Type AA crossover rectangular capacitor replacement

[henry4841]

I played around with crossovers years ago for my LaScala's and liked a crossover without the autotransformer better. Am I right with phase shift being a problem with the autotransformer? I did just put the AA back in mine for posterity though. I am on the downside of the mountain now. 

[mboxler]

I've been told that my T2A model isn't exactly accurate when it comes to phase, so I'm not sure.  I don't think tap 0 greatly affects the tweeter circuit (it's more the series 13uf/2uf).  That said, it looks like the tweeter circuit greatly affects the squawker circuit.   Again this is my Type AA, T2A model, with the squawker (14.2 ohm resistor) on tap 4 (-3.35db).

 

The red trace is the tweeter circuit disconnected.

 

Mike

 

 

[henry4841]

I am sensitive to phase shift. The more parts one adds to a crossover, 3rd, 4th, 5th order the more you have to consider phase shift. I remember one crossover that the mid was reversed on the crossover and years later engineers at Klipsch decided it was wrong and to change it. A simple 1st order, A and for mids AA, you do not have that much of problem with phase shift. Most involves the autotransformer. 

[mboxler]

10 hours ago, captainbeefheart said:

 

Yes same thing different math to show it.

 

BUT, we need the individual values of each "inductor"

 

So the halfway tap (0-3) we have each side totaling 11mH giving a total winding inductance of 42mH.

 

11/4=2.75mH

 

^^^This would be cutting the halfway points in half again down to a value of 2.75mH each "inductor", I just used 2.8mH instead of 2.75mH because that is what the actual autoformer measures because the total winding inductance is closer to 43mH-44mH.

 

Sorry I'm not much help.  If I understand you correctly, you are saying 1/16 of the total autoformer inductance equates to 3db (?).  I'm saying that it's purely the ratio of input turns to output turns that dictates voltage change.

 

In your example (4 2.8mh inductors connected in series and mutually coupled), each inductor would have the same number of turns.  Therefore, if you connect the input to the top of the first and the bottom of the fourth, and connect the output between the first/second inductor and the bottom of the fourth inductor, your total output turns will be 75% of the input turns.  As a result, your output voltage will also be 75% of your input voltage, or -2.5db.  The next tap, 50%, or -6db, and the last tap, 25%, or -12db.

 

If, in your model, you replace the 13uf capacitor with a wire, you should get the same input to output voltage changes as I just described (I think).

 

Mike

 

[captainbeefheart]

12 hours ago, mboxler said:

 

Sorry I'm not much help.  If I understand you correctly, you are saying 1/16 of the total autoformer inductance equates to 3db (?).  I'm saying that it's purely the ratio of input turns to output turns that dictates voltage change.

 

In your example (4 2.8mh inductors connected in series and mutually coupled), each inductor would have the same number of turns.  Therefore, if you connect the input to the top of the first and the bottom of the fourth, and connect the output between the first/second inductor and the bottom of the fourth inductor, your total output turns will be 75% of the input turns.  As a result, your output voltage will also be 75% of your input voltage, or -2.5db.  The next tap, 50%, or -6db, and the last tap, 25%, or -12db.

 

If, in your model, you replace the 13uf capacitor with a wire, you should get the same input to output voltage changes as I just described (I think).

 

Mike

 

 

Hopefully this helps better.

 

Input voltage is 10v (20db).

 

We need the whole winding to be 44mH, each 4 inductors are yes 1/16 of the total inductance because of the squared function. Once you get past -6db you need to halve the next inductor (so divide by 4 again) to get just a -3db past the -6db making the -9db. That is the tricky part to make the exact model work for each tap. I left them all at 2.8mH because I got my tap I needed for the model (-3db) and the whole winding would be modeled correctly at 44mH. If I needed the next tap down would also be correct at -6db. For this to make more sense for you and everyone else I am showing the model with the other -9db and -12db inductance values for the model to work exactly how it would in real life.

 

Starting at the top (20db), each tap going down will lower it -3db, so 17db, 14db, 11db, 8db. In the pictures, the graph is named at the top for which tap we are viewing.

[mboxler]

Please don't get me wrong.  If you are satisfied with your results then that's all that matters.

 

I assume you meant 44.8 total inductance (2.8 * 16) 🙂.

 

Your tests appear to confirm my math.  For example, tap 4 represents 75% of the total windings, therefore the output voltage will be 75% of the input voltage, or -2.5db.  I hope I read your plot correctly.  Likewise, tap 2 represents 37.5% of the total windings, resulting in a -8.52db drop in voltage.

 

I guess I'm too much of a perfectionist, and was trying to get your model closer to  -3db and -9db.

 

I was also reminding all that on the T2A, tap 3 is actually -3.35db.  I often wonder, regarding a new Crites Type AA crossover, that the combination of the 3636 autoformer on the -3db tap and the lower ESR Sonicaps might be the reason it seems brighter than a T2A/polyester cap combination.

 

Mike

 

 

 

[captainbeefheart]

3 hours ago, mboxler said:

I was also reminding all that on the T2A, tap 3 is actually -3.35db

 

Don't forget about the 13uF cap, at 1kHz it is 12 ohms. If you are looking at the crossover reduction of -3.35db the capacitor is also attenuating the input signal.

 

[henry4841]

3 hours ago, mboxler said:

the lower ESR Sonicaps might be the reason it seems brighter than a T2A/polyester cap combination.

Too much emphasis is put on low ESR on caps. When you reach an acceptable level of low ESR lowering it some more is not going to have any more effect. Not to say that Sonicaps are any lower then any of the good audio caps out there. Sonicaps are Bob Crites proprietary caps and he does not make them. They are some brand of cap that is readily sold by another name.  

[captainbeefheart]

49 minutes ago, henry4841 said:

Too much emphasis is put on low ESR on caps. When you reach an acceptable level of low ESR lowering it some more is not going to have any more effect. Not to say that Sonicaps are any lower then any of the good audio caps out there. Sonicaps are Bob Crites proprietary caps and he does not make them. They are some brand of cap that is readily sold by another name.  

 

They are made by "sonicraft".

 

I was playing around with the simulation and it looks like an ESR of 1 ohm might be audible as that's where you start to see an any  added attenuation. My speakers sound great with the stock capacitors and the last time I checked the ESR the 2uF caps were not above 1 ohm for frequencies of interest. They did increase as frequency was reduced down near 100Hz but that won't effect the sound so I didn't replace them. I too question the difference of .01 ESR for a polypropylene vs .1 ESR polyester can be audible. It could possibly help dampen at extremely high frequencies where the reactance of the caps are so low the ESR comes actually comes into play.

[Deang]

1 hour ago, henry4841 said:

Sonicaps are Bob Crites proprietary caps and he does not make them. They are some brand of cap that is readily sold by another name.  

Nope. They are wound for Jeff Glowacki at Sonicraft, who helped engineer them. The Sonicap is not rebranded or sold under another name. 

[Deang]

With a 3636 autoformer you can attenuate in 1dB steps. I knock off a dB on the midrange and use a 10uF capacitor. 

[captainbeefheart]

35 minutes ago, Deang said:

I knock off a dB on the midrange and use a 10uF capacitor. 

 

Many more capacitor choices out there for a 10uF value.

[Tizman]

I ended up with the stock AA crossover values for the capacitors.  The KHorns sound muddled overall as compared to my La Scalas.  I feel that the midrange could be louder as compared to the bass.  Perhaps I'll try changing the autoformer tap so that the squawker is 3 DB up.  Something is off.  The KHorn has deeper bass, more bass and better quality bass, but the La Scalas sound much more cohesive in the higher frequencies.      

[Deang]

6 hours ago, Tizman said:

I ended up with the stock AA crossover values for the capacitors. 

 

You're saying you changed out the capacitors for new ones?

 

Are those old autoformers really T2A's?

[Tizman]

On 12/12/2021 at 7:56 AM, Deang said:

 

You're saying you changed out the capacitors for new ones?

 

Are those old autoformers really T2A's?

Yes, T2As.  The crossovers are newer than the bass bins.  I replaced the capacitors for new ones of the same values.