# AC amplifier with fixed output range?

Lets say I have a sine frequency sweeping circuit (1kHz-10MHz) that goes down in magnitude, from say 3V to 1.5V when it reaches its maximum frequency. I've tried buffering it (and even terminating with 50 ohms before, its impedance), however the voltage still varies, and much so more when the frequency gets higher.

What I desire for example.. A small AC will have the specific gain to reach 5V peak, or a larger AC (unlikely) attenuate to reach 5V peak, so I get a flat frequency response.

How would I amplify it to specifically go from 0V to 5V with its centre point at 2.5V for example? Will I have to employ some sort of buck/boost converter, or can it be done with some form of feedback with an emitter amplifier? An op amp might have something, but those destabilise at 1MHz, however if that is all I can do I can upgrade them.

Additional info. for Jippie: - about this "sweep device". It is simply a function generator (Hantek 3x25) and it tends to vary its voltage lower at higher frequencies, even with terminated output due to the quality of the device or whatnot (even though the sine wave is perfectly suitable that comes out.) It could be anything however, an XR function generator, just some variables sweep of some kind (a VCO in the future?). My scope and its probes are 40x the highest bandwidth my gen can put out, so not to worry for that :P

-
Try to identify the cause of the frequency dependent amplitude. Are you sure your method for measuring the amplitude is in order, does your equipment have a better than 10MHz bandwidth? And how is your the bandwidth defined, -3dB? What is the cause that the current circuit is frequency dependent? You want to know at least the cause and work from there, otherwise your Automatic Gain Amplifier my suffer from the same issues with disappointing results in the end. –  jippie Jul 5 at 16:35

An automatic gain control (AGC) appears to be what you want. It will try and keep the output constant by using a variable gain cell (usually a JFET) driven by a voltage that represents the peak magnitude of the output. If the output voltage signal level is too small the gain cell is rapidly compensated to achieve the correct amplitude.

You might also consider using a high frequency boost filter just to raise the amplitude of the input signal so that things are levelled out.

-
amplifying the input is probably the best option, at least as a starting point. –  Vladimir Cravero Jul 5 at 14:20

Amplifying the signal will not change the flatness. Unless your oscillator circuit is slew-rate limited, reducing the output amplitude probably will not change the flatness.

I think your best bet is to redesign the swept oscillator to produce a relatively flat output.

It's possible to compensate for roll-off in response by using peaking circuits, but they only work to some degree (they can maybe double or a bit more the maximum frequency).

Amplifying a 1kHz-10MHz signal by 3:1 with ~5V out and with a flat response is not that hard, just use a suitable op-amp.. for example, a Micrel MIC912 is unconditionally stable with a gain of 2 or greater, has GBW of 200MHz and costs < \$1. You'll need supplies of maybe +8/-3 to get +5/0 out.

-