L lads

[henry4841]

An L-pad is nothing more then 2 resistors. The adjustable L-pad is one fixed resistor and one variable resistor. Basic simplest electronic component. You can make a resistor yourself at home with a carbon pencil and piece of paper. Autotransformers or transformers have significant more amounts of inductance and capacitance then a resistor. You have significant amounts of phase shift of the signal with an autotransformer. Some of the crossover networks on the old heritage lines have the + and - leads reversed going to the speaker because of this. In other words the signal has shifted significant enough to be close to 180 degrees from the input signal. The engineers at Klipsch have been using discrete L-pads for attenuation for years in all their products which should be enough said. You are not going to introduce anything that you can hear installing an L-pad other then the desired result of tailoring the sound for your room and taste. Install on the mids and highs and you can achieve more bass with reduction of efficiency of course. The biggest problem though is the highs where most install a L-pad because of the horn being too bright for their taste. It works and is cheap. Easy enough to remove as well. 

[ClaudeJ1]

10 hours ago, Chief bonehead said:

If the impedance is consistent, then an l pad will work because it will attenuate the freq response evenly. If the impedance if not consistent, then the lpad will not attenuate the freq response evenly. 
 

 

OK, so the Phase issue must be the reason why you no longer use Autoformers in your latest Heritage Line. Otherwise, the product contradicts the quoted words, unless you are using a lot of Zobel circuits in concert with resistors in your networks, aye?

[Edgar]

32 minutes ago, henry4841 said:

The adjustable L-pad is one fixed resistor and one variable resistor.

 

In a proper L-pad, both are adjustable, and they work as a resistor divider to keep the impedance seen by the amplifier constant.

 

Quote

Autotransformers or transformers have significant more amounts of inductance and capacitance then a resistor. You have significant amounts of phase shift of the signal with an autotransformer. Some of the crossover networks on the old heritage lines have the + and - leads reversed going to the speaker because of this. In other words the signal has shifted significant enough to be close to 180 degrees from the input signal.

 

In a proper transformer, the phase shift within its bandwidth is negligible. The reason that crossover leads are sometimes reversed has to do with the topology of the crossover itself -- 2nd-order crossovers invert polarity at the crossover frequency, so the lead reversal puts it back. (OK, so that's an oversimplification, but a good explanation requires math.)

 

The biggest problem with a transformer is cost.

[Bacek]

11 hours ago, Chief bonehead said:

If the impedance is consistent, then an l pad will work because it will attenuate the freq response evenly. If the impedance if not consistent, then the lpad will not attenuate the freq response evenly. 
 

 

Ok but while i remember correctly k55 impedance i k400  there is not more than 10% impedance variance in usable range. Does it this justify that large attenuation swing?

[henry4841]

52 minutes ago, ClaudeJ1 said:

OK, so the Phase issue must be the reason why you no longer use Autoformers in your latest Heritage Line. Otherwise, the product contradicts the quoted words, unless you are using a lot of Zobel circuits in concert with resistors in your networks, aye?

Look at the often criticized AL network. A case of throwing in more parts with the consequences they intel. I like simple.  

[henry4841]

It all really comes down to if one is not happy with their new Klipsch speakers because of being too bright either get rid of them with a loss or tailor the sound with L-pads. I hate to think of someone getting rid of such wonderful speakers when simply installing some L-pads will solve the too bright problem. Being too bright is a common complaint with horn speakers. Probably more to do with their room then anything else. 

[Wirrunna]

On 4/15/2021 at 4:10 AM, tigerwoodKhorns said:

This looks fancy.  Not sure what design it is and not smart enough to guess without doing a bunch of reading on the Google, but has mid and tweeter attenuation using the 3636. 

 

Just banged this into LT Spice, set the mid and hi as per mboxler's suggested values (see next post).

Looks like a La Scala 500 / 5800 network to me

 

 

[mboxler]

9 hours ago, Wirrunna said:

Just banged this into LT Spice, left the load for mid and hi as 10 ohms as the swamping resistor across the 3636 is what the network sees

 

Hey @Wirrunna

 

Actually, the network will see the 10 ohm resistor in parallel with the impedance reflected across taps 0 -5 of the autoformer.

In the case of the tweeter network, if the tweeter is running off taps 0 - 4, the impedance across taps 0 - 5 will be around 16 ohms.  16 ohms in parallel

with the 10 ohm resistor is around 6.1 ohms.  If using output taps 0 - 3, it will be 7.6 ohms.

 

Sorry if you already know this, but I stumbled upon a neat Spice directive, the .step param.  Instead of giving the resistor a fixed value (10 in your case), you 

can assign it {R_Value}.  Then, add a Spice directive (the .op tab on the top right), and paste something like this

 

.step param R_Value list 6.1 7.6

 

When you run your simulation, and click on the voltage across the resistor, you will get two plots, one for the 6.1 value and one for the 7.6 value.

I find this handy when simulating capacitor and inductor values as well.

 

Mike

 

 

[henry4841]

It gets a little more complicated when a transformer is involved then just measuring impedance of parallel resistors. Those that build and repair tube gear understand this from the impedance of  an output transformer. When you measure the impedance of a transformer, an auto-transformer is a transformer, there is the one measurement with just a ohmmeter. Then there is the impedance when one considers the audio signal. Anyone that reads the specs of a tube output transformer understands this. For example this Edcor transformer that is rated at 5K ohms. https://www.edcorusa.com/gxse10-5k  But if one measures it with an ohmmeter you will see 187ohms which is called DCR or the resistance at a dc voltage. https://www.edcorusa.com/gxse10-5k

 

The self banned former member of  this forum Al at ALK Engineering explains the concept of using a swamping resistor better then what you will find on the forum. No matter what is thought of Al he is a very talented, but very opinionated, individual who if memory serves me correct has an engineering degree so is to be respected. This is not to say the standard Klipsch crossovers are any worse or maybe even better. Just and alternative concept that allows adjusting attenuation using an auto-transformer forgoing using an L-pad along with the auto-transformer to tailor the sound.  

[Deang]

Nm 

[Langston]

On 6/25/2021 at 9:07 AM, mboxler said:

.step param R_Value list 6.1 7.6

 

I threw up my hands in surrender trying to accurately characterize transformers and autotransformers in LTSpice or elsewhere. I figured out how to get LTSpice simulations that matched measurements involving these gizmos connected to loudspeakers another way. All you have to do is measure the impedance of the transformer connected to the load (loudspeaker). Then use @mboxler 's Spice Directive to enclose that measurement inside a two-port component (two wires) of your choice. Of course you can do this with just the loudspeaker if you're not using an autotransformer.

 

Here are the directions. It's easy.

 

I then export the voltage response (includes phase) from LTSpice (click simulation window, File, "Export data as text"). I import this voltage response into my crossover design software and combine it with the acoustic response to verify the design without having to buy and build stuff first. The primary benefit of this for me (you can do all this without LTSpice of course), is that LTSpice makes it simple to see the wattage through resistors, voltage across caps and current through inductors. No more "what's that smell?" after Another Brick in the Wall at proper volume. Necessity is the mother of learning how to use a new tool. : )

 

 

[Edgar]

14 minutes ago, Langston said:

 

Of course you can do this with just the loudspeaker if you're not using an autotransformer.

 

Here are the directions. It's easy.

 

 

Very interesting!

 

I'm an EE, but I specialize in Signal Processing, not circuit design. So on the rare occasions when I use SPICE, I'm limited to Texas Instruments freeware "TINA". I'll have to see if a similar import mechanism exists there.

 

When I do use SPICE, it's generally to design Zobel networks. Having a circuit model for the driver is very helpful there. In the past I've done it manually, measuring impedance magnitude and phase and matching resonant frequency and quality factor as best I could. It works very well, but it's tedious.

[mboxler]

On 6/27/2021 at 12:15 PM, Langston said:

I threw up my hands in surrender trying to accurately characterize transformers and autotransformers in LTSpice or elsewhere.

 

Thanks for the link.  I had a DATS V3 sitting around and tried it out.  I attached the DATS to taps 0 - 5 of a 3636 and a 14.6 ohm resistor to taps 0 - 4, and was

able to use the exported file as an LTSpice sub circuit. I compared that to my 3636 simulation and was bummed that it didn't match.

As far as I can tell, the real 3636's inductance had no effect on the lower frequencies. 

 

I got the idea to change my 3636 simulation inductors from Mh to full Henrys 😮.  Reran the simulation and got an almost identical match.

I don't understand why it worked, but I was wandering if you could try this yourself and see what happens.

 

Mike

[Langston]

17 hours ago, mboxler said:

..the real 3636's inductance had no effect on the lower frequencies. 

 

Yep. In real life the 3636 is flat as a pancake through the audio spectrum, yet the sim. shows LF rolloff. I know that LTSpice assumes "typical" component parasitics in many cases, all of which you can change or override, but that's over my head and beyond my interest. I use this software for quick and easies, and move on to other methods that match measurements when it lies to me - such as with these autotransformers. I'm all in though if you do the work - I'd be glad to be your beta tester. : )

 

Beta test: I changed 3636 inductance values to full Henry units and didn't see any change (which I'm thankful for because I have dozens of LTSpice models where I used various units out of convenience).

 

See if the attached is what you're talking about..

 

Strikethrough Edit: I did this too quickly and didn't pay attention to the autoscaled Y-axis in LTSpice, which actually shows the autotransformer with a much flatter transfer function than reality. The LTSpice sim also shows quite a bit less attenuation than reality, further indicating it's assuming no parasitics. I matched the LTSpice scale of my measurement of the actual AX used for the sim with the same load and attached it below.

 

God bless you and your precious family - Langston

 

 

 

[mboxler]

52 minutes ago, Langston said:

See if the attached is what you're talking about..

 

Kind of.  If you were to plot V(Tap5)/I(V1) you will get the impedance across taps 0-5 of the autoformer.  The impedance plot of my DATS test was much closer to the

"Henry" vs the "MilliHenry" values.  I can't guarantee my DATS test was accurate, but the results were surprising. 

[Langston]

35 minutes ago, mboxler said:

If you were to plot V(Tap5)/I(V1)

 

Identical results again. Whatever you're doing wrong, I'm worse. : )

 

[mboxler]

7 hours ago, Langston said:

Identical results again. Whatever you're doing wrong, I'm worse. : )

 

Sorry, need to get my eyes checked (which happens to be this Thursday).  What I meant was drastically increase all the inductor values.  I changed all my mH to H.

In your case 5.83m becomes 5.83, or 5.83h.  I should have said multiply all the inductor values by 1000 (?).

 

In my screenshot, I placed a 13uf capacitor in series with each load, using a 14.6 ohm resistor across taps 0-4.  The plots are the voltage drops across the capacitor.

Gray is the autoformer in Herry.  Green is mH.  Red is the DATS sweep.

 

Sorry again for my poor explanation.

 

 

[Langston]

7 hours ago, mboxler said:

I should have said multiply all the inductor values by 1000 (?).

 

On the correspondence between grey/red trace, I think you got lucky. : ) I'm confident the math in LTSpice is correct and when it gives me different answers than what my measurements say, I know I screwed up. In this case I just don't have the knowledge and equipment to fully characterize a transformer correctly. I did study the issue out and found that it's (surprise) pretty involved, OTOH, simply dropping an impedance measurement (DATS is very good if used correctly) into a Spice Directive gives me results that are dead-on compared to real measurements. So why the bother characterizing this leaf when you can have the whole tree with much less effort?

 

If you do figure out this leaf, please let me know how you did it - it'll be fascinating. : )

 

God bless you and your precious family - Langston

 

[John Warren]

On 6/27/2021 at 2:15 PM, Langston said:

 

I threw up my hands in surrender trying to accurately characterize transformers and autotransformers in LTSpice or elsewhere.

[Langston]

@John Warren 's screenshot unlocked the door for me. Now I'm able to use autotransformers in LTSpice with more than enough accuracy for loudspeaker design. Thanks! : )

 

I'm getting better than 0.1dB correlation to measurements above 10Hz and phase is close enough as well. You do have to use measurement to derive the proper "K factor" for the Spice Directive to get the high frequencies right.

 

The error I was making was attempting to measure inter-tap inductance values. You have to do what John did and make the inductance (and DCR) measurements from each end of the winding to the tap of interest.

 

Here's my LTSpice file for anyone following along:

Klipsch T2A 3110A #1.asc

 

Simulation

 

Measurement

God bless you and your precious family - Langston