Level Headed

ã 1998 by Eddie Ciletti

Nominal Level (explained) plus an SV-3700 modification

Interfacing a DAT recorder with a mixer, a piece of outboard gear or a cassette deck isnít always quite as easy as it should be. In the analog-only daze, it was simply a question of "plus4," "minus10" and 0VU. This month, a gain structure modification for the Panasonic SV-3700 (post D-to-A converter) will help you understand ó and manipulate ó the "nominal" operating level. But first, a word from our sponsorÖ


Why are there so many different reference levels?  Quite simply, it's a matter of juice.  If a device runs from a 1.5 volt battery it can only deliver a maximum level of not-quite 1.5 volts peak-to-peak and that's with lots of distortion.  But assuming 1.5vp-p, that turns into a half a volt RMS which is about -8dBu, max.  You'll need 14 or so dB of headroom, now the nominal operating level is down around -22 dBu, which is in the neighborhood of a passive guitar or bass.  What's dBm?  Think of the "m" as the "meat" factor.  A device so specified can deliver its juice into a 600 ohm load.  The "u" in dBu implies that the load impedance is unspecified and is likely to be high, around 10k-ohms.


When the analog meter in Figure-1 indicates 0VU, the device to which it is connected will output its standard (nominal) operating level. For professional and consumer recording equipment, the standards are +4dBu and Ė10dBV, respectively. But what about the bar graph display above? On a stock SV-3700, Panasonic chose a point 18 dB below Full Scale (fs) as their reference. That itís not the same on every brand and model of DAT machine can cause problems when interfacing with analog equipment. On the DA-98, Tascam allows the user three choices: -20 dB, -18dB and -16dB.

The "location" of the reference level determines headroom ó the distance in dB before clipping (0dBFS) needs no translation. From the reference level to the noise floor is another story based on the number of bits, quality of D-to-A converter, circuit design and printed circuit board (PCB) layout. A tone recorded at the reference level will generate the "nominal" output.


One idiosyncrasy of the SV-3700 occurs when shorting either pin 2 or pin 3 of the XLR to ground. This old-fashioned unbalancing method applies to transformers but it can reduce headroom on some "active-balanced" circuits such as those used in the SV-3700. To "safely" unbalance, use ground and one signal pin and leave the other signal pin unconnected. The output will be 6 dB lower but the signal will not distort.

Here come the numbersÖ

The balanced output amplifiers of the SV-3700 can deliver a maximum signal of +24dBu (+18 dBu). Numbers in ( ) indicate the output when operated in the "safe" unbalanced mode. The gain structure is set to deliver +22dBu (+16dBu) when the recording level is 0dBfs. (Doing the math: 18 dB above +4dBu is +22 dBu.)  Remember that the amplifiers feeding pin 2 and pin 3 can only deliver +18dBu "max."  With only 2 dB of headroom, under normal circumstances, operation in the "unsafe" mode will attempt to add 6 dB of level, 4dB above "max."   Yee-ouch!  This means any signal recorded within 4 dB of Full Scale will distort the output amplifiers.  The Sidebar explains how electronic numbers are born...


To fix this problem, check out Figure-2, whichshows the complete output schematic for the right channel plus IC 905, the affected section of left-channel circuitry. Changing the value of R929 and R930 (left and right, respectively) from 5.6kW to 2.21kW lowers the output gain by about 4 dB. (This was the "nearest standard value" I had available.)  Table one indicates values that deliver precise dB increments. Use 1% metal-film for low noise.)

In order to deliver the nominal +4dBu at the XLRs, a tone formerly recorded 18 dB below full scale must now be recorded 4 dB louder. Nothing really changed on the record side of the A-to-D converter, but increasing the reference level by 4dB reduces the headroom by the same amount. (You still canít record above "0.")



FB Resistor
D dB

Max +18.5

Max Out = 0dBfs

Bal / unbal

R929 / R930
+18.00 ok
+17.5 ok
+17.0 ok
+16.0 ok
*use 2.21k*
Rounded for clarity


For the physical part of the modification, I chose to remove the original resistors and insert Vectorâ "inboard" pins, part number K26C. The whole circuit board has to be removed for this mod, but once the pins are press-fit into the PCB and soldered into place, any changes can be soldered from the top. Please note that I chose a fixed value. If adjustable output is desired, a back-panel switch or pot could be wired to the pins. Figure-3 is the PCB layout showing one resistor installed (R929 on the left) and just the pins in place for R930.


SIDEBAR: Getting Down To Dee Bees Ness

We nonchalantly misuse the decibel as if it were a voltage when in fact the dB is the logarithmic ratio of two voltages. A technician calibrates an audio device by using an AC voltmeter. Here, 0dBu is referenced to .775volts RMS. While this voltage may seem an odd choice, when applied to a 600W load (read: "vintage" gear) the power dissipated is 1 milliWatt, a nice, clean point-of-reference. An additional .455volts yields 1.23 volts or 4 dB. (For consumer gear, 0dBV is "cleanly" referenced to 1Volt RMS, 2.21dB higher than 0dBu.)

On the face of déjà VU (Figure-1) are the formulae for dB and Power. Donít wig out at the math. Even Windows has a scientific calculator so you can cypher better than Jethro Bodean! To the right are the "RMS" voltages and their corresponding "dee Bees." Voltage changes by a factor of ten every 20 dB, power numbers every 10 dB. Now you understand why most people just say "Plus 4."

Whatís sexy about an analog VU meter is its ability to display steady-state levels, especially at or near 0VU. Accuracy can be within 1/2 dB, unlike the Panasonic SV-3700 ó and most other machines ó where there is no "reference" designation and, as Figure-1 shows, the vertical segments are "locked" into blocks of three. Balance between channels or relative to the outside world can be as poor as 2.5 dB.

Go back to where you left off...

E-mail Eddie

Return to Eddie's Home page

Go to Eddie's Directory

Go to the TAPE Directory