Improving XDR-F1HD Audio Quality

New custom audio board for the XDR-F1HD

The oscillosope traces below illustrate different amounts of phase shift between the input and the output of the analog board.  A "fat" elipse means greater phase shift is occurring.  The "skinnier" the elipse is, the better the phase accuracy.  Phase shift is near zero at 1KHz and gradually increases with higher frequencies.  These examples are at 10 KHz.

A good option for the XDR is to completely replace the stock audio preamp circuit with a new audio board based on a high-bandwidth opamp.  This provides significant analog audio improvements:  

  • Uses the ultra-premium high-bandwidth Linear Technology LT1632 opamp
  • Uses Nichicon MUSE "Fine Gold" audio-grade coupling capacitors
  • Much flatter frequency response, within +0 dB / -0.2 dB from 20 Hz to 30 KHz
  • Distortion cut in half
  • Audio signal headroom is doubled, so no chance of clipping on "hot" HD stations
  • Improved "phase coherency" (reduced audio phase shift)
  • Resistors and capacitors in the audio path are hand-selected for matching Left & Right performance
  • Silver-bearing solder is used on all connections
  • There are no jacks or sockets
  • Only through-hole components used.  No surface-mount parts.
  • The circuit board is a high-quality 4-layer design with power & ground layers, silkscreened & solder masked.

Sony XDR-F1HD Audio Performance

Sony gave the XDR amazing reception capability, and its DSP section is very high quality.  But its analog audio preamp can be made much better:

  • Flatter frequency response (treble restored)
  • Reduced distortion
  • Improved S/N Ratio
  • ​Doubled headroom for "hot" HD stations (no clipping)

The audio analyzer plots below compare the unmodified XDR audio preamp to my custom audio upgrade board.  All voltage levels = RMS

THD + Noise: Stock XDR Audio Preamp (s/n 0D 130117)

THD + Noise: Custom xdrguy LT1632 Audio Preamp Board

Bench Testing New Analog Audio Board

Measuring % Distortion @ 1KHz, 0.35V Input Signal Level

Frequency Response: XDRGuy LT1632-Based Audio Preamp

Frequency Response: Stock XDR Audio Preamp (s/n 0D 130117)

Copyright © 2016-2017 xdrguy - All rights reserved  

​Measuring the XDR's FM Frequency Response from Antenna In to Line Out

Data suggests the XDR's analog audio circuit has drooping treble response.  What if Sony did that on purpose to compensate for a corresponding treble boost in the digital domain?  Seems unlikely, but not impossible.

To test the aggregate frequency response, I used my HP 8903B Audio Analyzer to generate a clean audio signal.  This was fed into a Panasonic FM Signal Generator and used as the modulation source during testing.  The Panasonic then generates an FM signal at the selected frequency on the FM band and modulates it with the HP's audio.  Think of broadcasting, say, a 1KHz test tone on 97.3 MHz.

Analog FM frequency response is usually stated as being 30 Hz to 15 KHz so that's the range I measured.

Note About FM Pre-Emphasis:

FM radio uses "pre-emphasis" at the broadcast side, and a corresponding "de-emphasis" at the receiver side.  It's kind of like the RIAA curve used for vinyl records.  In North America, the standard is 75 microsecond pre-emphasis.  I used data found here to calculate the expected deviation from flat response, then measured the actual deviation.  I used the data in their "Unlimited 75uS" column (that choice could be wrong, but there's little difference between Unlimited and Roofed in the audio band).  Note that their reference frequency (the 0 dB level) seems to be 50 Hz. They don't show any data below 50 Hz but I took measurements at 30 Hz anyway.

Test Setup:

  • HP 8903B Audio Analyzer reference output set to 2.3V RMS @ 50 Hz.  That Voltage level is accepted by the FM signal generator's external modulation input as being at a correct level, neither under nor over ideal.
  • Panasonic VP-8191A FM Signal Generator was set to use the HP Analyzer as its external modulation source.  I used 75 KHz modulation level.  The FM generator's RF output level was set to -55 dBm @ 97.3 MHz.  I believe -55 dBm corresponds to 65 dBf.  That was connected to Antenna In on the XDR, using a 50 Ohm to 75 Ohm impedance matching cable described here as a "Twelfth-Wave Transformer".  This signal generator is mono.
  • The XDRs were tuned to 97.3 MHz.  With a -55 dBm signal at its Antenna In, the XDR is well into the 3-bar signal strength range.

Test Results:

The table & graph show the measured overall frequency response for Unmodified and Modified XDRs, from Antenna In to Line Out, with compensation for FM de-emphasis happening in the XDR.  In the graph the 0.00 line is the reference, with deviation in +/- dB from there.  The treble dropoff on the Unmodified XDR from Antenna In to Line Out matches up pretty closely with the same measurements when confined to the XDR's analog audio circuit.  

Conclusion from this data: The XDR's treble droop is in the analog audio preamp circuit, not the tuner module or DSP section.