A/D Converters - part two

by Bob Hodas & Paul Stubblebine

We are back at Rocket Lab Mastering in San Francisco to listen to a new round of A/D converters. The new contenders include the Apogee AD-500, Drake PD 5050, and Vacuum Tube Logic. We also brought back the Pygmy and DCS from the first round of tests. It should be stated that we used updated versions of the DCS (900B) and the Pygmy for these tests. Listening procedures and material were the same as last time (see Mix, May '91) with a couple of exceptions. This time all material was recorded to a Sonic Solutions hard disk system since it is capable of recording up to 24 bits with no truncation. The Sonic was set up with no DC blocking, deemphasis or dither. Whatever truncation necessary occurred at the D/A side. For the D/A we used a VTL (20 bit Ultra Analog). Converters were recorded at the following bit rates: Apogee/18 bits dithered to 16, Drake/20 and 16 (w/ dither) bits, VTL/20 bits, Pygmy/16 bits, DCS/24 & 16 bits. Let us reiterate that these were blind listening tests. Each participant was assigned a letter of the alphabet during recording and identified only that way throughout the listening. Converters were compared for accuracy to the analog tape and not to each other. To make sure there were no level differences, all units were calibrated to within 0.1dB on a Sony digital meter. We also tested for polarity to make sure everyone matched up.

APOGEE: The Apogee AD-500 is the smallest and most versatile unit tested. It is smaller than a half space rack package making it easily portable. A rack mount adapter kit is available. The AD-500 uses Delta Sigma conversion technology. The front end is DC Servo, direct coupled discrete with no feedback and will accept levels from below consumer standard to +27dBu. The front panel is control packed but easy to read and operate. A three position switch selects Power Off/AES On/SPDIF On. Separate Left/Right pots control input gain and individual L/R multi-turn screwdriver Calibration pots allow precision level control. Signal may be routed to one or the other or both using internal jumpers. Two LEDs on each channel indicate levels of 12dB and 2dB below converter clipping. The Input Selector gives the same options for normal recording and recording with Emphasis. Positions are Off (digital black, all 0's), Mute (automatically calibrates the converter), On (normal conversion), Soft Limit (peak limiting), and Soft Saturate (emulates analog tape saturation). An LED indicates if you are recording with emphasis. Our listening tests were performed with this switch in the On position with no emphasis. A Sync Source Selector Switch gives one numerous options while recording. Crystal selects an internal crystal clock accurate to 10 PPM Word Clock locks to an external word clock reference and is capable of following from 32kHz to 54kHz, thus providing full vari-speed operation. The Apogee slaving clock is designed to have low jitter (timing irregularities). AES/EBU selects AES/EBU sync and termination with variable rates from 32kHz to 54kHz as well. The Optical position selects the fiber optic input with the same capability as AES/EBU. SPDIF selects that input with the same features as above (75 =.terminator is provided). Positions are selectable for PAL, NTSC video, and sync. All are tied to the sample rate switch. The Sample Rate Selector lets the user select 32, 44.056, 44.1, and 48kHz sampling rates. Three LEDs indicate Lock to 32, 44, and 48kHz. Also provided are positions to multiply and divide the incoming sync by 1.001 (the ratio between CD's and NTSC video). The rear panel houses left and right XLR input connectors and an XLR AES/EBU output connector. Two optical connectors.are provided for optical in and out as well as two BNC connectors for various sync sources. The 15 PIN HD D sub connector contains power as well as SPDIF and word clock access. Internally, jumpers allow a variety of grounding options as well as terminations for the BNC connectors. The Apogee operates pin 3 hot. Supplied with the AD-500 was a small bag containing an allen wrench, jumper pulling tool, jumpers, extra feet, and a BNC to RCA adapter. The Apogee AD-500 retails for $1995.

DRAKE: The PD 5050AD is a single space rackmount unit that houses a single A/D converter card (AES/EBU format). The rack itself is wired to hold a total of three cards and may be supplied with an optional D/A card and SDIF-2 card. The unit operates in 20 bit mode utilizing 128 times oversampling. When the 16 bit mode is selected, dither is applied to the input at the 16 bit level. Drake uses the newest UltraAnalog "A" version converter. Analog inputs are electronically balanced with a maximum input level of +18dBu. The front panel is a removable smoked grey plastic cover that allows one to view the PD 9375 converter card indicator lights. It is easily removed for calibrations. On the PD 9375 LEDs indicate Power On, AES/EBU Input Signal Present, Signal Errors, and Signal Close To Failure (both for AES/EBU external synchronising signal). Two screwdriver calibration pots can be adjusted for left/right input levels. Jumpers on the card allow one to set AES/EBU Delay Equalization from 0-200nS in 20nS increments in order to compensate for cable length (broadcast style runs). The PD5050AD is capable of varispeed operation within +or - 12.5% of the sample rate. The rear panel houses all necessary connectors for the A/D as well as the above optional cards. On the power side is an IEC AC receptacle, fuse and voltage selector. Three BNCs are supplied for SDIF-2 wordclock and left and right out. XLRs are standard for Audio Left/Right In and Out, two AES/EBU Outputs, AES/EBU Sync In, and AES/EBU Input. A D connector for a remote control is tied to a Local/Remote switch. Other backpanel switches tied into the remote are a Sample Rate Selector (48 or 44.1kHz), 20 bit/16 bit Selector, and a Sync/Repeat Selector (allows sync input to be repeated out on the digital output connectors). Although the Drake manual states that it operates pin 2 hot, we found that it was actually pin 3 hot. The Drake PD 5050AD retails for $5,089. The optional SDIF-2 card is $1,190 and the D/A card is $2,900.

VACUUM TUBE LOGIC: Contrary to the VTL line of products, this A/D does not contain tubes. It is a two space rackmount unit utilizing 20 bit, 128 times oversampling VTL uses a modified version of the original UltraAnalog converter. Three input stages are provided, an electronically balanced input, an unbalanced input, and an unbalanced -20dB low-Z input. The front panel is nicely laid out, clean and uncluttered. There is a power On/Off Switch with LED, a pair of Peak Program Meters (PPM), left and right Input Level pots for the unbalanced input stage, left and right DC Offset Trims, and a Sample Rate Selector switch (44.1kHz/48kHz). The VTL may be ordered without the sample rate switch, hardwired to 44.1kHz. The converter may also be purchased without the meters. We do not know what the pricing would be in this configuration. The rear panel houses the IEC power receptacle, fuse and voltage selector. A switch selects AES/EBU or SPDIF output formats. There are two SPDIF output connectors (UHF style) and two XLR AES/EBU outputs. The VTL may be ordered with different output combinations if so desired. There are XLR connectors for both Balanced and Unbalanced Audio Inputs and RCAs for the -20dB inputs. A switch selects either balanced or unbalanced operation. The VTL operates pin 3 hot. A couple of things stood out about the VTL. The first thing we noticed was the non standard SPDIF connectors. VTL claims the UHF connector is more robust and better shielded but everyone in the real world is going to have to buy adapters to get into it. We also found that the PPM meters were not properly calibrated. The left channel showed 90 while the right showed 100. VTL calibrates their meters to a Panasonic 3700 DAT which is only accurate to about 1dB. Fortunately the meters are easy to cal and instructions are given for the procedure in the manual. The final unusual item is that VTL uses a different operating reference level than everyone else. As supplied the unit gave a digital output of 9dB below full scale with an input of +4 dBm, while everyone else was set for a digital level of -14dBfs. This level can be internally adjusted though and VTL will set it up any way you request. In order to keep all units at the same level for our tests, we used the unbalanced input and calibrated the VTL using the front panel level pots. The VTL retails for $7,000.

LISTENING TESTS: We chose mostly the same performances as the last listening. Since there were seven converter configurations, (counting the convertors that can be switched for different bit lengths), for the sake of space we will only look at the top three contenders in any category. The first piece was "Waiting Song", a current folk-pop release by Barbara Higbie. Instrumentation was piano, vocal,electric bass, drums, congas, acoustic guitar, and percussion. Apogee The sound of the piano was harder and cooler than the original. The voice and the room sound also were cooled off. Bass response has power and extension, but the bass and kick drum sound blended into one instrument. The shifting harmonics in sustained piano chords are not as well defined as the original. Cymbals sounded very close to original. Soundstage and reverb tail were good.

Drake (20 bits): The low bass, even a bit of the mid bass, are definitely recessed. The top seemed smoothed off a bit and could be characterized as a "polite" sound. Good low level resolution and soundstage. Fairly close to the original.

VTL: The bass sounds a little wooly compared to the original, and the kick drum has less apparent impact. The midrange has a lively, dynamic quality, and the voice has a "believable" quality but was a touch fuzzy. There appears to be less reverb creating a slightly drier sound. Sounds alive.

The second piece was a mono narrator reading copy from a magazine. Half of this selection included some background noise that added dimension and demonstrated low level resolution quite well. Apogee The low frequency balance is correct (a plus!, very seldom did we hear the playback have bass that resembled the original). Low level resolution was good. This selection had sibilants that were hard for some convertors to handle without smearing. The Apogee exhibited only slight smearing on these sibilants. Overall the original had a sense of "natural presence" that was absent in the Apogee.

Drake (20 bits): The top end sounds softly rolled off slightly, but it handles the sibilants well nonetheless. Low level resolution was good. Good overall.

VTL: The tonal balance sounded lightened on the low end (voice pops were reduced a bit) and gently rolled off at the top. Sibilants were slightly less distinct than the original anda little fuzzy. Resolution of the low level sounds was somewhat diminished. Generally fairly accurate.

Next came the Slavyanka Chorus, a choir of twenty-five male voices, recorded with a Blumlein mic technique in the chapel of Saint Vincent's School, San Rafael. Apogee This recording (in the original) had an ability to nearly surround the listener, and the Apogee approached that same ability very closely. Unfortunately there were drawbacks in two other departments. The low bass was very noticeably reduced. And the reproduction of small dynamic shifts near the quiet end of the spectrum was not as fluid as the original. There was a slight loss of air resulting in a softer sound.

Drake (20 bits): Overall this came close to the original. There was a slight reduction in detail, a slightly diminished sense of the texture of the sound. The low bass was very close to correct, just slightly less dimensional. It handled dynamic shifts pretty well.

VTL: The very low frequencies and the very high frequencies were missing. The texture is harder to make out, noticeable on both a solo voice and the hall sound. Low level dynamics were not handled well. Not exceptionally accurate but pleasing.

Finally we heard a Mozart piece for full orchestra recorded with a stereo pair at the Oklahoma Mozart Festival. This was a very good test for resolution. The segment often had three or four interweaving lines which were quite easy to follow in the original. It was instructive to notice how well we could follow them in the various transfers. Apogee Resolution was moderately good, we could usually follow three lines. Reproduction of the hall sound was pretty good too. Unfortunately it came at the expense of some added edge on the strings.There was a bit of deep bass loss.

Drake Resolution was moderately good here too and the high end was close. The image was slightly smaller than the original, but otherwise a good representation. The bass, however, was noticeably diminished.

VTL: A little light on the low end, plus some added edge in the upper midrange. The image shrank a bit in all directions. Not as easy to follow the interweaving lines.

As mentioned above, the Apogee, Drake, and VTL were the top three choices in the group of seven. Even though we have some negative comments, one should not let that obscure the fact that these converters came close to the source. We would select the Drake as the most accurate of the three. It would be very hard to make a decision as to who placed second and third. Both the Apogee and VTL have different merits and drawbacks, so as always we would encourage individual listening for your specific needs. We did not believe that any one converter fit a particular style of music.

Based on features, the Apogee clearly offers more. For audio/video work, all the options seem to be covered. For strictly audio,this box looks like a problem solving tool for many important issues.

In our next round of listening which should be close to this article's publication time, we will look at the Sony, Wadia, Neve, new boxes from Pygmy, Apogee and possibly others.

SIDE NOTES: During our tests we took a side path for a little experiment. We recorded the DCS and Drake at their maximum and minimum bit rates to a Panasonic DAT. We discovered that in both cases, using maximum bit rates and letting the DAT do the truncation sounded preferable. These were the same results we got during the first A/D tests when we let the Dyaxis do the truncation. Another test we did compared the sound of a manufacturers A/D combined with its own D/A. We wondered if one manufacturer would have a sound advantage when covering both ends of the conversion. Did manufacturers who cover both ends tweak their boxes to operate optimally together? It seems not! We found that double teaming did not give anyone an advantage. The best sounding D/A gave a boost to all the A/Ds regardless of manufacturer. Our recommendation is to go for the best D/A you find regardless of who made your A/D.