KLH Model 5 crossover switch function study

[Carlspeak]

While working on the repair and component upgrading of the subject crossover, I became curious about exactly how the two function switchs affect the frequency response. The switch markings on the back of the speaker are rather vague (i.e. the 2500-7000 switch I call switch "A" has a 'lo, mid and hi' setting). The greater than 7000 Hz switch I call switch switch "B" has the same markings. But, what do they accomplish exactly acoustically?

The model 5 brochure claims "far more precise and repeatable adjustments than do the simple variable resistors offered for high-frequency adjustment of most multi-speaker systems."

I thought that a bit ambitions; claiming a 3 position switch could be more precise than a rheostat of the type used in the early Acoustic Research speakers.

Anyway, using a function generator, Tenma digital dB meter, spare cone midrange and dome tweeter (neither are KLH products), I generated 108 data points via separate dB measurements of each driver at 6 different frequencies (2-12 kHz @ 2kHz intervals) while changing the two switch positions in 9 different combinations of settings.

Obviously, because I used NON KLH drivers in the tests, I can't say with any certainty, what would occur with original KLH drivers. However, I expect the general up/down trends to be about the same.

Here is what I found. (Write me directly at the email address below and I will send the original M'Soft Excel file containing the raw data matrix, charts and additional details of the study). Please ignore the absolute dB readings. Only the trends are of value in this study. There is no explanation why the drop in dB level in the midrange at the 4 and 12 kHz levels. The function generator power setting was set constant for each driver over the frequences tested but was lowered to a starting level of 80 dB for the more efficient midrange driver.

1)At 2 kHz level there was virtually no change in output of either driver at the "A" lo switch position. 1-2 increase in dB level was measured at the mid and hi positions when switch "A" was moved to the mid and hi positions.

2)At the 4 kHz level, the tweeter increased about 7 dB going from lo to hi settings. The midrange stayed about the same or dropped a few dB at the "A" switch mid and hi positions.

3)At the 6 kHz level, the tweeter responded about the same with about a 3 dB increase going from lo to hi. The midrange driver dropped a bit again as in 2).

4)At the 8 kHz level, about the same results as the 6 kHz level.

5)At the 10 kHz level, there was about a 2 dB increase for the tweeter going from lo to hi. The mid dropped again.

6)At the 12 kHz level, the tweeter increased 3-5 dB while the midrange behaved the same as in 2)-5).

carlspeak@aol.com

It's all about the music

Carl

Carl's Custom Loudspeakers

[Guest pbda]

Let me guess, Carl: Are those my crossovers, by any chance? ;-)

[Carlspeak]

>Let me guess, Carl: Are those my crossovers, by any chance?

>;-)

YUP! :-) They will be done next week

It's all about the music

Carl

Carl's Custom Loudspeakers

[mrdb]

There should be no change in the midrange drivers' amplitude: the two switches affect only the tweeter output. See my schematic in the KLH Library: "MRDB's KLH Five Crossover Schematic". It was not clear to me from the description of your setup if you used an amplifier between the function generator and the crossover/drivers. If not, then the source impedance of the generator (probably 50 ohms) would have been loaded by the crossover and that load would vary depending on switch position, causing generator output level to vary and this could explain the midrange level variation. You mention something about keeping levels constant so I am not sure if this is your problem, but it is worth a look.

Bob

[dynaco_dan]

Hi Bob;

I just read your write-up of your KLH Five's in the KLH library.

Nicely done write-up.

Now maybe we can do a schematic for the KLH Twelve speaker system.

It will be interesting to see the differences between the Fives and the Twelves crossover, if there are any, if someone here can do it.

[dynaco_dan]

Hi Carl;

>While working on the repair and component upgrading of the

>subject crossover, I became curious about exactly how the two

>function switchs affect the frequency response. The switch

>markings on the back of the speaker are rather vague (i.e. the

>2500-7000 switch I call switch "A" has a 'lo, mid

>and hi' setting). The greater than 7000 Hz switch I call

>switch switch "B" has the same markings. But, what

>do they accomplish exactly acoustically?

Julian D Hirsch tested the KLH Fives, in Hifi/Stereo Review magazine, (date?).

He mentioned that the woofer, he stated 12", not 10", covers from 20 Hz up to 600 Hz.

The two mids are then covering from, 600 Hz to 3 KHz, where the single tweeter takes over.

He claimed that the two three postion switches adjusted the mids and tweeter about, 2 - 3 db in level, individually.

The mids operate from 300 Hz to 3 Khz and are adjusted by the mid switch.

By them labeling that switch, 2.5 Khz - 7 Khz, this would be the top end of the mids having some adjustment, just a hair below their cutoff.

I will re-read your write-up again, as I see that only the top end, between 2.5 Khz and 3 Khz, should be affected by the mid's adjustment.

The 7 Khz switch is obviously only for the tweeter.

This will be interesting to see why the crossover and the tweeter are differing, between 3 Khz and 7 Khz.

The tweeter is then taking over from 3 Khz, to at least the 14 Khz level that Julian mentioned in his review.

He did not clarify, if each increment or total adjustment, was his intention.

I will assume, that he meant, each increment.

He did comment, that the system went smoothly from 170 Hz - 14 KHz +/- 2.5 db.

Woofer distortion was, at 1 watt drive level, loud according to Julian, 5% @ 30Hz, 8% @ 25 Hz, 6% @ 20 Hz and typically less than 1.5% @ 50 Hz and above.

>

>The model 5 brochure claims "far more precise and

>repeatable adjustments than do the simple variable resistors

>offered for high-frequency adjustment of most multi-speaker

>systems."

>

>I thought that a bit ambitions; claiming a 3 position switch

>could be more precise than a rheostat of the type used in the

>early Acoustic Research speakers.

You are right, Carl, if they had just said, repeatable settings, then I would have agreed with them.

There is much more adjustment with the pots, warts and all, if and when, they are working correctly, but, the switches have a lot going for them, mainly consistancy.

If KLH, had used the Dynaco 5 position switch, there would be more selections of repeatable adjustments at least.

I do not find any repeatablity with the AR pots, at least not consistantly.

After reading his review of the KLH Five, if you covered over the KLH brandname, you could be reading an AR speaker review, they are very similar, which is a compliment.

>

>Anyway, using a function generator, Tenma digital dB meter,

>spare cone midrange and dome tweeter (neither are KLH

>products), I generated 108 data points via separate dB

>measurements of each driver at 6 different frequencies (2-12

>kHz @ 2kHz intervals) while changing the two switch positions

>in 9 different combinations of settings.

>

>Obviously, because I used NON KLH drivers in the tests, I

>can't say with any certainty, what would occur with original

>KLH drivers. However, I expect the general up/down trends to

>be about the same.

>

>Here is what I found. (Write me directly at the email address

>below and I will send the original M'Soft Excel file

>containing the raw data matrix, charts and additional details

>of the study). Please ignore the absolute dB readings. Only

>the trends are of value in this study. There is no explanation

>why the drop in dB level in the midrange at the 4 and 12 kHz

>levels. The function generator power setting was set constant

>for each driver over the frequences tested but was lowered to

>a starting level of 80 dB for the more efficient midrange

>driver.

>

>1)At 2 kHz level there was virtually no change in output of

>either driver at the "A" lo switch position. 1-2

>increase in dB level was measured at the mid and hi positions

>when switch "A" was moved to the mid and hi

>positions.

>2)At the 4 kHz level, the tweeter increased about 7 dB going

>from lo to hi settings. The midrange stayed about the same or

>dropped a few dB at the "A" switch mid and hi

>positions.

>3)At the 6 kHz level, the tweeter responded about the same

>with about a 3 dB increase going from lo to hi. The midrange

>driver dropped a bit again as in 2).

>4)At the 8 kHz level, about the same results as the 6 kHz

>level.

>5)At the 10 kHz level, there was about a 2 dB increase for the

>tweeter going from lo to hi. The mid dropped again.

>6)At the 12 kHz level, the tweeter increased 3-5 dB while the

>midrange behaved the same as in 2)-5).

>

>carlspeak@aol.com

>

>It's all about the music

>

>Carl

>Carl's Custom Loudspeakers

Thank you, Carl, for all of your work in doing this study.

[Carlspeak]

I did not use an amp between the function generator and the crossover.

The function generator's output impedance across the plus and minus clips was 20 ohms.

After re-assembling the crossovers into the cabinets, I ran some response curves using a special pink noise signal and a mike located about 4 inches from the baffle board and centered between the tweeter and mids. The woofer was not yet installed.

Attached is a scan of 5 superimposed curves I generated.

The results support your contention that the mids are not affected - only the tweeter.

Curves 1,2 & 3 show the increase in output in the 2-9 kHz range. Curves 4&5 show the bulging response of the tweeter as the switch was moved to the mid and hi positions.

It's all about the music

Carl

Carl's Custome Loudspeakers

[soundminded]

>While working on the repair and component upgrading of the

>subject crossover, I became curious about exactly how the two

>function switchs affect the frequency response. The switch

>markings on the back of the speaker are rather vague (i.e. the

>2500-7000 switch I call switch "A" has a 'lo, mid

>and hi' setting). The greater than 7000 Hz switch I call

>switch switch "B" has the same markings. But, what

>do they accomplish exactly acoustically?

>

>The model 5 brochure claims "far more precise and

>repeatable adjustments than do the simple variable resistors

>offered for high-frequency adjustment of most multi-speaker

>systems."

>

>I thought that a bit ambitions; claiming a 3 position switch

>could be more precise than a rheostat of the type used in the

>early Acoustic Research speakers.

>

>Anyway, using a function generator, Tenma digital dB meter,

>spare cone midrange and dome tweeter (neither are KLH

>products), I generated 108 data points via separate dB

>measurements of each driver at 6 different frequencies (2-12

>kHz @ 2kHz intervals) while changing the two switch positions

>in 9 different combinations of settings.

>

>Obviously, because I used NON KLH drivers in the tests, I

>can't say with any certainty, what would occur with original

>KLH drivers. However, I expect the general up/down trends to

>be about the same.

>

>Here is what I found. (Write me directly at the email address

>below and I will send the original M'Soft Excel file

>containing the raw data matrix, charts and additional details

>of the study). Please ignore the absolute dB readings. Only

>the trends are of value in this study. There is no explanation

>why the drop in dB level in the midrange at the 4 and 12 kHz

>levels. The function generator power setting was set constant

>for each driver over the frequences tested but was lowered to

>a starting level of 80 dB for the more efficient midrange

>driver.

>

>1)At 2 kHz level there was virtually no change in output of

>either driver at the "A" lo switch position. 1-2

>increase in dB level was measured at the mid and hi positions

>when switch "A" was moved to the mid and hi

>positions.

>2)At the 4 kHz level, the tweeter increased about 7 dB going

>from lo to hi settings. The midrange stayed about the same or

>dropped a few dB at the "A" switch mid and hi

>positions.

>3)At the 6 kHz level, the tweeter responded about the same

>with about a 3 dB increase going from lo to hi. The midrange

>driver dropped a bit again as in 2).

>4)At the 8 kHz level, about the same results as the 6 kHz

>level.

>5)At the 10 kHz level, there was about a 2 dB increase for the

>tweeter going from lo to hi. The mid dropped again.

>6)At the 12 kHz level, the tweeter increased 3-5 dB while the

>midrange behaved the same as in 2)-5).

>

>carlspeak@aol.com

>

>It's all about the music

>

>Carl

>Carl's Custom Loudspeakers

This project illustrates several points about speaker restoration. Because the midrange and tweeter are not KLH 5 model drivers, this speaker is not a KLH model 5, not anymore than a Dhalquist DQ10 is a larger Advent. That is not to say it may not be a better speaker, it's just a different speaker. The level switches on KLH Model 5 operate a low pass filter for the high range and a bandpass filter for the midrange both being in the tweeter circuit. The design and performance of such a circuit depends on the electrical characteristics of the driver which includes both its resistance and its inductance, in fact its complex impedence as a function of frequency. The use of generic crossovers or of a crossover designed for one driver but connected to another can have unpredictable results, often those not exactly intended by the original crossover design. Often designers will start with a standard design such as a Butterworth or Linquist Riley configuration and then tweak it so that it conforms to no standard design but optimizes it according to their own criteria.

Measuring speakers in a meaningful way is also a very difficult problem because how the measurement is made can strongly affect the results. Often measurements are made one meter from the drivers on axis in an anechoic chamber using a calibrated microphone whose response is subtracted out, the drivers are measured individually and the results spliced or presented separately on the same graph. I think this is how AR did it. One problem in measuring drivers simultaneously or assuming their integrated response is that in the crossover region, there will be phase cancellations and reinforcements which not only depend on frequency but on the relative locations of the drivers and the microphone. Close miking can also be affected by diffraction of the baffleboard. KLH model 5 is less prone to this than model 6 because model 6 has side panels which project beyond the front baffleboard creating far more early reflections where model 5 doesn't. Speakers like AR9 use sound absorbing material to further reduce baffleboard reflections.

Finally, the sensitivity of the replacement drivers is critical to maintaining timbral balance of the overall system, even a db or two difference between the original and replacement drivers can have a noticable effect.

In the 1960s when Model 5 was designed and built, the tools available for speaker designers to tweak their designs were far fewer and more primitive than we have today. A project to reverse engineer a speaker like KLH model 5 using contemporary parts is far easier if you have a properly functioning model 5 as a reference and availability of digital active crossovers networks and a 1/3 octave equalier available. This makes selection of the replacement drivers and the ultimate design of a suitable passive crossover to match the optimal performance of the drivers much faster and easier. Once the optimal filter FR is determined, a passive equivalent to obtain the same or similar results is simply a matter of plugging the driver parameters and the desired FR into a software program and it will design a close match for you which you can further tweak.

[Cjj2d]

I have a set of KLH Model 5 crossovers being recapped. one crossover, the two rotary switches are busted and need replacing. This has been an impossible search thus far to find either a used set or even a suitable replacement of any sort.

Is bypassing the switches an option? Would this just take the adjustability away and rely on the amp? I am anxious to get these suckers back to producing great music and have been on the hunt for some switches for a couple months with no luck.