Jump to content

The Amazing Autotransformer


Langston

Recommended Posts

2 hours ago, mikebse2a3 said:

If you take into account the efficiency of the compression driver/horn (example a TAD TD4002 is good for 110db with 1w input) and the maximum listening level expected in the listening room then 2.7w might be plenty for many home listening rooms. 

 

Spot on. I was wondering if someone would point this out. : )

 

2.7W in dB is 10log(2.7) = 4.3dB that you add to the 110dB sensitivity for 114.3dB maximum peaks. That's loud, but for midrange peaks I'm aiming for at least 120dB. The orchestral stuff I like most hovers around 20dB or more crest factor (peak to RMS), which means I'm in the upper 90's (loud as heck) RMS. I go there sometimes. It's also a little rude to pony up for a pair of drivers like that and adjust their effective max. wattage rating down to 2.7W. : )

 

God bless you and your precious family - Langston

Edited by Langston
Link to comment
Share on other sites

52 minutes ago, Langston said:

orchestral stuff I like most hovers around 20dB or more crest factor (peak to RMS), which means I'm in the upper 90's (loud as heck) RMS

PWK himself told me to make "headroom" 17 db, with unbridled confidence, since all he ever listened to were his own Symphony Recordings (some of which I heard at his house on his Reel to Reel and 2PH3 array and they were outstanding!). I just round up to 20 to be safe. Good point.

  • Like 1
Link to comment
Share on other sites

4 hours ago, Deang said:

So, maybe tell us more about your test set up.

 

Alright you whippersnapper, I re-ran the measurements again this morning, but only at the -9dB step on the 3636 so I could demonstrate a couple of other things.

 

The measurement setup:
APx515 output -> AHB2 (bridged, more power Scotty!) -> 3636 (0-5, 0-2 taps; -9dB) -> 10Ω 10W resistor across 0-2 output taps -> APx515 input.

 

Notes:
Adding a second 10Ω 10W resistor across 0-5 input taps slightly reduces distortion in the 3636 simply because it received slightly less voltage. The swamping resistor did increase distortion slightly in the amp due to the increased load. Thus I chose to skip the swamping resistor for all measurements until this post. The swamping resistor in parallel with the 3636's input also caused a slight increase in HF rolloff; immaterial in this application. I have (6) samples of the 3636, (2) made in 2016 (yellow tape), (4) made last month (white tape). They all measure virtually identically, thus I just used one installed in a jig for all measurements.

 

2075755319_MeasurementSetup.thumb.jpg.f81047e6a6064083ad9d3ac690790adb.jpg

 

The following Max Output plot was done entirely at the -9dB tap setting and adds measurements with the swamping resistor and 1dB into saturation (about 3% THD).

 

2068177712_3636-9dBMax.thumb.png.e62e9c1f9b24f6c14d227261b9e79591.png

 

Next zooms into the attenuation provided by the 3636 at each of the above drive levels and bandwidths.

 

237258781_3636-9dBGain.thumb.png.99f54470a3a4ec263c59b63e0c35bdc8.png

 

And a final effort to put a nail in the coffin of the autotransformer phase error myth.

 

912960577_3636-9dBPhaseandRelativeLevel.thumb.png.f6c7bb4cf641222ebaa76fbce461d72a.png

 

Options:
1. If you really want autotransformer behavior in your passive crossover and need more than 8dB attenuation, use a pair of inductors in an L-Pad type arrangement that can deliver the power you need to the driver without saturating. You loose adjustability but you retain the other benefits.

 

2. If you want to use the 3636 and need more than 8dB attenuation, use an active 2nd order (adjustable Q) shelving filter to reduce the passband further. There's another benefit to this in that near the crossover from low to mid/high, the shelving filter will also be pushing the nasty bass horn resonances further down.

 

3. Don't attenuate anything passively, skip passive impedance compensation, and don't use the passive network for any kind of EQ whatsoever. Design the passive network for proper high and low pass function alone, then use active filtering to do everything else. This will give you almost all the benefits of bi/triamplification while still being able to use a single (low output impedance) amplifier. Of course the loudspeaker will be useless without the active processing, like the new Jubilee. I stole this idea from the great Dave Gunness and used it extensively in concert production. Cut my amp racks approx. in half.

 

4. Be a sissy and use resistors.

 

5. Stop the insanity and go fully active, like the new Jubilee. Active will probably be cheaper by the time I'm done and may very well be what I end up with, but I wouldn't have learned anything.

 

PS: I've been cheating the whole time I've been writing these posts and using fully active DSP bi/triamplification to design and tune my Klipschorns. Once I decide on the endgame for the top hat, I'm hoping for a passive solution with flat impedance around 6Ω from 40Hz to 20kHz. The Klipschorn is the most exciting loudspeaker system I've ever heard in my life. I always thought this was possible, but not until a couple of weeks ago did I experience it.

 

God bless you and your precious family - Langston

Edited by Langston
  • Thanks 1
Link to comment
Share on other sites

3 hours ago, Langston said:

And a final effort to put a nail in the coffin of the autotransformer phase error myth.

 

Perhaps I don't understand what this means.  I believe you demonstrated that the autoformer itself doesn't change phase.  But when used with a series

capacitor, as @Deang mentions above, won't the inductance of the output taps increase the phase shift of the circuit,  creating a 2nd order high pass to the 

squawker?  A capacitor/l-lad circuit will still be first order.  Again, I believe the inductance of the T2A is close to half that of the 3636, so the T2A

phase shift increase would be greater.

 

Mike 

Link to comment
Share on other sites

2 hours ago, mboxler said:

Perhaps I don't understand what this means.  I believe you demonstrated that the autoformer itself doesn't change phase.  But when used with a series

capacitor, as @Deang mentions above, won't the inductance of the output taps increase the phase shift of the circuit,  creating a 2nd order high pass to the 

squawker?  A capacitor/l-lad circuit will still be first order.  Again, I believe the inductance of the T2A is close to half that of the 3636, so the T2A

phase shift increase would be greater.

 

You're absolutely right! The autotransformer would do just that and each tap would present yet a different inductance (one of the three parts of impedance) to the passive network that is loaded by the AX/driver combination. Thus different caps, etc., will have to be used due to the AX and changed again every time you choose different attenuation taps.

 

Unless..

 

.. you use a swamping resistor and turn the AX/driver system into a resistor that barely changes value even as you change taps. Then who cares about the AX and driver with all their weird internal inductance/capacitance/resistance when all the passive network sees is what it thinks is a very close approximation to a constant value resistor? I wouldn't consider dealing with a moving target to load my passive network, thus it's AX plus swamping resistor or no AX for me.

 

This plot illustrates what I'm saying - and please forgive me if I'm missing your point! And I'm still going to do those measurements you and Dean mentioned. : )

 

God bless you and your precious family - Langston

 

2108186793_A55G3636Z10R.thumb.png.c35f472b940ae6c90ce7c4ec0d7ce8fb.png

  • Like 1
Link to comment
Share on other sites

I need to review all this but......what are we talking about?

A 1" driver (k55) on a K400 horn?

Really I mean who gives a sheet?

Now a bms 4592 on a tractrix horn, now we're talkin

7 hours ago, Langston said:

The Klipschorn is the most exciting loudspeaker system I've ever heard in my life. I always thought this was possible, but not until a couple of weeks ago did I experience it.

Ever heard Jubilees?

  • Like 1
Link to comment
Share on other sites

4 hours ago, babadono said:

Ever heard Jubilees?

 

Never! I'm only two weeks old with the Klipschorn. If Roy keeps the stunning bass slam of the Klipschorn while increasing LF extension with those ports, I'm going to rob a bank. Simple as that. Also, down here in the South it's considered bad manners to ask something like that without the possibility of a solution. Is there a pair setup somewhere in the US in a proper room where the room modes have been dealt with at the listening position? : )

 

Also: I haven't done the tuning for my 1 inch A-55-G/K-400 and DE120/Dave Harris Tractrix horn yet, so I haven't heard the Klipschorn as designed, well as a highly modified 3-way I should say. I never used the OEM woofers other than measuring their TS parameters (they're in good shape) - I installed the Crites cast frame woofers. Silly little 15 inch things with a tiny 150W rating (I'm used to pro audio) that almost knocked me on the floor. Those corner horns are brilliant.

 

As of now I've only listened to the system as a two-way with Dave Harris' large wood Tractrix horn with the Faital Pro HF206 and B&C DE750. I also just got the B&C DCX464 with the big horn they made for it, but haven't even done the raw measurements on it yet. I assume that'll be the winner, but I've been wrong so many times. This is also taking longer than expected because everybody I know is coming over to hear the Klipschorns.

  • Like 1
Link to comment
Share on other sites

Thanks Mike for the 3636 and T2A LTSpice component files. As you know LTSpice assumes ideal components, but also allows the addition of parallel and series L/C/R values to make the model act like the real thing. I measured a 3636 sample and got slightly different inductance and resistance values between taps than you did, but both models missed real measurements too much for my taste. I've attached my version of the 3636 component file that you can use to get my inter-tap values.

Crites 3636.asc

 

Still, the only reason to bother with such an exercise is if you don't use a swamping resistor on the input of the AX. With a 10Ω resistor across the AX inputs, the specific AX impedances become useless information because the resistor overwhelms everything downstream of it. The resistor is Borg - it assimilates everything - but in this case inductance becomes futile, not resistance! : )

 

So back to the real world that I know and trust - measurement.

 

I decided to measure the voltage applied to the A-55-G driver input terminals simultaneously with the nearfield acoustic output from the K-400 horn (mic almost touching the grill cloth). Dean asked for the -3dB (0-4) attenuation setting, but I decided to use -6dB (0-3) to make transfer function defects, if any, more apparent.

 

I don't include a series capacitor in this group of measurements because it would make things too busy. Next time. It doesn't reveal anything IMO.

 

Conclusion:
The 3636 is transparent at audio frequencies. All it does is reduce voltage like a volume control by the amount selected without messing with the time domain (phase, etc.). The max output of the 3636 decreases as attenuation increases, the specifics of which were shown here.

 

1797764532_A55Gvs3636.thumb.jpg.611427b3413d11b5db3d0a95dcdf2889.jpg

 

Blue traces are the reference measurement without the 3636, i.e. with the amp directly connected to the driver. Pink traces add the 3636 set at -6dB between amp and driver without a swamping resistor. Again, each measurement includes two traces of the same color for the electrical input and acoustic output of the driver.

 

633391175_A55G3636VSPL.thumb.png.21a4cb1c55f3513f409b17de8582c334.png

 

Now what I've done is take those same traces and force each to intersect at a mid-point (I chose 900Hz). I also smoothed the curves to 24th octave. This makes it much easier to see if the frequency response varies when the 3636 is inserted in the circuit.

 

1976758335_A55G3636VSPLRelative.thumb.png.426974ba5d1d1b1cad607b8980059028.png

 

Next is the identical plot except this time I include a third pair of (Green) traces that add a 10Ω swamping resistor on the input (0-5) of the 3636.

 

1492999718_A55G10R3636VSPLRelative.thumb.png.93e64280b03eb1a2ac10d0def93ca664.png

 

Now the final nail in the coffin to dispel the AX phase error myth. Again, all three pairs of traces are shown.

 

1474823805_A55G10R3636Phase.thumb.png.735a3167daeb7a9bd452491a14fb22d4.png

 

God bless you and your precious family - Langston

Link to comment
Share on other sites

12 hours ago, Langston said:

Thanks Mike for the 3636 and T2A LTSpice component files. As you know LTSpice assumes ideal components, but also allows the addition of parallel and series L/C/R values to make the model act like the real thing. I measured a 3636 sample and got slightly different inductance and resistance values between taps than you did, but both models missed real measurements too much for my taste. I've attached my version of the 3636 component file that you can use to get my inter-tap values.

 

Thank you for measuring!    I've been trying to determine if the inductance differences between the 3636 and the T2A are noticeable using stock Klipsch 

schematics.    I'm also considering building an AK-3 to replace my AK-2 and don't know how different the parallel 5mH inductor would be 3636 vs T4A.

It probably won't make much of a difference but it's fun to run the simulations. 

  • Like 1
Link to comment
Share on other sites

On 6/19/2021 at 9:40 PM, Langston said:

 

I'm only two weeks old with the Klipschorn. If Roy keeps the stunning bass slam of the Klipschorn while increasing LF extension with those ports, I'm going to rob a bank. Simple as that. Also, down here in the South it's considered bad manners to ask something like that without the possibility of a solution. Is there a pair setup somewhere in the US in a proper room where the room modes have been dealt with at the listening position? : )

 

 

To get the Jubilee listening experience, you could contact Coytee, who I believe lives in Tennessee, and has had Jubilees for a long time.  There's also Mikebse2a3, whose room is totally set up.  I think he's also in Tennessee.  It it's not practical to visit them, I expect that they'd know who to put you in touch with to hear their Jubilees.  Most owners are intensely proud of their speakers, and would be happy to share a listen with another Forum member, especially someone who really has a clue about speakers.  Klipschorns are great speakers, but bi-amped Jubilees (they're all bi-amped) with driver time-alignment through the electronic crossover may really impress you.  There's also a few Jubilee owners in Texas, but that's a bit of a hike from Florida.

  • Like 1
Link to comment
Share on other sites

On 6/19/2021 at 9:40 PM, Langston said:

Also, down here in the South it's considered bad manners to ask something like that without the possibility of a solution. Is there a pair setup somewhere in the US in a proper room where the room modes have been dealt with at the listening position? : )

Sorry, my bad. But yes there are as @Islander has pointed out.

  • Like 1
Link to comment
Share on other sites

Well, then, might as well finish by measuring the T2A a bit more. Universal Transformer knows what they're doing. Here's a screenshot of the tap to tap measurements of my pair of T2A's as well as the data from the LTSpice file that Mike gave us. There's a lot of difference between samples.. Or is there? : )

 

2137202257_T2ALDCR.png.32d5ad499a1bb3ed789b16c2c8e9d09c.png

 

T2A L DCR.xlsx

 

Since these things are basically L-Pads using inductors1, it isn't inter-tap absolute values that have to match, what has to match are the ratios between them because that's what causes the gain reduction. And these ratios match nicely. PWK did not compromise his life's work with the autotransformer.

 

Small signal measurements of my pair of T2A's at each attenuation step with a 10Ω load in place of a driver.

 

1492315891_T2AGain.thumb.png.ecd67207fc7de9ec7736c3c4f6c15a4f.png

 

Now I take all (8) measurements and force the magnitude traces to intersect at 1kHz to see the overall variation between every setting of each AX. Phase is also shown, but it is not normalized. I've seen worse, but not from an autotransformer. : )

 

791512350_T2ATF.thumb.png.58d686c8b278980a2dc9e9143b9fd869.png

 

God bless you and your precious family - Langston

 

1 Nope, it ain't. An AX is a continuous winding on one core with one or more taps into those windings with its behavior determined by the ratio of turns or voltage or current, depending on how you want to measure it. I'm learning. : )

 

1986366120_AutotransformerRatios.thumb.png.24d1328bfeb54a0ddbd1205158a6efdb.png

Link to comment
Share on other sites

1 hour ago, Langston said:

Since these things are basically L-Pads using inductors, it isn't inter-tap absolute values that have to match, what has to match are the ratios between them because that's what causes the gain reduction. And these ratios match nicely. PWK did not compromise his life's work with the autotransformer.

 

 

Again, thanks for measuring.  I realize the inductance is not important to you.  That said, I believe there must be an error with your T2A #1 measurements(?).

For example, taps 0 - 1 represent 1/4 of the total turns of the autoformer, and therefore the inductance should be 1/16 of the total inductance.

If you were to measure the inductance of taps 0 - 5, I doubt if it's 8.96mH * 16 or 143.36mH.  

 

Mike 

 

 

  • Like 2
Link to comment
Share on other sites

1 hour ago, mboxler said:

I believe there must be an error with your T2A #1 measurements(?)

 

Sure enough, the battery on the (calibrated) Keysight U1733U was dead when I tried to use it for a second go at it. The new battery made the results quite different, which has been edited in the post. Apparently it has to source a fair amount of current to make larger inductance measurements. The Audio Precision measurements don't rely on batteries. : )

 

Thanks! - Langston

Link to comment
Share on other sites

3 hours ago, mboxler said:
:

Since these things are basically L-Pads using inductors ...


They are impedance multipliers, which means they are nothing like l-pads. An autoformer doesn’t waste power by turning it into heat - well, not until you add that swamper. 

Link to comment
Share on other sites

2 hours ago, Deang said:

An autoformer doesn’t waste power by turning it into heat

 

That is a fact. My stomach turned when I wrote that. Input and output power are almost equal with the autotransformer vs. a resistor L-Pad where power is dissipated to ground. And you're right - the swamping resistor is an expense - an expense I'm more than willing to pay given what I get out of it.

 

Re: the Chief, I'd love to meet you - but I don't have the patience for fishing. Let me know if I'm a problem and I'll disappear.

 

At this moment I'm half way through a Crosby Stills and Nash album using the Klipschorn as intended as a 3-way (triamplified), and cheating further with the Crites A-55-G/K-400 and DE120 Dave Harris Tractrix. Very nice, but falls short of the two 2-ways (biamplified) I've tried the last two weeks, even at low levels. At higher levels It can't flesh out the lower frequency meat that the 2 inch drivers deliver even though the crossover points are identical. L/R location is much more toward the loudspeakers, whereas the two 2 inch / Harris Tractrix 2-way setups I've tried result in the loudspeakers largely disappearing. Nevertheless I could easily live with this "classical" 3-way system. The ear/brain system adapts. "Helplessly Hoping" just started. Phenomenal. : )

 

God bless you and your precious family - Langston

Link to comment
Share on other sites

  • Klipsch Employees
11 minutes ago, Langston said:

 

That is a fact. My stomach turned when I wrote that. Input and output power are almost equal with the autotransformer vs. a resistor L-Pad where power is dissipated to ground. And you're right - the swamping resistor is an expense - an expense I'm more than willing to pay given what I get out of it.

 

Re: the Chief, I'd love to meet you - but I don't have the patience for fishing. Let me know if I'm a problem and I'll disappear.

 

At this moment I'm half way through a Crosby Stills and Nash album using the Klipschorn as intended as a 3-way (triamplified), and cheating further with the Crites A-55-G/K-400 and DE120 Dave Harris Tractrix. Very nice, but falls short of the two 2-ways (biamplified) I've tried the last two weeks, even at low levels. At higher levels It can't flesh out the lower frequency meat that the 2 inch drivers deliver even though the crossover points are identical. L/R location is much more toward the loudspeakers, whereas the two 2 inch / Harris Tractrix 2-way setups I've tried result in the loudspeakers largely disappearing. Nevertheless I could easily live with this "classical" 3-way system. The ear/brain system adapts. "Helplessly Hoping" just started. Phenomenal. : )

 

God bless you and your precious family - Langston

That’s not a klipschorn anymore. Mod it all you want but it’s not a klipschorn.....by the way fishing is awesome!!

  • Like 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...