The Sony MD Microphone Pre-amplifier

The first portable Sony recorder was the MZ-1, released in November 1992.  It weighed nearly a kilogram, used an early crude ATRAC, and ate batteries.  I don't know anything about its mic pre-amp.  Skipping a few years to 1995, they'd settled on a general body shape which was to take them through the next few incarnations – the MZ-R3, MZ-R30, MZ-R50 – these recorders were robust, with later improved ATRACs.  With each iteration, the recorders got slightly smaller & lighter, with improved battery life.  They all had barely adequate microphone pre-amplifiers, including a high/low sensitivity switch so as to enable recording louder sounds without distortion.  The R50 to date is arguably Sony's highest achievement.

Then in October 1998 came the MZ-R55.  With this recorder, Sony abandoned their body shape to ape Sharp who had produced a recorder with an MD sized footprint.  For the first time they also introduced a budget model based on the R55 internals – the MZ-R37.  This had no pre-amp sensitivity switch, and was the start of the rot.

In October 1999 the MZ-R90 emerged.  Ridiculously small, Sony's new flagship trounced the competition in the size wars.  However again the pre-amp sensitivity switch had been omitted, meaning it was now completely impossible to record loud sounds on Sony's "best" recorder without resorting to external devices.  They produced a second budget model with the same problem – the R70.

(After writing, the MZ-R900 appeared with the sensitivity switch reinstated.)
 
 

Inside The Sony Portable MD Microphone Pre-amplifier:

One channel of the typical Sony microphone pre-amplifier redrawn in *extremely* simplified form.

From the MZ-R30 to the MZ-R55, Sony have used a custom dual op-amp - TLV23621 – in a non-inverting configuration.  There have been variations from model to model but the bare essentials remain largely the same.  The TLV23621 is powered from a single 2.5V rail, and the quiescent non-inverting input voltage is naturally split between the supply rail and ground, in this case by two 47kW resistors.  Quite probably the R90 has a similar set-up.

Bias voltage to power back electret condenser microphones is supplied through a 6.8kW resistor.  The bias voltage is blocked from the input by a capacitor which has varied slightly, but generally in the vicinity of 0.47uF.

A 10uF capacitor is connected between the rails.  This means the positive rail is also ground with respect to AC signals (DC biasing is unaffected).  Therefore the input sees three parallel resistances to ground – the 6.8kW resistor, and the two DC bias 47kW resistors.  Thus the input impedance is around 4.6kW.

The feedback network consists of a 33kW resistor between the output and inverting input, and a 680W and 6.8kW resistor in series from the inverting input to ground.  (There are also various capacitors in this network for blocking DC and setting HF & LF roll-off but they are not germane to this discussion.)

What of the gain switching? It is a transistor that optionally bypasses the 6.8kW feedback resistor, removing it from the feedback network.  When the 6.8kW resistor is in the network, the gain of the op-amp is ~5.4.   When it is bypassed, the gain is ~50.

We would hope that Sony's custom op-amp is capable of rail to rail operation given the pitiful supply – if so, the maximum input would be ~164mV RMS unclipped with the 6.8kW resistor engaged.  Empirical measurement confirms this – several units have been measured to clip at around 150mV to 170mV RMS.  This is barely adequate for live sound – a high output microphone will overload this stage, but if a microphone is chosen with extremely low sensitivity, it might just scrape through..

The output of this pre-amp goes directly to a large custom IC containing further gain control, and the ADCs.  But if clipping occurs at the pre-amp, nothing downwind can save it.
 
 

Imagined R90 circuit.

As for the newest – the R90, it has no gain switch.  The older circuits clip at ~14mV RMS at the high gain setting.  The R90 clips at a measured 28mV RMS.  This means the gain of the R90 is ~25, a compromise between the sensitivity settings of the older models.  It probably has a 1.5kW or similar value in the bottom half of the feedback loop.  28mV is hopelessly inadequate for any live sound situation.  Thus the user is forced to use external work-arounds to remedy this situation.  The R70 has the same problem.  The R37's performance is unknown to me but is probably similar to the R70 & R90.
 
 

Conclusions:

Sony has abandoned the ability to control the gain of the initial input stage, either to save money or space. This has resulted in a microphone pre-amplifier that is much too easy to overload and useless for any significant volume.  Thus users must buy & use external devices to compensate, losing all the size advantages of the new tiny MD recorders.  This is frustrating, entirely unnecessary, and will be a factor in at least some consumer's decision not to buy Sony.
 
 

Recommendations:

It is recommended that the old style of feedback network with the selectable gain be readopted.

In addition it is recommended that new resistor values be substituted such that the low sensitivity gain is ~1.5 and the high sensitivity gain is 23.  This could be achieved by replacing the 680W resistor with a 1.5kW , and the 6.8kW resistor with a 68kW.  This would result in two usable ranges with even more headroom and would be a better match for commonly available microphones.  However in making this recommendation, I don't know how stable the TLV23621 is at close to unity gain.
 
 

Relevance For Live Recording:

A live performance can be quite loud.  How loud can you record without distortion?  It depends on the sensitivity of your microphones, and the headroom of your microphone pre-amp.  For reference I have spreadsheeted up a table.  Look across the top to find the sensitivity of your mic, either in dB, or V/Pa.  Then look down the side to find the headroom of your pre-amp in mV.  Where they intersect is the maximum dB sound pressure level you can record cleanly.  For a rock concert I would want to be able to record at least 130dB cleanly, and preferably even 140dB.  (You can by-pass all this foolishness by using a battery box and the line input - it will take several volts without overloading - the microphone will give up before the line input does.)

Anyway, let's have an example.  Say you've an R90, and a Sony MS 907 microphone with a sensitivity of -45dB.  The R90 clips at about 30mV, so we go along there until we find -45.  But there's no -45.  Oh my goodness, we'll have to use our initiative and assume it's inbetween -44 & -46, and that we can record up to about 108dB SPL with impunity.  For this to be more useful, we need to know the clip points of other manufacturers besides Sony.  The only rows you need to use for Sony are the 30mV and the 150mV rows.
 

Get your own Free Home Page