Autoformer vs Speaker impedance Curve

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.

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. 

I have started a new topic to explain how conventional output transformers differ from Autoformers. 

https://forum.audiogon.com/discussions/autoformers-the-benefits-in-matching-amp-to-speaker

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.

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. 

 

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.

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).