp, which most probably is totally unjustified for the normal situations. Even in your situation it may not be justified at all. But at least you now may have a reason to not spend 14K GPB.

I think with your help, I'm on the road to convincing myself to buy another Magtech... and save >10K UKP! Thank you!

But any additonal thoughts (from Peter or anyone else) would be really welcome. I have to make a decision as to whether I buy the Soulution pretty quickly.

Cheers, Mani.

[manisandher]

A slightly different topic, but I think I'll keep it in this thread...

Yesterday, I was testing some other old amps that I have lying around, and I ran out of interconnects. I went down to the basement and pulled out the only pair of XLR interconnects that I could find - 'PureNote Epsilon' interconnects. These have ultra-high purity silver conductors housed in a stiff a metal sheath. I bought these ICs years and years ago, tried them, didn't like their forward/bright sound and put them away for I guess 8-9 years! I convinced myslef that silver was a bad metal for audio. But you know what, they sound incredible now, with much better low-level detail than any other ICs I have. And they just sound 'faster' - better transients and overall boogie factor.

Does anyone have any idea as to why a silver cable might sound 'faster' and more detailed than a copper cable? I'm not sure I buy into the simple "better conductivity" argument, because you can always just increase the amount of copper in a cable for significantly less cost...

Mani.

[Flecko]

Hi Mani,
have a look at this amp: http://www.audionet.de/main/elektronik/endverstaerker/max/charakteristik/page.html?L=en
I did not heard it but they have a very good reputation to have an excelent value for money. I like their philosophie and this is the amp I would look first for, if I would try to finish the search for an high end amp. 400W into 8ohm and a bandwidth of 0-500kHz should be fast enough, plus a dampinfactor of about 10000 should control every speaker. New "only" 12500€ and used it goes for about 8000€. Here you actually find one for 8300€ 3 years old: http://www.audiomarkt.de/_markt/
Greetings
Adrian

[manisandher]

Hi Adrian, thanks for this - much appreciated. I've looked into Audionet before but have never heard any of their gear. I remember the owner of Wilson Benesch recommending Audionet with my then Chimera speakers, which were hard to drive. I'm not sure how easy it is to get hold of, and sell if/when necessary, an Audionet in the UK.

From a purely technical point of view, I guess what I'm looking for is the following:

- high voltage output (is there any other good audio amplifier that operates at 90V and not 30V?)
- at least 100W into 8 ohms
- power that doubles with every halving of impedance (down to at least 2 ohms) - to keep control at frequency extremes with the Quads
- ~500V/µs slew-rate (and therefore high bandwidth) - although many people would say this is 10 times higher than is necessary...
- low THD* at 100W ouput
- something that won't depreciate massively and that I can sell if/when necessary

* Unless it's a true single-ended output amp. You see, you can theoretically reverse the speaker polarity and use the amps HD to cancel that of the speaker, which should have a similar HD spectrum to a single-ended amp. But, I've never come across a single-ended amp that has all of the other requirements above.

Any thoughts anyone?

Mani.

[PeterSt]

Briefly :

Actually, I think the slew rate is 31V/0.33µs = ~90V/µs.

I didn't look it up again, but I think the amp bears 31VRMS. If I (again) don't make mistakes, this is 42V-pp (from bottom minus to top plus).
Slew rate (I think), is measured from zero to plus (or from zero to minus, but that shouldn't make a difference; btw fall time (opposed to rise time) can be different again). So, 42-pp would be 21V-p (from zero to top only). And now *that* rises in 330ns ...

If this is wrong and slew rate would be measured from bottom minus to top plus, then my figure has to be doubled. So, then it would be 133V/us. You can always count on me being wrong.

~500V/µs slew-rate (and therefore high bandwidth) - although many people would say this is 10 times higher than is necessary...

a. This is not related to bandwidth, but the sheer being able to follow transients;
b. 10 times higher than necessary is the common electrical engineer's rule.

Ad a. If the frequency of the high transient (think synths) becomes so high that the fall time is longer than the next rise is to start, yes, then it is related to bandwidth.

An additional notice : the NOS1 also is rated 10 times higher than necessary; Only last week I tried with 2 times only, and it didn't sound good anymore ...


I never understood Stereophile (DAC) measurements. John Atkinson is "able" to judge a random 16/44.1 CD player or DAC to be not worth while for even playing CD's, when the dynamic range is worse than 130dB or something. Well, I don't know what dither he applies around the player (or expects it to be active somewhere) to achieve something like 130dB which inherently has 96dB only, but as I said - I just don't understand it (and looked HARD for the explanations, which I can't find anywhere).

Peter

[Flecko]

- high voltage output (is there any other good audio amplifier that operates at 90V and not 30V?)

Shouldn't have an high power amp also have an high output voltage? Because P=U*I,  and I=U/R so P=U^2/R. R is constant for the speaker. Hence you need more power to have a higher output voltage.
Also if you think it the other way around: An Amp deliveres an high output voltage but does not have the power to deliver the current, it would make no sense to have the high voltage.

- ~500V/µs slew-rate (and therefore high bandwidth) - although many people would say this is 10 times higher than is necessary...

Does anybody know how to calculate slew rate from bandwidth and vice versa? (My preamp has bandwidth of 0-1.5Mhz )

Corrections: Ups... It must be Mhz not Ghz haha

[Flecko]

a. This is not related to bandwidth, but the sheer being able to follow transients;

I read an article once, where it was said there is a connection between slew rate and bandwidh. I think it must.

[Flecko]

and here it is...http://www.amplifier.cd/Verstaerker/Wald_und_Wiesen_hifi_Verstaerker_Test/what_are_open_loop_slew_rate_bandwidth.htm

I do not have the time to read it myself at the moment, I should work now....

[manisandher]

R is constant for the speaker.

If only! My Quads have pretty much 4 ohms in the bass, 8 ohms in the mid, and 2 ohms in the treble (falling to 1 ohm at ~18KHz). The current required in the treble will be 4x that of the mids (8x at ~18KHz). Of course, this assumes all sorts of things which I'm sure aren't really valid... reactive vs. capacitive loads, etc.

Another complication with my Quads is that being electrostatics, they are not driven by current but by voltage. The Soulution outputs 31V RMS, which I'm sure is fine for 'normal' magnetic speakers. My Magtech outputs 62V RMS, which the Quads should prefer.

EDIT: This voltage I guess is dependent on the operating voltage of the amp. High voltage parts are more expensive, so it's not in an amp manufacturer's interest to have a high operating voltage... unless they were designing an amp for electrostatics 

Mani.

[earflappin]

Mani, regarding the silver interconnect question you raised, I have also done a lot of experimentation over the years with silver cables.  Initially they have typically wowed me with their sense of speed, greater detail, etc.  But then after listening for a few days I end up concluding they are too thin and lacking in tonal balance.  And I go back to a copper based cable (or at least a hybrid).

I remember George Cardas telling me that his conclusion after much experimentation with silver was that they were like a beautiful woman with no soul.  YMMV.

[manisandher]

I remember George Cardas telling me that his conclusion after much experimentation with silver was that they were like a beautiful woman with no soul.

I love it!

I'm sure many of us have had a similar experience - I know I have. But this simply goes totally against the 'human bias' and 'double-blind' argument. I mean, 8-9 years ago, I knew I had silver ICs in place, I knew they were 5x the price of my copper ICs, and yet I still returned to the copper ICs, for much the same reasons as yourself.

Today, the silver ICs still sound 'thiner' than my copper ICs. But for the first time, they sound more right. Why should this be? If I had to guess, it's because my new mains circuit has gotten rid of all the hash that in the past has been accentuated by the silver ICs. Just a guess...

Mani.

[PeterSt]

a. This is not related to bandwidth, but the sheer being able to follow transients;

I read an article once, where it was said there is a connection between slew rate and bandwidh. I think it must.

Up to some degree, sure yes. I mean, in order to suffice a certain bandwidth, first the slew rate has to be sufficiently high to not let mix up the subsequent "samples" (but notice this is about analogue). When this is met, next the slew rate can still be very much higher, while the bandwidth does NOT increase. Something like : nice that we have an overly high slewrate, but in order to really comply to that bandwidth which is "implied" by the slew rate, many more things have to happen. Like inherent bandwidth in other places of the device, it will be more hot (and burn without special precautions) and probably a few more things I don't know myself. Anyway, if you look at datasheets of devices with slew rate and bandwidth, you will see that both are totally unrelated, BUT, a device with bandwidth of 200Khz for sure will not have a slew rate of 1uV/uV because that will allow for a frequency of 1 (I guess). So, when you work this out, the 200,000 needed implies for 200,000uV/uV which is 200mV/uV at 1V output (am I doing this right ?). So, for e.g. 30V this takes 6V/uV. Is the needed bandwidth 1GHz ? than multiply this by 5. Now we have 30V/uV and we come close to reality.
But we are not there, because any transient will have some sustain (otherwise it's nothing), and don't ask me what to add for that. But it will be something like 1/1GHz sec to be realistic. Now things change a bit, because that 1/1GHz already consumes the time it is given, which makes the rise time infinitely high. Also the fall time has t be added to the lot.
After this has been done, the logical conclusion should be that an original 1GHz bandwidth decreases to less because of the "overhead" needed. Thus, rise time consumes, fall time consumes, and the sustainment of whatever is the length of the pulse also consumes (should be a wary formula on the latter).
One can maybe also say : multiply whatever you think the needed bandwidth is by 10 (only for the slew rate number !), and you're good. But this is from another angle, and a not much official one.

Peter (hoping he's not too much off on this all)

[manisandher]

Referring to amps, I think 'telstar' had an interesting take on things. His feeling was that you need 1V/µs for every 1W of amplifier power (presumably at 8 ohms). My Magtech amp is 500W at 8 ohms, and coincidentally has a slew rate of 500V/µs. The Soulution is 120W at 8 ohms, so I guess all other things being equal, it 'should have' a slew rate of 120V/µs... no?

Does this make sense, or does it just confuse things further?

Mani.