Autoformer Attenuation

[mark1101]

"...trying to understand the audible implication?"

Well, that would certainly be the one I'm interested in.

The effect of hysteresis was well known and understood during the time PWK was designing his filters - I have trouble with the idea that a mental giant like Paul Klipsch wouldn't have accounted for it while making choices that would be critical to his design.

I did enjoy that last post though, interesting stuff to think about.

Well he didn't think time alignment was important either, on speech and music, even though a Khorn's bass is 8 milliseconds behind the tweeter. PWK did a lot of really great things, the Khorn bass bin being one of them, but he wasn't perfect. Hey he was human. He farted in his couch during the demo I was litening to, but it didn't smell though. I was on the other couch.

Claude,

You are a very detail oriented guy.

[babadono]

Dean,

I believe I replied on the previous thread. I am using the BMS 4592 mid on a GotHover eliptrac horn. I believe it is an 8 ohm version. I have ALK ES crossovers and have the attenuation set to -13.7dB. This on a KHorn.I also have a Pair of Lascalas in the same room and use only the bass horn at the moment with the Mid and Hi output from the crossovers terminated into load resistors. And I still could use more bass. Either I really like bass or my room absorbs / cancels it out. But anyway the BMS/eliptrac combo is a beast, hard to keep up with.

Eric

[Steve R]

Not sure if it helps but here go's .I'm currently working on a KHorn build that has V-Trac mid horns with JBL 2482 drivers using JBL diaphram's ,JBL 2405 tweeters triamped. I have been trying out a BSS Omni drive 388 ,the mids are set at -10 db. They seem to be good at that level ,but it's a work in progress and not ready for prime time just yet.

[ClaudeJ1]

"...trying to understand the audible implication?"

Well, that would certainly be the one I'm interested in.

The effect of hysteresis was well known and understood during the time PWK was designing his filters - I have trouble with the idea that a mental giant like Paul Klipsch wouldn't have accounted for it while making choices that would be critical to his design.

I did enjoy that last post though, interesting stuff to think about.

Well he didn't think time alignment was important either, on speech and music, even though a Khorn's bass is 8 milliseconds behind the tweeter. PWK did a lot of really great things, the Khorn bass bin being one of them, but he wasn't perfect. Hey he was human. He farted in his couch during the demo I was litening to, but it didn't smell though. I was on the other couch.

Claude,

You are a very detail oriented guy.

I was just trying to add some humor to the whole thing. How did I do? LOL

[DrWho]

The effect of hysteresis was well known and understood during the time PWK was designing his filters - I have trouble with the idea that a mental giant like Paul Klipsch wouldn't have accounted for it while making choices that would be critical to his design.

PWK was quite emphatic about maximizing efficiency - it's even the first of his cardinal rules. If you put efficiency at the top of the list, then you would right away rule out resistors because the autoformer is more efficient....

And then in the 70's, the distortion was less than that of the amplifiers of the time - why worry about an artifact swamped by the best of the rest of the system chain? Maybe he figured the increase in efficiency helped offset some of the amplifier's distortion?

[mike stehr]

Can't a guy just take one of those autoformers, simulate a primary and secondary load, (whatever it may be) connect a frequency generator to the primary and a scope to the secondary tap of choice?

Then run square waves at the voltages used at speaker levels, monitoring the scope for ringing and overshoot from the secondary tap?

I know it has nothing to do with hysteresis. I've done this to check linearity/bandwidth of tube output transformers. A tube OPT that has bad ringing characteristics within the audio band (or slightly above) will result in bad sound, unless proper feedback compensation is used.

If the autoformer has some ringing within the audio band over the range of voltages used in a passive crossover network application, that could affect sound quality?

Maybe the ringing is out of band, or there isn't a high enough voltage to induce any ringing used in a crossover?

I've been told that transformer/autoformers don't make very good voltage dividers...but I have no clue...

[Deang]

The testing you just described is exactly what Al did before he started using them. He was convinced they would ring, but they didn't - I believe the tests were done using the 3619.

Well, "...distortion is inversely proportional to efficiency". So, anything that increases the sensitivity of the system and so reducing the distortion exhibited by the drive units at any given power level -- should be seen as desirable.

[DrWho]

Well, "...distortion is inversely proportional to efficiency". So, anything that increases the sensitivity of the system and so reducing the distortion exhibited by the drive units at any given power level -- should be seen as desirable.

Unfortunately, distortion is NOT inversely proportional to efficiency - they are not intrinsically related properties of nature so any correlation between the two is just that: a correlation, not causality. I might even suggest that PWK would agree with me on that.

When PWK made that statement, he was referring to the efficiency of the mechanical / acoustical transducer....and specifically to the doppler effect as a limiting source of frequency modulation distortion. A perfect driver can't do better than the limits imparted by the doppler effect - which can actually be shown mathematically. That mathematical relationship is absolutely causal, but its causality is limited to the very specific condition that the math describes. Improving the coupling of the driver to air so that its motion is reduced (aka increasing its efficiency) does in fact reduce the doppler distortion of the system.

However, the doppler effect is not the only source of distortion in a system. It is absolutely possible to have higher distortion with a higher efficiency system....just look at the horn throat distortion generated from an undersized throat. You're getting way more SPL per Watt, but it sounds like frying bacon garbage. PWK would have been aware of this when he made his statement - the problem is he was trying to describe an incredibly complicated concept at a layman's level....and now people are misquoting him and misunderstanding how nature actually behaves.

In fact, and I hesitate saying this because I want to stress that distortion and efficiency aren't inherently related, but in the audio electronics world we usually see the opposite. Any time you reduce the power consumption of an analog circuit, you almost always see the distortion increase. And the reason I say almost always is because sometimes it goes the other way too. What matters are the exact specifics of the topology being discussed.

At the end of the day, the distortion generated by the motion of the loudspeaker diaphragm has nothing to do with how much energy was consumed to create the voltage at its terminals. PWK was referring to diaphragm motion - not the non-ideal effects of resistors and magnetics. PWK is an acoustical / speaker transducer and mathematical genius, but I find other teachers are far more equipped to discuss electrical circuits. I don't trust PWK with anything happening to the other side of the loudspeaker terminals.

[DrWho]

Can't a guy just take one of those autoformers, simulate a primary and secondary load, (whatever it may be) connect a frequency generator to the primary and a scope to the secondary tap of choice?

Then run square waves at the voltages used at speaker levels, monitoring the scope for ringing and overshoot from the secondary tap?

I know it has nothing to do with hysteresis. I've done this to check linearity/bandwidth of tube output transformers. A tube OPT that has bad ringing characteristics within the audio band (or slightly above) will result in bad sound, unless proper feedback compensation is used.

If the autoformer has some ringing within the audio band over the range of voltages used in a passive crossover network application, that could affect sound quality?

Maybe the ringing is out of band, or there isn't a high enough voltage to induce any ringing used in a crossover?

I've been told that transformer/autoformers don't make very good voltage dividers...but I have no clue...

At audio frequencies, any ringing we see will be intrinsically related to the complex frequency response. I mention it because its far easier to interpret the frequency response of a system than its square-wave response. It's also easier to generate accurate sine waves than it is to generate accurate square waves.

Did you ever measure the frequency response of these tube OPT's? You should have seen some aberrations near the ringing frequency (the shape of aberration dependent on what kind of signal source you used).

[John Warren]

"...distortion is inversely proportional to efficiency".

wrong

[erik2A3]

Dean: gosh all of this goes back a ways, doesn't it? You mentioned someone had concluded, and then shared that conclusion with you, that resistors used for mid--horn attenuation were 'better' (a term we all know is hugely subjective as well as often ambiguous) than a multi-tapped choke. Years ago, when we were throwing that idea back and forth, I had palyed around with networks with both our K-horns and La Scalas (now only La Scalas) that were, for all intents and purposes the same as a factory type 'A,' except for the manner in which squawker output was attenuated. By the same as the 'A' I'm referring to the order of slope and crossover frequencies. Naturally, there were some minor changes that needed to happen because of the resulting alteration of reflected impedance with the autoformer not being present. Without putting a 'better' or 'worse' value judgement on my impressions (which might and probably would be different from that of lots of others), and keeping in kind that one's acoustic memory (at least MY acoustic memory) can by virtue of our own humanity often be faulty and not always consistently reliable, my impression of both fixed and variable (resistive) L-pads was what I perceived as a reduction in overall midrange opacity. Describing sound quality with these sort of descriptors is notoriously difficult because people have different ideas about what words mean. What I mean by less opacity has mainly to do with sharpness of attack of transient elements, where notes and leading edges of various instruments had more of an etched or sort of stringent quality. Some might well ask whether etched and sharper qualities are ultimately musical, and my own response would be that it would be a matter of magnitude. having played drums for years in both rock-fusion bands, as well as jazz trios and quartets, I can say that the sound of an acoustic drum kit can sound incredibly brash, even wincingly bright. The same can be true for violin and many other instruments. The L-pad seemed to reduce that quality for me. I would say, then, that whether one attenuation approach is better than the other would in part depend on one's priorities -- what one listens for and likes to hear as far as reproducing music at home. As you know, I also experimented with various orders of slope and complexity and more common midrange bandpass, Zobel networks (definitely not for me), but always seem to prefer simple, 6dB/octave designs, whether an autoformer is used to attenuate or resistors.

[tigerwoodKhorns]

Dean: gosh all of this goes back a ways, doesn't it? You mentioned someone had concluded, and then shared that conclusion with you, that resistors used for mid--horn attenuation were 'better' (a term we all know is hugely subjective as well as often ambiguous) than a multi-tapped choke. Years ago, when we were throwing that idea back and forth, I had palyed around with networks with both our K-horns and La Scalas (now only La Scalas) that were, for all intents and purposes the same as a factory type 'A,' except for the manner in which squawker output was attenuated. By the same as the 'A' I'm referring to the order of slope and crossover frequencies. Naturally, there were some minor changes that needed to happen because of the resulting alteration of reflected impedance with the autoformer not being present. Without putting a 'better' or 'worse' value judgement on my impressions (which might and probably would be different from that of lots of others), and keeping in kind that one's acoustic memory (at least MY acoustic memory) can by virtue of our own humanity often be faulty and not always consistently reliable, my impression of both fixed and variable (resistive) L-pads was what I perceived as a reduction in overall midrange opacity. Describing sound quality with these sort of descriptors is notoriously difficult because people have different ideas about what words mean. What I mean by less opacity has mainly to do with sharpness of attack of transient elements, where notes and leading edges of various instruments had more of an etched or sort of stringent quality. Some might well ask whether etched and sharper qualities are ultimately musical, and my own response would be that it would be a matter of magnitude. having played drums for years in both rock-fusion bands, as well as jazz trios and quartets, I can say that the sound of an acoustic drum kit can sound incredibly brash, even wincingly bright. The same can be true for violin and many other instruments. The L-pad seemed to reduce that quality for me. I would say, then, that whether one attenuation approach is better than the other would in part depend on one's priorities -- what one listens for and likes to hear as far as reproducing music at home. As you know, I also experimented with various orders of slope and complexity and more common midrange bandpass, Zobel networks (definitely not for me), but always seem to prefer simple, 6dB/octave designs, whether an autoformer is used to attenuate or resistors.

Do you happen to have a schematic for the resistor based networks?

[Deang]

Mike, I quoted PK in response to the following statement:

And then in the 70's, the distortion was less than that of the amplifiers of the time - why worry about an artifact swamped by the best of the rest of the system chain? Maybe he figured the increase in efficiency helped offset some of the amplifier's distortion?

I was simply using the quote to confirm your conjecture (the latter statement). Really though, you have to go back a little further than the 70s to find those kinds of amps -- some of the best sounding amplifiers ever built were built in the 70s (though there were handful of some very bad ones). Seriously, it would take a pretty bad amp to surpass the distortion exhibited by drivers. I appreciated all of the keystrokes in that one post, but I'm surprised that you would think I didn't know most of that.

@John, if contained to the original context in which PK applied it, it's true. Mike brought up the example of the problem associated with long, narrow throats when the levels are driven up, something you are intimately familiar with. So sure, high efficiency doesn't always mean lower distortion.

@Erik, good post, I enjoyed it. We don't always agree, but I always understand where you're coming from. The fact that you've actually tried these different methods lends quite a bit of credibility to your opinions, which is why I finally stopped beating up on you. I do sometimes question your ability to identify nuances in the sound due to the known issues with some of your hearing, but the fact that you can clearly articulate those differences from your listening sessions tells me something I've suspected for some time -- that the brain may play a bigger part in the listening experience than we give it credit for.

Building any of the simpler circuits with resistors instead of an autoformer is not difficult.

Using an autoformer allowed PK to use smaller value capacitors at the primary position. It's actually possible that this approach may have saved him money. For the Type A and Type AA, he would have needed 26mfd of capacitance -- I have no idea how much that would have cost him in comparison to using the autoformer and smaller capacitors. Honestly, all we have so far in this thread is a lot of speculation. I asked what I thought was a pretty simple question -- why did this really smart guy who didn't spend a dime more than he had to, pick this relatively expensive part to accomplish his means? We know he would spend more if it resulted in lower distortion or higher efficiency. Resistors soak up energy, which is why they get hot. The distortion question may not be completely settled at this point, but we do know that autoformers don't waste power. Actually, PK addressed the, or if you prefer "a" distortion issue in "The Problem with Attenuators". Whatever is discovered regarding the hysteresis issue, it won't do away with the measurements and resultant conclusions found in the article.

I offered to provide a unit for testing. I believe in testing. So, until we have some results indicating that the autoformer is inferior to resistors, I'm going to continue to use them.

In my other thread, there is an article suggesting that hysteresis may have a euphonic effect. It seems unlikely that PK wouldn't have compared the two methods, and I don't think he would have picked the method which sounded worse, especially if the measurements supported what he heard.

"For my part, I'm still using ears as measuring tools. The ear

says it is right or wrong; the X‐Y recorder may aid in showing

why." -- Paul Klipsch

Edited December 7, 2013 by DeanG

[DrWho]

Dean, I'm referring to the distortion of the amplifier's output transformer versus the autoformer in the xover....which comes directly from PWK in one of his articles. I was never talking about amplifier performance versus driver.

[Deang]

Ah, I see.

Let me know when you get some free time, I'll be happy to ship an autoformer or two for you to play with.

[John Warren]

@John, if contained to the original context in which PK applied it, it's true. Mike brought up the example of the problem associated with long, narrow throats when the levels are driven up, something you are intimately familiar with. So sure, high efficiency doesn't always mean lower distortion.

strike two...

[Deang]

John,

If you have something to say, please say it.

Like PWK said, he would only be proven wrong if and when there was a change in the laws of physics.

[DrWho]

I've seen John bring up THD comparisons in the past, but the Doppler effect on a multi-tone signal doesn't show up in a THD measurement.

[DrWho]

Ah, I see.

Let me know when you get some free time, I'll be happy to ship an autoformer or two for you to play with.

Let me get some measurements of Bob's autoformer completed first and make sure we're all in agreement there. Sounds like we should be comparing a 12dB pad? Ironically, the 10dB range is one of the hardest to do with resistors for a variety of boring technical reasons

Should we put a price limit on the resistors? And how much power we looking for? I'll need to bring a nice amplifier too work too...

[greg928gts]

FWIW, I'm using V-Trac horns and BMS4592ND drivers in my Khorns with John Warren's networks, mid resistors at -10db.

Greg