My Thoughts and Know-how

23 Feb 2022

I bought one speaker out of curiosity. I wanted to see the value the speaker delivers for just $199 (in 2022!).

B2031A

What a great first impression unboxing and taking it apart!

These things are designed (thanks, Genelec!) and built (thanks, Behringer!) really well: the cabinet is dense and solid, the driver frame is cast, the circuits have good layout and soldering. Speaking of circuits: 3 TDA7293 chip amps power it, fed by a toroid transformer. At the ±42V supply voltage, the power these chips can deliver is over 300W. This is not the 20-year old B2031A design you see reviewed online: on the outside it looks the same, but the PCB is a new revision of the now discontinued B3031A (which I have had for 10 years and like a lot).

The B2031A electronics.

I particularly like the front facing bass ports and the down facing rear connectors that make placement easy.

But then you turn it on... Right out of the box, my B2031A sounded bright. Sure enough, a sweep confirmed the tweeter is 4-5 dB too loud above 4 kHz. They say it is factory calibrated... Did the technician stick the microphone in the woofer?

17 April 2017

I put my barely-taken-out-of-the-box JBL Charge 2+ bluetooth speaker to the test and I was quite impresed!

JBL Charge 2+ frequency response.

The DSP changes the response with volume. This review has detailed measurements.

8 August 2016

The JBL LSR310s sub is so good I had to take it apart. I hesitated buying it since I wasn't a fan of the downfiring and not grill-protected woofer, but having the LSR305s I decided to be brand loyal and wow: -3dB at 32Hz and quite flat above. It is audible with uncolored bass lower, though at much diminished SPL, which my equipment cannot properly measure, so let's say -10 dB at 27 Hz.

Board shots follow, let's go thorugh!

2 June 2015

The JBL LSR305 has a space age looking waveguide that indeed does provide subjectively omnidirectional tweeter pressure, while the wide flared port has practically no wind noise. The resulting sound lacks a bit of detail, but is otherwise outstanding at its $130 price point. Surprisingly, the speaker does not offer over-power protection or any power saving capabilities, both of which could have been implemented using the existing chips. There is also a peculiar bug, which might be destructive. I am particularly interested in the design because of the switched mode power supply, digital filtering and class D power amplification.

Here are the guts, let's dive in!

24 May 2014

How do we add clipping protection, overpower protection and automatic standby to a studio monitor? ...Using a microcontroller! ;-)

Completed board.

Functionality

Block diagram.

The input and output of the LF amp are fed to the differential ADC input of the microcontroller (ATTiny85). When clipping occurs, the output fails to follow the input causing the MCU to mute the power IC (LM4776) until the input signal decreases to a lower level. Similarly when the HF amp overpowers the tweeter (I set the threshold to 15W), the system is also muted.

30 Jan 2014

There are numerous advantages to self-powered bi-amped speakers, including more efficient and quality cross-overs, amplifier-to-driver matching, driver response equalization, and lower distortion. These benefits, however, come at the cost of making these systems complex - a professional studio monitor has 10+ amps, split among the filter sections, protection circuitry, gain stages, etc. These in turn require at least low voltage and high voltage power supplies, leading to a complex design that is hard to build as DIY, more prone to failure, has higher distortion and noise, etc.

Here I present a system of my own design with minimal number of components, yet delivering all benefits listed above. It is based on the popular in studio monitors 4th order (24 dB/octave) Linkwitz-Riley cross-over since it allows for phase aligning the tweeter and woofer at the cross over-frequency. The novelty here is in adding significant filter gain in the pass-band region, which allows us to use the power amplifier as the active element of the active cross-over. This eliminates the need for any opamps and extra voltage supplies. The resulting filter is imperfect (in fact each channel has many poles and zeros), but due to the strength of the power amplifier (compared to an opamp) a wider range of resistor values can be chosen to qualitatively eliminate the imperfection. The frequency response is further shaped by frequency dependent impedence in the feedback loop to boost the bass and cut the supersonic frequency content.

Here is the complete circuit diagram:

A Bi-Amp system with just one active element - a stereo power amplifier IC.

The schematic is concise because components play multiple functions. The block diagram should clarify what is happening.

Block diagram


Bode plot

11 March 2012

A problem of the first revision of this amplifier was the limited precision at low volumes. Although the PWM of the ATtiny45 is just 8-bit, I realized that it can produce arbitrary precision output. Timer1 can be clocked at 64 MHz from the high frequency PLL, which combined with the double-buffered counter compare value and the overflow interrupt, can get 9 and even 10-bit precision in the audible range. Lets say we want to DAC the value 300/512. We set OCR1A to 255 (which will hold the output high) and on the timer overflow, we write 300-255 = 45. Writing this 9-bit value takes 2 timer cycles.

5 Feb 2012

This article is about an obsolete version of the amplifier. For the new revision, go here.

Class D amplifiers are very efficient and could be simple to build. They operate the power stage in a binary mode and pass the delivered pulse train through a low-pass filter (which might involve the speaker itself) to filter all frequency components introduced by the switching of current.

I designed a 70W bridged amplifier based on the Atmel Tiny (attiny45) microcontroller. Vcc is 12V and it requires no heat sink, nor has a feed-back loop of any kind. The output power is limited by 2 factors:

• Precision in the digital domain. Making the gain very high will hurt quality at lower levels.
• I_sd of the power MOSFETs. The ones I had are rated at 6A, potentially delivering P = I^2 R = 144W at 4 Ohm or 72W at 2 Ohm..