Prism Sound Orpheus FireWire Interface

This prodigious Orpheus has the potential to rescue FireWire audio from an underworld of mediocrity.
Author:
Publish date:
Social count:
0

Like the harp-wielding Orpheus of Greek mythology – born to the wisest of muses Calliope – the modern-day Orpheus that now sits before me also comes from impressive lineage: high-end converter, preamp and test/measurement product manufacturer Prism Sound.

Prism's first foray into FireWire audio interfacing was the high-resolution, very high-end ADA-8 XR 16-channel conversion system ($10–12k). Trailing in the ADA-8 XR's footsteps, the Orpheus puts Prism's revered sonics and conversion integrity into a user-friendly package, with a comparatively accessible price of $5,000.

Image placeholder title

Features

The 1U Orpheus follows a common FireWire audio interface I/O configuration eight analog recording inputs and eight monitoring outputs, optical I/O (for TOSLINK S/PDIF or 8-channel or 4-channel SMUX-ed ADAT signals), digital I/O on coax RCA (for S/PDIF or AES3 signals; AES adapters included), MIDI I/O, Word Clock I/O (on BNC) and parallel FireWire 400 connections. A high-quality real-time sample rate converter is included for use on S/PDIF input or output paths.

All line-level inputs and outputs are available on TRS 1/4-inch connectors and the first four recording inputs also accept XLR mic-level inputs (via combo jacks). Inputs 1 & 2 also as double high-impedance instrument inputs on two front-panel 1/4-inch jacks. Also on the front panel are two stereo headphone outputs that share the same monitoring source (user-selectable; more below), but feature independent amplifiers/level controls. Perhaps the most prominent feature of the front panel is the oversized, LED-ringed and stepped rotary encoder, assignable to control the level of any or all of the analog plus digital 1/2 output pairs. This inclusion makes the Orpheus a capable two-to-eight-channel monitoring controller. Finally, a meter panel provides selectable input or output metering of the eight analog channels plus stereo digital, as well as source/locked status and channel 1 – 4 input settings.

The Orpheus is supported on Macs with OS X 10.4 or later (PPC or Intel platform) and on Windows PCs with XP or Vista. Multiple units can be daisy-chained, with the maximum number of units determined by the desired sample rate (six at 48kHz, one at the Orpheus' max sample rate of 192kHz). Please refer to the company's website for full compatibility and expected performance vs. latency ranges.

Image placeholder title

In Use

Though it walks like a duck and looks like a duck, the single-space Orpheus FireWire interface is head-over-tail-feathers above the flock. Based on my use, and as the bench tests confirmed, the Orpheus is clearly not a product of bandwagon marketing mentality, and is no Prism Lite. The high plane from which Prism typically operates is evident throughout the Orpheus, from its controls, rock-solid rear panel connectors and overall physical implementation to the number of thoughtful and high-end features incorporated under the hood.

This high plane is also evident where it counts most: Sonic quality. Though it's near impossible to properly A/B FireWire interfaces, there was a noticeable overall improvement in clarity and detail with the Orpheus over an RME Fireface 800, and a fairly dramatic difference over a PreSonus FirePod. Of course, there is a dramatic difference in cost as well – about 3 and 7 times, respectively. The mic preamps, employed on various recordings of acoustic guitars, vocals and a spate of percussion instruments, displayed very good accuracy and low noise performance, plus superb transient handling – perhaps not on the level of the company's knockout Maselec MMA-4XR (see PAR review, January 2007), but definitely in a similar class.

While the Orpheus control panel won't win any graphic design awards for attractiveness, it is quite robust and intuitive. The panel provides for user settings to be saved and recalled, and current settings can be stored in the flash memory of the hardware unit for standalone duty. The input tab provides mic gain adjustment via faders and line sensitivity via -10dBV/+4dBu switches, plus a set of per-channel buttons engages (via hardware relays, in a testament to path purity) Prism's progressive Overkiller limiters; corresponding limiting action is displayed on the hardware meters. Each of the first four channels includes a selectable 80Hz high-pass filter and phase reverse switch, and the first two also include the RIAA de-emphasis filter for vinyl ingest. A much-appreciated inclusion is built-in M-S matrixing on pairs 1/2 & 3/4, in both mic- and line-input modes. Fast Facts Applications
Studio, post-production, broadcast, remote recording

Features
Multichannel FireWire interface with a maximum capability of 18 concurrent input and output channels plus stereo headphones; 8 analog inputs and outputs, including 4 high-quality mic preamps; optical and coax digital I/O for S/PDIF, AES3, ADAT and SMUX signals; MIDI and Word Clock I/O; real-time S/PDIF sample rate converter; two Hi—Z instrument inputs; 2- to 8- channel monitor level control.

Price
$5,000

Contact
Prism Sound | 973-983-9577 | www.prismsound.com

The interface mixing and monitoring facilities of the Orpheus are the finest I've seen. Beyond supplying the usual means to route and monitor a low-latency input mix (>0.5ms at 48kHz and proportionally lower as fs increases), the control panel includes a dedicated "foldback" mixer for each of the analog output pairs as well as for the S/PDIF and headphone outputs. Each mixer provides full fader mixing and panning of the analog and S/PDIF input channels plus DAW output. Enhancing the monitoring facilities, the headphones source can be derived from its dedicated input mixer or quickly stepped across any of the aforementioned output pairs – a great help in monitoring/adjusting performers' phone mixes or ensuring stem mixes to a video deck contain the correct information etc. Note that ADAT-format signals are not accessible in any of the mixing/monitoring features of the Orpheus.

There are only a couple things on my Orpheus wish-list, but at the top would be provision for control surface access to the many mixer layers and their respective fader, pan and mute controls – I wouldn't be asking if Prism hadn't made the mixers so damn useful!

The non-traditional glow-stick metering is a bit outside of my taste, and takes some getting used to: they do not display level gradients but instead change intensity and colors at defined thresholds. This may prove to be problematic for a visiting engineer looking for a trad level readout.

Of more fundamental importance to me is parallel routing of channel 1 to 4's XLR and TRS inputs so that either can be routed to the mic preamps or the line inputs. As it stands the XLR is auto-selected as mic and sent to the preamp and the TRS is sent to the line amp. This implementation makes use of the first four channels on a patchbay an impossibility if you want to be able to variably use the inputs for line and mic signals, depending on your current needs. ["This architecture was chosen in the interest of improved signal integrity," offers Prism Sound. — Ed.]

Summary

Having had the pleasure of reviewing a few Prism Sound products for Pro Audio Review in the last several years, it is no great admission that I am a big fan – the Prism products are of the highest technical integrity and overall sonic quality. The Orpheus maintains this highroad with its fully balanced analog paths, measurement instrument-grade conversion and as transparent sonic throughput as I have encountered in a FireWire interface. For those with the financial wherewithal, the Orpheus could be one of the best upgrade investments a DAW-based studio can make.

PAR Studio Editor Stephen Murphy was engineered and produced almost 41 years ago via vintage, all-analog techniques. His website is www.smurphco.com

- - - - - - -

BENCH TEST
Prism Sound Orpheus FireWire Audio Interface

A/A Measurements
INPUT LEVEL FOR 0 dBFS internally

Line inputs 6.12 V, 18.0 dBu
Mic inputs 1.22 mV, -56.0 dBu

OUTPUT LEVELS @ 0 dBFS
Line outputs 6.11 V, 17.9 dBu

Headphone outputs
High impedance load 4.66V, 15.6 dBu
50 ohm load 1.13 V, 3.3 dBu, 25.5 mW

INPUT IMPEDANCE
Line inputs 14.2K
Mic inputs 5.2K

OUTPUT IMPEDANCE
Line Outputs 51 Ohm Headphone outputs 155 Ohm

FREQUENCY RESPONSE
44.1 kHz Fs +0, -3.0 dB 20 Hz – 22.1 kHz
96.0 kHz Fs +0, -3.0 dB 20 Hz – 47.8 kHz
192.0 kHz Fs +0, -3.5 dB 20 Hz – 68.0 kHz

TOTAL HARMONIC DISTORTION
At 0dBFS signal level
22 kHz measurement filter
44.1, 96.0, 192.0 kHz Fs < 0.0016%, 20 Hz – 20 kHz

80 kHz measurement filter
44.1, 96.0, 192,0 kHz Fs < 0.008%, 20 Hz – 20 kHz

LINEARITY ERROR
44.1, 96.0, 192.0 kHz Fs < +0.5/-0.5 dB 0 to -120 dBFS < +5.0 dB @ -130 dBFS

SIGNAL TO NOISE RATIO
Line inputs
44.1, 96.0, 192 kHz Fs

Wideband 62.0 dB
20 kHz BW 113.6 dB
A weighted 116.8 dB

Mic inputs
Wideband 61.0 dB
20 kHz BW 72.7 dB
A weighted 75.0 dB

DYNAMIC RANGE
44.1, 96.0, 192.0 kHz Fs 116.8 dB

QUANTIZATION NOISE
44.1 kHz -113.0 dB
96.0 kHz, 176.0 kHz –111.8 dB

CHANNEL SEPARATION
44.1, 96.0, 192.0 kHz Fs

Ch1 > Ch2, Ch2 > Ch 1 > 120 dB 20 Hz – 4.0 kHz
110 dB @ 20.0 kHz

A/D Measurements
FREQUENCY RESPONSE

44.1 kHz Fs +0, - dB 20 Hz – 22.0 kHz
96.0 kHz Fs +0, -3.0 dB 20 Hz – 48.0 kHz
192.0 kHz Fs +0, -3.0 dB 20 Hz – 79.4 kHz

TOTAL HARMONIC DISTORTION
At 0dBFS signal level
20 kHz measurement filter
44.1, 96.0, 192.0 kHz Fs < 0.0015%, 20 Hz – 20 kHz

Fs/2 kHz measurement filter
44.1,kHz Fs < 0.0012%, 20 Hz – 20 kHz
96.0 kHz Fs < 0.0012%, 20 Hz – 20 kHz
192.0 kHz Fs < 0.015%, 20 Hz – 20 kHz

LINEARITY ERROR
44.1, 96.0, 192.0 kHz Fs < +0.5/-0.5 dB 0 to -120 dBFS < +2.0 dB @ -130 dBFS

SIGNAL TO NOISE RATIO
44.1 kHz
Wideband (Fs/2) 114.5 dB
20 kHz BW 115.7 dB
A weighted 117.6 dB

96.0 kHz Fs
Wideband (Fs/2) 107.3 dB
20 kHz BW 110.7 dB
A weighted 113.0 dB

176 kHz Fs
Wideband (Fs/2) 78.4 dB
20 kHz BW 112.8 dB
A weighted 113.0 dB

DYNAMIC RANGE
44.1 kHz Fs 118.1 dB
96.0 kHz Fs 114.4 dB
192.0 kHz Fs 113.0 dB

QUANTIZATION NOISE
44.1, 96.0, 176.0 kHz Fs -114.0 dB

CHANNEL SEPARATION
44.1, 96.0, 192.0 kHz Fs > 124 dB 20 Hz – 20 kHz

D/A Measurements
FREQUENCY RESPONSE

44.1 kHz Fs +0, - dB 20 Hz – 20.8 kHz
96.0 kHz Fs +0, -3.0 dB 20 Hz – 44.3 kHz
192.0 kHz Fs +0, -3.0 dB 20 Hz – 78.0 kHz

TOTAL HARMONIC DISTORTION
At 0dBFS signal level
20 kHz measurement filter
44.1, 96.0, 192.0 kHz Fs < 0.001%, 20 Hz – 20 kHz

Fs/2 kHz measurement filter
44.1, 95.0, 192.0 kHz Fs < 0.008%, 20 Hz – 20 kHz

LINEARITY ERROR
44.1, 96.0, 192.0 kHz Fs < +0.5/-0.5 dB 0 to -120 dBFS < +2.0 dB @ -130 dBFS

SIGNAL TO NOISE RATIO
44.1 kHz
Wideband 61.0 dB
22 kHz BW 115.0 dB
A weighted 116.6 dB

96.0 kHz Fs
Wideband 62.3 dB
22 kHz BW 113.0 dB
A weighted 115.0 dB

192.0 kHz Fs
Wideband 62.3 dB
22 kHz BW 112.5 dB
A weighted 115.0 dB

DYNAMIC RANGEFig. 1 A/A mode frequency response at 44.1, 96.0, & 192.0 kHz Fs. Red = 44.1 kHz, Magenta = 96.0 kHz. Blue = 192.0 kHz.
44.1 kHz Fs 116.0 dB
96.0 kHz Fs 115.0 dB
192.0 kHz Fs 114.0 dB

Image placeholder title

QUANTIZATION NOISE
44.1 kHz Fs -116.0 dB
96.0, 192.0 kHz Fs -114.4 dB

CHANNEL SEPARATION
44.1, 96.0, 192.0 kHz Fs > 120 dB 20 Hz – 20 kHz

D/D Measurements
See commentary for D/D measurements

Note: Measurements were made at 44.1,96.0, & 192.0 kHz sample frequencies and with bit depth of 24 bits triangular dithered. Channels 1 & 2 were measured with all unused other channels muted. With all unused channels unmuted, SNR's were about 5-10 dB worse.

Commentary on the Bench Measurements: PrismSound Orpheus FireWire Audio Interface

Fig. 2 A/A mode THD+N at 192.0 kHz sample rate as a function of signal frequency and measurement bandwidth. Red/Blue = Ch 1/Ch2 at 22 kHz measurement BW, Magenta/Cyan = Ch1/Ch2 at 80 kHz measurement BW. The PrismSound Orpheus is a member of an increasing number of multi-channel audio interface units that connect to a computer via a FireWire connection. A software application, the Orpheus control panel applet, included with the unit was used to set up the various modes for performance testing with an Audio Precision System 2722. The control panel applet was run on the same computer that controls the Audio Precision system.
Testing was done first in the analog/analog I/O mode using the mixer controls in the control panel applet, then the performance was tested for A to D, D to A, and D to D modes.

Image placeholder title

Fig. 3 A/A mode typical THD+N as a function of signal level at 192.0 kHz Fs. Both channels measured. Red/Magenta = Ch1/Ch2. A/A performance
Frequency response for the sampling frequencies of 44.1, 96.0, & 192.0 kHz is shown plotted in Figure 1. With frequencies extended down to 10 Hz, the response at 10Hz was down about 0.08 dB.

Image placeholder title

THD+N in a 22 kHz measuring bandwidth as a function of signal frequency and sampling frequency was little different for sampling frequencies of 44.1, 96.0 & 192.0 kHz. A typical result is plotted in Figure 2 for a 192.0 kHz sample rate for both the 22 & 80 kHz measuring bandwidth. Amount of THD+N with measuring bandwidth increased to 80 kHz was variously greater due to increased noise from the greater bandwidth and was the greatest at the 192 kHz sample rate shown in the figure. Fig. 4 A/A mode deviation from linearity of a 1 kHz test signal at a sampling frequency of 192.0 kHz. Red/Magenta = Ch1/Ch2.

Image placeholder title

THD+N of a 1 kHz test signal as a function of level down from 0 dBFS full scale was pretty much the same for the three sample rates tested. A typical example is shown in Figure 3 at a sample rate of 192 kHz and for both channels.

Deviation from linearity was quite good and similar for each of the three tested sample rates. An example of the deviation from linearity is shown in Figure 4 for both channels at the 192.0 kHz sample rate.

Channel separation was close to being the same for the two testing directions and for the three sampling rates. A typical result is shown in Figure 5 for both directions at the 44.1 kHz sample rate.

Fig. 5 A/A mode typical channel separation vs. frequency at 44.1 kHz sample rate. Red/Magenta = Ch1 > Ch2/Ch2 > Ch1. A/D performance
Frequency response for the three testing sample rates of 44.1, 96.0, & 192.0 kHz was similar in shape and bandwith to that of the Analog/analog response shown in Figure 1. At 10 Hz, response was down about 0.04 dB.

Image placeholder title

THD+N vs. signal frequency and sample rate in a 20 kHz bandwidth, like in the analog/analog mode, was similar in amount at the different sample rates. However, the amount of THD+N over much of the audio band was a bit lower as might be expected – the analog/analog mode being the composite effect of an A/D and D/A in series. This can be seen in Figure 6 for the 96.0 kHz sample rate. As the sample rate went up from 44.1 to 96.0 kHz, the Fs/2 measurement bandwidth THD+N amount went up a little and at 192.0 kHz, went up very significantly to about 0.015%.

Looking at how THD+N varied with decreasing signal level from full scale for a 1 kHz tone in a 20 kHz measurement bandwidth, again results were quite similar for the three sample rates. A plot of this for the 192.0 kHz sample rate is shown plotted in Figure 7. Of note, the amount of distortion is lower, again — no surprise — than in figure 3 for the analog/analog mode.
Fig. 6 A/D mode THD+N at 96.0 kHz sample rate as a function of signal frequency and measurement bandwidth. Red/Blue = Ch 1/Ch2 at 22 kHz measurement BW, Magenta/Cyan = Ch1/Ch2 at 80 kHz measurement BW.
Input/output linearity, not shown, was very good for the A/D converter section and, as is quite frequently the case with many measurement parameters, was very similar for all three tested sample rates.

Image placeholder title

Channel separation proved to be virtually the same for both tested directions and was better than 124 dB from 20 Hz to 20 kHz for both testing directions and all three sample rates.

D/A performance
Frequency response in the D/A mode looked like Figure 1 in general shape with the exception that there was the onset of a much sharper roll-off starting at about 88 kHz. At the low end, response was down about 0.04 dB at 10 Hz. This, along with the 0.04 dB loss in the A/D more would account for the 0.08 dB overall loss at 10 Hz in the A/A mode.

THD+N vs. frequency and sample rate at full scale, as in the other modes, was essentially the same in a 20 kHz measurement bandwidth. The appearance of the plot, not shown, is similar to Figure 2 for the A/A mode. Surprisingly, in the 80 kHz measurement bandwidth, the highest THD+N was at the 44.1 kHz sample rate.
Fig. 7 A/D THD+N as a function of decreasing signal level and measurement bandwidth at a sample rate of 192.0 kHz. Red/Blue = Ch 1/Ch2 at 22 kHz measurement BW, Magenta/Cyan = Ch1/Ch2 at 80 kHz measurement BW.
THD+N of a 1 kHz signal with decreasing level below full scale was similar in appearance to Figure 3 for the A/A mode with the amount of THD+N being a couple of dB lower below – 10 dBFS. Again, THD+N in a 20 kHz measurement bandwidth was about the same for all three sample rates.

Image placeholder title

Deviation from linearity was similar in nature to Figure 4 but with less deviation below –120 dBFS.

Channel separation in the D/A mode was very similar to that in Figure 5 for the A/A mode suggesting that the majority of crosstalk between channels is in the D/A converter.

In a final test of the D/A converter, a 500 Hz sine wave jitter at 1 UI (Unit interval) was imposed on the 1 kHz 0 dBFS digital test signal output to see the effect of this in the distortion of the converted audio. As in just a few of the D/A converters that I have made this test on, the Orpheus completely rejected the effects of the added jitter, which would show up as strong sidebands at 500 and 1500 Hz, as shown plotted in Figure 8. Fig. 8 D/A THD+N mode spectrum of 1 UI 500 Hz added jitter to 0 dBFS 1 kHz digital test signal

Image placeholder title

D/D performance
When I first started looking into the D/D behavior of the Orhpeus, I discovered that the SNR was nowhere near the expected threshold of the Audio Precision, which is usually around –140 dBFS at the 44.1 & 96.0 kHz sample rates. This equates to about 23 bits of resolution. It was more like 106 dBFS. That is when I figured out that unused channels that were not muted were contributing noise to the FireWire buss. Muting all unused channels pretty much got the D/D noise levels down to the Audio Precision threshold. This obviously affected the other modes and all measurements on them and the D/D mode were with only channels 1 & 2 active with all others muted.

Needless to say, the D/D frequency response was dead flat right out to the Audio Precision generator limit at each sample rate.

THD+N vs. frequency and sample rate was at –140 dBFS from 20 Hz – 20 kHz for sample rates of 44.1 & 96.0 kHz and at about –137 dBFS at the 192.0 kHz sample rate. These data with the measurement bandwidth set at Fs/2.

Signal to noise ratios, dynamic ranges, and quantization noises were all essentially down to Audio Precision Thresholds.

Finally channel separation was generally better than 150 dB with a few strange anomalous peaks at lower frequencies that reached –140 dBFS.

Bascom H. King
August 30, 2008