Problems with putting two amplifiers in series?

Hi - I'm trying to learn as much about amplifiers as possible. My understanding is that, say in a non-inverting op-amp topology, your circuit's bandwidth goes down as the gain of your amp goes up (I think your bandwidth is divided by your gain? so a 10x amplifier has 1/10 of the bandwidth of a 1x amplifier?).

So let's say your goal is 100x gain. You could have a single op-amp amplifier with 100x gain (and one hundredth the bandwidth, I think), or two 10x amplifiers in series with each other (and one tenth of the bandwidth?). Would the two stage amplifier have better bandwidth?

Thanks!

Would the two stage amplifier have better bandwidth?

Click to expand...

Under the assumption, that you are using standard OP with fixed compensation and voltage feedback, this is correct.

FvM said:

Would the two stage amplifier have better bandwidth?

Click to expand...

Under the assumption, that you are using standard OP with fixed compensation and voltage feedback, this is correct.

Click to expand...

Hi FvM - Can you tell me what disadvantages having two amplifiers in series (and I suspect 'series' is the wrong word, but I don't know the right one) like this cause? When would you want two (or more) gain stages instead of one?

You would cascade gain stages for the reason you have described - to get more gain for a given bandwidth (or vice versa). If you want a gain of 1000 at 100MHz you have to look at practical ways of achieving that. Using 3 or 4 stages would be more practical than 1 or 2.

Keith

You can also use a two stage amp, where the second stage is a simple common source, but with miller compensation. Using a capacitor and resistor between the two stages, you can move both poles and zeros around. The attached PDF sums it up well.

Good luck.

Can you tell me what disadvantages having two amplifiers in series?

Click to expand...

I don't see a disadvantage in cascading amplifiers, except for higher part count. The first amplifier stage
should have a gain of at least 5 or 10 to avoid an additional noise contribution of the second stage.

It can be very beneficial to use multiple stages with different characteristics. A first gain stage with low noise & a gain of maybe 10 as FvM says, followed by noisier, lower current opamps for subsequent stages can be the optimum for a low power design.

Keith

The main problem with using multiple stages is maintaining stability. Anything more than 2 stages and you would have problems compensating for the multiple poles in the system.

The main problem with using multiple stages is maintaining stability.

Click to expand...

This would be true only with overall feedback. But the discussion is about cascading amplifiers with individual
feedback, as far as I understand. It's the way, most systems with amplifiers, either RF or AF, are build.

Noise is a parameter that none of you are discusing.
..
More stages equal more noise.
..
Every stage has a certain amount of noise that is added to the circuit path. The amount of noise is not linear. In other words you don't just add the noise. If the first preamp/amp has a noise floor of say -60dbmo and the introduced noise is say 5% the output would have 5% x the gain of noise in the output, if the input was 100mw and the amp gain was 10db then you would have 1Watt out. Add a second stage with the same gain and figures and your noise would be different because you may be over loading the next amp, or the noise figure may be higher due to the increased signal level.
..
Noise is the main reason that amp stages are limited. However if you are using perfect amps in a perfect noise figure environment - have at it. Put 10ea 3db amps in line to get 1,536 watts with a 1 watt input.
Dave

Noise has been mentioned in two of the posts above. It is not a problem. Your comments don't make sense to me. If you cascade two identical gain stages with a gain of 10 the contribution to the final output noise by the second stage will be 0.5% of the total.

To put some numbers to it, assume two gain stages with 10nV/RT(HZ) input noise and a gain of 10. The first stage would contribute 1000nV/rt(Hz) to the output noise. The second stage would contribute 100nV/rt(Hz). These would be combined by sum of squares to give a total noise of 1005nV/rt(Hz). Not a big penalty compared to trying to amplify in one stage with 10 times the bandwidth.

Keith