A network for bi-amping

[Al Klappenberger]

I have always said that if you have to ask about bi-amping, don't try it. It is a bag of worms! Here's a little glimpse inside the bag.

Some of the advantage to bi-amping are:

1 - The ability to connect the woofer directly to an amplifier doing completely away with the losses involved in the large inductors required at low frequency.
2 - The ability to easily adjust the crossover frequency and digitally correct for time delay difference caused by the long path through a folded horn woofer.

The complications are:

1 - Two amps are required.
2 - An active crossover is ABSOLUTELY required AHEAD of the two amplifiers. Trying to connect the output of your preamp into both amplifiers is a VERY bad idea!
3 - Establishing the correct phasing between the two amps is difficult to do.
4 - Each amp may have a gain control and the active crossover my also have its own gain controls. These may interact and setting them may cause confusion.
5 - The amp used for the high frequencies must not have any turn-on or turn-off "thumps" or the squawker driver could be destroyed in an instant.
6 - A passive crossover is still required to operate the squawker and tweeter. The efficiencies of these two very with manufacturer and model and either one or the other may require an attenuator to make it match the other. This is the problem I am going to try to illustrate here.

The way most of us run our speakers is through a single amplifier using a passive crossover to divide the sound into three ranges. The the case of the Khorn, the woofer gets the extreme lows below about 400 Hz. The squawker gets everything between 400 Hz and 6000 Hz. The tweeter gets everything else from 6000 Hz up. The woofer is always the least efficient driver requiring the other two (squawker and tweeter) to be cut down to equal the woofer.

Here's the block diagram of a 3-way speaker using a bi-amp scheme to allow a high power amp to run the woofer and a smaller one, maybe a vacuum tube type, to run the high frequency section.

[Al Klappenberger]

I plotted several well known drivers on the same grid to illustrate the level matching problem. The traces compare the sensitivity of the Klipsch K77M and the Crites CT125 tweeters to the K55V (solder lug version) squawker on a K500 (Belle Klipsch) horn. I also used a B&C DE10 for one of the tests. It's not plotted but has higher sensitivity than even the K77M. The Crites CT125 is the lowest.

The thing to notice here is that using the K55V squawker with K77M tweeter would require an attenuator in the line to the K77M tweeter because it is more sensitive then the K55. If you used the Crites CT125, it's the K55 squawker that would need to be attenuated to make the frequency response flat. In addition to figuring out where to put the attenuator there is the problem of making sure the drivers are all the correct impedance to match the passive filters in the crossover. The K55V is 16 Ohms (13 Ohms actually) and most tweeters are 8 Ohms. This involves transformers and the correct value and placement of swamping resistors. All this will depend on what combination of squawker drivers and tweeters you are using. Setting this combination to equal the woofer is a simple matter of adjusting the gains of the two amplifiers and active 400 Hz. crossover.

[Al Klappenberger]

The first of the two cases involved in equalizing the two drivers is when the tweeter is more sensitive then the squawker. The examples in the plots are when using the K55 squawker with the K77M or B&C DE10 tweeters. In these cases an attenuator is required in the tweeter channel. I offer the ES5800 network and the new AP15-6000 I'll describe later. Both of these have a transformer attenuator in the tweeter channel. Both are also designed to be used with 8 Ohm drivers. The K55V is actually 13 Ohms so it can't be connected directly to the 8 Ohm squawker channel of either network without an additional swamping resistor connected across the driver to properly terminate the filter. This needs to be about 22 Ohms. The plots below use the ES5800 network with the K77M and B&C tweeters and the 22 Ohms resistor. The tiny amount of power drawn from the amp by the 22 Ohm resistor is totally insignificant. I can assure you that it will not raise your electric bill!

[Al Klappenberger]

The opposite case is when the squawker is more sensitive than the tweeter requiring the attenuator to be in the squawker channel. Here is where the Universal network could be used as it has the transformer in the squawker channel, runs the tweeter wide open and has an input usable for bi-amping. It is also the reason I have always suggested the Crites CT125 was perfect for use with the Universal network. The K77M tweeter should be attenuated with an external attenuator when used with the Universal network. Looking at the previous plot should show you why.

Using the Universal network when inputting to the bi-amp input leaves the 400 Hz woofer to squawker section of the network idle (1.3 mHy inductor and the two 24 uF caps). It's not used at all. The AP15-6000 network used in the next plots is simply the 6000 Hz squawker to tweeter section of the Universal network built on a separate board. Either the tweeter attenuator or the 3619 transform can be included on the board depending on which driver needs to be attenuated. When outfitted with the tweeter attenuator it is usable with the AP12-350 or AP12-500 woofer to squawker gentle-slope networks through a single 3-wire cable in case you ever want to convert back to using a single amp.

The top plot of the two below shows what happens when the K77M is connected to the new AP15-6000 or Universal networks along with the K55 squawker. The 3619 transformer provides attenuation to the wrong driver! The K55 just can't keep up with the hotter K77M. Even if the 3619 transformer was set to 0-5, completely bypassing it, the K77M is still hotter. The lower plot is with the Crites CT125 tweeter. It's just right with the transformer set to 5.5 dB attenuation (X-4). Here the 10 Ohm swamping resistor across the transformer matches the squawker driver to the 8 Ohm filter just as it does in the Universal network. After all, the AP15-6000 IS THE UNIVERSAL NETWROK 6000 Hz SECTION!

[Al Klappenberger]

Here is a photo of the AP15-6000T with the 3619 transformer mounted on it for use when the squawker needs to be attenuated rather then the tweeter. Normally, I mount the 3619 on the woofer to squawker network and feed the squawker channel output back to it through a 3-wire cable. That's to allow 2-way operation where there is no tweeter. Everybody knows the values of the parts because they are exactly those of the Universal network.

[Al Klappenberger]

Here is the layout drawings showing the two versions of the AP15-6000 network. One has the transformer attenuator in the tweeter channel. The other is wired identically to the Universal network to attenuate the squawker. The part values are shown but are identical to those used in the Universal network.

Al K.

[Al Klappenberger]

Here's the plots comparing the ES5800 to the AP15-6000 / Universal networks.