Constant power - tubes vs. solid state amps and wild-impedance speakers

microstrip said:

Sorry, only today I managed to post the promised ESL 63 impedance graph.
As you see, it is level dependent and will raise with power - 160 mV at 8 ohm is only 3mW .
Low bass impedance is low only at powers that are not significant.

.

Click to expand...

Thanks, Micro...this illustrates -hehe- you prior post, splendidly.

microstrip said:

Martin Logans are known to be incompatible with OTLs, and in my opinion with typical tube amplifiers - but I have no experience with recent models with active woofers.

Both the old Prodigy's and the Monoliths sounded best with high power SS.

But other people can have different opinions!

Click to expand...

I have not tried the Atmaspheres with older Logan models, but my on my ML Spires, which have a powered woofer, Atmasphere A1's sounded AMAZING. That was the best I had ever heard Logans sound, including the CLX. And the Atmaspheres driving the Spires were one of the few amps that CLEARLY outclassed my CJ Premier 350. On the Soundlabs A1s (non-px at the time), the A1's were rolled off in the highs and had flabby bass using the smaller Atmaspheres. On the other hand, I have heard the Atmaspheres A1 drive Soundlabs U1PX quite well.

I did not have my VTL 750s at the time, but currently I am getting a tiny bit of rolled of highs with them driving A1 px's. The Ayre monoblocks and the CJ Premier 350 have cleaner highs, but musically the VTL smokes just about any amp I have ever heard.

I wonder if bi-amping solves these problems...

Sorry to resurrect an old thread, but this topic is one I’ve been pretty intimate with for the past decade or so.

I have a set of MartinLogan’s that I’ve worked on over the years. See my site for details.

Regarding amplification challenges, yes, these definitely present huge challenges. But there are SS amps that deal with them just fine.
The key seems to be to use designs that flow enough current into the wildly swinging impedances AND that gracefully handle the back-EMF these speakers can generate.
Specifically, in speakers where the factory crossovers have been stripped out and the panel is run directly coupled to the amp, like I do with my Monoliths and custom center channel, the amp loading is pretty severe.

I’ve found that the Sunfire class-G amps are able to handle the punishment and drive these speakers cleanly to astounding SPL levels with minimal THD. The Sunfire design is able to produce the current flow into low impedances needed to drive the ESL effectively.

Using an active-crossover and full bi-amping, each panel gets 800w/4ohms of juice, and when the >10Khz dip down to 2ohms occurs, the amp is pulling massive amperage from the powerline to generate close to 1,600w

I’d also recommend looking into the sanders Sound System amps, and reading Roger Sanders great posts on this topic of amplification in this thread.

Bottom line is that SS can do ESL justice, you just need to use the right designs.

JonFo said:

Sorry to resurrect an old thread, but this topic is one I’ve been pretty intimate with for the past decade or so.

I have a set of MartinLogan’s that I’ve worked on over the years. See my site for details.

Regarding amplification challenges, yes, these definitely present huge challenges. But there are SS amps that deal with them just fine.
The key seems to be to use designs that flow enough current into the wildly swinging impedances AND that gracefully handle the back-EMF these speakers can generate.
Specifically, in speakers where the factory crossovers have been stripped out and the panel is run directly coupled to the amp, like I do with my Monoliths and custom center channel, the amp loading is pretty severe.

I’ve found that the Sunfire class-G amps are able to handle the punishment and drive these speakers cleanly to astounding SPL levels with minimal THD. The Sunfire design is able to produce the current flow into low impedances needed to drive the ESL effectively.

Using an active-crossover and full bi-amping, each panel gets 800w/4ohms of juice, and when the >10Khz dip down to 2ohms occurs, the amp is pulling massive amperage from the powerline to generate close to 1,600w

I’d also recommend looking into the sanders Sound System amps, and reading Roger Sanders great posts on this topic of amplification in this thread.

Bottom line is that SS can do ESL justice, you just need to use the right designs.

Click to expand...

ESL's love power. Period. There are much better sounding amps than what Sanders makes, but they also cost more money. One just has to listen.

amirm said:

Here is then what we know:

1. Most power is required at low frequencies. In the example I had, below 80 Hz, the impedance is 16 ohms or lower. So in this situation, the SS amp wins.

2. Highest frequencies require the least power. So even though the impedance of that example rose up 16 ohms at 20 Khz, the power advantage is not as material there (assume tube wins at 16 ohms).

3. The peak is at 90 to 100 Hz. So in mid base tubes would do better.

4. In the critical mid-bad frequencies of 2 to 3 Khz where the ear is the most sensitive, impedance is actually quite low at 6 ohms. So again, looks like SS does better which is strange because most people like tubes for their midrange performance.

So overall, I like the point he is making but having trouble realizing it with the one graph I found.

Click to expand...

The Quad is in fact a great example, but the first thing to understand is that most of our amps have no problem with 8 ohms at all. What you will find in practice that a SS amp will not keep up in the bass. That peak in the bass will kill all the power!

OTOH, the decreasing impedance of the speaker below the peak is not a problem for most OTLs on account of the fact that the bandwidth of the speaker itself becomes an issue below 30Hz.

At any rate, Martin Logans are notorious for dropping their impedance below 1 ohm at 20kHz, so perhaps we can look at those too?

Click to expand...

Martin Logan has been working to make their speakers compatible with transistors. The problem here is that the capacitive aspect of the impedance curve still causes a transistor amp to make too much power at high frequencies without enough at lower frequencies. IOW just morphing the impedance lower as they have done does not fix the problem. The solution to driving MLs properly is to get a set of ZEROs and use a tube amp (or amplifier with a constant-power characteristic). Then you can get the constant power requirement that the speaker needs, while at the same time having drive-ability.

When we look at this the ELS is just impedance friendly at low frequency and the matching transformer must correct this - in short you need to equalize the impedance from the ELS for the amplifiers.

Click to expand...

As long as the impedance is above a certain minimum amount, in the case of an OTL (or other amplifier with a constant-power characteristic) the power curve will appear to be constant. So the vast peaks that are characteristic of some ESLs simply is not a concern, so along as the dips are not below the amplifier's ability to cope. This is why the matching transformer is important- its job is to place the impedance curve within a window that makes the amplifier useful.

This is why the ESL 63 has been for so many decades to be an ideal match with OTLs, as its impedance curve was kept above a certain minimum value. The peaks don't matter.

The issue with feedback as a design element in an amplifier is tricky to say the least. Its nice to use it to reduce distortion and output impedance in theory, but in practice it can be seen to not be very effective.

The issue with distortion is that if the amp has a propagation delay (and all amplifiers do) then the loop feedback will not arrive back at the input in order to correct the signal that generated it. Its usually just slightly too late. As frequency is increased (since the propagation delay is constant with frequency), the problem is exacerbated. At any rate the result of feedback is that the 5th, 7th and 9 th harmonics are slightly distorted, often less than 0.01%, but since the human ear/brain system uses these harmonics in determination of sound pressure, a small amount like that is readily audible. Audiophiles have terms for such distortion: bright, harsh, hard, clinical, etc.

Now if you want something to sound real, since our ability to gauge volume is one of the more important aspects of human hearing, its pretty important to *not* alter the odd harmonics. Loop feedback flies right in the face of that. IOW its not a 'religious' ideal, its simply something that does not work.

Of course eliminating feedback makes getting rid of distortion a lot trickier! But that is not to say it cannot be done. So what this 'constant power' concept is about is understanding how the amplifier/speaker interface is likely to behave with amplifiers that have been optimized to obey human hearing rules.

The CLS Impedance urve form Stereophile.

Atmasphere said:

The Quad is in fact a great example, but the first thing to understand is that most of our amps have no problem with 8 ohms at all. What you will find in practice that a SS amp will not keep up in the bass. That peak in the bass will kill all the power!

Click to expand...

Hi Atmasphere,

Can you please clarify what you mean by this? I have found most tube amps I have heard, except for the large VTL's, have problems producing the bass. Powerful SS amps, such as Bryston or Krell, however, seem to do this very well.

caesar said:

Hi Atmasphere,

Can you please clarify what you mean by this? I have found most tube amps I have heard, except for the large VTL's, have problems producing the bass. Powerful SS amps, such as Bryston or Krell, however, seem to do this very well.

Click to expand...

Originally Posted by Atmasphere
The Quad is in fact a great example, but the first thing to understand is that most of our amps have no problem with 8 ohms at all. What you will find in practice that a SS amp will not keep up in the bass. That peak in the bass will kill all the power!

ack said:

Hi Ralph,

I drive my MLs with Spectral amps, and I know I am not the only one - the fellow I sold my DMA-250 amp to last year bought it to drive his ML Prodigies after listening to a friend of his drive his own Prodigies with the Spectral 360 monos to astounding results (I can attest to all of this, based on my own experience with my Odysseys). In addition, my dealer keeps talking about a beautiful setup another customer of his has, driving ML ReQuests with a DMA-260. So what do you think is going on here? Lots of negative feedback? I know for a fact that Spectral use modest amounts of feedback (probably not your ordinary SS design). What gives?

Click to expand...

Martin-Logans have a very low impedance, generally so low that most tube amps are not happy with them. This is because they want the speakers to work better with transistors- there is a bigger market. Its a challenge though, due to the decreasing impedance with frequency, by 20KHz some of their speakers are only 0.5 ohms! A lot of transistors won't double power into that impedance, so in a way they sort of get away with it. To drive ML with tubes, you need a set of ZEROs (http://www.zeroimpedance.com) to allow the amp to do the job. IME they are *not* power hungry.

caesar said:

(in reference to a comment about Quads) Hi Atmasphere,

Can you please clarify what you mean by this? I have found most tube amps I have heard, except for the large VTL's, have problems producing the bass. Powerful SS amps, such as Bryston or Krell, however, seem to do this very well.

Click to expand...

Quads have a high impedance peak that makes it hard for many amps to make power. When you have a transformer-coupled tube amp, the transformer actually does what it says- transforms impedance. So if you are on the 8 ohm tap and the speaker has a bass peak of 45 ohms, the tubes (which might be looking for 3000 ohms to make power) are now looking at more like 15,000 ohms; there isn't enough voltage in the amp's power supply to allow the amp to make power into that impedance.

Transistors have a similar issue but without the transformer. That is why so many people wind up putting really high power amps on the Quad, even though it does not take that much to make it play easily.