Autoformers, The Benefits in matching amp to speaker

Much thanks for the information!
Well I am certainly not literate in EE so a bit fuzzy on all of this but I did recently purchase some MC601 monos but have not heard them yet. Is your definition of the issue the same one as Mac has in their design?

I would be happy to comment on particular Mac designs using Autoformers. If anyone has a question and can provide a link to a schematic that would be a big help. Im not having much luck finding the SS amp schematics. 

I did find an early Mac SS amp that they admitted ran in class B, that was not such a good idea and I doubt they do that now.

Ralph,

I agree your amplifier is fine for the Maggies as the impedance is constant however the lowness of it is hard on the tubes. I look at low impedances are like driving a short. I did put a one ohm tap in the RM-200 and it will indeed output 100 watts into a 1 ohm load with good damping. So why did you stop making the Autoformers?

I am curious in reading you white paper and posts here that while you admit that an ESL can have a 10 to 1 impedance range you feel that it is appropriate to run such a speaker with high impedance drive. I was working for Beveridge when we did the model 2 and 3 speaker and was involved in the specification of the transformer for both(I was not winding yet). While we came up with something that kinda worked for the system 3 we found that system  2 was impossible due to the bass extension down to 30 Hz while the system 3 only required 200 Hz.

The original Beveridge direct drive amplifier produced 1500 VA (similar to watts). We could not make a transformer or hardly find a high enough current ampifier at the time to produce 1500 VA. This was in 1978. I did find a Mitsubishi amp that produced 60 amps but getting all those VA through the transformer was another challenge. When I questioned Bev as to why one needs 1500 VA he said. "go play some trumpet music and measure the current". I measured 1500 VA! This is why I make 5,000 volt direct drive amplifiers for ESLs.

I think he is and I'm not sure what he means by "your poor design"?    

realdaeal, I don't understand this either  

in OTL design manufacturer don't mention about the specs about the tubes they used which is the most important in OTL designin,.transconductance of the tube is the sweetness of an OTL amplifier.they only built since the parts are cheap.

The dynamics of transformers are very good. Just think how many transformers are in the recording chain from the one in the MIC to the final output.

bdp24     I consider the old Quad (which I own, in addition to LFT-8b's
and Magneplanar Tympani T-IVa's) still the gold standard in that regard.
But the ET's play a lot louder!  

I agree the 57s are the gold standard for midrange due to the virtually massless diapragm and excellent step up transformer. I used them to voice my ESL. 

I tried to understand the OP, but I'm not getting it or something. How does a transformer only act on 180 degrees of a wave form? That doesn't make sense. And if the transformers is making the load easier to drive, how is it not taking away the control the amplifier has over the load? I get it with the shrinking class A envelope, but all that really does is drive a class A amp into class AB with a little lump of distortion at the envelope edge. And isn't a transformer really just feedback in the form of degeneration?

Although at first read is appears that kosst_amojan is indeed confused but as I am considering his points his answer does beg some clarification on my part. Along with some assumptions and further questions. So here we go. Thanks for your reply and I hope this clarifices this for others.

The transformer acts on the entire signal not just 180 degrees. What I was saying is that in a push pull amplifier there is one half of the primary used for the positive going signal and the other half used for the negative going signal. Making those equal is difficult above 45 KHz where feedback can cause ringing on one half and not the other half of a square wave. In an Autoformer the entire primary is used all the time.

In the application of making a speaker easier to drive we use the transformer in a step down configuration, the step up would be a rare but still valid usage. Stepping down the voltage by a factor of two doubles the current and current is what is lacking in all OTL amps. There is plenty of voltage. Although this discussion is based around impedance I would like readers to consider voltage and current. Impedance is just a simple way to combine the two into one term. 

By presenting a higher impedance to the amplifier the amplifier has MORE control over the speaker, control=damping. 

The  Aurtoformer actually extends the class A region when used to increase the impedance as in this discussion. Similarly going to a lower tap on a tube amp extends the class A region and can reduce the distortion by a factor of 5-10.

A transformer is not a form of feedback, Perhaps you are talking about what ARC and some others do when they use the secondary for cathode degeneration. I tried that in the early RM-200 and later found it was not so useful. 

09-01-2018 3:38pm

I'm building an active system where the midrange will be driven by a 2A3 SET. SETs like constant, high impedance. And need high efficiency too. Many high end midrange drivers have relatively flat yet low impedance. Would an autoformer be a good solution?

I would hope your 2A3 amplifier has a measured tap that is equal to the impedance of the driver you select. Low power ampifiers need this more than ever at 4 and 8 ohms. 

The popular Zero is just too much iron for a 2A3 to deal with. You would be hooking up a 200 watt transformer to a 2.5 watt amp. I do have a nice small autoformer that I made for my two tube OTL that would do the job nicely. Very low loss, easy to drive, 4 taps to match any driver.

To Lewinski and others thinking about multi amping please consider this.

The crossover in a typical two or three way speaker is entirely responsible for widely varying impedance, distortion and is just a bad idea in a high end system. Its fine if you need a speaker you can carry around and hook up anywhere, but we aren't doing that in our stationary home systems. We have the option of doing something much better.

Dynamic drivers have very flat impedance over their useful range. I know and hope I will get some disagreement on this, so go ahead.

For all the money people invest in expensive speakers and amplifiers there is a much better solution. If you don't care to make your own speakers then ask the maker whose speaker you like to "hold the crossover and give me direct terminals to the drivers".

Then get a good active crossover or have one made. Many of us offer them. Choose some amps approiate for those drivers like Lewinski is doing. You will have a great system, no impedance problems, no crossover saturation or the EQ tricks that are going on in most of them to obtain good frequency response at the expense of flat impedance.

I would hope that many of the readers here are tired of searching for the right amp for their speaker. I think that's a fools errand because the information is hard to find, some amplifier makers don't tell the truth, and most speaker makers assume you have a big SS amp with infinite damping, 60 amps of current just like they do. 

Here are a few examples of what you can do.

Drive your woofer with some inexpensive big SS amp, don't sweat the midrange, there wont be any here.

Drive the high efficiency midrange with a small tube of class A SS amp. Small amplifiers always sound better than big amps used at low power. 

Drive the tweeter similarly or use the midrange amp with a simple 6dB per octave passive crossover consisting of a choke and cap. Properly done these will not affect the drivers of the load. At these frequencies both the choke and cap are small, easy to obtain and have vanishingly small losses. Such a crossover will sum perfectly.

If you are into ESLs, DIrect drive is the way to go. 5,000 volts at 1/4 amp is 1250 VA and equivalent to a 1250  watt SS amp which is what many ESLs need.

tomic6011,278 posts09-03-2018 7:46pm

Roger - echoing the appreciation for joining this community and hopefully you will see value and continue.
i worked atProgressive Audio and we carried the Bev 2, magic when they worked !!!!

The model 2 and 2sw were great speakers and  the beginning of High End audio. I fixed the reliabity problem in the amplifiers and did the electronics for the 2SW. However the woofer itself was rather poor quality, had mass added, leaky cabinet, not a great woofer for such a nice speaker.

I enjoyed being Chief Engineer under Bev and we had a lot of fun there.

bifwynne2,044 posts09-03-2018 5:43pmRoger, your comments about speaker crossovers is very interesting. Passive crossovers present another issue: phase coherence problems. I tried to deal with the issue of phase coherence caused by higher order crossovers with a DEQX device. I inserted the DEQX between my linestage and amp. While the tonality seemed better, my system lost a unit or two of blood (metaphorically speaking), … something went missing. The DEQX twisted the signal to be phase coherent but took something away. I wound up taking the DEQX out of the signal path and decided to live with the phase issues but at least the music regained its former sparkle.

I have my doubts about digital crossovers and have no use for them. I was speaking of analog crossovers or simple RC filters that can easily be added inside the amp or in a little box. 

"An Autoformer or any transformer cannot fix a difficult load and only affects stability in a poorly designed, on the edge amplifier." 

What I am saying here is that a difficult load stays difficult. The transformer just  puts less stress on the amplifier but at the expense of less power available. 

An analogy might be the transmission in a car. When going up a steep hill (difficult load) we downshift to take the load off the engine but cannot go as fast. 

As to stability we note if we don't downshift the car sometimes shutters, engine knocks. That is like instability, which down shifting also cures. 

A transformer is very much like a transmission in that the proper gear ratio gives the most power for the given situation and is kind to the engine.

Should have just bought an automatic car to begin with. (The right amp to do the job)

But automatic transmissions still have gears. If there is a right amp for every speaker then does one have to get a new one with every speaker change?

I've designed all my amplifiers to be as universal as possible to sound and measure well with a wide variety of speakers. Music Reference owners tend to keep their amps for a long time, you don't see them often and there are many cases of Sellers Remorse.

I do this by reasonable choices of damping factor around 10, distortion below 1% at full power and typically 0.1% at listening levels. The RM-9 and RM-200 provide ample current, and are easy on the tubes. A resaonable amount of feedback is used to achieve these things. I don't care to use so much as to get the distortion down to 0.001% or the damping up to 100. Those will result in an unstable amplifier.

I would like to say one important thing about damping. Paul Klipsch said many times "Who cares if the amplifier is 0.1 ohm or 0.001 ohm output impedance. The woofer is a 6 ohm resistor, so how much do you think you are going to damp that". This thought of the amplifier having a grip on the woofer is another misconception so lets look at another way 

The following is intended to dispel some of the misconceptions on how dynamic speakers work. You may find it hard to accept.

Certainly a low damping amplifier is going to produce boomy bass in many speakers though not in all. The extent to which it does this is determined by the impedance peak at the bottom end. One peak for sealed boxes and two peaks plus a rather deep dip between the peaks in a ported speaker. What happens at these peaks. To the extent that the impedance rises to 20 or 30 or 50 ohms the voltage of a low damping amplifier will rise as much at 6-10 dB, a very noticable amount and a very narrow peak. It is this rise in voltage and rise in efficency at resonance that produces the peak.not that the woofer is out of control. It is perfectly in control and resonates because the air load on the cone now equals the mass load of the cone. If one designs a woofer without a impedance peak then damping doesn't matter in the bass. 

Here is the impedace curve of a well designed driver from Fostex. Note how flat the impedance curve is; only 8 to 10 ohms over the usable range.Of course the bass peak has to be handled by the proper enclosure. These curves are open baffle. Also note the 90 dB efficiency!

https://www.madisoundspeakerstore.com/approx-4-fullrange/fostex-fe108ez-4-full-range-sigma-series/

Now what happens in the rest of the range? The other drivers, if operated in their range of comfort, will present constant impedance in that range.Once you accept the fact that all dynamic drivers are mass loaded devices you will understand that. The impedance peaks and dips in a mulitway speaker are entirely due to driver overlap, driver underlap, and components added to voice the speaker. 

Here is a nice tweeter  https://www.madisoundspeakerstore.com/soft-dome-tweeters-peerless-vifa/peerless-dx20bf00-08-3/4-text...

One might now see that using a woofer down to its resonance is not such a good idea as that is beyond its usable range. We don't do that with  midranges or tweeters.

I use a lot of opamps in test equipment I make but not in audio except for a servo. Too may transistors in there.

I have a pair of modified Acoustat amps available with an all tube driver stage. No transistors anywhere in the signal path. I am also making new amps from scratch of my own all tube design.

I might eventually pick up an RM-10 for my Quad ESL's, the best amp in the world for that speaker.

Thanks for the compliment. The RM-10 is perfect for the Quads and you don't need the voltage limiter boards because it puts out just the right voltage. The damping is good so you will get the true response of the speaker. Perhaps this combination is so popular is because the Quads were my living room speaker when I developed the amplifier. Though I didn't do any particular things for the Quad and it ran the Vandersteens well too. 

Then I somehow learned that Roger would perform "triode mods" on the RM-9. This consisted on adding a switch for each output tube socket, permitting it to be switched to triode at the cost of 1/2 power. I had this done, switched all those sockets to triode, and went straight to heaven. That made a great-sounding amp into something really special, and the Vandy's fully conveyed that triode magic.

Triodes have inherently low output impedance which allows one to use little or no feedback. DIstortion is also lower. The industry only went to beam tubes and pentodes to get the most power, and it appears people still buy power as evidenced by the rediculously high power SS amps out there. There is no benefit in using an amp at 1% of its power level and many people are. 

Many people are suprised how loud they can play their speakers with the dimunitive RM-10 which I routinely pick up with one hand by grabboing the transformer cover. 

I will make triode amps all day long when people come to realize that many of them don't need 100 Watts. One really needs to look into that with something as simple as an AC voltmeter.

I was not previously aware of this ampifier or what they were up to.  It appears from the Stereophile measurements that NAD is compensating for the interacion of the switching amps output filter by modifying its drive signal while still in the digital domain. 

Here are the measurements, to see what is going on here may take some time and thinking , it was not immediately obvious. If there are questions I will be happy to look further into it. 

https://www.stereophile.com/content/nad-m2-direct-digital-integrated-amplifier-measurements

I found this interesting in the Levinson #53 review.  Michael Fremer notes:

 The transistors switching the positive and negative voltage rails to the load in Levinson's "class-I" topology are each connected to the load via a large air-cored inductor and to the opposite voltage rail via a diode.
  
Read more at https://www.stereophile.com/content/mark-levinson-no53-reference-monoblock-power-amplifier#FaZxXFFTa...

I am not sure how any of this applies to autotransformers, I will just say that neither the NAD or Levinson have any nor would likely benefit from one except for the following: 

One problem with switching amps is that the output filter is optimized for a particular resistive impedance. Unfortunately our speakers are neither resistive or constant so the fix is only nominal. I would love to see a swicthing amp's response to driving a 10uF capacitor which is typical of many electrostatic speakers. 

Since we are developing a class D maybe I can chip in.
The filter is there to reduce the amplitude of the switching component; IOW the switching frequency of the amplifier which is present at the output of the amp.

The switching frequency is never completely eliminated; what is still there is called 'the residual'. If the filter is doing its job, the residual will be a pure sine wave between about 1/2Volt to maybe 3 volts or so.

Thats much switching residual would bother me and not pass CE approval. 

While the capacitance of an ESL will affect the residual, the amplifier will otherwise behave as a voltage source, even if it employs no loop feedback.

Several uF will cause a large reduction of the residual. However the problem I am concerned about it the severe falloff in response due to the output filter. WIthout feedback I would think not a very good voltage source 

Class D amps, like all other amplifiers, have greater distortion into lower impedance loads. So an autoformer can help reduce distortion (while also decreasing output power) and it will attenuate the residual somewhat; a lot depends on the design of the autoformer as to how effective it might be in this regard.

I find class D amplifier are not like all other amplifiers and have special considerations which are obvious in published measurements.