EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: suspension on October 16, 2022, 06:11:31 pm
-
This article talks about using various Analog Devices chips to build an output stage of a function generator which allow precise control of output magnitude and gain through a control voltage (https://www.analog.com/en/technical-articles/design-of-a-powerful-signal-generator-output-stage.html (https://www.analog.com/en/technical-articles/design-of-a-powerful-signal-generator-output-stage.html)). In many of the cheap/entry level function generators in the market, I see a different approach to set gain and magnitude using relays and passive components. I would like to understand basic principles behind such a VGA, specifically to understand basic structure/block diagram, how they achieve high resolution (say 5mv per step for a total gain of 40db), how to make sure they work foe high frequencies (>5MHz atleast), if they use any digitally controlled POT's and anything else that is important.
-
The internal implementations of something like a variable gain amplifier is going to vary dramatically. On a board, implementing it myself I'd likely go with a digipot as you've suggested, but in a chip I'd likely use switches capacitors to control my gain, there are many reasons for doing that mostly related to chip design (Things like resistors take stupidly large amounts of space on chip so if an application where you can replace them out with capacitors which are typically smaller you do). It's an area I'm admittedly rather weak, but if switched fast enough a capacitor can look like a resistor. So you use a differential amplifier with a bunch of controllable switched caps. (This is my guess mind you)
For getting to higher frequencies a matter of making sure you spec an amplifier with a high enough gain bandwidth product. From a design standpoint internally it's about making sure you make sure your corners are at substantially higher frequencies (Easier said than done). But in my limited experience my op amps are terrible, but getting better through careful design and caring a lot about accidental parasitics. Wish I had some more to say on that, but unfortunately there's not much to say.
For level of control that's just a matter of having a lot of on off switches in the circuits guts to turn on and off the switched capacitors to change the effective resistance. I suppose they might be doing it by changing their internal clock frequency, but that would be harder control in my admittedly limited experience.
Now someone with more experience please come correct any misconceptions I've inadvertently spread.
-
Most VGAs use Barry Gilbert's transcondcutcance cells. See datasheet for AD600 and www.analog.com/media/en/analog-dialogue/volume-26/number-2/articles/volume26-number2.pdf#page=3 (http://www.analog.com/media/en/analog-dialogue/volume-26/number-2/articles/volume26-number2.pdf#page=3)
see attached from AD. I dont know I'd be buying an AD600! Probably an AD8369 these days.
I'd parallel up some fast opamps for the output stage. eg LT6301, a beast for the least.
-
Thanks for info. I wonder if digipots's would create various non linearities and other issues such as suppressed bandwidth. As per the transcondcutcance cells, I was wondering if this is possible outside of a chip due to large number of component count involved.
-
new AS394 matched dual transistors in any case
they dont mention max ft but it must be ~200MHz
-
The bandwidth of digipots is low. Internally its a stack of analog switches with all their internal capacitances contributing to high overall input/ output capcitance.
see https://www.analog.com/en/analog-dialogue/articles/digipot-specifications-and-architecture.html (https://www.analog.com/en/analog-dialogue/articles/digipot-specifications-and-architecture.html)