0
\$\begingroup\$

Does anybody know if there are ADCs with programmable threshold levels?

The continuous signal is a radar signal which is to be digitized with an ADC with controllable qunatization levels.

I am not looking for a software or etc. I want to know if such ADC is available or not.

\$\endgroup\$
  • 2
    \$\begingroup\$ Welcome to EE.SE. Please tell us more about what you are trying to accomplish, and about the signal that you are working with. \$\endgroup\$ – Nick Alexeev Jul 30 '14 at 7:06
  • \$\begingroup\$ If I understand the question correctly, you would use software to take the result from the ADC to choose your "trigger" levels, not the ADC itself. If you need to "customize" the input, that is typically done with an op amp. I agree with @NickAlexeev, more info is required in defining your intended use. \$\endgroup\$ – Enemy Of the State Machine Jul 30 '14 at 7:17
  • \$\begingroup\$ There still seems to be some confusion between thresholds and quantization levels. It might be worth adding why you want to control it, you can increase quantization errors just by dropping bits but I guess that's not what you're after. \$\endgroup\$ – PeterJ Jul 30 '14 at 7:46
  • \$\begingroup\$ @PeterJ The reason is that the dynamic range of the input signal is varying with time, so a fixed ADC would not be optimal. The uniform quantizer is not optimal, also. Quantization levels should be adjusted to match the (time-varying) characteristics of the input signal. \$\endgroup\$ – user50360 Jul 30 '14 at 7:53
  • \$\begingroup\$ Most AD architectures (dual slope, successive approximation) are inherently linear, so I don't think you will find what you want. It would be technically feasible in in a flash AD converter, but only for a low number of thresholds, hence a very low number of bits. \$\endgroup\$ – Wouter van Ooijen Jul 30 '14 at 8:05
1
\$\begingroup\$

The continuous signal is a radar signal which is to be digitized with an ADC with controllable qunatization levels.

A lot of ADCs have an externally controllable reference voltage input - this controls what the full scale range for the ADC is. For example, some ADCs might use a +2.5V reference with an ADC input range from 0V to +2.5V - if it is a 10 bit device, each quantization step is 2.5V/1024 = about 2.5mV.

If a different reference were used (say) 1V, the quantization level will be about 1mV.

So, choose an ADC that can work with a variation in reference voltage that suits your application OR use a fixed reference and process the signal to the ADC with a programmable gain amplifier - you could even use a multiplying DAC to give you very fine gain adjustments making it almost the perfect programmable gain amplifier.

Another alternative is to the use highest resolution ADC you can find and then not worry about quantization levels assuming it is far greater than what your application demands.

\$\endgroup\$
1
\$\begingroup\$

A lot of ADCs have a PGA (programmable gain amplifier) up front. Or you can add one. For example, the ADS1115 PGA can have a gain of 2, 4, 8 or 16, so you're effectively reducing the quantization increments by those ratios.

You could also attenuate the input voltage by similar ratios using a programmable attenuator, but usually the maximum input voltage of an ADC is close to or equal to the supply rails, and there is little advantage in doing so, since it will be preceded by other circuitry that runs from similar rails.

If you really need more dynamic range you might want to look at preceding the ADC with a log amplifier.

\$\endgroup\$
0
\$\begingroup\$

After thinking things over with the AGC, I think a better approach would be to use an ADC with a programmable attenuator. Amplify the input signal so that your weakest expected signal will give you a reasonable usage of the input range (say, 10dB below full scale) without attenuation. When the input signal goes over that level, then you switch the attenuator to reduce the input level. Each time it goes over your limit, you switch to a higher attenuation. When the input signal is too weak, then you switch to a lower attenuation. This is in effect an AGC.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.