Control amplifier gain via voltage?

[Korken]

Hi all,

I am working on an ultrasonics based imaging project at my university and I am in need of a functionality I have yet to use myself.
What I'm in need of is a variable gain amplifier. I want to change the gain of the amplifier in inverse proportion to the distance of the traveled sound wave.

Does anyone know what these type of amplifiers are called?
I have found the "Variable Gain Amplifier" however this seems to be for RF (100MHz+) applications and very expensive, are there any alternatives?

Thanks!

[Alex Eisenhut]

Ah, I've used that very function on Sonatest equipment years ago!

How about

http://www.digikey.ca/product-detail/en/AD603ARZ-REEL7/AD603ARZ-REEL7DKR-ND/3897133

You can also use

http://www.digikey.ca/product-detail/en/AD633JRZ-R7/AD633JRZ-R7TR-ND/617832

AD8036 is also useful for the clipping function.

http://www.digikey.ca/product-detail/en/AD8036ARZ/AD8036ARZ-ND/620585

What's "expensive" for you? Can't you beg and plead for freebies from the rep?

[Alexei.Polkhanov]

Hmm, define "expensive"? There are regular amps with AGC - automatic gain control like these
http://www.digikey.ca/product-search/en?pv405=236&FV=fff40027%2Cfff8026e&mnonly=0&newproducts=0&ColumnSort=0&page=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=25. So called Video Amplifiers are usually very good choice. 

Even if you don't find one that you like with AGC - you can use one like OPA548 by adding Digital Pot(entiometer) into feedback:
http://www.digikey.ca/product-search/en/integrated-circuits-ics/data-acquisition-digital-potentiometers/2556350?k=digital%20potentiometer

[vk6zgo]

Use a dual gate MOSFET.

These were commonly used to control IF & RF gain stages with AGC voltages in  HF & VHF Radios.
Look up a few AGC circuits in the ARRL Handbook or on line.

[T3sl4co1l]

The preferred method is to have a constant transmitter power (as much as you can operate at) and AGC (automatic gain control) on the receiver.  If the transfer curve of the AGC stage(s) is useful, the AGC voltage can be used directly as an indicator of signal strength (the circuit tells you how much gain it had to use, rather than your having to manually program it from an outside loop).

Examples include just shifting the bias on gain stages (works best for BJTs at very weak signals, where the distortion introduced is small and constant), using a DC coupled mixer (e.g., a single balanced mixer like the old CA3028) or OTA (operational transconductance amplifier, essentially an op-amp with variable gain), or any of the AGC / VGA / PGA ICs you'll find out there.

Tim

[Korken]

Thanks for the input!

Alex Eisenhut:
These are in my "expensive" area (I should probably have been more specific in my question).
The end application is an 8-channel ultrasonic array (microphones) so the end system will need 8 of these amplifiers.

Alexei.Polkhanov:
This is in the range of acceptable price.
I was looking at an potentiometer based way as well, but it would be difficult to get the 1/distance^2 gain (or rather, I didn't find a good way).

But now I will have to get in touch the ACG amplifier. This will be much fun!

vk6zgo:
Will do!

T3sl4co1l:
>> The preferred method is to have a constant transmitter power (as much as you can operate at) and AGC (automatic gain control) on the receiver.
This is exactly what I am doing.

>>  If the transfer curve of the AGC stage(s) is useful, the AGC voltage can be used directly as an indicator of signal strength (the circuit tells you how much gain it had to use, rather than your having to manually program it from an outside loop).
Could you expand on this? How can you get the correct gain curve from the AGC directly without an outside loop?

[Precipice]

From a practical point of view, are you sure you want / need to do this?
In some applications, losing small objects when they're distant is fine. Useful, even. For navigation purposes, getting good return on near stuff to avoid hitting it, and big far stuff for mapping, works nicely.
The massive return signal you get off a wall is going to swamp fine details in front of it anyway.
Just because something is far away, doesn't necessarily mean you're not going to get a massive return signal off it. There's almost no useful information in the returned amplitude signal, if you're working in free space with arbitrary objects. (The same object flat-on will return hundreds of times more signal than the same object with a 10 degree rotation). Try just turning the gain up at all times. Accept clipping is going to happen. Trimming the system for 'perfection' just means that your software's going to fail when faced with the inevitable imperfect data from the real world.

Of course, if you're in a closed system, with nicely coupled transducers and you're looking for subtle reflections, as in medical imaging, then yeah, ramp that gain!

[T3sl4co1l]

When you say "I want to change the gain of the amplifier in inverse proportion to the distance of the traveled sound wave", it sounds like you want to program the signal path gain based on an outside variable or expectation, but that won't account for variations in reflectivity (how much sound is being reflected at that distance?) or noise and interference (if something louder than the reflection is present, it will overload your amplifier).

What you should have is the signal path working all on its own, where an error amplifier controls the AGC / gain level so that the output amplitude is constant.  The AGC function must be one-to-one (more AGC = more gain).  To make use of the AGC level, it should be a smooth and linear function as well, usually logarithmic.  Then, reading the AGC voltage is an indication of input signal strength, which you can use to relate to signal range or whatever.

The FM analogy is using a PLL to both tune and detect the signal; if the oscillator is linear, the changing control voltage is the signal directly.

Tim

[yramgu]

Tl026 is quite cheap, would it match your needs?

[hamdi.tn]

8 channel analogue acquisition and a changeable gain, i think the STM32F373 have both , embedded Delta Segma ADC and changeable gain, not sure how many channel it have, check datasheet.

[David Hess]

I would use an operational transconductance amplifier like the LT1228, a Gilbert Cell mixer, or analog multiplier like the AD633.  This can also be done with a pair of matched JFETs.

[dannyf]

I would use a dual gate mosfet. It also has the added advantage of being neutral to the direction of the amplitude: the gain goes up with higher voltage on the 2nd gate initially, and after the peak, the gain goes down as the voltage on the 2nd gate continues to increase.

[hamdi.tn]

To correct my last post it's STM32F373 not 323

[Korken]

Precipice:
>> From a practical point of view, are you sure you want / need to do this?
Short answer: no. During the design discussions it came up as a "good thing to have", hence I'm looking into this.
The ADC we are planning to use (MAX11049) has already 16-bits so I do not believe we need to do anything in the analog part, digital gain (in needed would probably do the trick).
And also, the analog frontend will be a modular board for other projects so in case someone after us might want it as well.

This is more a feasibility study you could say.

T3sl4co1l:
Indeed we do not know the outside effects, this is for the experiments to show. We can only estimate the "worst case" when all sound energy is reflected from objects (absorption in materials is not a problem in the end result though).
But I like to plan for the future, in case we might need it. But still as said above, with 16-bits of range I don't think we will have any problems.

Or, maybe just have a constant gain input stage. To use if the dynamical range of the input is lower/larger than the dynamical range of the ADC to allow some tuning of the system dependent on the environment.

hamdi.tn:
I did have a look at this, but the choice was made to use FPGA + ADC to do the math in there (to future proof the system for others to use).
The ADC is quite nice though (STM32F373), used it in a different project.

David Hess & dannyf:
Sadly this is outside my area of knowledge, would you care to explain what you had in mind?

[David Hess]

I would use an operational transconductance amplifier like the LT1228, a Gilbert Cell mixer, or analog multiplier like the AD633.  This can also be done with a pair of matched JFETs.

David Hess & dannyf:
Sadly this is outside my area of knowledge, would you care to explain what you had in mind?

I am not sure what there is to explain.  The first three examples I gave directly implement variable gain controlled by an analog signal using variable transconductance.  The datasheets for the LT1228 and AD633 contain examples of how they are used and there are application notes as well.  The LT1228 uses an input current to control gain while the AD633 and similar analog multipliers operate with symmetrical voltage inputs and themselves are usually based on Gilbet Cell multipliers which provide the basic multiplication function.

The modern trend for ultrasound imaging has been to implement the gain control digitally.  Analog Devices makes ultrasound imaging ASICs which include this function.