# How to control the gain of FET/BJT/Op-Amp amplifier by input voltage range?

Assume I have an input voltage ranging from 5-10 MHz and want to amplify it. The case is pretty simple as any device as FET,BJT or high frequency op-amp can do the amplification well but I have a problem: the input frequency range (amplitude) changes by the frequency as it is 2.5 volt pk-pk at 5MHz and reduces to 5 milli volts pk-pk at 10 MHz. I need a constant output amplitude ( at any voltage ). I tried a high-pass filter to reduce the high range at lower frequencies but unfortunately it is not that precise and some changes remains ( even with good frequency calculations and active filters) and some distortion at extremities noticed. Is there any standard way to do that? Any helps appreciated in advance.

• How much gain do you need to add? – Matt Young Aug 28 '13 at 22:13
• The gain is not important itself. For example if I can have a fixed output range at 5 volts pk-pk between 5-10 MHz, that's ok. This means I need a gain=2 at 5 MHz and gain= 1000 at 10 MHz !! this is what I can't manage. – Aug Aug 28 '13 at 22:16
• Is your goal to have a particular gain or a particular output signal level? It isn't at all clear from your question. In either case, what kind of accuracy do you need? Your signal falls by 54 dB in one octave, which implies something like a 9th-order filter. Is this signal a single tone (amenable to AGC of some sort) or a broadband mixture (requiring a filter)? Can you give us some context for this problem? – Dave Tweed Aug 28 '13 at 22:34
• With this sort of range in gain, stating only two frequencies is not helpful. Basically 10mV to 2.5V is 54dB in one octave and doable with filters and some AGC but you'd need to specify across the range of frequencies what is acceptable to be sure. – Andy aka Aug 28 '13 at 22:38
• If a device produces a 2.5V p-to-p signal at 5Mhz, but at 10Mhz only 5mV, then 10Mhz should be considered outside of its operating range. If an amplifier (in closed loop) had this behavior, you would not consider 10 Mhz to be part of its useful "midband", but way past its upper corner frequency. The right solution here is to design an oscillator whose frequency you can double from 5 MHz to 10 Mhz, without losing 54 dB in the process. Why don't you post the schematic of what you have so far. – Kaz Aug 28 '13 at 23:02

## 2 Answers

Adding as an answer because comment dialog became too long:

AG: A simple set of experiments with a DDS module (AD9850 prebuilt DDS module, less than $5 shipped internationally) should address your concerns. Creamy smooth low THD pure sine waves with an extremely fine frequency resolution of under 0.05 Hertz. AG: No point debating in theory when practical experiments are so easy, you won't even have to make a PCB. OP: I looked at at the AD9850 data sheet. That looks great but needs a 125MHz input clock to produce half frequency (62.5 MHz). What is the best input circuit for that? a common Colpitts/Hartely can produce the required clock? Is there any suitable IC for that? If that works, can be definite solution for my question. AG: You would not need to look for any external "good clock source" - The device uses any standard oscillator crystal: For 10 MHz maximum output, a 20 MHz crystal is sufficient, but use a 40 MHz for some room to grow. Crystals are a few cents each. AG: Also, the link I provided is a ready-made AD9850 signal generator board with crystal and all associated parts already assembled and tested, so you can initially experiment with it to see how well it meets your requirements. OP: I got a little confused. the datasheet requires a clock source as CMOS level( squre 0-5 volt wave) or a sine wave with 1/2 volt bias. Are you sure a crystal alone works fine? Yes, an XO works fine as a clock source. See the first application schematic in the datasheet: Also see this answer to another question, for more discussion on DDS ICs and sine wave generation. • Thanks. I searched the web for that and think that may work. I ordered a module and waiting for that. I'll give it a try and post the result here. – Aug Aug 31 '13 at 6:53 • : I tried the module. It works fine but the main problem remains: It needs low pass filter and low pass filters lose amplitude range by -20dB/decade. I set your answer as main answer here and started a new thread for the new problem. Please have a look at the new thread: electronics.stackexchange.com/questions/80900/… – – Aug Sep 1 '13 at 10:56 The standard approach for generating precision signals in this frequency range is a direct digital synthesis (DDS) chip. Analog Devices has a wide selection of chips of this type, and so do several other manufacturers. If this is for a medical experiment, one of their DDS evaluation boards might be all you need. • We first started by trying ICs like MAX038, AD7569, 74LS624 ,.... Unfortunately they are all based on a PLL or relaxation oscillators that basically convert other wave forms to sine. This has even more than 5% error in wave form at specific frequencies. That's why finally we decided to make a clean sine wave as De-Novo. – Aug Aug 28 '13 at 23:16 • No, the chips I'm talking about use digital synthesis, and produce sinewaves whose noise and distortion is on the order of -70 dB or better. The frequency and amplitude can be controlled very precisely. (BTW, the AD7569 is an ADC -- how is that relevant here?) – Dave Tweed Aug 28 '13 at 23:30 • @Aug A simple set of experiments with a DDS module (AD9850 prebuilt DDS module, less than$5 shipped internationally) should address your concerns. Creamy smooth low THD pure sine waves with an extremely fine frequency resolution of under 0.05 Hertz. No point debating in theory when practical experiments are so easy, you won't even have to make a PCB. – Anindo Ghosh Aug 29 '13 at 6:59
• @AnindoGhosh: +1 for "creamy smooth" :-) – Dave Tweed Aug 29 '13 at 10:46
• I looked at at the AD9850 data sheet. That looks great but needs a 125MHz input clock to produce half frequency (62.5 MHz). What is the best input circuit for that? a common Colpitts/Hartely can produce the required clock? Is there any suitable IC for that? If that works, can be definite solution for my question. – Aug Aug 30 '13 at 8:52