dbx 700 digital audio processor Page 2

The decoding process is identical to the coding. We simply use the logic highs and lows to switch the Ipos/Ineg charge pumps. The integrator output is then a kind of zigzag approximation of the original input. Of course, the zigs and zags are at 644kHz, so it's easy to filter them out. We're then left with (we hope!) the original waveform. Of course, the actual circuit is rather more complex. A different type of integrator, which produces less quantization noise, is used. The noise level still isn't low enough for critical use, so compression (similar to the system used in MTS) is placed ahead of the encoder. There is even digital compression; if the encoder produces more than nine "highs" in a row, the compressor is told to squash the signal even further. Of course, all the compression is undone in playback.

But the nicest feature of a companded delta modulation system is its noncritical level setting. The encoder has to be severely overdriven before it goes into slew-rate limiting, and be badly under-driven before noise becomes objectionable. You don't know what a relief it is not to have to worry about clipping the encoder when recording a live performance. (I did it a few times with the Nakamichi DMP-100, but the clipping was so brief as not to be readily audible.)

Of course, every advantage has its tradeoff. The nature of delta modulation imposes a limit on the maximum slew rate that can be handled. For a given signal amplitude, doubling the frequency doubles the slew rate.footnote 3 So although delta modulation can accommodate very high frequencies, the maximum input level varies inversely with the frequency. dbx gets around this problem by weighting the record-level display correspondingly, to prevent over-recording.

Physical features
The 700 is a professional product. Though light enough to take into the field, it's intended to sit in an equipment rack; its basic configuration, therefore, is a unity-gain, line-in/line-out device. All input and output sockets are XLRs. As shipped, the line outputs are unbalanced. They can be converted quickly to balanced operation with a pair of (supplied) resistors.

The front panel is imposing at first, but the layout is completely logical: each function has its own module, and they proceed from left to right, in input-to-output order. A few minutes' perusal of the instruction book will make you an expert.

Mike preamps are an option, and plug into the first slot on the left. Once installed, a toggle switch selects between line and mike. (There is no mike/line mixing.) A second switch selects 48V phantom microphone powering. A stepped rotary control sets mike gain from 20 to 60dB, in 10dB steps.

The record input module is next. A switch at the bottom chooses between record and playback—unlike the Sony PCM units, the 700 cannot simultaneously encode and decode. The 700 has a pair of conventional analog outputs that can feed the video recorder's audio tracks (either linear or Hi-Fi). In playback, these supply quick, easy cueing, since there's no need to wait for the decoder to lock in. A second switch determines whether these analog outputs will be straight or compressed. (Oddly, this compression is not dbx, and there is no playback expansion.)

The record inputs are unity gain. However, there are two options. You can move a toggle switch to ADJ, and turn a knob to vary the gain over a +10 to –60dB range. If only a small change is needed, you switch to TRIM, and use a screwdriver to turn a trimpot. Its range is ±10dB.

The input module has CLIP lights preceding and following the gain stage. dbx claims that a single cycle at 10kHz is enough to light them, if the level is too high. Keep an eye peeled, and lower the gain (at the appropriate point) if either comes on.


The next module, naturally, is for playback. It, too, is unity-gain, and it has the same ADJ and TRIM options as the recording module. There is, however, only one CLIP light, at the input. At the bottom of the module is a headphone jack (with gain control) that's live during both record and playback. A toggle switch selects DIGITAL or EDIT, which switches the headphones from the decoder output to the analog cue tracks during playback.

The last "module" is actually a cover plate for several cards underneath. All the audio and video displays are here. The audio-level display is an LED bargraph with three modes, chosen by a toggle switch. The mode is indicated by red, yellow, and green LEDs.

RECORD LEVEL (weighted) is just that. The scale is from +20 to –40dB; 0dB does not correspond to unity gain. The weighting takes into account both the pre-emphasis of the compander and the limitations of the delta-modulation encoder.

CALIBRATION has a 15dB range (from +5 to –15dB), unweighted, in 0.5dB steps. 0dB is line level. This mode is for setting levels and balances.

SIGNAL LEVEL runs from +20 to –100dB, in 4dB steps, unweighted. It's used during playback to examine the dynamic range of the music or the noise floor of the equipment feeding the 700.

On the other side of this panel are the three LEDs that comprise the video display. The red VIDEO UNLOCK indicates that there is no video input, the yellow STANDBY shows that there is a video signal but the decoder hasn't yet locked in, and the green VIDEO LOCK comes on when the delta demodulator is fully functioning.

Below these three lights is a red ERROR CORRECT LED. Unlike the PCM-F1, the 700 has the common decency to let you know when it's fixing things. And, unlike PCM units, there is no interpolation. Either the bad bit is properly corrected, or it's missed altogether. As explained, an occasional uncorrected error wreaks no havoc.

This light is a great way to check for tape dropouts. (With Sony ES-HG,there's an error correction about every two seconds; with Maxell HGX, they occur about every 30 seconds!) In any case, the circuitry reconstructs the signal and feeds it to an error-corrected output on the back panel, for dubs.

Fortunately, the 700 doesn't look like a Christmas tree. Only three lights are normally on: power, display mode, and either the record light or one of the playback status lights. Any other light indicates an exceptional condition—48V powering, nonstandard gain, error correction, or clipping. A nearsighted person can stand 20 feet away and still know just what the 700's doing.

My first 700 was DOA; it recorded, but emitted only a low buzz during playback. dbx said it was a show unit that hadn't been checked, and replaced it. The second unit (reviewed here) recorded and played properly, but had a flaw in the display circuits. The left display showed a continuous "ghost" signal. Since 80% of everything I've reviewed in the last six months was defective, broke down, had some design flaw, or featured a combination of these problems, I wasn't fazed by this. I did, however, have a minor problem with the input sockets (both mike and line). My XLR plugs had to be pushed in very hard before they would lock in place, and the left-channel plugs wouldn't lock at all. (There was no problem with the outputs.) This appears to be due to some subtle (!?) incompatibility between XLR brands. I wedged in the plugs, then dressed the cables so they wouldn't be knocked loose. I had no problems after that.

The instruction book was up to dbx's usual superb quality: friendly and informal without silliness; clear and simple without inaccuracies. It was a pleasure to read, something I can't say about most manuals. The ability to write literate manuals is common among New England hi-fi companies. It's hardly surprising, considering the area's literary tradition.

Sound quality
The sound quality was checked in the best possible way—by using the 700 to make live recordings. On one outing, I brought another VCR and the Nakamichi DMP-100 (nee Sony PCM-F1) for comparison. I made up a cable so the mikes could feed both digital processors simultaneously. (I didn't use an external mike preamp. I wanted to see how each processor sounded with its own preamp, since this worst-case situation is the way most of our readers would use these products.)

Before going any further, I want to dissociate myself from the opinions about the PCM-F1 and DMP-100 expressed in Stereophile's "Recommended Components" listing in Vol.10 No.3. Although I switched from analog recording to the DMP-100 because of its superior sound quality, I've never claimed, publicly or privately, that it was "99.7% perfect." It ain't. Comparing tapes of the same performance, it was immediately obvious that the Nakamichi DMP-100 and the dbx 700 did not sound alike. The most striking difference was in instrumental timbre. On the DMP-100, almost everything was lighter and brighter in texture, even bass percussion instruments! This was especially noticeable on violin, which sounded as if it were all strings and no wood. Nor did the 700 fully capture the "body" sound of the violin, but it came a lot closer. There was a more solid sense of the fundamental, and less exaggeration of the overtones.

Another difference showed up on brass. The DMP-100 was noticeably grundgy. The 700 had a deliciously smooth brass tone that in no way lacked "flatulant blattiness" or attack.

Of course, "different" is not "better." Was the DMP-100 adding something, or the 700 taking something away? I checked this by comparing the direct output of the 700 with the same signal sent through the DMP-100. (The latter has simultaneous encode/decode.) The DMP-100 made the 700 recording sound like the DMP-100 recording, lightening and brightening the instrumental texture. The proper conclusion, then, is that the dbx 700 is the more accurate of the two processors.

The obvious question arises: are these sonic differences due to fundamental differences between PCM and PDM? Unfortunately, there's no way to tell. They could just as well arise from differences in the analog circuits that surround the digital processors. The only way to know for sure would be to swap the analog sections, an impractical task.

To sum up . . .
To say that I am delighted with the dbx 700 is an understatement. It's more accurate than the DMP-100 (and, by implication, the PCM-F1). Although bulkier, it's not especially heavy, and the power supply is built-in. (The DMP-100/PCM-F1 needs an outboard unit.) Best of all, I don't have to chew my nails worrying whether or not I'm going to clip the processor during loud passages. On sound quality alone, it goes right into Class A, with the Nakamichi and Sony dropping into Class B.

I also like the 700's open architecture. In principle, any module can be modified or upgraded, something not really practical with other digital processors. (If you've dismantled a PCM-F1, you've seen its tightly wadded innards.) Think you can design a better-sounding line module? Go to it! Further, since delta modulation uses off-the-shelf components, the 700 will probably be repairable well into the next century.footnote 4

Of course, $4975 is a bit pricey for a digital processor, even for our well-heeled readers. (Price is not a consideration for entry into Class A.) Still, it's a fine-sounding unit, and any serious amateur or professional recordist should give it a careful listen. Highly recommended.

Footnote 3: Simple math. Doubling the frequency doubles the number of waveform transitions in a given time. To make twice as many transitions, we have to move (slew) twice as fast. Right?

Footnote 4: As I write this, dbx tells me that they are selling the 700 "out of stock." They will make at least one more production run of the 700, but I would suggest that if you want a 700, now is the time to buy.

dbx Inc. (1987)
dbx by HARMAN (2021)

thethanimal's picture

I really enjoyed the technical descriptions in this article; it’d be great if Stereophile could bring some of that technicality back in future articles. I haven’t thought about Coulombs in around 15 years so it was a nice mental workout. Not being an electrical engineer (but an engineer) I’m still catching up on the physics and theory of the hobby, but to my mind PDM sounds like it was a precursor to DSD. Any truth to that? Some technical history articles could be a good idea for a new column. (Unless some of that is already covered in print but not the website; I must confess to being a website-only reader.)

John Atkinson's picture
thethanimal wrote:
to my mind PDM sounds like it was a precursor to DSD. Any truth to that?

Yup (though there are differences in the details)

The dbx 700 was largely the work of the brilliant Robert Adams, who went on to design digital audio products for Analog Devices - see https://en.wikipedia.org/wiki/Bob_Adams_(electrical_engineer)

John Atkinson
Technical Editor, Stereophile

georgehifi's picture

The Sony PCM F1 a couple of days ago, now the dbx 700, hopefully next the first mass retail CD player the Sony CDP-101 and 701 that I had both of and made my ears bleed!?!!Z
All this goes to show the youngsters of today what us older folk had to suffer through at the age of digital.
We all knew it had the potential, but my god we had to suffer for it with that early stuff. Hope you youngsters apricate it.!!!

Cheers George

cgh's picture

I agree with that comment: these old articles are really enjoyable... and comforting, like reading the back of classical music LPs, written during a time when writers didn't need to treat lay people like simple idiots.

I especially love the history, like Coulomb's story in that footnote. Reminds me of Faraday, who famously locked himself in a cage of his own making; or Gauss, who famously invented a "degaussing" device to help tame his wife's hair during the stormy and humid German spring time, only to accidentally kill her during an electrical storm.