Home | Projects | RealAudio™ | About me

Introduction

Update January 2003:

I have decided the analog PCB modifications as originally presented are not acceptable. I have left the text (mostly) viewable for those curious, but cannot at this time continue to recommend modifications to the output buffers.


The Technics SH-AC500D was a (reasonably) low cost Dolby Digital and DTS surround sound decoder produced for a few years by Technics. The unit has since been discontinued, but still shows up occasionally on the used market.

Although the market for surround sound processors has been taken over by the integrated receiver category (at least in the low to moderate price range), I bought a used SH-AC500D because I was converting to an active speaker system which placed the active crossovers and amplifiers at each speaker, so a surround processor and amplifier combination would not be useful to me. Surround processors/preamp combinations are available from a few other manufacturers, such as Mark Levinson/Proceed, Integra Research, Adcom, Lexicon, and Meridian, but I could not justify the expense of a high end processor to start my surround sound system. That left the Technics as the only reasonably priced surround processor for consideration at the time.

The processor is internally organized as four printed circuit boards, one of which holds only a power transformer. The remaining three boards handle the tasks of digital signal input, digital decoding and digital to analog conversion, and analog output buffering and switching. Given the organization, it makes sense to consider separately the three potential modification areas:

Analog board modifications

I have to sheepishly rescind my previous recommendation to lower the output impedance of the output buffer. While I still advocate low output impedance as a general principle, I did not realize the extent to which the performance of the muting control circuit in the SHAC-500D relied on the specific output resistor values to properly mute while changing inputs or decoding modes. Consequently, the modifications I performed resulted in clicks and pops on the output which I have decided are too obtrusive to be acceptable.

The analog board is pretty straight forward, and doesn't have too many areas that just beg for improvement. The first tweak to do is reduce the output impedance. The factory configuration has a relatively high output impedance, specified as 1kOhm, but looks to be closer to 800 Ohms.

A damped inductor is formed by connection a 39 Ohm resistor in parallel with a 4μH inductor (30 gauge magnet wire on a 0.25 inch diameter form, 39 turns in about 0.5 inch length).

The damped inductor is optional. On outputs which are known to have plenty of phase margin, you can sometimes use only a damped inductor, which provides a very low output impedance at low frequencies, which rises to approximately 39 Ohms at higher frequencies (above a couple of MHz). Since the output circuit of the SH-AC500D includes muting transistors to ground, I was not comfortable using such a low impedance device as the only output isolation from the last buffer op-amp, since without access to the microcontroller source code it was difficult to make sure that the muting transistors would never mute while the output op-amp was trying to drive current.

These changes should make the processor slightly less sensitive to the cables used, and allow driving reasonably long cables if needed.

If the processor is driving a balanced input, you should connect identical components from the "cold" side of the balanced line to the ground reference on the analog board for additional noise immunity. This balances the output impedance, but leaves the signals driving the output asymmetrical. The asymmetrical output drive means that you do not get the reduced crosstalk that a symmetrically driven line gives you (which should be inconsequential in typical home theater installations), but the balanced impedance still gives you a reduction in the common-mode to differential-mode conversion of induced noise (assuming you are driving a balanced input on the other end).
Note that for this to be effective above DC, you should match the capacitors to ground on each leg, not just the resistors.

The second tweak after the output impedance is lowered to a reasonable value is to jumper around the CMOS switch at IC401 if you are not going to use the line inputs. This change will likely be subtle, but the CMOS switches do add a very slight amount of noise and distortion, so if you don't need them, might as well avoid the slight degradation.

Digital input board modifications

The digital input board is only used for straight forward input switching, but one glaring omission is that the SH-AC500D only uses a common mode choke on the coaxial inputs, not a transformer. Using a true digital input transformer, such as a Pulse Engineering or Schott, should give better noise immunity on the digital input, and allow the input jack to be configured for proper shielding. The configuration as shipped forces any noise currents on the shield of the coaxial cable onto the PCB ground.

Digital processor board modifications

The digital processor board is implemented using a Cirrus/Crystal Semiconductor CS4926 multi-channel decoder and a CS4226 codec for the digital-to-analog conversion. The codec is an all-in-one type of device which doesn't offer much in the way of tweaking, but one very interesting item to note is that the decoding of Dolby Digital or DTS is done separately from the conversion to analog, which leaves three sets of two channel digital audio accessible between the two chips. Not for the faint of heart, but conceivably the digital signals could be routed to a digital audio tranmitter, and used to drive a higher quality external DAC.


Page last updated: 10 January 2003


Home | Projects | RealAudio™ | About me

Valid XHTML 1.0! Valid CSS! Created with VIM!