Peachtree iDecco D/A integrated amplifier Measurements
Sidebar 4: Measurements
To perform the measurements on the Peachtree iDecco, I mostly used Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system (see the January 2008 "As We See It" and www.ap.com); for some tests, I also used my vintage Audio Precision System One Dual Domain and the Miller Audio Research Jitter Analyzer.
Before I did any testing of the iDecco, I ran it at one-third power into 8 ohms for an hour, which imposes the maximum heat stress on an amplifier with a class-AB output stage. At the end of that time, the iDecco was hot but not bothered.
Looking first at the iDecco's performance as a digital decoder, a full-scale 1kHz tone clipped the amplifier's output stage with the volume control set to 2:30. The level from the variable preamp outputs with the tube in-circuit was 848mV in this condition. The maximum level from the line-level output jacks was 2.06V, sourced from a low impedance of 10 ohms, and while both the loudspeaker outputs and the fixed-level line outs preserved absolute polarity for digital sources, the variable preamp outputs inverted polarity with the tube. Other than when noted, I continued the digital testing from the fixed line-level outputs with the volume control set to its minimum, to avoid stressing the iDecco's amplifier output stage.
The iDecco locked to S/PDIF datastreams with sample rates ranging from 32 to 96kHz, but not to data with sample rates greater than 96kHz. The top two pairs of traces in fig.1 show the iDecco's D/A frequency response with 44.1kHz data (cyan, magenta traces) and 96kHz data (blue, red) with the rear-panel Filter pushbutton set to Fast. The response is flat and extended at both frequency extremes, at least until the inevitable steep rolloff just below half the sample rate. By contrast, the green and gray traces in fig.1 show the response with 44.1kHz taken from the variable preamp outputs with the tube operating, the volume control set to 2:00 and the Filter set to Slow, offset by 1dB for clarity. There is now a slight (0.3dB) imbalance between the channels and the low-frequency response is down 3dB at 11Hz. At the other end of the spectrum, the effect of the Slow filter is to roll off the top-octave output a little early, the response being down 3dB at 19.5kHz.
Fig.1 Peachtree iDecco, frequency response at 12dBFS into 100k ohms from fixed outputs with data sampled at 44.1kHz (left channel cyan, right magenta) and 96kHz (left blue, right red), and from variable outputs with volume control set to 2:00 (left green, right gray). (1dB/vertical div.)
Testing the DAC's resolution with a swept bandpass filter while it decoded a dithered 1kHz tone at 90dBFS gave the traces shown in fig.2. The top pair of traces were taken at the variable preamp jacks, again set to 2:00, with 16-bit data, and the bottom pair with 24-bit data; you can see that the increase in bit depth drops the noise floor by 10dB or so in the treble. There is a slight bump at the 60Hz AC line frequency, but this is sufficiently far down in level not to be an issue. Repeating the analysis, this time from the fixed-level jacks with an FFT technique, gave the traces shown in fig.3. The increase in bit depth now drops the noise floor by 18dB, which implies that the iDecco's ESS 9600 Sabre chip has at least 19-bit resolution. However, some harmonics of the AC frequency can be seen with the 24-bit data (blue, red traces), and a regular series of distortion harmonics is also unmasked by the lowering of the noise floor. DAC linearity error with 16-bit data (not shown) was vanishingly low to below 100dBFS, and the iDecco's noise floor was low enough to allow the three DC voltage levels that describe an undithered 16-bit tone at 90.31dBFS to be readily resolved (fig.4). With undithered 24-bit data, the result was a noisy but otherwise well-defined sinewave (fig.5).
Fig.2 Peachtree iDecco, 1/3-octave spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with 16-bit data (top) and 24-bit data (bottom). (Right channel dashed.)
Fig.3 Peachtree iDecco, FFT-derived spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with: 16-bit data (left channel cyan, right magenta), 24-bit data (left blue, right red).
Fig.4 Peachtree iDecco, waveform of undithered 1kHz sinewave at 90.31dBFS, 16-bit data (left channel blue, right red).
Fig.5 Peachtree iDecco, waveform of undithered 1kHz sinewave at 90.31dBFS, 16-bit data (left channel blue, right red).
When it came to distortion, the two line-level outputs varied dramatically. The blue and red traces in fig.6 show the spectrum of a full-scale 50Hz tone at the fixed-output jacks. The second harmonic lies at 90dB (0.003%), the third at 80dB (0.01%), and the fourth at 110dB (0.0003%). By contrast, while the third harmonic remains at the same level from the tubed variable-output jacks, the fourth harmonic has risen to 90dB and, more significant, the second has risen to 50dB (0.3%). In both cases, a picket fence of very-low-level spuriae is also visible. The primary difference between the two outputs is that, with the variable jacks, the signal passes through a 6922 tube. It is the "bent" transfer function of this tube that generates the even-order distortion.
Fig.6 Peachtree iDecco, spectrum of 50Hz sinewave, DC10kHz, at 0dBFS into 100k ohms from fixed outputs (left channel blue, right red), and from variable outputs with volume control set to 2:00 (left channel cyan, right magenta). (Linear frequency scale.)
As can be seen in fig.1, the slow in "Slow Filter" refers to the rate of the reconstruction filter's ultrasonic rolloff. It is generally felt that a slower rate of rolloff sounds better, but the downside is that there is less rejection of the ultrasonic images that result from the digitizing of the signal. This can be seen in fig.7, which shows the spectrum of the iDecco's fixed outputs while it decoded 24-bit data representing an equal mix of high-level 19 and 20kHz tones. The primary ultrasonic image of the two tones is suppressed by just 12dB, and other aliasing spuriae are folded down into the audioband. Switching this filter to Fast gives the spectrum shown in fig.8: the ultrasonic images have dropped significantly in level, as have the audioband aliasing spuriae. The primary intermodulation product, at 1kHz, has risen slightly, to 96dB (0.0015%), but this is still negligible in absolute terms.
Fig.7 Peachtree iDecco, HF intermodulation spectrum, Slow Filter, DC24kHz, 19+20kHz at 0dBFS into 100k ohms (linear frequency scale).
Fig.8 Peachtree iDecco, HF intermodulation spectrum, Fast Filter, DC24kHz, 19+20kHz at 0dBFS into 100k ohms (linear frequency scale).
The iDecco offers three choices of digital input: S/PDIF (on coaxial and TosLink connectors), USB, and from an iPod plugged into the top-panel dock; a second rear-panel pushbutton selects between Wide and Narrow receiver PLL bandwidths. The Wide setting is to allow the iDecco to successfully lock to digital sources with poor-tolerance clocks; however, the downside is that this gives rise to increased levels of jitter. Fig.9, for example, shows the spectrum with the 16-bit J-Test signal fed to the iDecco via TosLink. Though data-related jitter other than the sidebands at ±229Hz is at the residual level, there are strong sidebands at ±1.8kHz, and a significant widening of the central spectral peak due to random low-frequency clock variations. The jitter level was 519 picoseconds peakpeak, according to the Miller Analyzer. This is still low in absolute terms, but switching to the Narrow setting lowered the jitter level to below 200ps, narrowed the central peak in the spectrum, eliminated the high-frequency sidebands, and reduced all data-related sidebands to the residual level of the test signal (fig.10), which is superb performance.
Fig.9 Peachtree iDecco, Wide receiver, high-resolution jitter spectrum of analog output signal, 11.025kHz at 6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz, 16-bit data from SYS2722 via 15' TosLink. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).
Fig.10 Peachtree iDecco, Narrow receiver, high-resolution jitter spectrum of analog output signal, 11.025kHz at 6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz, 16-bit data from SYS2722 via 15' TosLink. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).
The iDecco takes data from an iPod in digital format; repeating the jitter test with the 16-bit J-Test signal playing on an iPod Classic 160GB gave a low 310ps of jitter. While there were no data-related sidebands present in the spectrum (fig.11), there were sidebands of unknown origin present at ±91 and ±816Hz. Feeding USB data from my MacBook gave just 260ps of jitter, from sidebands at ±1423Hz (not shown). The iDecco offers excellent rejection of jitter, though with both iPod and USB sources there was significant widening of the central spectral peak, again due to the presence of random low-frequency jitter. Examining the iDecco with the Mac's USB Prober utility identified the USB receiver as a "USB Audio DAC" from "Burr Brown from TI" operating in adaptive isochronous mode, and indicated that it accepted 16-bit data with sample rates of 32, 44.1, and 48kHz only.
Fig.11 Peachtree iDecco, Narrow receiver, high-resolution jitter spectrum of analog output signal, 11.025kHz at 6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz, 16-bit data from iPod Classic. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).
Moving on to the iDecco's amplification section, the output from the speaker jacks preserved polarity (ie, was non-inverting) for digital inputs, but inverted absolute polarity for analog signals fed to the Aux input via the tube. The maximum gain for analog input signals was fairly low for an integrated amplifier, at 28.7dB into 8 ohms. The Aux input impedance was usefully high, at 48k ohms at low and middle frequencies, dropping slightly but inconsequentially to 42k ohms at the top of the audioband.
I had to scratch my head a bit when I examined the iDecco's output impedance. I perform this measurement by comparing the output voltage with the amplifier driving 8 or 4 ohms with how much that voltage rises when I remove the load. A simple Ohm's Law calculation then gives me the amplifier's output impedance (including the cables in use, of course). However, when I disconnected the load with the iDecco, the output voltage dropped, suggesting that the Peachtree actually has a negative output impedance of about 0.2 ohm. This can be seen in the plot of the amplifier's frequency response (fig.12), where the output with the 2 ohm load (green trace) is higher than it is with 4 or 8 ohms. This is unusual, and suggests that the amplifier circuit uses a degree of positive feedback. However, the variation in response with our simulated speaker (gray) is very low.
Fig.12 Peachtree iDecco, frequency response at 2.83V into: simulated loudspeaker load (gray), 8 ohms (left channel blue, right red), 4 ohms (left cyan, right magenta), 2 ohms (green). (1dB/vertical div.)
Note the low-frequency rolloff in fig.12, which, at 3dB at 55Hz, is even more extreme than that seen in the response from the preamp outputs (fig.1, green and gray traces). I was puzzled by this, but after some thought I repeated the response measurements at various settings of the volume control. The traces in fig.12, taken with the volume control at its maximum, are repeated as the bottom pair of traces in fig.13. The other traces in this graph were taken at progressively lower settings of the volume control, and you can see that the iDecco's low frequencies are well extended at volume-control settings of 12:00 and below. It looks as if the iDecco reduces its low-frequency bandwidth as you increase the volume beyond 12:00, and that this is performed in the tubed preamp section.
Fig.13 Peachtree iDecco, frequency response at 2.83V into 8 ohms with volume control set to (from right to left): maximum, 3:00, 2:00, 12:00, 9:00 (left channel blue, right red, 1dB/vertical div.).
Fig.12 indicates that the iDecco's ultrasonic response extends well beyond the 200kHz limit of this graph. This correlates with very short risetimes on its response to a 10kHz squarewave, but there is a small degree of overshoot with very-high-frequency ringing into lower impedances (fig.14). Channel separation was good rather than great, at 87dB RL and 59dB LR at 1kHz, as was the wideband, unweighted signal/noise ratio (taken with the volume control at its maximum but the Aux input shorted) at 67.5dB ref. 1W into 8 ohms. This was primarily due to low-level spuriae at 120Hz and its harmonics that I couldn't eliminate by experimenting with the grounding between the iDecco and the Audio Precision test system. A-weighting the S/N ratio, which discounts the effects of LF and HF noise, thus improved the result to 83dB.
Fig.14 Peachtree iDecco, small-signal 10kHz squarewave into 4 ohms.
Peachtree specifies the iDecco's maximum output power as 40Wpc into 6 ohms (14.8dBW) rather than the usual 8 ohms. However, fig.15 shows that the amplifier didn't clip (defined as 1% THD) until 40Wpc into 8 ohms (16dBW) and 53Wpc into 4 ohms (14.2dBW). The low-power distortion is around 0.03%, but increases in a linear fashion with increasing power, this due to the mainly second-harmonic distortion introduced by the preamplifier stage (figs.16 and 17). Intermodulation distortion with an equal mix of high-frequency tones at a level just below visible clipping on the oscilloscope screen (fig.18) was primarily the low-order difference product, at 50dB (0.3%), again this stemming from the tubed preamp stage; but the power-supplyrelated spuriae can also be seen in this graph, admittedly at a low level.
Fig.15 Peachtree iDecco, distortion (%) vs 1kHz continuous output power into (from bottom to top at 1W): 8, 4 ohms.
Fig.16 Peachtree iDecco, 1kHz waveform at 1W into 8 ohms (top), 0.155% THD+N; distortion and noise waveform with fundamental notched out (bottom, not to scale).
Fig.17 Peachtree iDecco, spectrum of 50Hz sinewave, DC10kHz, at 20W into 8 ohms (left channel blue, right red; linear frequency scale).
Fig.18 Peachtree iDecco, HF intermodulation spectrum, DC24kHz, 19+20kHz at 20W peak into 8 ohms (linear frequency scale).
Peachtree's iDecco may be budget-priced, but it packs a lot of functionality into a small package. Its D/A section is particularly impressive technically, especially regarding its resolution and rejection of jitter, but you need to take the output from the fixed line-level jacks to get the maximum performance from it. As an integrated amplifier, the iDecco's performance is dominated by its tubed preamp section, which introduces even-order harmonic distortion much like that of a classic tubed amplifier.John Atkinson