L56 crossover error

[wa3drc]

Hello All,

I recently acquired some L166 and L56 speakers. Both needed re-foaming, but the L56's were good enough to try first. (I already have another working pair of L166's that I have had since ~1976) I thought they sounded dead and then after some internet searching see a few complaints about this issue.

While I was re-foaming I did some searching and found schematics of the crossover networks for both speakers. I put the schematics into LTSpiceIV. The response of the L166 crossover looked as I would have expected, BUT boy was I surprised with what I saw for the L56 network! The L56 circuit has a huge notch (-40 dB) at 1.34 kHz. This is due to a series LC (L3 and C5) at the tweeter terminals of the crossover. Along with the notch comes a large jump in phase at the notch frequency as the circuit changes from inductive to capacitive with increasing frequency.

I didn't find any discussion of this behavior when searching. I am wondering if anyone knows what JBL engineering might have been thinking when they put this network there??? Is it possible the speaker response has a huge spike they were trying to counter?

Anyway, I decided to remove this network and listen... I cut the L3 inductor wire at the connection to the positive tweeter terminal. (You can always solder it back) The response in LtSpice looks far more normal when modelling this. I listened for hours after this, and feel they no longer have a void in the middle. Please let me know if anyone else agrees. I did not notice any "peaking" that JBL may have been trying to notch either. I am leaving my speakers with the LC removed.

Please let me know if anyone knows a reason for this series LC being there. I can send frequency response plots and/or LTSpice schematic files to anyone interested.

thanks,
Ed

[fpitas]

How did you model the impedance of the drivers? They probably (certainly?) aren't a simple resistive 8 ohms.

Francis

[wa3drc]

How did you model the impedance of the drivers? They probably (certainly?) aren't a simple resistive 8 ohms.

Francis

good point! I did not create a complex model for the load, I just used 8 ohm resistive. I would have to get an audio generator and try to get a magnitude and phase response of the speaker, I guess. that is why I just cut the lead and listened...

Ed

[rdgrimes]

I had a pair of L56 for a while and never liked them, the L15 have similar issues. I always attributed this to the low crossover point to the 034. Oddly, the L46 does not have this problem and sounds terrific.

[fpitas]

You might also measure the frequency response of the speakers, using Holm Impulse or whatever program you prefer. Nothing beats an actual measurement.

Francis

[wa3drc]

attempting to add simulated response

[fpitas]

A little hard to see, but that tweeter notch looks to be below the crossover frequency (2kHz crossover? something like that).

Francis

[wa3drc]

Francis,

thanks for pointing that out.

yes, true. i went back and took some more measurements. The -3db point for the sub low pass occurs at 530 Hz and the -3db point for the tweeter high pass is at 2.3 kHz. That leaves a pretty good gap.

It's probably not too easy to move the LP filter point up without getting a new inductor and I don't know if the 10" would respond there anyway. Similar concerns on the other side about moving the tweeter HP point down, but easier with a cap change and maybe more probable that the tweeter has some life down a bit lower.

What do you think?

Ed

[fpitas]

Francis,

thanks for pointing that out.

yes, true. i went back and took some more measurements. The -3db point for the sub low pass occurs at 530 Hz and the -3db point for the tweeter high pass is at 2.3 kHz. That leaves a pretty good gap.

It's probably not too easy to move the LP filter point up without getting a new inductor and I don't know if the 10" would respond there anyway. Similar concerns on the other side about moving the tweeter HP point down, but easier with a cap change and maybe more probable that the tweeter has some life down a bit lower.

What do you think?

Ed

I'm not familiar with those drivers, or at least not familiar enough to know what you can and can't do with them. Maybe somebody else here has some illumination.

Francis

[bigredplane]

attempting to add simulated response

I'm confused in JBL N56 schematic what you have mark as L3 they marked L2 and what you marked L2 they have it marked L3.Which one did you cut out the .3mh or the 1.0mh coil?

[ivica]

attempting to add simulated response

hi,

But if You put some resistor 'in series' with the 1mH coil, I believe that this is a kind of Zobel network to compensate the driver impedance.
You can read more about of such compensation....

regards
Ivica

[wa3drc]

When I translated the L56 crossover schematic into LTSpice, I made no attempt to keep the reference identifiers the same as the JBL schematic. The component schematic/netlist should be accurate with the addition of 8 ohm resistive loads to model the sub and tweeter.

The little stub left unconnected was from another simulation I was trying which shorted out capacitor in the output series LC. I actually soldered wires across that capacitor (C5 in the L56 schematic) and brought the wires out the bass port. I then open and shorted the ends to see if I could hear any difference, as shorting them removed the sharp null at 1.34 kHz in the simulation. If there was any difference, it was almost inaudible. As pointed out by Francis, this is probably because this resonant null is well below the cutoff frequency of the high pass filter.

I am still thinking that the "hole" between the low frequency cutoff and the high frequency cutoff is too wide. The woofer frequency either needs to move up or the tweeter need to move down, or a combination of both. I am just not sure of the actual capability of either driver to make that choice.

Ed

[rdgrimes]

I am still thinking that the "hole" between the low frequency cutoff and the high frequency cutoff is too wide. The woofer frequency either needs to move up or the tweeter need to move down, or a combination of both. I am just not sure of the actual capability of either driver to make that choice.

Ed

The spec crossover for the 034 in the L56 is 2200Hz. In the L46 its 3000Hz, in the L86 its 3700Hz.

One can easily hear the result of the low crossover in the L56, and its not good. The harder you push it the worse it sounds. The L46 comes pretty close to eliminating that issue at 3000. The L86 sounds great even when pushed very hard.

If you really want to spend the time/money to "fix" the L56 you'd first need to address the 034's inability to go below 3000. Either by finding a woofer than can cover up to 3000 or finding a tweeter that can go lower. I think the L56 was intended for folks who wanted that extra thump at lower levels. The hole in the mids is not fixable with those drivers.

[wa3drc]

Dear Rdgrimes

Your last reply was very insightful, thanks...

So it is probably hopeless, but I may try to eek a little more out of the woofers by making a change to L1. It looks like reducing that to 0.6mH will raise the cutoff to 2.1kHz. If the woofer has no response up that high, I am probably no worse off.

At this point I'm just playing around. I got these for free and after re-foaming the woofers they look brand new. I still have my L166's for the main stereo and the L56's are just going into my workshop.

Ed

[wa3drc]

Hello All

I haven't given up on this yet. I am learning as I go...

Since the last post, I learned that I may have been chasing a red herring. The snubber circuit I was complaining about is probably to reduce a self resonant peak in the tweeter response and as already pointed out is below the cutoff frequency anyway. So I will be reconnecting L3 to the tweeter terminal.

I am still trying to extend the high frequency end of the woofer and have obtained some 0.6mH inductors to replace L1. I think I want to build a microphone circuit and use something like TrueRTA to chart the stock performance of the speaker before I continue.

Ed

[ivica]

Hello All

I haven't given up on this yet. I am learning as I go...

Since the last post, I learned that I may have been chasing a red herring. The snubber circuit I was complaining about is probably to reduce a self resonant peak in the tweeter response and as already pointed out is below the cutoff frequency anyway. So I will be reconnecting L3 to the tweeter terminal.

I am still trying to extend the high frequency end of the woofer and have obtained some 0.6mH inductors to replace L1. I think I want to build a microphone circuit and use something like TrueRTA to chart the stock performance of the speaker before I continue.

Ed

Hi wa3drc,

There is one possible solution to 'amplify' the response of your bass-driver, in the way that you use L-C low-pass filter but in stead of using driver impedance to be about 8 Ohms, recalculate the network L, C elements as the driver impedance is about 3 Ohms, and then recalculate the response (using SPICE) using L, C but in the simulation put that the bass impedance is 8 Ohms. Let try with these:
C=37.5 uF, and L=0.675 mH, and Bass R=8 Ohms.....
You will get almost +6dB at about 930Hz, while -3dB would be at about 1480Hz, and -6dB at 1670Hz ( all the data are not exact values).

regards
ivica

[fpitas]

Hello All

I haven't given up on this yet. I am learning as I go...

Since the last post, I learned that I may have been chasing a red herring. The snubber circuit I was complaining about is probably to reduce a self resonant peak in the tweeter response and as already pointed out is below the cutoff frequency anyway. So I will be reconnecting L3 to the tweeter terminal.

I am still trying to extend the high frequency end of the woofer and have obtained some 0.6mH inductors to replace L1. I think I want to build a microphone circuit and use something like TrueRTA to chart the stock performance of the speaker before I continue.

Ed

I have and use TrueRTA, but it doesn't feature time windowing. You may want to get a program that does, so you can perform time windowed measurements to take the room out of the equation. Otherwise you'll need to measure outdoors away from reflecting surfaces, or in an anechoic chamber. I use HolmImpulse, which is free and works well other than a tendency to crash occasionally, but there are others around that work fine.

Francis

[4313B]

Hello All,

I recently acquired some L166 and L56 speakers. Both needed re-foaming, but the L56's were good enough to try first. (I already have another working pair of L166's that I have had since ~1976) I thought they sounded dead and then after some internet searching see a few complaints about this issue.

While I was re-foaming I did some searching and found schematics of the crossover networks for both speakers. I put the schematics into LTSpiceIV. The response of the L166 crossover looked as I would have expected, BUT boy was I surprised with what I saw for the L56 network! The L56 circuit has a huge notch (-40 dB) at 1.34 kHz. This is due to a series LC (L3 and C5) at the tweeter terminals of the crossover. Along with the notch comes a large jump in phase at the notch frequency as the circuit changes from inductive to capacitive with increasing frequency.

I didn't find any discussion of this behavior when searching. I am wondering if anyone knows what JBL engineering might have been thinking when they put this network there??? Is it possible the speaker response has a huge spike they were trying to counter?

Anyway, I decided to remove this network and listen... I cut the L3 inductor wire at the connection to the positive tweeter terminal. (You can always solder it back) The response in LtSpice looks far more normal when modelling this. I listened for hours after this, and feel they no longer have a void in the middle. Please let me know if anyone else agrees. I did not notice any "peaking" that JBL may have been trying to notch either. I am leaving my speakers with the LC removed.

Please let me know if anyone knows a reason for this series LC being there. I can send frequency response plots and/or LTSpice schematic files to anyone interested.

thanks,
EdThe conjugate is there to mitigate the resonant frequency of the dome... without it the high pass filter wouldn't behave as intended. If that 1.0 mH coil is a small square coil then it also has the standard 7.5 ohm DCR value that JBL used on many of their domes.

And yes, if you are modeling a filter using a standard resistor load instead of the impedance of a transducer you leave out the conjugate filter(s) that are intended to directly affect specific transducer impedance.

Photo of the two conjugate coils JBL liked to use. Both have a DCR of ~ 7.5 ohms. 1.0 mH for the domes and 3.5 mH for the midranges.

[SEAWOLF97]

One can easily hear the result of the low crossover in the L56, and its not good. The harder you push it the worse it sounds. The L46 comes pretty close to eliminating that issue at 3000. The L86 sounds great even when pushed very hard.

I mostly agree with RD on this one.

Have had 2 pairs of L56's . LF had poor power handling , bottomed out when being pushed a bit.
Had L46 , much better.

L86's are gems. Found no issues. Should have kept them.

[4313B]

Ditto

G.T. did the best he could with what he had to work with though.

I just couldn't hang with anything less than the L96 at that time. L96's on top of B380's, you know, just for fun.

I'm glad people liked the L86. I thought it was a very nice offering using very affordable components.

I replaced too many 034's in the 2-ways.