Using tube amp with electrostatic speakers.
What occurs to the sound of the speaker when the transformer saturates or begins to saturate? Is transformer saturation a voltage or current issue?Distortion- Its pretty obvious. Its a power issue- both current and voltage (of which neither can exist without the other). Are you saying:Not needed for the Toroid 1, needed for Toroid II 2. That the concern of transformer saturation does not occur (at higher input signals) because: We're talking here mainly about the Toroid 1: Bass energy can saturate the core of the part, because it was built with intention to optimize higher frequency performance. In practice it does not saturate with low frequency energy simply because its impedance is so high at lower frequencies (and there is a capacitor in series with its input) that most amps can't make any power at those impedances. The Toroid II: The crossover of the resistor and capacitor prevent lows from entering the input of the transformer; otherwise it too would saturate because its core is too small just like the Toroid 1. |
@lewm Yes, the tricky bit is that the transformer impedance increases as frequency goes down. So with a given capacitance with increasing impedance, the crossover point moves lower- there is no set crossover point! This is why the resistor is there, so that the crossover point is not a moving target- the resistor dominates the formula of the -3db point. Removing it theoretically would open the transformer up to saturation caused by too much bass, but in practice I've not heard of that happening. At any rate Dr. West seems to have it sorted with the newer Toroidal 2 version, and thanks goes to you for pointing out the problem with his simulation from years earlier. |
For the example given above where 10 ohms and 36uF values of R and C give you a X-over point of ~440 Hz, removing the R entirely would give you a high pass -3db point at ~4400 Hz. This would leave a big gaping hole in the midrange, because the bass transformer does not work much above 2kHz, based on my earlier measurements of my speaker. I am sure you are aware of all this, which is why I think I might be misunderstanding the nature of your recommendation. You could remove R and preserve the x-over point, if you add more C, in this example.The issue here is that the old toroid was not set up right. As you know, the crossover is a bit of a moving target because the impedance of the transformer is on a slope with respect to frequency (increasing as frequency goes down, which means that the existing capacitor in the crossover is actually allowing it to go lower than 500Hz- the resistor was there to sort of 'enforce' that there was a set crossover frequency). In practice (rather than theory) removing the resistor on the older original toroidal backplates (which were replaced by the toroidal 2 about 4 years ago) works out quite well. The big concern Sound Lab had was low frequency saturation of the toroid, but in practice its actual impedance was so far off that this isn't a problem with most amps. We have a number of customers that have done this and been very happy. No-one has reported a gap in the midrange because its not there. |
@jazzdude99 Andy, If you want to get an idea of what the Sound Labs can do with the new backplate, there is a resistor that can be removed from the old toroidal backplate that gets you about 90% of the way there. The resistor is usually composed of a bunch in series/parallel to get the wattage up- the result is a resistor with a 200 watt rating. Think about how much amplifier power is being absorbed to need to be rated that high to survive! Anyway, in the older units that resistor can be removed and an instant improvement is heard with any amp and the speaker is easier to drive. Our customers feedback is that they can do with our MA-1 (140 watts) what used to take a set of MA-2s (220 watts) before. Its more than just an efficiency improvement, its also a load that is a lot easier for any amp. |
ESLs on paper have low efficiency, but that is because the standard measurement is made at 1 meter. If the ESL is a large panel, a good deal of the sound might go past the microphone without being picked up. So in reality many ESLs (and planars in general) are much more efficient than their specs suggest. |
From your suggestion I might just stick with my SS amps. They have ample power (Innersound 800 ESL monoblocks (800W@8 ohms)). My journey to this thread was based on understanding if a lower power tube amp could be used on a speaker such as Soundlabs. I believe I understand the tradeoffs you have outlined.On a set of Sound Labs, a 200 watt tube amp can do what an 800 watt solid state can do. |
Normally you would use a higher tap. This is assuming the amplifier uses loop negative feedback, which most do. The feedback will help the amp to boost its output into the lower impedance. Since there is almost nothing for power requirements, it should be able to do this easily. So its really whether you use the 4 or 8 ohm tap and I would start with the 8 ohm tap. |
If you plan to use Martin Logans with a tube amplifier, you will want to investigate getting a set of ZEROs (www.zeroimpedance.com). These will help that tube amp play the speakers without high end rolloff. |
Are you PAYED BY THEM?No- I have no financial relationship with Paul Speltz or his company. He just makes a good product that I believe in. The latter statement is false when the loudspeaker has lower impedances nor is it limited to tube amplifiers (see below). It turns out that on almost any tube amplifier employing an output transformer that when using the 4 ohm tap, bandwidth (particularly in the bass) is reduced and distortion is increased, while total output power goes down slightly. It will be found that the transformer runs warmer too (this is where the loss in output power goes- it is used to heat the output transformer). This is because nearly all output transformers are less efficient into their lowest impedance tap. This is not only measurable but is also audible and that's as real as it gets. How many people use Zero’s anyway?Turns out that the ZEROs are used in many installations that don't involve OTLs. Some of the Martin Logans have impedances at 20KHz of only 0.5 ohms. That's a problem for most tube amplifiers even though there isn't much energy up there its likely you will hear a rolloff or power will be limited or both. The ZEROs solve the problem easily. Paul Speltz has a letter from Steve McCormick which states that while Steve's amplifiers can double power form 8 ohms to 4 ohms quite easily, they actually sound better driving 4 ohms through a set of ZEROs. Although power is cut in half (8 ohm power), its a simple fact that all amplifiers including solid state amplifiers like those of Steve McCormick *sound better* into higher impedances because they make less distortion. Martin Logan has always made ESLs, and ESLs traditionally have their best combination when used with tube amplifiers. This has been the case since ESLs were developed in the 1950s. However what ML recognized was that the transistor amplifier market is much larger than the tube market, so to make their speakers work with solid state better (greateer sales), they set up their matching transformers (similar to an output transformer in a tube amp- all ESLs have them) so that the speaker presents a 4 ohm load or less to the amplifier. This makes them an often very difficult load for a tube amplifier even if it has a 4 ohm tap, since with any ESL the impedance curve varies by about a 10:1 value from 20Hz to 20Khz. So a four ohm ESL will have about a 0.4 to 0.5 ohm impedance at 20KHz (Sound Labs, which have a maximum of about 30 ohms in the bass, tend to be about 3 ohms at 20Kz but can be as low at 1.5 ohms depending on the setting of the Brilliance control). The thing is, impedances like this are not really all that great for solid state either, especially with less expensive solid state amps that might lack for heatsink area. So a set of ZEROs can put them in much more comfortable (read: cooler running) territory! Plus they will have audibly and measurably less distortion. Now ESLs tend to be very fast and detailed loudspeakers, so if you can reduce distortion the speaker will reveal that very quickly by sounding smoother and more detailed. This is because the human ear/brain system converts distortion into tonality, and the distortion we are reducing by using the ZEROs is going to be the higher ordered harmonics which audiophiles call 'bright' and 'harsh' because of how the ear/brain system reacts to them. Now I mentioned that ML wanted to make their speakers more drivable with solid state by keeping their impedance low. To give you an idea of what is meant, look at a Quad ESL 57 or 63, or a set of Sound Labs. These ESLs have much higher impedances in the bass region- anywhere from 30 ohms (Sound Lab) to as much as 95 ohms (ESL 57). Solid state amps have troubles making power into those impedances and this is in the bass region where most of the power demands exist. So ML's solution was to reduce the impedance. To further illustrate, in the case of a Sound Lab, if using solid state usually people use amps that make 400-600 watts into 8 ohms. In the case of 600 watts, what will happen is that into 30 ohms of the Sound Lab ESL in the bass region the amp is going to make slightly more than 150 watts at full power. what this means is a tube amp of only 150 watts can keep up with a 600 watt solid state amp on a set of Sound Labs. Do you see the problem? That is why ML went for much lower impedances and that is why the ZERO is such a handy problem solver when working with them. |
One last thing, most ESL's are really not very neutral. Almost all can benefit significantly from digital EQ from the midrange on up. I find that ESLs can be quite neutral! You are correct though that if there is a problem on the amplifier side, it will be from the midrange on up that it most likely manifests. |
^^ to further build on that, many transistor amps can't continue to increase power into low impedance loads like 0.5 ohms. Further, the speaker cable itself becomes a significant factor when dealing with impedances like this!! This works in ML's favor to help tame high frequencies, because the impedance curve of any ESL is not also an efficiency curve. What I mean by this is that if you put a watt into the speaker at its four ohm impedance level for a given output you should not be seeing 8 watts into the 0.5 ohm impedance level to get that same output level. It is this 10:1 impedance ratio which is why tube amps are usually favored on ESLs as they tend to be less bright and there is a better ability to play bass. The brightness in this case is a simple frequency response error caused by a solid state amplifier's tendency to act like a voltage source (which is another way of saying that it tries to put out the same voltage regardless of what the load impedance is). By setting the impedance as low as they have, ML has limited the ability of most amps to be too bright, but I suspect various users will find that a lot depends on the amp and the speaker cables. |
Ralph, I’d still like to get your opinion on McIntosh amps with autoformers. Just in general, do you think they are a good sounding amp compared with other amps in their price ranges? Do they implement their autoformers well? Would they be a good choice for driving speakers with difficult impedance loads? Theoretically, of course. We all know that good sound depends on many factors in a given system.I was not aware that Mac was still making those; in the old days their autoformers worked quite well. But with those older amps I would not be asking them to drive anything much below 4 ohms. The only thing difficult about the MLs is usually their impedance. Beyond that its my opinion that its a Bad Idea to make any amplifier work hard by making it drive a 'difficult load'. What happens is you get more distortion, usually of the type that make the resulting sound harsher and less detailed. This is why I've recommended the ZERO so often in the past, as it allows the amplifier to drive an easier load. I don't think the Macs have impedance taps below 4 ohms but usually they do employ a fair amount of negative feedback so they should work alright on ML loudspeakers if not pushed too hard. |
Those effects will become increasingly pronounced as speaker impedance in that octave decreases relative to amplifier output impedance, everything else being equal.Al, as I mentioned earlier, the impedance curve of any ESL is not a graph of its efficiency (unlike a typical box speaker; this is due to the fact that the impedance curve is based on a capacitor). The fact of the matter is the efficiency curve looks a bit different and often ignores the impedance curve. This is why an amplifier that acts as a voltage source will tend to sound bright on an ESL and might not make so much bass. none of it negates the fact that ESLs and MagneticsThis statement is really questionable. If it were completely true a 15-watt amp would not do so well on Quad ESL57s (but they do- Quad used to make a 15-watt tube amp that was made for the ESL-57). Or a 60-watt amp on Quad ESL63s and Accoustats (we've sold lots of our M-60s to such owners). I've seen 70-watt amps play Sound Labs quite nicely. We sell a good number of our MA-1s (140 watts) to Sound Lab owners. We also have a number of customers running M-60s on Magnaplanars (although they use a set of ZEROs to allow the M-60 to deal with the impedance). The reason the Sanders needs such a powerful amp is that the speaker is inefficient. This is a conundrum with ESL designers- how to place the diaphragm... closely spaced so its easier to drive (might not handle excursion well)? Widely spaced to handle excursion (will be really inefficient)? A speaker that **requires** a 900 watt amp to drive it is next door to criminal (and FWIW the Sanders does not seem to need that sort of power). The last thing you want to do with **any** amplifier is make it work hard (such increases audible distortions that make the presentation harsher and less detailed; exactly the opposite of what you want on an ESL). If you've set up your system properly, your amps won't be working hard most of the time. |
But in my experience, a good dose of juice gives electrostatic loudspeakers (to stay on topic) a more effortless quality better kick in the lower registers....I've yet to hear a solid state amp play bass properly on an ESL, in particular Sound Labs and Quads. On transistors can make thump but have no definition (output impedance is too low). If you don't understand that last statement then you probably listen to a transistor amp. Transistors fail to make bass power on many ESLs (the ML and Sanders being exceptions as their overall impedance is set really low) simply because of the 10:1 impedance curve. Its simple physics- on many ESLs they simply run out of voltage (which is why the bass power is limited). At the same time the speaker is overdamped in the bass so the combination is weak bass without definition. Some people like it, but in the real world there is no such thing as 'tight' (thump but no body) bass. That's an audiophile thing; but if you listen to real, live music you won't be hearing 'tight' bass. |
by far these panels are far better controlled with the High Current of a Solid StateSome of this bit of text appears to be a bit open to interpretation; if my read on it is correct than the basic implied statement here seems to be false. A high current amp does not do most ESLs justice. This is simply because there are no high current requirements. Here's a bit of math: A Sound Lab, reproducing a tone at about 30Hz, with 200 watts to reach the desired sound pressure, will have a current flowing though it of about 2.6 amps. That's not a whole lot. The current myth is debunked. All that really happens with 'high current' amps (which is often a misleading descriptor in its own right) is a tendency to be bright, due to the 10:1 impedance curve exhibited by most ESLs. Dynamics: in 90% of audiophile conversation usually the word 'dynamics' refers to distortion rather than actual dynamics. Usually the distortion involved is higher ordered harmonics (5th and above) and often in trace amounts (as far as a distortion analyzer is concerned). The human ear/brain system converts such distortion into tonality and also uses it to determine how loud a sound is, so an amplifier that makes greater amounts of higher ordered harmonics often tends to sound louder and also sometimes more 'dynamic' than an amplifier that does not make these harmonics in such quantity. Its a long-known fact that tubes make less of these harmonics. However, the impedance of the speaker also plays a role. A 350 watt solid state amp driving a Sound Lab will make about the same power as a 90 watt tube amp due to the impedance curve of the speaker. So the final statement of the quoted text also appears to be highly questionable- a lot depends on the ESL being driven!! |
