SSM2019 monolithic pre-amp chip

May be of interest to any DIY’ers:
A new chip from Analog Devices which is getting very close to theoretical limits for low noise.

The SSM2019 is a latest generation audio preamplifier, combining SSM preamplifier design expertise with advanced processing. The result is excellent audio performance from a monolithic device, requiring only one external gain set resistor or potentiometer. The SSM2019 is further enhanced by unity gain stability.

Key specifications include ultralow noise (1.5 dB noise figure) and THD (<0.01% at G = 100), complemented by wide bandwidth and high slew rate.

Applications for this low-cost device include microphone preamplifiers and bus summing amplifiers in professional and consumer audio equipment, sonar, and other applications requiring a low noise instrumentation amplifier with high gain capability.


other mfgrs say that the theoretical (at room temperature) usable bits are 20. Current chips doing 24 bit recording gives 4 bits of space for computations to avoid damaging the signal.

so what is the true edge of this new chip over current ones?

do we gain anything that we can really take advantage of ?

cds only use 16 bits. what devices can actually use more bits?
sacd never got any traction. dvds? if so which of 15 different “standards” of dvds? or do we wait for the format wars to settle down first?

It’s an analogue pre-amp chip. It doesn’t use “bits”.

yeah… I was wondering how the bits were related to this chip…

I was looking at the example circuits (particulary the phantom power one) in the datasheet… I see they use +18V/-18V for voltage supply… is this the standard supply voltage for this kind of audio amps?

Sorry if it’s a dumb question (I confess I didn’t search much about it) but this sounds like a lot of different voltages in the same circuit… 0V-48V for phantom, -18V+18V for the preamp, what are usual voltages for line-in? 2V-3V? Should I have some buffer stage between the preamp and the line-in of the soundcard? Ok maybe I’m just making a big mess in my head and I should go read more :slight_smile:

“Standard”? What’s that? :laughing:
There is no standard voltage for this kind of chip - the “standard” is whatever it says on the data sheet.
A typical application for this chip might be for the inputs on a pro-level mixing desk. For such applications it does not matter too much what the specification is as the manufacturer can easily provide whatever voltage is required. Solid state devices will usually run well (though a little below optimum performance) at lower supply voltages - details are usually available in the data sheets. Care needs to be taken with some active filters though as filter frequencies may vary depending on the supply voltage.

A line level signal is in the order of 1volt and amplifiers typically need to have a higher supply voltage than the maximum output swing. Amplifiers will often become non-linear as they approach the rail voltages, so some degree of overhead is usually required. This chip, with +/- 18v, provides a lot of overhead.

There are specialised chips designed for battery powered devices that use much lower supply voltages and low operating current. Surprisingly some are even able to produce output voltage swings that are greater than the supply voltage :astonished: Where low voltage/current requirements are not an issue, higher voltage/current devices can be used with the design focus firmly on sonic performance.

I think that the SSM2019 has been designed as an improved, but otherwise direct replacement for an older device (SSM2017?) This type of design practice is very common and has obvious advantages for manufacturers.

48v is a standard of sorts for phantom power. Just about everyone agrees that phantom power is 48v, but in practice you will find that the actual real-life voltage can vary widely. The specification of 48v should not really be exceeded as there is no guarantee that a higher voltage will not cause expensive damage to equipment. Many phantom powered devices will work happily down to 12v or even lower (the AKG C1000S may be run directly from a PP3 9v battery, but can also be used with a full 48v phantom supply). Again there is no guarantee that equipment will work at low voltage, and most mixing desks provide phantom somewhere in the range of 40-48v depending on design and load.

Even more with a valve condenser microphone - high and low voltage for the valve, another fairly high voltage for the condenser polarisation,in addition to whatever voltages are required by the solid state electronics. Valve mics will typically use a dedicated PSU with a multi-pin connector rather than running everything from 48v phantom (another reason why they are better suited to the studio than the stage).

DC regulators with floating outputs can be very useful for deriving multiple voltages from a single supply.

Thanks for the reply steve. Looks like this mic preamp project might not be that much straightforward as I initially may have thought… (well… no surprises here :stuck_out_tongue:) Nonetheless I think I might order the components soon and start working on it before the enthusiasm starts to fade away… :slight_smile:

they still make analog ???

let me rephrase the question
after you use this whizbang preamp chip
do you really gain any bits over any other typical chip
after the next step where stuff gets digitised ?

the way i understand the a/d/a chip specs
you can only use 20 bits because of thermal noise
they give you 24 bits
i hope they already use analog preamps that are good past 20 bits

if you go from 22 to 24 bits for example
you really have not gained anything useful
certainly not when you burn a cd

Imagine you have a microphone that has a dynamic range of 145dB and a full scale signal of 1 volt. In such a situation a noise level of 0.0001v is pretty small.

Now imagine that we have a microphone that has the same 145dB dynamic range, but a full scale signal if 0.1 volts. If the microphone pre-amp has the same 0.0001v noise level, then this is ten times worse than for the first microphone and is starting to eat up a sizeable amount of our dynamic range.

Now bring it down to a real-world scale where the full scale output of the microphone is in the region of 0.001 volts. In this case the microphone pre-amp noise of 0.0001v represents 1/10th of full scale which is clearly unacceptable.

Extremely low noise microphone pre-amps have been around for many years by using high precision discrete components and a typical price tag in $1000’s. What’s so cool about this chip is that it offers these incredibly low noise levels in a single chip that cost $5. Of course there is a lot more to a really good pre-amp than just the noise level, or even just the gain stage, but a high quality, low noise gain stage is one of the essential parts.

Why does my CD player sound so much better than plugging a computer into my amp and playing the CD on the computer? It’s only 16 bit data in both cases, but data does not make the air move.

i have a compressor on my stereo
nothing plays with more than 27db or so range – ever
no more blasting highs waking the baby
no more inaudible lows masked by the traffic outside
i can finally enjoy the classical music that i bought

d/a is what is hard
a/d is the easy part
that is why computers dont sound as good driving your stereo

I have a collection of the THAT audio chips which claim similar specs. Ever try to Google “that?”




Very similar indeed.

From the product page:

The ICs are available in a variety of packages and pin configurations, making them pin compatible with the Analog Devices SSM2019 and SSM2017 (discontinued), and the Texas Instruments INA217 and INA163.

May I’ll just get both and do a head-to-head comparison? :slight_smile:

From the specs looks like SSM2019 has more distortion, but that might also give it a warmer sound… hard to guess which one would sound better… THAT chip is slightly more expensive, but not a big difference…

I used to work in the manufacture of semi-conductors for the MOD - they have extremely tight specifications, but that did not mean that the components were necessarily any better than lower spec. “off-the-shelf” equivalents. It just meant that the specifications were higher and in the case of MOD products it meant that the specifications were guaranteed. The specifications are so close here as to be within natural variability and the marginal differences could easily be accounted for by how the measurements are made and the degree of certainty. Having said that, there is little doubt that both of these chips (assuming that they are not identical) are high quality devices.


Dream on. Semiconductor Distortion, unless specifically tailored to the purpose, always sounds terrible. Vacuum tubes have that graceful, non-linear characteristic and internationally famous overload sound. Semiconductors have crossover distortion, transient intermodulation distortion, and surgical clipping which sound like somebody poking your ears repeatedly with ice picks.

If you get far enough down into the theory books, most microphone transformers are in the circuit to electrically match the microphone to the amplifier. This latest generation of chips match the microphone without going through coils. It’s an impressive feat, although I will say that making something with these chips is a challenge. In order to get the stunning noise specs, they leave out all the stuff you need to make an unconditionally stable amplifier, and at 60dB gain (1000) any electronic errors at all are immediately fatal.


I have a Shure SM57 and I want to connetct it to the amplifier, which configuration figure in datasheet should I use???
Single-Ended, Pseudo-Differential or True Differential???

Thank you very much!

True differential is the only configuration that will reject electrical interference on the microphone cable. Noise arrives single ended and the microphone signal arrives differential. If both the microphone and the noise are single ended, that’s the end of the show – there’s no way to get rid of the noise.

If you have a very short microphone cable in an electrically quiet room, it’s not going to make any difference.


True differential.
The input impedance on the two inputs should be matched as closely as possible for maximum CMR (common-mode rejection).
The SM57 is designed to be used with a low impedance load, so the overall input impedance (across the two inputs) should be somewhere around 500 to 600 Ohms.
If you are building a microphone pre-amp from scratch, it may be worth considering including switchable high/low input impedance as many condenser microphones benefit from a substantially higher load impedance (2.5 to 10 kOhm)

<<<The input impedance on the two inputs should be matched as closely as possible for maximum CMR (common-mode rejection).>>>

I don’t know that I agree. Common Mode Rejection is a function of the current match between the two sides of the balanced line. Impedances are irrelivant, but they are required to match. See: unshielded copper telephone lines.

Noise is a completely different matter. There, impedances are critical, and they don’t necessarily have to match. Mic-In on commercial sound desks is 1200 ohms and they’re expecting a 150 ohm microphone for best performance.


Let me hit that again. Interference Rejection depends on the two sides of the balanced line to be actually balanced. The start and stop impedances don’t matter, although I attended a lecture by the Jensen Transformer people who make a good case for maximum trash rejection when the destination is high impedance. No current, no signal.

There are actually companies wired like that, NBC Washington being one.