Market for an A/D converter with 150dB dynamic range?

(snip)

So, why do you think your 150 dB dynamic range converter will not have these same "analog limitations?"

-a

The rec.audio crowd all claim to have golden ears and can hear the difference between 44, 88, and 192KHz. Then they listen to 128kbit mpeg.

Do you mean like this?

"Because of this, logarithmic ADCs are very common in voice communication systems to increase the dynamic range of the representable values while retaining fine-granular fidelity in the low-amplitude region. An 8 bit a-law or the =B5-law logarithmic ADC covers the wide dynamic range and has a high resolution in the critical low-amplitude region, that would otherwise require a 12-bit linear ADC."

Or maybe this:

"An eight-bit exponential converter has been constructed with a dynamic range of eight decades and a relative error of about three percent. With the same technology, logarithmic A/D converters and floating-point D/A and A/D converters can be built."

-- Joe Legris

K2-W op-amp front end?

martin

[Bollocks]

You need to brush up on your marketing skills.

In case you missed it we all have mobile phones these days and ring tones with a dynamic range of 150dB are not duh riguer.

DNA

A 24-bit A/D has a dynamic range of about 16 MILLION (in volts). Square that if you dare to get the power ratio.

So if the highest voltage out was 16 volts, the lowest would be one microvolt.

Problem is, if you calculate the thermal noise across say a 10,000 ohm input across a 200KHz bandwidth (for those golden ears), it's going to be a whole lot more than a microvolt, like nearly 20 microvolts peak to peak.

So even if you had a full 24-bit D/A, you can't read out the bottom four bits due to thermal noise.

Also it takes a heck of an amplifier to play that kind of music. If you're listening to really gentle music at the milliwatt level, say exercising the bottom six bits, then you put on The Hole, you need like

2^36 times more power, which will burn out your local power plant.

You missed out how much CO2 it would produce, and how the planet would slow down each time a modern popular "tune" was played.

Nice toy in a BMW, but it would probably demolecularise the cocaine the driver had on him, (and the driver hopefully)

martin

Joe, not really. But it is interesting to see alternatives. what I mean is a linear converter.

Genome, could you elaborate a bit? are you saying cellphones are using a wide range dynamic range A/D converters? I've never heard of it.

There is a market for large dynamic range converters in studio applications. This is because most modern music is a mixture of mono sources. Mix two source, you get 3db more noise, etc. etc. For the end user, 16 bits is fine, but I wouldn't turn down 24/96k.

I suppose it needs a lot more clarification. I'd like to know about distortion versus signal amplitude, about spurious responses versus signal amplitude, and about noise floor (which may also vary with signal amplitude). It would be a pretty decent converter indeed if it has a broadband noise floor that's 150dB below full scale, and the bandwidth is at least audio (apparently what you're interested in).

Actually, I'd be happy with an 18 bit converter that will run at

150Ms/s and has distortion and spurious products worst-case 120dB below the input signal level, and broadband noise 90dB below a +10dBm full scale...

Cheers, Tom

I could be wrong but I don't see the logic here. With either analog or digital mixing. Please explain. As long as you don't do something stupid, like attenuate the incoming signals so they're down in the noise, where is the extra noise coming from? And what kind of signal source has that kind of dynamic range? And what medium or amplifier or listener can handle or listen to anything more than a 60db range?

One possible application would be a "digital radio". The signals hitting an antenna can easily span 120db, if you're close to an AM radio station yet want to listen to Radio Antipodal. Let me know when you have that 24 bit 60 MHz converter ready. Preferably with 12 bits of exponent, 12 bits of mantissa. For under $5.

No analog signal is noiseless, so every time you double the number of channels the noise goes up another 3 dB.

RF / IF AGC is a simpler way to improve the dynamic range. reduce the gain for local signals, and ramp it up to full gain for DX reception.

--
Service to my country? Been there, Done that, and I\'ve got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

See the post from Michael A. Terrell.

The individual signal is a mic'd instrument, electric instrument, or voice. Granted, the individual signal won't have 150db dynamic range, but you are constantly mixing audio signals,so you need to keep the noise "build up" down. Pro-audio pushed for the 24 bit converters, at least ADC, for their end of the business. They need to do a mix down and deliver a signal that has a noise level suitable for the final 16 bit product. And what medium or amplifier or

I want one too.

Still makes no sense. If you mix two analog signals and keep the total signal level constant, how does the noise level go up? If that were true, if you took a bazillion signal generators putting out a very clean sine wave, you're telling me there's going to be a bazillion *

3db of noise?

Indeed, but that requires analog front-end filtering, analog multiplication for gain control, analog frequency conversion.

With a 150db A/D we could simply hook a wire to the input and digitally extract any signal we wanted, no analog front end or tuning required.

Any noise will add when signals are mixed, so yes any residual noise in those generators would go up +3 dB every time the number of generators was doubled. Soon, the mixing losses would have the desired signal so low its useless because they have to be impedance matched. (Look up "Directional Coupler")

Turn on a fan, you get a little noise. Turn on another identical fan (total = 2), the noise doubles. (+ 3 dB). Turn on two more identical fans (total = 4), the noise doubles again (+6 dB). Turn on four more identical fans (total = 8), and you're up to (+9 dB) from a single fan. Its the same, whether the noise is mixed electronically, or acoustically.

--
Service to my country? Been there, Done that, and I\'ve got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

You are not getting the point. If you mix two audio sources each of one volt of signal and one millivolt of noise, you get TWO volts of signal and TWO millivolts of noise.

Of course you also turn down the overall gain, so as to not overload the next stage. So you have ONE volt of signal and ONE millivolt of noise. Not any more noise in proportion.

And why do you think that is? - it has to do with the laws of physics and how noise is related to resistance. The only ways to increase the actual dynamic range to get somewhere near the theoretical is either to cool the analog, reduce the source impedance or increase the maximum signal handling. The last is the most practical but you would probably need a bipolar process to get you 10V rms input.

Ian

Ok, whatever you say! That works when stacking receive antennas, where they all receive the same signal.

OTOH, I'd never let you design a large audio mixer board. There is thermal noise generated by the resistors in the summing network, and the output voltage doubles only if everything is exactly in phase. been there, done that. Why do you think there are switches for each channel on a large audio board? To disconnect unused channels, to lower the overall noise floor.

When you mix video signals, they have to be under one degree of phase shift at the color burst frequency which is 3,579,545 Hz for NTSC.

--
Service to my country? Been there, Done that, and I\'ve got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida