Autoformer vs Speaker impedance Curve


Autoformers vs speakers with wild impedance curve swings (for instance; MC601 amp paired with B&W 802D3 speakers).

There’s a wealth of information about tube amp audio transformers interaction with speaker impedance, but I can’t find anything regarding Autoformer and speaker impedance/phase curve relationships. 

Can any techies enlighten me? 

Thanks!

(I tacked a similar post onto the end of a 10 year old thread but thought I might get a few more hits with a new thread.  Sorry for the redundancy)



73max
I design and manufacture audio transformers for all Music Reference amplifiers. I also make OTL amplifiers with and without autotransformers which actually extend the frequency response in the high end. 

Here are a few things to note. An autotranformer is not like a standard output transformer in that there is no need for HV insulation and the winding scheme is totally different. Where we interleave sections in an output transformer we can mulitfilar wind in a autoformer which reduces the leakage reactance markedly.. 

Also, in a 2 to 1 (4 to 1 impedance ratio) autoformer half the current is direct and half is transformed. So they are smaller and lower loss. 

As to McIntosh, their use is very clever and wise. If you note they drive the autoformer typically at the 2 ohm tap. This allows them to use lower rail voltages and stay out of the second breakdown region that all bipolar transistors suffer from. 

This is just a glance, if you want more just ask. As many of you know I have been doing this since 1980 and I find the art of transformers facsinating. Every design brings new challenges.
How fortunate are we in having two of the very best hi-fi amplifier designers/engineers working today (Ralph Karsten of Atma-Sphere and Roger Modjeski of Music Reference) sharing their knowledge with us here? That products of the quality they offer are made by not only such talented engineers, but also such nice guys, I am very grateful for. 
I note that this discussion has many questions concerning the use of an autotransformer with high damping solid state (voltage paradigm) amplifiers. I would say there is little to gain and a lot to lose using one. Here's why. This does not apply to the McIntosh amps and I previously stated their reasons for using autotransformers in their SS amps. 

First they place an almost zero DC resistance to the SS amplifier. That means a small offset voltage will create a large offset current which may be large enough to bring a torroid close to saturation. It takes very little DC to bother a non-gapped core. This is why tube amps should never have a torroid output transformer. 

Second, modern SS amps already have good current capability and low output impedance. They don't need any help

Third, good autoformers will do nothing to modify the wild impedance curve and reactive load that some speakers present. Some discussion of load line limiting in SS amps needs to enter here.

Heres what they will do. 

High output impedance OTL amps like the Atmasphere will benefit as their ideal load is typicaly higher than 8 ohms. My single pair OTL likes 64 ohms so I autoform it down to 16, 8 and 4 ohms and wrap 6db of feedback around the autoformer. 

All OTL amplifiers suffer from lack of current though there is plenty of voltage. Hey they are tubes... However the Futterman family has inherently low output impedance so the autoformer does not help with impedance but does exchange this excess voltage for improved current. A 2/1 autoformer will double the current available to the speaker. 


+1 Eric (bdp24).

We are privileged to have highly experienced and accomplished designers of outstanding audio electronics, such as Ralph and Roger, sharing their knowledge with us here.

Best regards,
-- Al 
@ramtubes
Hello Roger, welcome to the forum. Thanks for your posts!
We used to make an autoformer we called the Z-Music autoformer, which had taps for 4,3,2 and 1 ohm loads. With it, Steven Stone of TAS was able to listen to his Apogee Full Range loudspeakers with tubes for the first time. Although the Full Range had a very low impedance (not much more than 1 ohm) they were otherwise easy to drive as they had no serious phase angles and were reasonable efficiency.
I think you and I are on the same page here- if an amplifier is acting as a voltage source, it really doesn't matter if it has an autoformer in its output section, especially as in the case of the Mac, which includes the device in its feedback loop. If its a voltage paradigm device, it will make constant voltage into all (practical) loads by definition.



ramtubes
Second, modern SS amps already have good current capability and low output impedance. They "don’t need any help"

And I’ll add to that, they will also make that amp sound worse if used, as I’ve proved to myself with 3 or 4 amps that were very capable without the autoformers, they made the amps sound like they were cushioned, like looking through opaque glass instead of clear glass.

They are for amps "that need help", but I say save your money and get the right amp instead, or change the speaker to one that the amp in question is happy with.

Cheers George
A twist on the original question.  Numbers and graphs aside, what we hear is the ultimate arbiter. 

With a tube audio transformer:

What happens to what we hear when the speaker impedance dips hard at a specific frequency? Will that “frequency” be heard louder or softer?

...and the opposite. What happens to what we hear when the impedance spikes?  Will that frequency be heard louder or softer. 

Same question, but with an Autoformer?
For both transformers and autoformers, softer on dips and louder on peaks, ***relative to the response that the particular speaker would provide if the amp were an ideal voltage source, meaning a voltage source having an output impedance of zero.*** However in the case of an autoformer (which would be used in a solid state design) the differences will be extremely small, and essentially negligible with most speakers. Not because of the difference between an autoformer and a transformer, but because of the differences in output impedance between tube amps and nearly all solid state amps. As I said in my post dated 8-16-2018:

A major factor contributing to this, and what is probably the most major factor in many cases, is not the output transformer itself, but the interaction of the output impedance of a tube amp with the speaker impedance variations that you are referring to. In contrast to nearly all solid state amps, most tube amps have output impedances that are a significant fraction of speaker impedance, usually somewhere between a large fraction of an ohm and several ohms. That in turn causes the voltage divider effect to have significant effects on tonality, to the extent that the speaker’s impedance varies as a function of frequency.

In the case of McIntosh solid state amps which use autoformers that particular effect is essentially negligible with most speakers, because as a consequence of being solid state their output impedance is much smaller than the output impedance of most tube amps. (Although that certainly does not mean that an amplifier having low output impedance is necessarily the best match for a given speaker, in terms of tonality). For example the MC302 has a specified damping factor of "greater than 40," which for the 8 ohm tap theoretically corresponds to an output impedance of less than 8/40 = 0.2 ohms.
Regards,
-- Al
Thanks Al.  Guess that explains the bloated bass when I run 802D3s on my tube amp. 
Guess that explains the bloated bass when I run 802D3s on my tube amp.
No- it doesn't. If the amp is able to act as a voltage source, then it can make enough energy to be flat on the 4 ohm load of your woofer array. Since you say the bass is bloated, its doing that, so the bass bloat is something else. One possibility is that like at amps, yours is making more distortion into 4 ohms. The extra distortion could easily be perceived as bass bloat, since the 2nd harmonic is likely predominant. 

It could also be the power cord, as a cord that limited the amplifier's ability to replenish its power supplies might come off with more distortion as well. This effect is quite measurable- I've seen power cords rob a tube amp of nearly 30% of its total power!
Thanks atmasphere. Maybe bloated isn’t the correct word. The bass is markedly more pronounced from my tube amp (60 Watt push pull, KT88, “some” local negative feedback) than my SS CA-2300. It’s also seems “looser” or less defined.

Nonetheless, it seems reasonable to my novice brain that if an increase in speaker impedance increases the sound level “heard”,  that the bass should sound louder where the impedance spikes. No?
73max OP
bloated bass when I run 802D3s on my tube amp
60 Watt push pull, KT88, “some” local negative feedback

Without global feedback to keep the output impedance low, you probably have quite high output impedance giving you very mediocre damping factor, and no control over the bass hence your term "bloated" compared to the control the Classe has over the bass.

You could always bi-amp with the tubes on the mids/highs and the Classe on the bass, but you’ll need a passive volume control on the input of louder of the two amps to reduce and set it so it’s the same gain as the other amp.

Then use your master volume control on the main pre or dac to vary the whole (both amps) volume.

Cheers George
The bass is markedly more pronounced from my tube amp (60 Watt push pull, KT88, “some” local negative feedback) than my SS CA-2300. It’s also seems “looser” or less defined.

Nonetheless, it seems reasonable to my novice brain that if an increase in speaker impedance increases the sound level “heard”,  that the bass should sound louder where the impedance spikes. No?
If the tube amp is able to act as a voltage source, then it should not make any more bass than the solid state amp. In your speaker, there are two bass impedance peaks, which represent the box resonance with the port. If the amp is behaving as a voltage source, it will make less power, not more, into these peaks. Otherwise the speaker is a fairly benign load for a tube amp- our amps would do fairly well on that speaker (seems to me we have customers with them too).

Solid state amps often have what is called 'tight' bass, but I've yet to encounter tight bass at any show I've attended. IMO/IME tight bass is a symptom of an over-damped speaker. Here is an older article, written by the head engineer at Electro-Voice back when the industry was trying to sort out the voltage rules (EV and Mac lead the charge on that one):

http://www.dissident-audio.com/Loudspeakers/CriticalLSDamping.pdf

As you can see from the article, not all speakers need high damping factors and there are some that need really low damping factors. That is still true today and is why equipment matching is still an on-going conversation!
73max OP 
Guess that explains the bloated bass when I run 802D3s on my tube amp.

From Stereophile:
" 802 D3's 8, because the bass was entirely devoid of bloat or emphasis or resonance. Despite this, there was no lack of authority when that was called for, and the speaker's reproduction of deep bass was formidable."
This was achieved with Parasound Halo solid state, and Theta Dreadnought, also solid state. 
The big MacIntosh 303 with output transformer was kindly referred as being leaning towards the power, which to me "kindly" says "bloat" (remember MacIntosh are huge advertisers in Stereophile.)   

 
Also this is said about the design of the 802's bass alignment.
"the 802 D3's low-frequency alignment is free from underdamped boom."


  And this is the load the 802's present to the amp. especially in the bass
The magnitude drops to 3 ohms between 100 and 130Hz, and again between 670 and 770Hz.
There is also a combination of 4 ohms and –64° at 69Hz, implying that this speaker does require an amplifier that is not upset by a low effective impedance.
This last paragraph will represent a EPDR load close to 2ohms to the amp.

As I said before with what you have without resorting to buying anything else, go with the bi-amping, your Classe on the bottom for control of the the bass and whatever the the tube amp you have on the mids/highs.

Cheers George
 
  
This is an interesting article.   http://www.dissident-audio.com/Loudspeakers/CriticalLSDamping.pdf

I like what he is doing in figures 2 and 5. However this is just about the woofer. There is no consideration to speakers that have a wide impedance variation over their range. In 1954 speakers had simpler crossovers without the frequency response modifying parts we see today. With at woofer and a tweeter all you need is a capacitor for the tweeter and wind enough inductance in the voice coil so its impedance rises markedly with frequency. 

Once again I encourage people to take off their damping blinders and look at what is going on in the interaction between impedance of the speaker vs the amplifier.

atmasphere. 

  It could also be the power cord, as a cord that limited the amplifier's ability to replenish its power supplies might come off with more distortion as well. This effect is quite measurable- I've seen power cords
rob a tube amp of nearly 30% of its total power!


I would like to know how to measure that. 30% is hard to imagine. I test amplifiers all the time with ordinary power cords and have never seen this. This reminds me too much of Bruce Brission's white paper which was a total embarassment to his distributor who promptly dropped his line.
I would like to know how to measure that. 30% is hard to imagine.
+1
Back in 2010 I asked Ralph about that statement in this thread.  Here are excerpts from our discussion:

Atmasphere: Power cords: a 2V drop across a power cord can rob a tube amplifier of as much as 40% of its output power! Cripes! You're trying to say you can't hear that?? [That comment was addressed to someone else].  So this is very measurable and audible as well. On lesser transistor amps, a power cord will be less audible as the drop across the cable is reduced, but a class A transistor amp will easily bring out cable weaknesses.

Almarg: Ralph, could you provide a technical explanation of why that would be so? I don't doubt your statement, but I'm interested in understanding why that would occur.... Re your other points, all of which strike me as excellent, I think that it should be stated that none of those points NECESSARILY mean, to cite an example, that a $2,000 power cord will outperform a $200 power cord in any given system.

Atmasphere: Al, the reason a power cord can have this effect is simple. If there is a 2 volt drop in a power cord, the filaments of the tubes will run cooler and the B+ will be reduced. Since this is a voltage, the result is we get less voltage output out of the amp. Less voltage=less power. Depending on the amp this can be pretty profound. and I have seen it with my own eyes. I do agree though that that does not justify a $2000 power cord, but it **does** justify one that has decent connectors and conductors that will not heat up at all. That has to cost something, probably not $2000 though. One thing about audio is that if there is a phenomena, there is also snake oil for it.

Almarg: Thanks, Ralph. That would also seem to say that the value of the ac line voltage at each listener's location can be a very significant variable in the performance of a tube amp (assuming it does not have regulated filament and B+ supplies). Which in turn emphasizes how easy it can sometimes be for extraneous variables to lead to incorrect sonic assessments.

Atmasphere: No doubt! But it extends to anything that can draw significant power- and bigger transistor amps can! Imagine the peaks just... not... making it.

This taught us a lesson... when we set up an amplifier for test, we test the AC line voltage from the IEC connector. The meter on the variac (a bit of test instrumentation) cannot be trusted. 

A 2 volt drop in a power cord of typical length seems like a lot, of course.  But if the amp draws say 5 amperes a resistance of 0.4 ohms, including the resistance of the connectors and the contacts as well as the combined resistance of the hot and neutral conductors, would do it.  I'd imagine that could happen in some situations.

Regards,
-- Al  
I am enjoying the debate among experts, but am still not understanding how an autoformer works. The last question from the OP was a good one.  I Think including ss adds to the confusion and would prefer explanations stick to tube amplifiers where they are more practical.

How does an autoformer work? Does the speaker cable mate with the windings? How is the signal changed at this point?
@recluse---look at the 8-29 and 8-31 postings by ramtubes (Roger Modjeski of Music Reference) above.
I would like to know how to measure that. 30% is hard to imagine. I test amplifiers all the time with ordinary power cords and have never seen this. This reminds me too much of Bruce Brission's white paper which was a total embarassment to his distributor who promptly dropped his line.
Set up the amp on a variac with an appropriate speaker load and sine wave input. Measure the voltage drop across the power cord with the amp at full power. A 3 1/2 digit DVM is sufficient for this.  Then boost the variac voltage to compensate for the voltage loss. You may have to do this several times because as you increase the output power of the amp, the load on the variac increases too.

This came to our attention with our MA-1 amplifier back about 1998 or so. Bascomb King had measured the amp as only 100 watts where we get 140. I was wondering where the measurement error was and found it easily enough. We spec the amp at 120 volts (which is what Bascomb claimed he had) since these days that's more common than 117V. We normally set the incoming voltage by measuring it at the IEC connection, since there is typically a voltage sag in our variac. We have IEC connections on all of our products since we know that audiophiles will want to use or audition different power cords from what we supply; the idea that power cords make an audible difference is nothing new- its been around for decades (this is also why our MA-2 employs *two* power cords per chassis, to reduce the effect of the power cord). So the power cord isn't regarded as part of the amplifier in our test. Bascomb King didn't do it that way- thus the difference. By our measurements he was seeing about 117 volts at the AC input of the amp.

So it follows that the heavier the load on the power cord, the more audible the differences between power cords will become. This ignores the higher frequency aspects of any power cord's current delivery of course, but cynics of the fact that power cords can make a difference are usually silenced by the simple revelation that they never bothered to measure the power cord's voltage drop! It is easier to measure the effects the cord has on the product with which it is being used.
For example on our MP-1 preamp, which has heavily regulated power supplies, even though we can measure differences between power cords, we can't measure any differences in the performance of the preamp, nor can we hear any differences. So we can also conclude that regulation in the circuit plays a role here as well.
Ralph, would I be correct in thinking that the low B+ voltages that must be used for the 6AS7G power tubes in your amps, compared to the much higher B+ that is supplied to the tube types used in most non-OTL amps, would tend to increase the degree to which output power capability is sensitive to AC supply voltage variations?

Thanks.  Best regards,
-- Al
@almarg
No. If that were the only factor transistor amps would be more sensitive than tube amps in this regard, and I don't see that happening (although transistor amps don't have a filament circuit to cool off). I suspect Kirchoff's law might be in play there. But we use triodes and they are usually more susceptible to the actual B+ level due to reduced plate resistance and lower mu.
bdp24  

Thanks for citing those replies. I hope I have covered all if it but new questions are welcome. 

Several posters think a autoformer is evidence of a bad amplifier design. I don't think that at all though we have to be aware how a particular amplifier responds to a particular load. OTL amplifiers with reasonable number of tubes want to see 16 ohm and above speakers. So my recommendation is either get a high impedance speaker or get an autoformer to make your speaker high impedance. 

Several European radio manufacturers built OTL amps into common table radios. In their wisdom they used a 600 ohm speaker. Thats really cool.

McIntosh's use of an autoformer recognized that transistors like low impedances. If someone came to me and said "what impedance would you want to design an ideal SS amp"? I would say 1-2 ohms. That is exactly what McIntosh did.
So far I see this discussion has entirely ignored the resistance of the wire back to Hoover Dam. You can very easily measure this by plugging in a 1000-1500 watt space heater and noting the voltage drop. I do this when I look at buying a new house. Its so easy, the agents are amused, and you find out the actual resistance of your wall socket .... back to Hoover Dam. Sadly the lake is almost dry so I have to find another source to cite. I just like the sound of HOOVER DAM. Nice clean water powered electricty. 

The best I have found is a drop of about 3 volts at 1200 watts so that is 10 amps and the resistance is therefore 3/10 or 0.3 ohms. A poor house might have 10 volts, really bad wiring, which would be one ohm back to the Grid. It is pretty amazing to consider how you can get 0.3 ohms over all that distance, but transformers and 500,000 V transmission lines do the trick.

Now everyone knows their voltage fluctuates and frankly a good amplifier doesn't much care. The power loss with voltage in any well designed amplifier is related to V squared. If we loose 2 % voltage we loose 4 % power not 40%. Something is not right here. Good tubes operate with at plus or minus 5% filament range with no problem at all. No loss of power, none. Weak tubes are another story, but not good fresh tubes. I do this test on every batch of RAM tubes running the heater from 5.7 to 6.9 volts. The change in characteristcs is vanishingly small, within experimental error. 

However, now and then I do find a bad batch. I test in batches of 100 or more and one time I found some KT88s from Russia that went all over the place with filament variation. The provider of these tubes said they had sold thousands of that lot to... well im not gonna embarass anyone, but you know, the big guys, and nobody complained. 

So back to nuts and bolts. I don't believe this post of 40% and if it is true you had better go buy some kind of regulatiing power conditioner. The big one from Monster with the motor driven variac is quite good and I have used one at locations where I have large and frequent line variations. Otherwise I'm turning the variac all the time. My daytime voltage averaged 117 and 125 at night.

The resistance of a 5 ft, 16 ga power cord is 0.04 ohms not 0.4. Keep in mind that the primary of the power tansrformer in a typical large amplifier is wound with 50-100 feet of 16 ga magnet wire and and then another equal amount in the secondary. That's 10 to 20 times the resistance of a $5 power cord. The total equivalent resistance of a 500 watt transformer, referred to the primary is often 1-2 ohms.

Do what you will with power cords, but what are you going to do about the wire all the way back to Hoover Dam.

Bascom King (just one b please) is fine friend from Santa Barbara. I have visited his home and lab many times as he has visited mine. His lab is in an out building and I doubt with high current lines. We should all use a variac to measure amplifier power to make it a level playing field. In addition to that I like to run the line up and down at least 5% to see what happens to the bias and distortion. Often bias is very unstable which explains why high line can be a problem. Some lines these days run 125V which may be a problem for on the edge amplifiers. I have seen amplifiers (not mine) where a 5% rise in line causes a 10-20 % rise in bias current which added to the voltage rise causes a 15-30% or more rise in dissipation and possible thermal runaway. 

Theres a lot more to say about this but lets at least get real about the numbers. I don't see how Kirchoff applies though Thevenin is how one analyzes these situations. Kirchoff is for circuits with lots of nodes and branches. This is simple series resistance. 

 




almarg8,371 posts09-14-2018 1:08pmRalph, would I be correct in thinking that the low B+ voltages that must be used for the 6AS7G power tubes in your amps, compared to the much higher B+ that is supplied to the tube types used in most non-OTL amps, would tend to increase the degree to which output power capability is sensitive to AC supply voltage variations?

This is actually a good thought and I agree with you. Because the saturation (minimum voltage at some current) of the 6AS7 is going to be a large (say 50) percent of the B+. Much larger than in typical amps where is is 20-30%.

The RM-10 uses 720 volts on the EL-84 plates and the saturation is around 100 V which is only 14%. This is why it can produce 40 watts per channel. (50 at the plates).
Roger, I recommend you try some measurements on this bit yourself. Its obvious to me that you have not. Its simple enough to do. Bascom is a nice guy and I've always had respect for him, but he screwed this one up. People are human. 

Its a common question as to what about the wiring in the wall and such? Part of that is what substations are for; to make sure that power at a local level is the correct voltage. In tandem with that are the power transformers mounted on power poles that serve local nodes. The one on the pole in my back yard serves 5 houses. The AC power sent to that transformer is at a higher voltage; this is done to minimize current loss thru the power lines. As you might know, a different waveform is employed on high voltage transmission lines (like the ones which come from the Hoover Dam; that is why you can hear them on the AM radio if you get close to them) for this same reason. This of course is in a nutshell.

Once in the house, voltage drops can indeed occur, but the Romex found in most homes is very high performance. The problem is how to get the power out of the wall, since Romex as you know is solid core and very stiff; thus being illegal for use in a power cord. IOW the real challenge is how to build a power cord that does the job properly and still be flexible and legal. Its doesn't have to be expensive- it just has to be right. That is why its worth it to replace a power cord, even though there is wiring in the wall going all the way back (in effect) to the Hoover Dam.

We employ a simple bias mechanism that allows our amps to autobias and so respond to changes in wall voltage. It won't do for example to have the AC voltage to be very high and possibly damage power tubes! So the amp reduces the bias level when the AC line voltage is high and increases it as the line voltage drops.