three-stage amplifier?

[carpenter]

I have a sinus signal 1MHz, amplitude 10uV - 1V and need measured it on ADC.
The question is how to solve the amplifier?
My intention is to use a three-stage amplifier assembled with 3 x AD8021 amplification every stage 10x.
Stage switched with signal relay.
I've never done anything like this, I would welcome any suggestions on solving.
Particularly those with the reference to the similarly designed amplifier.

 

[betwixt]

I can't provide a design but these are my thoughts about it:

1. You haven't specified the maximum voltage or resolution of your ADC so I'm going to assume it's 5V and 10 bits. That gives you a resolution of just under 5mV which is obviously far too poor to measure 10uV so you definitely need some amplification.
2. Even with a single stage of x10 gain you can't measure 1V without going well over range so you need some sort of gain switching to cope with such a wide voltage range. The lowest gain needs to be about x5 and the highest about x500,000. There is no way you can do that without selecting different amounts of gain.
3. To measure accurately, at 1MHz you can use a standard precision rectifier circuit to convert the signal to DC.

So you need to implement some sort of auto-ranging and use it to switch the amplifier gain so the output from the precision rectifier falls somewhere below but close to 5V so you can take the most accurate measurement.

You also have to consider the background noise and inteference levels which could be high in comparison to your 10uV signal. Careful screening and filtering on the amplifiers will be needed.

Brian.

 

[carpenter]

For specific.
ADC is high speed 16bit but that's not important.
For simplicity, assume the full range of the ADC 1V
For amplifiers assume.
If input to amplifier is
- 1000mV amplification is 0 all amplifier is bypassed
- 100mV amplification is 10x 1 stage On 2,3 Off
- 10mV amplification is 100x 1+2 stage On 3 Off
- 1mV amplification is 1000x 1+2+3 stage On

For 10uV amplifier return 10mV and ADC return 655 decimal.
AD8021 is High precision , high speed Amplifier with noise 2.1 nV/√Hz.
Why is the gain for each AD8021 only 10x? Because I need minimise Phase Error

 

[zorro]

I assume that you want to convert the waveform, not only to measure the signal. Is this the case?

In that case one possible solution can be as follows. (Maybe this is similar to your idea, but I didn't understant it completely.)

You can have several ADCs operating simultaneously, each one of them converting the input signal with different gains. The several gains can be obtained with amplifier stages in cascade.
Strong signals are well converted by the first ADC (that with the lower gain) and saturates the others.
Now, if you reduce the amplitude of the signal, the first ADC deals with a weak signal and quantization and noise become important, but the second ADC converts it without saturation.
There is no need of switch. Digital processing decides which of the ADC is giving the best signal.
Number of converters depends of the total dynamic range, number of bits of the ADCs and noise allowed.
It is a way to increase the total dynamic range.

Another solution that can be useful is a logarithmic amplifier; it compresses a big dynamic range to a reduced one.
The log amplifier can be with detector or "true". It depends of you application. E.g.: Are you interested in conserving the waveform or just to measure amplitude of a sinusoidal signal?

Regards

Z

 

[carpenter]

Accurate and fast ADC are not expensive, just as they need a fairly large area on the PCB, so I thought something like this
Think of it as a quick sketch

I do not know

 

[zorro]

Hi Carpenter,

do you intend to use several 2P2T relays?
If the gains are chosen correctly, this configuration would give 8 different gains (and ranges). If 4 ranges are sufficient, you can solve it with only 2 relays. In any way, I would avoid relays.
Electronic switches, a variable gain amplifier or an electronically-controlled attenuator can fulfill the need.
The other solution is to use several ADCs like I said in post #4.
Regards

Z

 

[carpenter]

That is why I am trying to find some sample solutions.
For example, the relay is not on the market many relay capable of switching MHz frequency with sufficient accuracy and minimal distortion
The electronic switch, I also wondered, but have significant resistance to which it is to be considered when using the 50 ohm impedance line.
VGA, I could not find the corresponding VGA.
Electronically-controlled attenuator + transistor amplifier.
I can not judge, but I thought it also.
This solution is used in some devices Agilent.