I bought these speakers in 2000. I was unhappy with the way they reproduce low frequencies. I decided to add a subwoofer. I used the fact that the lower part of the housing of the speaker is empty and isolated from the upper part in which the speakers operate. I cut a hole in the side of the bottom part and installed a 25 cm driver.

It is Vestra PW-250-2154 woofer.It has a 19 cm coated-paper membrane, foam external suspension, extruded metal sheet basket, ventilated 4.5 cm coil, the diameter of the magnet is 12 cm, 80 W, 4 ohms, the self-resonance (Fs) at 27 Hz.

I damped the interior of this newly created enclosure space with bitumen mat and felt glued to the walls and filled it with fleece. I cut out the rear wall from the top part of the enclosure. Inserted a reinforcing crossbars into the lower and upper housing. The upper part of the housing is entirely filled with a damping material (with fleece filled pillow, which...

All measurements done with Panasonic WM-61A microphone with Eric Wallin preamplifier and E-MU0202. Single driver frequency response without any crossover. The microphone is located at a distance of 1 meter from the driver on its main axis.
Vifa 3075 dome tweeter:

Vifa TC 3520 woofer:

Vestra PW-250-2154 woofer with it's crossover, microphone located 10 cm from the driver:


The complete speaker after modification.
Measurements done at my usual listening spot, 2.5 m from the speaker, microphone positioned 50 cm closer, farther, higher and lower around the spot. Obtained frequency response curves are strongly smoothed, because changing the position of the microphone produces a large change in the jags of the curve, and I wanted to show invariant features of the speaker frequency response:

It is pretty flat, and the bass is very nice now....

Finished, front:

back:

what's inside:

This device is built around DIYINHK ES9018 DAC board, with USB to I2C CM6631 converter, also from DIYINHK. USB is the only signal input. Power amplifier uses two LM4780 chips, one per channel in mono-parallel configuration. Power supply for the amplifier is made out of two HP printer power supplies. DAC power supply is more complex, but also uses three, a little bit modified, 12V supplies, wall plug in type, general use ones. Volume regulation (attenuation to be exact) is entirely digital, provided by ES9018 dac chip. It is controlled by PIC16F84 microcontroller via I2C. There is also 3-digit attenuation level display and rotary encoder is used for setting the attenuation level. Details are described in next posts.

DAC:
On the DIYINHK DAC board we can find ES9018 chip and ADP151 1.2V linear regulator (VDD) factory soldered. You have to populate the rest of the board yourself. I have soldered:
- ceramic and electrolytic noise filtering bypass capacitors,
- various resistors, mostly in the I/V stage,
- I/V stage op-amps (4xAD797)
- output signal low-pass filter capacitors.
The board is designed for 6 op-amps with SE output. I wanted differential output, so I have done some modifications, following the ES9018 demo board datasheet.

The I/V stage and power amp schematics (one channel):

When the main power is switched off, the relay shorts the power amp inputs in order to avoid oscillations, generated by I/V stage during discharge of its power supply bypass capacitors, to pass to the power amp.

A/V stage section of the DAC board – one channel, ES9018 chip side (top side in my case)....

POWER AMPLIFIER:
It works in parallel configuration in order to drive 4 ohm speakers easier. Such configuration has quite low input impedance, around 5kOhm, and low gain of 3.7. This, with the 3x20k input signal resistor divider, gives the maximum output power of only 1W for single sine.
The two pairs of 100k, multi-turn potentiometers set the symmetry of each amplifier. It has to be adjusted manually. In order to do it, you have to supply in-phase (single ended) signal to the input of each amplifier and turn the potentiometers until the signal at the output of each amplifier drops to zero (or rather goes below the noise floor). Unfortunately it also affects the voltage at the amplifier output. There is a lot of trial-and-error work needed to achieve zero signal and zero volts at the output. It is good to do it separately for each of the amplifier from the pair by unsoldering output 0.1 ohm resistors. Signal generator and frequency spectrum meter comes in handy for this...

POWER SUPPLIES
Power supply for the amplifier is made out of two HP 0957-2093 printer power supplies (32V, 2.5A output), removed from their original plastic boxes and fastened together with metal bolts, nuts and plastic isolating spacers. This way they form one unit, producing +32V and -32V needed for the LM4780. Additionally there are two banks of six 1500uF/35V low ESR noise bypass capacitors in parallel at the output of the power supply. They greatly reduce switching noise, which has a peak at a frequency of around 30 kHz. There are also four 2200uF/35V caps soldered directly to the amp boards.
I have chosen the switching mode power supply instead of a classic 50Hz transformer because. SMPS is lighter, smaller, voltage regulated, requires less decoupling capacitance, does not produce any audible hum or spurious magnetic field around it . It’s main disadvantage is production of HF noise, but, as I am going to show later, it is not a big issue when the output is properly...

The post-regulation schematics:

ThreeTHX203 SMPS boards soldered to the main power supply board. There are also additional 230V input filter, output bypass caps and gyrator’s TIP122 visible:


The other side of the main board – LT1763 regulators, 5V main power daughterboard with heatsinks:


Details of the 5V main power daughterboard. The top heatsink is for the LM1763, the two on the bottom are for TIP122 (gyrator) and LM7805 (5V for volume controller):


ADM7150 3.3V regulators on separate board with Rubycon ZL 4700uF/6.3V low ESR bypass caps (board bottom):


(board top):


This photos show the how dac power supply is mounted inside the housing:

Graphs showing the dac power supply output noise. You can see the SMPS switching frequency noise peaks at 26kHz and 32kHz. Their frequency rise a little when more current is drawn from the supply. It seems that the dac playing 1kHz is an example of the hardest load to regulate for the supply:







VOLUME CONTROLLER:
It’s a PIC16F84 running a program written by me. It reads rotary encoder and accordingly writes values, via I2C, to the ES9018 attenuation registers. It writes single value to all 8 registers every tick of the encoder. After 10 seconds of no encoder activity, it stores the value to the internal EEPROM of the microcontroller.
The controller board has its own 3.3V regulator, the display board has 1.8V regulator for LEDs
The display module is a separate I2C dumb reader. It catches the attenuation values transmitted...

HOUSING:
Typical steel box consists of four pieces bolted together. Dimensions: 440 x 100 x 255 mm (W x H x D). Notice the "Acqua di Gio" volume knob


Bottom piece. I made two holes to let air flow through power amp heatsinks. There are corresponding holes in the top piece. The black plastic pieces are heatsink mountings:


Top cover and side walls form one piece. I glued same aluminum mesh into heatsink airflow holes in order to protect the inside of the housing.


Detailed view of front piece inner side. Display protecting glass cut out of old picture frame and painted with red transparent paint. Total DIY:


MEASUREMENTS:
Output of the power amplifier loaded with 3.9Ohm resistor, connected to Left line input of the E-MU 0202 with two 10uF unipolar electrolytic caps. E-MU...