MC352 into varying-load speaker?


Subject says it all... how well does the autoformer handle a nominal 4 ohm speaker which is not flat? I can't seem to get a good understanding of this situation with my limited technical knowledge of this sort of thing.

Speakers are Usher 8871 in a large space. They are rated at 90db, 4 ohms but seem to be much hungrier. I onced owned a Mac 6500 (200 w into 4 ohms) that while sounding great did not satisfy with classical music owing to power limitations.
kck
As with any speaker that uses a passive crossover, some of your power is lost in the crossover itself. On top of that, speakers that use contouring networks, notch filters, impedance compensation, etc... all end up sucking up more power. While the end result might be a speaker that measures very well in terms of amplitude linearity ( flatness of frequency response ), the sound produced is typically lifeless and lacking in "dynamic jump factor". That is, unless one feeds them with GOBS of power.

As far as impedances varying with frequency, all speakers do this to some extent. SS amps without output transformers deal with this by varying their available power in large fashion. That is, big impedance swings result in big differences in available power. Tube amps and / or SS amps with output transformers are also susceptable to impedance swings, but the transformer itself acts as part of the load. This tends to stabilize the available power to some extent, but not enough to provide equal power at any given impedance. The output variance that does occur with impedance changes are what making playing with the various output taps on the amp worthwhile. Not only can this change the available power output, but also the tonal balance and distortion characteristics.

With that in mind, Usher's are designed by Joe D'Appolito. Joe is a speaker guru and i'm sure that the crossover is relatively complex and parts heavy. This tells me that the impedance may not swing all that wildly compared to some speakers, but that it may take considerably more power than the speakers that are simpler in design. If it were my system, i would not hesitate to play with the output taps and see what worked best for your specific speakers and listening tastes. Sean
>
Better off with the autoformer than not. The 352 will be much more powerful than the 6500. Arthur
Sean, thank you for the excellent analysis and background info. I will try to talk with someone at Usher Monday. I do not have the Mac but am contemplating purchasing it used, but at its weight and cost it is something I would like some more knowledge on before committing to.
Did you imply that even with a rating of 4 ohm the Ushers just may sound better on the 8 ohm or 2 ohm taps and the only way to know is to try it?
Also, without getting too much into the search function again (I am really tired of doing that) do I recall correctly that at times you might have been a little critical of the Mac design philosophy? Forgive me if mistaken, but if not I would like to understand why, with a view to gaining knowledge prior to purchase.
I drove Infinity kappa 7's with my 352 for years. The Kappas are all over the impedance road. I have heard from 2-32 ohms. At least that is various dealers told me. But the 352 came through flawlessly. The autoformers seem to adapt with no problem.
Kck: I am not a fan of "autoformers" in the least. To me, they are blasphemous. If you really want to know my thoughts on them, fire away with questions. Just don't expect "politically correct" answers. Sean
>
Sean, confused. Your first post in this thread seemed to imply that autoformers could work. But the blasphemy comment is pretty strong in the opposite direction.

Not knowing where to begin picking your brain, perhaps you could summarize the idea, if it's not asking too much. Essentially, the question would be: "Do autoformers hurt, help or are neutral to sonic quality under most/average circumstances, and in what circumstances do they veer from this stated path"?
Kck: Autoformers act as a buffer. If you have an anemic amp and a speaker system that presents a wild and highly reactive impedance as a load, that "buffer" could be a good thing. Otherwise, that "buffer" acts as a sonic veil between the amp and speaker interface, lowering the sound quality. Using an autoformer with a good amplifier would be equivalent to running hundreds of feet of small gauge highly inductive speaker cable. Would you ever consider doing this under normal circumstances??? Sean
>
The autoformer doesn't work like Sean thinks it does.. Mcintosh output transformers are bifilar wound which is a complex technique used to nearly eliminate the inductance associated with the wire. We use it in the lab for special EMI filters (different core setup though). The coupling is just about perfect with a larger bandwidth than the output transistors. Just like tube amps - only better because not all manufactuers have the ability to bifilar wind their trannies.
Aball: Has Mac always used a bifilar design or is that what they've gone to on more recent models? While i saw the test results / bandwidth on one of their more recent models, and it looked pretty impressive for an output transformer coupled model, i don't remember the older models being anywhere near as linear over that wide of a bandspread. As such, either their transformers have changed or the designs of their amps have improved with age.

Either way, the transformer still acts as a buffer. Rather than directly dealing with the input of the speaker and the reflected EMF that it would present, the output stage sees the transformer first with the speaker as a secondary load. As i mentioned, this increases the stability of the amp as it always sees a somewhat consistent load. That "consistent load" is also what gives these amps their "house sound" and electrical characteristics. If you like them and it will do what you want, buy it and enjoy it. Sean
>
Yes, all autoformer McIntoshes use bifilar winding (trifilar on some amps). McIntosh patented it in 1947 for their "unity-coupled" circuit, originally for tube amps of course but then they applied it to solid state in the early 60s and have ever since. I have performance data for a MC2255 and the bandwidth is nearly as wide as today's models.

I have tried Mc amps with autoformers and without, concluding that there is actually little sonic difference between them (when comparing appropriate models). The autoformer mainly acts as a stabilizer, as you said. I currently use a direct-coupled McIntosh amp myself. I like them either way. Arthur
The MC352 is a stereo amp which has two 225 watt amplifiers per channel that are fully balanced through the output Autoformer.The two amps in each channel are floated above chassis ground. The autoformer is a transformer with a coil of wire that ends in a ground point. The two amps feeding the autoformer produce one half of the sine wave each and are combined in the autoformer. The amps crossover point will float to maintain balance and if one of the amps is shorted out the other amp will complete the sine wave at its 225 watts. This design is called Quad balanced by McIntosh and will result in a signal to noise ratio of over 124 dB. This design of course uses common mode rejection.
Like a direct coupled amp the MC352 will double power if say a 2 ohm load is put on the four ohm tap. Heat and current draw will also go up. Unlike a direct coupled amp the MC352 will not halve power as the impedance is increased. With the MC you get all the power you pay for regardless of the speaker used.
The output autoformer allows the amp section to be designed for maximum linearity and efficiency which may result in say an impedance of 2.37 ohms. Since there are not a lot of 2.37 ohm speakers the autoformer is then taped to yield the 2,4, and 8 ohm point. This results in an increase in power over compromised direct coupled designs.
The autoformer has a greater bandwidth and power capability than the amp section which drives it. The MC will put out rated power to 100 KHz and swing high output current. The autoformer is an audio 'transmission' to match the amp to a variety of loads. In no way is a properly designed and built autoformer a 'buffer'.
ron-C
Appreciate all your inputs and thanks to Ron for joining in. I have a related question, that of Damping Factor for this amp, which seems fairly low at 40... what is your take on this? My present amp, 200 w into 8 and 400 into 4, boasts a DF of 400 and demonstrates that by exceptional basss control.

If I am on the wrong track please correct me... as stated above, I am interested in this amp but do not have the tech knowhow, hence the questions.
Dampening factor, slew rate, phase shift, and many other parameters are often mentioned as the key to great amplifier sound. You can find amps that excel in any one of these specs or not and the amp is still regarded as sounding good. The MC352 will put out over 400 watts which is enough for most speakers and may be to much for some. I suggest you try the amp with your speakers and to see if you are satisfied.

Ron-C
MC352 driving my Aerial 10T's sounds sweet. The 10T's are less efficient than the Usher's. The large-ish woofer requires an amp with some power and control. The MC352 sounds great with either Verve's Ella Fitzgerald re-releases, or hammering out some electric rock.

At absurd listening levels, the 10T's can suck the 352 dry. The fact that the 352 runs clean all the way into clipping, is a major compliment to its design and implementation. I do lust after the new 500W mono's though.....

I have heard "better" amps, but no where near this price range.
Ron-C: Thanks for the education and background. I learned something today. I wasn't aware that this was basically an amp that was bridged internally.

You made mention of the fact that the autoformer is linear to 100 KHz or so, which was wider in bandwidth than the amplifier itself. Out of curiosity, what is the rated power bandwidth of the amp itself?

Outside of that, i based my response on how the "average" transformer coupled amplifier is designed. I obviously overlooked the fact that Mac has done things differently and done so for a long period of time. Having said that, please look over my response below. Some of this is conjecture based on logical assumptions, but if i'm wrong in these "guesstimates", i'd like to be corrected.

Kck: Damping factor has to do with the impedance ratio between the output stage of the amp and the input impedance of the speaker. The closer that the output impedance of the amp is to the input impedance of the speaker, the lower the damping factor. This is why damping factor is rated at a given impedance i.e. you have to have a specific speaker load impedance to compare the output impedance of the amp to. Industry standard is typically 8 ohms, but i guess that an unscrupulous manufacturer could rate it at 16 ohms and not clarify the rating until asked : )

As a side note, the lower the damping factor, the more likely that the amp can be "modulated" by the reactance of the speaker. This is especially true if the speaker is of a low impedance design with a lot of reactance / long throw woofer. This is why many tube amps get "mushy" bass when coupled to very large, long throw woofers. The woofers themselves are generating enough reflected power that it modulates the output of the amp. The amp looses control over the woofers, resulting in "slop". In severe cases, the loss of control in the output stage is coupled back up into the earlier driver stages of the amp, causing further non-linearities to occur. In this type of situation, the negative feedback circuit ( if used globally ) is getting a real work-out.

While i'm guessing here, i would have to assume that the autoformer is somewhat responsible for the damping factor that was quoted above. If such is the case, that tells me that there is a certain amount of loss involved through the autoformer due to increased series resistance. The higher series resistance is what increases the output impedance of the amp, lowering the damping factor of the amp.

While the autoformer may be wound in a manner that minimizes inductance, the use of hundreds of feet of smaller gauge wire in series with the output of the amp WILL act as a "buffer". The increased series resistance associated with such a design will "soak up" or "absorb" signal, both going out to the speaker and / or from the reflected energy that the speaker tries to push back towards the amp. While this is not the "classic" definition of a "buffer", the lack of direct contact between the output stage of the amp and the speakers themselves could be interpreted as "buffering" the load. Sean
>
A great discussion that is a little over my head, but I am getting the gist. Hope Ron can respond. Anyone else want to have a go then?
Sean,
The frequency response of the MC352 is 20Hz to 20 KHz + 0, -.25 dB or 10Hz to 100 KHz + 0, -3DB. Since the amp is underrated at 350 watts per channel it will actually do the 350 at 100 KHz. With todays modern high speed output transistors this bandwidth is obtainable where the devices were the limiting factor until the early 90s. The McIntosh output autoformer uses a grain oriented steel core with a very large winding made up of groups of wire. The actual gage is high and these are terminated at about 12 gage total for each tap. Unlike a tube amp where the transformer is converting an impedance ratio of 125 to 1 in a McIntosh unity coupled tube amp, or 250 to 1 in other tube amps, the solid state autoformer is 4 to 1 or less depending on the tap.
If an amp is to be reliable and stable something will have to be placed across the output section, either a cap or choke coil or speaker wires with little boxes of Zobel networks in them. The autoformer is less intrusive and has the advantage of impedance matching. The drawback is cost and weight while the advantage is cool operation, long life, and output impedance flexibility.
In the case of the Quad Balanced amps like the MC352 the two amp sections per channel are balanced in the autoformer and not to chassis ground. If a direct coupled amp is 'balanced' it is actually bridged to chassis ground and will have a max signal to noise ratio of -112 db. In the case of the quad balanced amps the autoformer allows the ground point to float and -124 dB SN is achieved.
The McIntosh autoformer will introduce 0 degrees of phase shift at 20 Hz and less than 3 degrees at 20 KHz. The average volume control on a pre amp will introduce about 15 degrees of phase shift.
McIntosh has never made a big deal out of dampening factor. The old amps from the 50s typically had dampening factors of 10 or 20 while the new ones are rated at higher, 40 in the case of the MC352. The main concern with deep bass performance is usually power and lack of phase shift.
I have an old MI200 tube amp running a double 12 inch sub, This is a mono, 200 watt amp using transmitter 8005 triodes at 1000 volts. This amp has a dampening factor of 10 yet very few transistor amps will deliver the clean effortless bass of this amp. A low dampening factor does not mean the tail will wag the dog.

Ron-C
Ron: Thanks again for all of the info. This has been very educational for me, at least pertaining to Mac products, design philosophy and implimentation. I do have to take issue with some of the comments that you made and a few more questions if you don't mind.

"the solid state autoformer is 4 to 1 or less depending on the tap."

I'm not sure as to what the various impedance taps are on this specific amp. As such, it makes it rather difficult to calculate the appr output impedance of the amp. If you can provide me with what impedance taps are available on this math, i can reverse engineer the appr output impedance of the amp. Otherwise, if you can simply state the output impedance, that would be good too : )

"If an amp is to be reliable and stable something will have to be placed across the output section, either a cap or choke coil or speaker wires with little boxes of Zobel networks in them. The autoformer is less intrusive and has the advantage of impedance matching."

Depending on the purpose and hinge frequency of the Zobel being used, i'm quite certain that the autoformer is far more "intrusive" electrically speaking. Given that most amps only suffer from instability at frequencies well above the "directly audible" band, the use of a Zobel tuned well above that range doesn't come into play with what we would hear in the least. Can the same be said about the autoformer with the same amount of confidence?

In other words, if the autoformer were removed and the circuit revised accordingly, would the sound or electrical performance ( other than S/N ratio ) change? From experience, i know that one can remove a Zobel from what many consider to be an "unstable" amp and obtain the same performance, given that the load itself doesn't contribute to the instability of the amp.

"If a direct coupled amp is 'balanced' it is actually bridged to chassis ground and will have a max signal to noise ratio of -112 db. In the case of the quad balanced amps the autoformer allows the ground point to float and -124 dB SN is achieved."

This is a 12 dB increase in signal to noise ratio, which is equivalent to reducing the noise floor by 16 times. If there are no drawbacks sonically or electrically ( other than cost ), i would consider this a worthwhile design attribute. Then again, i'm not directly familiar enough with this design to know all of the pro's and con's. I am learning though : )

"The McIntosh autoformer will introduce 0 degrees of phase shift at 20 Hz and less than 3 degrees at 20 KHz. The average volume control on a pre amp will introduce about 15 degrees of phase shift."

I find this statement very interesting and wonder what / how the "average" 15* phase shift figure was arrived at? I have a preamp that was available to the public in 1974 that produces +1* / -3* of phase shift from 20 Hz to 20 KHz. While it used very high quality parts ( Mil-Spec Allen Bradley pots ) and a very advanced design for its' time, one has to assume that more modern designs that take advantage of the increases in technology should be able to at least equal if not supercede this spec.

"A low dampening factor does not mean the tail will wag the dog."

This statement is in direct contradiction with what every "bench jockey" or EE that has any form of experience with real world conditions will tell you. There is a direct correlation between output impedance / load impedance ( damping factor ) and the load stability of the amp. Not only can an insufficient damping factor alter frequency response, it can also play games with the amount of phase shift and distortion involved. This is evidenced by John Atkinson's test results and comments in many of his published works found in Stereophile's monthly reviews.

"I have an old MI200 tube amp running a double 12 inch sub, This is a mono, 200 watt amp using transmitter 8005 triodes at 1000 volts. This amp has a dampening factor of 10 yet very few transistor amps will deliver the clean effortless bass of this amp."

I think that you answered your own question here. That is, so long as the amp has sufficient current to sustain the high plate voltage mentioned while under full load, the amp itself will never run out of dynamic headroom and / or have any problems dealing with the much lower voltage levels that one would encounter with reflected EMF generated by the drivers. So long as there is always more forward voltage present, the lesser reflected voltage can't "modulate" the output of the amp*. This is why the amp delivers "thunder" while retaining "great control". The low damping factor would would come into play if the amp was voltage or current deficient, hence the lack of drawbacks of such a design with this specific "high dynamic reserve" aka "high headroom" unit.

Other than that, i appreciate the time that you've taken to respond to this thread and to my questions in particular. Out of curiosity, are you a "Mac enthusiast" or an employee / former employee of Mac itself? Sean
>

*For those that are interested in learning about this specific subject ( amplifier headroom vs reflected emf and the effects it has on driver control ), i would recommend reading the white papers that Bob Carver wrote up for the original Sunfire subwoofer. Bob does a good job of keeping things relatively easy to understand while trying to explain what is a very complex subject.
I may be able answer at least one of those questions, Sean... I believe Ron here is Ron Cornelius, present Product Manager of McIntosh Labs and de facto spokesman for the company.
Sean,

On the 4 to 1 point. If the amp is terminated at 2.3 ohms before the autoformer and the autoformer has taps for 2,4, and 8 ohms, the autoformer does not have a very high matching ratio. The 2.3 ohm point is where the output stage is most linear and needs the least feedback for maximum performance. This is not so much of an issue with the autoformer but would be without it as there is not a lot of 2.3 ohm speakers.
I did not say that a pre amp has phase shift but the audio-taper pot usually does. Fully balanced designs can avoid this but fully balanced volume controls are complex.
Yes the actual power of an amp is the main factor in controlling bass. 10 feet of speaker wire will usually swamp most dampening factor specs. If the autoformer limited the output of a Mc amp they would not put out high power but they obviously do. A properly designed autoformer will pass all of the watts. Our MC1201 amp is rated at 1200 watts. It will sine wave about 3000 when a two ohm load is on the 8 ohm taps. On a ten millisecond tone burst under the same load it will swing 5000 watts. This is a lot of power. If one is aggressive with this amount of power we have seen voice coils pulled off drivers and the solder melted on crossover boards to where the parts fell off. The autoformer does not limit the amp's output.
McIntosh amps will roll off above 100 KHz by design. If an amp starts to pick up very high frequencies say from induction on the speaker wires this will get into the feedback network and may cause ultrasonic activity. Output transistors do not like very high power at very high frequencies. This is why amps are protected with coils or caps on the output stage.
In explaining the McIntosh engineering philosophy I am only responding to this thread. Most high end audio products are well designed and many companies other than McIntosh make fine amplifiers. The autoformer, Sentry Monitor, and Power Guard are unique to the McIntosh design and have tangible advantages all at a cost of course.
Thanks again,
Ron-C

PS- Yes I work for McIntosh and have various titles my favorite though is Historical Advisor. Now about that MC60..
Ron: I appreciate your responses. While we could continue down this path for quite some time, i'm glad that you've at least taken the time to share what you did with us. It is enlightening to know that there are still manufacturers around that don't have much to hide and are willing to discuss the design decisions and technology that went into their products. It is obvious that the 1201 is built VERY sturdily and can pass a lot of current. I do have further questions that i'd like to ask you about this and other Mac designs, but i think that you've been quite gracious with what you have provided so far.

With that in mind, i'd like to pass on some further observations that are slightly off the subject, but i think need attention brought to them. Given our "basic agreement" on most points after clarification, i don't think that you'll disagree with the points that i'm going to make. I could be wrong though : )

"Yes the actual power of an amp is the main factor in controlling bass. 10 feet of speaker wire will usually swamp most dampening factor specs."

I agree with this statement and have stated so in previous amplifier / speaker cable / speaker related threads. This has to do with the fact that most speaker cables are very poorly designed and present such a high nominal impedance to the amp. Length of speaker cables isn't nearly as critical or detrimental to the amplifier / speaker interface IF the speaker cables themselves are of a very low nominal impedance. In this regards, and i know that he's speaking on his own and not for Mac, i disagree with the comments that Roger Russell has posted on his website to a very great extent.

"If an amp starts to pick up very high frequencies say from induction on the speaker wires this will get into the feedback network and may cause ultrasonic activity."

This further proves the points that i've tried to make in this specific post and other threads. Just one more reason to avoid speaker cables that are inductive in nature. That inductance is also what is responsible for the higher than desired nominal impedance of the speaker cable. Chalk another one up for low inductance, low impedance speaker cables. That is, so long as they are properly terminated with an impedance compensation network : ) Sean
>
Thanks Ron for the explanations! Cool hearing about the details involved. Arthur
I have read this complete thread and am very intersted as I also own a 352. I have done my own experiments with a couple of signal generators and an oscilloscope. Simply, what I found was amazing. The waveform or beat harmonics of the two signal generators did not change all the way up to where it was so loud that the low frequencys shook the walls. The various wave forms varied only in amplitude and never in shape. The 352 did not change its face, not one bit. Probably not to good on the speakers but overall a good experiment. (I wore hearing protection) Last but not least, if there is ever a tornado around here, I am running to hold on to the Mc. That puppy wont budge!

-Mark