L212 crossover schematics

[DHL]

I am restoring my L212 speakers, and I need the schematics for the cross over networks. I tossed the crossovers years ago when I was planning to tri-amp the L212 with active cross overs, but never got around to completing the project. I found something on this forum under "biasing" posted back in 2004, but the schematics published are very hard to read and appear to have numerous errors. The good news in this post was frequency-impedance plots that showed resonant frequencies for the 066, LE5-9, and the 112A. With this info I can design my own first order 3 ways if needed. This post also had the Zobel networks for each of the drivers which I plan to keep. But the main crossover components seem to be way off what they should be. I am assuming cross over frequencies at 750/6000 (8:1 spread).

Please point me in the right direction if you have these posted in a sticky somewhere.

Also, what is the thinking about applying a third order Butterworth 3 way instead of the first order networks originally used by JBL? Some speaker design guides strongly suggest first order networks be avoided at all costs, as you never get enough attenuation near the resonant frequency because there is only two and max three octaves from the cross over frequencies for these drivers. Third order networks are of course expensive, esp if polypropylene film caps are used and air core inductors.

Thanks in advance for any input.

[Odd]

Here is a link to L212 crossover schematics.

[4313B]

I found something on this forum under "biasing" posted back in 2004, but the schematics published are very hard to read and appear to have numerous errors. The good news in this post was frequency-impedance plots that showed resonant frequencies for the 066, LE5-9, and the 112A. With this info I can design my own first order 3 ways if needed. This post also had the Zobel networks for each of the drivers which I plan to keep. But the main crossover components seem to be way off what they should be. I am assuming cross over frequencies at 750/6000 (8:1 spread).

Please point out what you think are errors in the following thread.

N212 biased network - JBL L212 Loudspeaker System

I can see and read all the schematics just fine on my little laptop. There were three versions of the L212 network.

Also, what is the thinking about applying a third order Butterworth 3 way instead of the first order networks originally used by JBL? Some speaker design guides strongly suggest first order networks be avoided at all costs, as you never get enough attenuation near the resonant frequency because there is only two and max three octaves from the cross over frequencies for these drivers. Third order networks are of course expensive, esp if polypropylene film caps are used and air core inductors.

You can even try LR filters if you want.

The guides suggesting "avoiding this or that at all costs" are usually dealing with run of the mill drivers that can't behave themselves outside of their intended bandwidths. JBL transducers are often reasonably well behaved well outside their intended bandwidths. All the drivers in the L212 system are reasonably benign.

The JBL L212 is an example of JBL's advancement in filter technology during the mid and late seventies. Many of the systems from that era, that John Eargle, Greg Timbers, Gary Margolis, Mark Gander, D.B. Keele, David Smith, and Co. had a hand in, advanced JBL filter technology.

[DHL]

I just got schematics via EMail from HCG customer support. I also was able to see the tech papers posted here under the technical info section for the L212.

The one "error" that I note between the post you reference and the JBL tech notes is the impedance of the various drivers. In the post, the impedances are near 5 or 6 ohms. In the JBL tech specs they are closer to 10-12 ohms. This, of course, makes a huge difference when computing the values for the crossover components.

For example, take the 112A bass driver. With the Zobel network in place, the poster (4313B) has the impedance near 5-6 ohms at 750-1K Hz, which is near the crossover frequency. If you use the cited inductance of 2.5 mH from the schematic you get a corner frequency of 382 Hz, nearly an octave off the 800 Hz cited in the specs. If you use a nominal speaker impedance of 12 ohms, you get closer to 800 Hz. The early JBL data shows the driver at 10-12 ohms, but I do not know if that data includes the impact of the Zobel network or not. Seems to me this network must be included to get the whole system to operate properly.

Next, look at the 066. With the 4 mFd cap and a 10 ohm impedance, the corner frequency is 4000 Hz. Specs say 3000 Hz. With the Zobel network in place from the "4313B" post data, driver impedance is cited at under 5 ohms. This pushes the corner frequency up to 8K Hz. Again, I don't know if those plots are accurate, but they cannot be consistent with component values in the schematic.

And finally the LE5-9. If I use a nominal value of 10 ohms as the load following C1 and L2, and use the calculators for a first order 3-way at 800 hz lower corner frequency, I get a capacitance for C1 of 22.4 mfd, nearly double that of the 12 mfd shown. This means that the actual lower bandpass corner frequency is nearly an octave higher than the 800 Hz cited in the specs, and that of the 112A driver. Similarly, the inductance of L2 is computed at about .44 mH for a upper corner frequency of about 3KHz. The value in the schematic cites L2 at 1 mH, pulling the upper corner frequency to near 1.5 KHz, an octave off the 3KHz spec. This is also strange as it puts both the upper and lower corner frequencies of the bandpass section AT THE SAME frequency of 1.5 KHz. The data plots from JBL show the bandpass section of the 3-way having -6 dB points of 750 and 3000 Hz, as spec'd. But I don't know how they got that data with these components in the crossover.

I am concerned that if I build the crossover as shown in the spec sheet, I will have huge holes near the cited crossover frequencies near 800 and 3K. With the 066 pulled up to 4-8K and the bass driver pulled down to 400 Hz, AND the bandpass narrowed to a hump at 1.5K. Yikes!

[4313B]

The one "error" that I note between the post you reference and the JBL tech notes is the impedance of the various drivers. In the post, the impedances are near 5 or 6 ohms. In the JBL tech specs they are closer to 10-12 ohms. This, of course, makes a huge difference when computing the values for the crossover components.

Those impedance curves were of the actual drivers mounted in the L212 enclosure.

For example, take the 112A bass driver. With the Zobel network in place, the poster (4313B) has the impedance near 5-6 ohms at 750-1K Hz, which is near the crossover frequency. If you use the cited inductance of 2.5 mH from the schematic you get a corner frequency of 382 Hz, nearly an octave off the 800 Hz cited in the specs. If you use a nominal speaker impedance of 12 ohms, you get closer to 800 Hz. The early JBL data shows the driver at 10-12 ohms, but I do not know if that data includes the impact of the Zobel network or not. Seems to me this network must be included to get the whole system to operate properly.

Next, look at the 066. With the 4 mFd cap and a 10 ohm impedance, the corner frequency is 4000 Hz. Specs say 3000 Hz. With the Zobel network in place from the "4313B" post data, driver impedance is cited at under 5 ohms. This pushes the corner frequency up to 8K Hz. Again, I don't know if those plots are accurate, but they cannot be consistent with component values in the schematic.

And finally the LE5-9. If I use a nominal value of 10 ohms as the load following C1 and L2, and use the calculators for a first order 3-way at 800 hz lower corner frequency, I get a capacitance for C1 of 22.4 mfd, nearly double that of the 12 mfd shown. This means that the actual lower bandpass corner frequency is nearly an octave higher than the 800 Hz cited in the specs, and that of the 112A driver. Similarly, the inductance of L2 is computed at about .44 mH for a upper corner frequency of about 3KHz. The value in the schematic cites L2 at 1 mH, pulling the upper corner frequency to near 1.5 KHz, an octave off the 3KHz spec. This is also strange as it puts both the upper and lower corner frequencies of the bandpass section AT THE SAME frequency of 1.5 KHz. The data plots from JBL show the bandpass section of the 3-way having -6 dB points of 750 and 3000 Hz, as spec'd. But I don't know how they got that data with these components in the crossover.

I am concerned that if I build the crossover as shown in the spec sheet, I will have huge holes near the cited crossover frequencies near 800 and 3K. With the 066 pulled up to 4-8K and the bass driver pulled down to 400 Hz, AND the bandpass narrowed to a hump at 1.5K. Yikes!

I'll try and find the original impedance curves and import them into LEAP so we can visualize this.

[DHL]

I have never heard of the term "acoustic crossover" when discussing passive component crossovers. Every design guide and speaker building document refers to the electrical cossover points in the network, as these are supposed to control the acoustic responses of the drivers. Yes, these are commonly "tweaked" in the finished assembly, but by TWO OCTAVES at each crossover point? What do you do when applying an active crossover system? I guess now I am glad I never proceeded with my project to tri amp these drivers.

I did check briefly the impedance of the 066 (my 121A and 112A are out for re-foaming) and it matches the measurements you posted. I get 5-6 ohms, not 10 or 12.


No matter. I will build the stock JBL crossovers as I am planning to sell the system in any case. Just wanted to confirm those values are correct.

[DavidF]

I have never heard of the term "acoustic crossover" when discussing passive component crossovers. Every design guide and speaker building document refers to the electrical cossover points in the network, as these are supposed to control the acoustic responses of the drivers. Yes, these are commonly "tweaked" in the finished assembly, but by TWO OCTAVES at each crossover point? What do you do when applying an active crossover system? I guess now I am glad I never proceeded with my project to tri amp these drivers.

I did check briefly the impedance of the 066 (my 121A and 112A are out for re-foaming) and it matches the measurements you posted. I get 5-6 ohms, not 10 or 12.


No matter. I will build the stock JBL crossovers as I am planning to sell the system in any case. Just wanted to confirm those values are correct.

The values are correct, rest assured.

An electrical filter taking formula values based upon a constant impedance will rarely work as predicted. The driver has its own natural frequency response that, when combined with the electrical filter, will result in the crossover acoutiscal response. If you have two drivers with flat response about two octaves above and below the crossover point, then you may get an acoustical response approximating the predicted response. Hard to do in practice.

[DHL]

Please look at the attached data:



This is a voltage (not SPL acoustic) BODE diagram of the L212 plotted in 1977. These, I assume, were taken to document the Zobel network applied to the 066 to tame the impedance variations near resonance.

Note that the low pass section has a -6 dB crossover point of 900 Hz (not 400). Note that the high pass section has a -6 dB point at 2900 Hz (not 6000). Note that the band pass section has descernable upper and lower -6dB points of 3K Hz and about 700 Hz. This plot could not have been made with the components in the current schematic.

If you use the 4 mfd capacitance with a 6 ohm impedance 066, you will not have a -6 dB electrical response at 3K Hz. The Zobel components serve to smooth the impedance but their effect is still two octaves away from the crossover point. They would have to increase the impedance of the 066 to 10-12 ohms at 3K to get a -6 dB corner frequency of 3K with a 4 mfd cap. And they don't do that.

Likewise, if you use a 2.5 mH inductor with a 6 ohm impedance 112A (from the data measurements posted in the bias post), the -6dB corner frequency of the electronic response is 381 Hz.

So either there was a lot of further modification done after this plot in 1977 was made, or something still does not jive.

And to the point of acoustic response vs. electronic, of course it is the goal of every driver designer to have a linear relationship between acoustic output and drive voltage. This goal is not always achieved, but the best drivers come very close. In fact, I assume that is the reason for the conjugate networks in the first place; to smooth out and level the impedance curves so that the system response will behave more like an ideal first order system should. From the data plots in the bias post, the driver impedance looks very flat, both before and after the crossover points, so that is not the issue.

[4313B]

I did check briefly the impedance of the 066 (my 121A and 112A are out for re-foaming) and it matches the measurements you posted. I get 5-6 ohms, not 10 or 12.

Now run another impedance curve with the L-Pad (set at the 12 o'clock position) in front of the 066 and let us know what you find.

[4313B]

This is a voltage (not SPL acoustic) BODE diagram of the L212 plotted in 1977. These, I assume, were taken to document the Zobel network applied to the 066 to tame the impedance variations near resonance.

It's just a voltage drive. I would have to pull the original to be able to read the notes for context.

Note that the low pass section has a -6 dB crossover point of 900 Hz (not 400). Note that the high pass section has a -6 dB point at 2900 Hz (not 6000). Note that the band pass section has descernable upper and lower -6dB points of 3K Hz and about 700 Hz. This plot could not have been made with the components in the current schematic.

It's just a voltage drive, meaningful in context. It isn't meant to confuse but it can be confusing out of context.

If you use the 4 mfd capacitance with a 6 ohm impedance 066, you will not have a -6 dB electrical response at 3K Hz. The Zobel components serve to smooth the impedance but their effect is still two octaves away from the crossover point. They would have to increase the impedance of the 066 to 10-12 ohms at 3K to get a -6 dB corner frequency of 3K with a 4 mfd cap. And they don't do that.

Likewise, if you use a 2.5 mH inductor with a 6 ohm impedance 112A (from the data measurements posted in the bias post), the -6dB corner frequency of the electronic response is 381 Hz.

Listen to the system and pick out the alleged peaks and dips. Measure it and see what is going on.

So either there was a lot of further modification done after this plot in 1977 was made, or something still does not jive.

There were three crossover versions, all mere Zobel (JBL traditionally referred to them as conjugates) tweaks.

Interestingly there was a fourth unofficial version that further addressed the baffle step response but David Smith can't remember the details.

And to the point of acoustic response vs. electronic, of course it is the goal of every driver designer to have a linear relationship between acoustic output and drive voltage. This goal is not always achieved, but the best drivers come very close. In fact, I assume that is the reason for the conjugate networks in the first place; to smooth out and level the impedance curves so that the system response will behave more like an ideal first order system should. From the data plots in the bias post, the driver impedance looks very flat, both before and after the crossover points, so that is not the issue.

There is acoustical and electrical. Acoustical and mechanical are often used interchangably. An example of an acoustical/mechanical filter would be the L212 0.4 cubic foot sealed box for the 112A/H which rolls the driver off at approximately 12 dB per octave. This mechanical interaction can be represented by an electrical equivalent circuit.

One can pair acoustical and electrical filters to achieve higher order filters. For example, pairing the 12 dB/octave acoustical roll-off of a compression driver with a 12 dB/octave electrical filter to achieve a 24 dB/octave filter.

I guess now I am glad I never proceeded with my project to tri amp these drivers.

Too bad. That could have been fun. You could have whipped up some nice 24 dB/octave LR active filters, put some nice little amps behind them and possibly ended up with something rather spectacular.

No matter. I will build the stock JBL crossovers as I am planning to sell the system in any case. Just wanted to confirm those values are correct.

The schematics are correct.

Too bad you don't have the SFG ferrite version (112H). I'd buy the whole lot off you in a heartbeat.

[4313B]

Exactly.

Sometimes Greg would also write on the plot what he set the L-Pads at for that specific plot. Sometimes he would write "driver loads", sometimes he would write "standard test fixture", sometimes he would just turn the L-Pads off and use them for the loads.

Context.

All the crossover schematics for field service used the standard JBL test fixture. A service center would also have the various high pass filters required to test the high frequency transducers.

[DHL]

It very likely was -not-. These voltage curves are typically run into resistive loads (JBL had at least one
documented standard, IIRC); as such, they are relatively easy to compare and measure in a test fixture.
They also do -not- represent acoustically measured response curves for each driver with the passive filter
elements inserted; those would be separate measurements.

For reasons already stated above, the combined electrical response (into the actual driver impedances,
which are -not- plotted in a crossover "voltage curve") and driver voltage-in vs. SPL-out acoustic
response is what defines the relative levels, in-band response, and effective crossover points.

If this is true, then what were the "standard" resistive loads used as substitutes for the drivers? 8 Ohms? If you use 8 Ohm resistors you will not get the response in the 1977 plots with the values in the schematic. If they were 10-12 ohms then maybe that would explain it.

Plus I doubt what you say only because you would not get the frequency dip and aberration in the 066/high frequency section WITHOUT THE DRIVER. A resistor combined with the Zobel network would give you a negative spike in the response due to the comb filter effect of the network. I think the plot is with drivers, not resistors.

If you folks have an actual crossover connected to the drivers, then reproducing the 1977 plot with actual driver impedances would resolve this issue. That plot should show the new cross over frequencies and confirm the values depicted in the schematic.

[DHL]

Now run another impedance curve with the L-Pad (set at the 12 o'clock position) in front of the 066 and let us know what you find.

Sorry, I don't have either the L-pads or the cross overs.

[DHL]

4313B: "Too bad. That could have been fun. You could have whipped up some nice 24 dB/octave LR active filters, put some nice little amps behind them and possibly ended up with something rather spectacular."

That's what I originally thought. I liked the sound of the L212s and used them for many years, but they needed EQ to sound "right" to me. I have moved away from electronics with OP Amps to Class A discrete and you just cannot find reasonably priced active cross over networks with class A discrete circuitry. Plus the cost of six power amps became prohibitive vs what the L212s give you.

Well, what about using a 3rd order network set up for actual driver impedances and the original cross over frequencies? The driver impedances don't vary that much in the vicinity of the cross over points, and the steep slopes of the cross overs networks would minimize this effect more than a few octaves either side of the cross over frequency in any case. You might not even need the Zobel networks if the cross over frequencies are chosen far enough from the resonant frequencies.

[4313B]

I posted alot of the L212 information here years ago: L212

It does explain the first two network versions. The voltage drives were done with 8 ohm dummy loads and I cannot read what the L-Pads were set at. L-Pads are capable of producing a 10 ohm load with an 8 ohm load behind them at certain settings.

The L212 Service Manual details all three network versions: