Nagra D open-reel digital recorder Page 2
Each input level control consists of a dual pot: an inner Fade control, normally set to its "calibrate" position to allow for lowest noise/highest overload headroom; and an outer sensitivity control, marked in dB and ranging from -34dB to +3dB. It's this outer control that adjusts the analog signal input to the A/D converter; the fade control actually supplies a DC voltage to an auxiliary ADC that supplies control data to an internal DSP chip. This topology is said to maximize input dynamic range, reduce the amount of active circuitry in the signal path, and allow for on-location fades to be modified after the event.
[As the same circuitry is used for all analog inputs, line- or mike-level, there is the possibility of overload with very "hot" signals. Nagra offers attenuating balanced cables for situations like this. I had to make use of these cables when I recorded Stereophile's 1998 Rendezvous CD. Even with the Nagra-D's input attenuator set to its maximum +3dB setting, Art Baron's close-miked trombone was still overloading the Nagra's ADC.]
The left-side panel of the D carries all the input ports: four balanced analog channels via XLRs (each switchable between transformerless line and phantom-powered microphone), two pairs of two channels of AES/EBU digital, again via XLRs, time-code/sync—the Nagra-D can interface with just about every video machine made—and an RS-422 port to allow the machine to be controlled by a PC. The sync port can also be used to allow two machines to be run in parallel as an 8-track recorder.
The right-side panel carries the output ports: four balanced transformerless analog channels via XLRs; two pairs of ¼" stereo headphone jacks; and two AES/EBU twin-channel digital outputs, again via XLRs. On the front panel are rotary polarity-inversion switches for the four analog inputs, as well as rotary switches offering a Flat position, and two bass-cut EQ settings, "LFA" and "Speech." All the analog XLRs are wired with Pin 2 positive.
Internally, the machine is a surprising amount of empty space. Circuit boards with mainly surface-mount components are mounted behind the arrays of input and output sockets; the mechanical tape and scanner drive and tape transport components are mounted immediately beneath the machined top plate; in fact, most of the weight seems to come from the rechargeable battery. Construction quality is to gasp at.
D is for Directory
The standard 5" reels of tape—Ampex/Quantegy 467, BASF/Emtec 931, or equivalent (footnote 3)—are capable of storing either two or four channels of 24-bit digital data (over 16 gigabits). In two-channel mode, each reel lasts two hours; four data channels give just over an hour's worth of recording time at a 44.1kHz sampling frequency (48kHz and 32kHz are also available). The 5" reel hubs are asymmetrically mounted on circular inserts; by rotating these and fitting a larger cover, the Nagra-D can be used with 7" reels, which doubles the recording time. The data channels are grouped into pairs—1+2, 3+4—and in 4-channel mode, it's possible just to record tracks 1+2, laying down 3+4 on the tape at a later time.
The D's tape format's radical departure from a standard recorder is made apparent when the user first threads a reel of tape: the machine starts winding or rewinding depending on whether the tape is loaded heads or tails out. It's looking for a Directory on the tape—like a computer floppy diskette, each reel of tape can be given a directory file at its start that describes not only the contents of the tape, its time, date, and sampling frequency, and the status of the machine's inputs and faders; but also the points where ADC overload or uncorrectable data errors occurred. If the Nagra doesn't find a directory, either because the tape is blank or because it was recorded without one, it asks the user if he or she wants the tape formatted. For a blank tape, pressing the Execute key is equivalent to answering "yes," while if the tape has been recorded without a directory, pressing Escape (Down Arrow and Side Arrow together) means "no." If a tape has been recorded without a directory, it cannot be added at a later date.
When a recording is finished, the Nagra-D usually rewinds the tape to the directory area to update the information. Under the time pressure of live recording, however, the recordist will generally spool the tape off in a forward direction, storing it tails out. Under those circumstances, the directory won't be updated. However, the Nagra-D remembers the directories for the last four tapes that were loaded on the machine, and the next time a particular tape is loaded, its directory will be updated.
When the Nagra-D is connected to a PC running the NADCOM software, the external computer can be used to control all aspects of the recorder's operation, as well as to monitor 38 different internal operating voltages, currents, and electrical conditions (screenshot A). Remote diagnosis of a Nagra-D is therefore possible via modem.
Screenshot A: NADCOM operating parameters screen. The data panel at the bottom indicates the state of the digital inputs and outputs; whether the machine is locked to an external source or set to be the "Master" clock; the sample rate and tape speed; and the incidence of corrected errors (ECC). The panel at the top indicates the machine name and serial number and its operating software version.
During recording or playback, a screen shows time code and the four channel levels as bargraphs (screenshot B). The tape's directory is also accessible; screenshot C shows a typical display, with take number, duration, date, time, absolute time code of the start, title, dropouts, and the maximum peak level reached during the take, including ADC overloads. A subsidiary screen shows the number of corrected and uncorrected "fatal" errors over the duration of the tape (there should be very few, even none, of the latter) (footnote 4), while you can zoom in on an individual take (screenshot D) to see when input overloads have occurred. (As explained earlier an "overload" is defined as a level greater than -1.16dBFS; it doesn't necessarily mean that clipping has occurred. When recording I also monitor the levels with an outboard Dorrough meter, which takes an AES/EBU input.)
Screenshot B: NADCOM interactive tape control screen. The transport controls are the bottom, absolute time code at top right, and the four level meters at top left, with the gray bars indicating peak hold. The metering is noticeably jerky if the NADCOM software is run under Windows 95/98/ME; when the PC is booted up under MS-DOS, it runs in realtime.
Screenshot C: NADCOM tape directory screen. This is from one of the session tapes for the Silverman Beethoven project and shows the duration of each take, the time and date it was recorded, its absolute time code reference, and the peak level reached. The arrow on the left shows which take is currently being played, with a real-time display visible on the right. The take info has to be entered via the PC, of course, and can be done at the session or later.
Screenshot D: NADCOM individual take screen, showing detail of screenshot C's Take 1. The horizontal gray bar shows the take duration, with the vertical arrow indicating current position. The musical notes indicate where in the take "overloads" occurred.
Footnote 3: My preferred tape is BASF/Emtec 931, which has very low shed characteristics, with almost zero corrected errors. I had a real problem in 1996 or so with Ampex 467 tape, which shed a lot of oxide, with corresponding head-cleaning problems. I am told that more recent Ampex/Quantegy tape is fine in this respect.—John Atkinson
Footnote 4: v2.0 of the NADCOM software allows a Nagra-D to be used to record at 88.2kHz and 96kHz, each high-speed channel's data being stored on two of the four normal-speed tracks using dual AES/EBU datalinks, one for each channel. I used the Nagra-D in this mode to make Hyperion Knight's Rhapsody in Blue and Robert Silverman's Beethoven Piano Sonatas recordings.—John Atkinson