Photos 1 and 2 show the front panel and the component side of the circuit board of the Sansui N-115 crossover.

This addendum was inspired by the floor standing design vs a possible elevated or tilted installation, the latter of which would place the listener's ears close to 'on axis' to the high frequency speakers in the system.

When measuring responses, the two distortion components between the octave of 1khz to 2khz was noticed and that seriously aroused my curiosity. Most of this diatribe is the result of that curiosity.

It has been said that curiosity killed the cat but satisfaction brought it back. It is realized that these speakers are some 55 years old and of little interest to most people. In my defense, I recently (Feb. 2026) read an article on a group of scientists who, for some reason, analyzed Edison's carbon filament light bulb that had been burning for over 120 years.  What they discovered was that unbeknownst to Edison, the carbon filament was being converted to graphene, unknown to Edison or anyone else at that time and wouldn't be discovered and analyzed for another 100 years. Curiosity.

That full article can be found HERE

The system response, on axis, has a 10dB rise between 1500hz and 6500hz, 2.12 octaves. Perhaps Sansui engineers took this into account knowing the treble section would be some 17" below one's ears. At a listening position 7 feet from the speakers, this equates to about 11.4 degrees above the treble section which will subdue the 10dB peak. (see fig. 1 below)

FIGURE 1a

Here, the red curve shows the response at a distance of about 7 feet with the listener seated. The black curve is the on axis response. Both are with the system on the floor, not elevated and both are measured with 1 watt applied.

FIGURE 1b

This set of curves is described as follows.

Lower three curves: Red-Clear Violet(Grey)-Normal Blue-Soft

Top curve: GREEN-with 2mh in woofer and displaced +2dB for clarity as it is coincident with the red curve.

It can be seen that there is no appreciable difference between the red and green curves. This is likely due to the three high frequency sections being high pass. The upper frequency outputs from these three sections would mask the slight attenuation of the 2mh low pass filter in the woofer section.

FIGURE 2

A screen capture of the responses of the SP-3500. Neither this or the above responses are impressive from a technical point of view. This resulted in designing a new crossover, Xover 5, the result of which is briefly described in figs. 3a and 3b

FIGURE 3a

Xover 5  Flat within 5dB from 50hz to 12khz with exception of the dip below 2khz

This crossover, fig.3b was designed to smooth the response which was influenced by helping a fellow AudioKarma member with his SP-3500 pair. In the end, he retained the original crossovers after replacing the capacitors. The washing of the W114 annulus brought f_s to about 40hz which enhanced the bass output to his satisfaction as well as retain the original configuration, and that's what matters, the listener's satisfaction.

FIGURE 3b

Click the image for a larger view

The primary modification that resulted in a flatter response is due to making the high frequency sections band pass while the secondary was in making the woofer section low pass. the original crossover frequencies were retained.

Details, details, details

To satisfy a curiosity, responses were taken on each driver, through the N-115 crossover. All but the driver under test were disconnected. The aim here was to check the rolloff of the filters. Two were as expected and two were not. The two in question were the horn squawker and horn tweeters. Each capacitor was lifted from the foil on one end and measured. The results are shown below.

15uF 50v 14.62uF 0.54 ohms ESR -2.53%

4.7uF 50v 4.74uF 0.49 ohms ESR +0.85%

4.7uF 35v 5.38uF 0.51 ohms ESR +14.47%

2.2uF 50v 2.66uF 0.50 ohms ESR +20.91%

The two lower capacitors were replaced and the result was again, not as expected. In other words, it made no difference despite the 14% and 21% increases. The results are explained in figure 4.

The variance in capacitance from 4.7uF to 5.38uF in this case will cause the crossover frequency to shift lower by some 500hz, from somewhere around 4.2khz to 3.7khz. The variance from 2.2uF to 2.66uF would cause the crossover frequency of the tweeters to shift by about 1.5khz, from 9khz to 7.5khz. In most cases, this would not be a problem.

The impedance of the tweeter pair is 7W at 6500hz, which would require a 2.47uF capacitor to provide a 2nd order filter around 6500hz, yet this crossover uses a 2.2uF., again possibly due to the inductive attenuators. 

Let's delve into that further at the risk of complicating things more.

The figures' nomenclature has been changed here to alphabetic as this was added after the rest of the page and renumbering all the figures and their references in the text could result in serious errors.

FIGURE A

This shows the impedance of the parallel pair of T-104 tweeters; it is clearly 7W at 6500hz

FIGURE B

This is the response of the tweeter pair at 1w1m. At 6500hz, the level is 93dB and at one octave below that, 3300hz, the level is 82dB, an attenuation of -10dB, close enough to the required 12dB for a 2nd order filter and considering the following microphone accuracy.

+/- 1dB 20hz to 10khz      +/-2dB 10khz to 20khz

The slope from 8khz to 4khz is about -4dB. Perhaps Sansui's engineers were taking into account the effect of the inductive attenuators which also alter the crossover from 2nd order high pass in CLEAR to 2nd order partial bandpass in NORMAL and SOFT positions

See the italicized text in the section titled Linearizing the Impedance.

FIGURE 4

Note the reduction of THD below 1500hz. In the original circuit, it's as high as 17% below 1500hz. With the simpler filter (red trace), THD below 1500hz drops to under 3%. (orange trace at bottom) 

Click on image for a larger view

FIGURE 5

This response is that of the S106 horn with a 3khz 2nd order filter, an attempt at reducing or eliminating the nasty distortion peak centered at 1700hz. That attempt obviously failed.

In this figure, RED-with lens, ORANGE-THD

GREEN-without lens, Light Blue-THD

The very light violet (barely visible because it is mostly superimposed on the orange trace is 2nd harmonic without the lens. (Its presence can be seen below 300hz) The DARK VIOLET is 3rd harmonic without lens.

My S.W.A.G. (Scientific Wild Ass Guess) is that the lens in not the cause of the distortion; it merely alters that which is already there. The following images try to show a relationship between the distortion and impedance.

Note that in this set of curves, there are two components of distortion in the 1khz to 2khz octave. These characteristics can be seen in the impedance curves of fig.7

Click on image for a larger view

FIGURE 6

First, a calibration, for whatever it may be worth. CLIO doesn't specify the error tolerance for impedance, inductance and capacitance.

Using a 10W 0.1% resistor, it's impedance (inductive reactance) was measured. The vertical scale is 0.1W per major division and 0.02W per minor division.

The impedance of this resistor is measured as 10.06W. Since the actual resistor can be anywhere between 9.99W and 10.01W, an error window of -0.7% to +0.5%, or a variance of +/-0.6% exists.

FIGURE 7

This is the impedance of the S106 horn. RED is with the lens and GREEN is without the lens. The effect of the lens is apparent, although negligible outside the octave between 1khz and 2khz. Note also the two peaks within that octave, corresponding extremely close to the distortion peaks seen in fig.5

It should be noted that these peaks, as previously stated in fig.5, are not caused by the lens; they are caused by the horn.  

This set of impedance curves are those of an Electro-Voice 1824M horn driver. the GREY trace is the impedance of the driver only, without a horn attached.

 The RED trace is with a 500hz exponential horn attached. The effect of the horn is similar to the effect of the lens in that both reduce the impedance peaks. It can be safely assumed that if a 300hz horn were used, the peak seen in the RED curve at 750hz would be further reduced.

In any case, the purpose of this study was to determine the cause of the two distortion peaks and it seems that the culprit has been exposed.

The horn does it.

These attenuation circuits are interesting in that they change the type of the high pass section. As as shown in fig.9, the filter section is second order, high pass. However, when switched to N or S, it becomes a first order band pass with a second order low pass.

The use of an inductor to attenuate will maintain a more constant impedance due to the impedance of the inductor being frequency dependent whereas the impedance of the  L-pad, being resistive, isn't. Klipsch also used a tapped inductor in the type AA crossover to adjust the midrange level. This was fixed by soldering. Wharfedale, as early as the dirty thirties had a multi tapped inductor (auto-transformer) to adjust output level. This was later replaced by a six position rotary switch which placed a resistor in series and another in parallel with the speaker. Back then in was called a compensated volume control trade named as the Truqual. Today, we know it as the L-pad.

FIGURE 9

Sansui T-103 cone tweeters  as used in fig.9

In each set of curves, RED is position C, GREEN is position N and ORANGE is position S.

The curves on the right, fig.11 are with the 11W resistor, the value of which was found using a 25W potentiometer and multiple sweeps to obtain as close as possible the linear impedances above 6khz.

FIGURE 10

FIGURE 11

FIGURE 12

System impedance above 1800hz hovers around 5W +/-1W very likely due to the inductive attenuators and their parallel resistors.  The hump around 700hz is the lower crossover point. The humps centered at 3khz and 7khz are likely caused by the crossover sections being switched into bandpass sections. Again, see the italicized text in Linearizing the Impedance. 

Red-Clear    Green-Natural    Black-Soft

Click on image for a larger view

The horizontal red line is explained in fig.6 

FIGURE 13

1w1m, midrange horn section only, showing the effect of the three modes of attenuation

 Clear(black), Grey(THD)

Normal(red), Pink(THD)

Soft(green), Blue(THD)

Referring to the italicized text in Linearizing the Impedance, we see the effect of the 2nd order low pass section in the red and green curves. In the normal position, this section is a second order high pass section. When switched to Clear or Soft modes, part of the inductor is in series and the remaining parts are in parallel, rendering this section in the Normal and Soft modes as second order band pass with a low pass section applied. This is seen in the red and green curves with their steep slopes below 2khz.

Click on image for a larger view

Individual responses of the woofer(red), cone mid parallel pair(green), horn mid(orange and horn tweeters parallel pair(black), all measured through the N-115 crossover with 2.83V applied. Only the speaker being measured was connected to the crossover outputs and all measured with both attenuator switches in CLEAR mode.

The woofer is not low pass filtered, it's all pass. See photo 8 below

Ignore the irregularities below 100hz as that is affected by the HVAC fan running (60hz hump) as well as ambient noise (rise below 30hz)

Click on image for a larger view

PHOTO 8

Part of the foil side of the N-115 crossover board. Just right of center are the input terminals, (IN) negative on top and positive below it. The woofer (W) terminals are at bottom center. The negative terminal is on the right and the positive terminal is on the left. The positive of the input pair is directly connected to the positive of the woofer pair.