5

I am making a frequency generator using the AD9833 (programmed from the PIC16F876). I wish to use a Programmable Gain Amplifier to adjust the signal level before being sent to a power amplifier (common collector coupled to class AB Push Pull). I have found several PGA's which fit the bill, for what I want, BUT...

The output from the AD9833 is single ended. All the PGA's I have found which fit my requirements use differential input. So my question is:

To use a differential input with a single ended device can I just ground Vin- and put my signal on Vin+ ???

This does stand to reason, but is there anything I have overlooked?

Tim M
  • 1,999
  • 1
  • 19
  • 27
  • What PGA device do you have in mind? – Andy aka May 15 '13 at 12:58
  • If you aren't bound to the 9833, you could consider using the AD9834 instead, which offers some amplitude control. – Chintalagiri Shashank May 15 '13 at 13:02
  • The PGA I am considering is the AD8251. Datasheet link: http://www.analog.com/static/imported-files/data_sheets/AD8251.pdf I had considered the AD9834 but to alter the full scale control I would need an DAC to change the voltage, and board space is tight, by using a PGA I get gain control and buffer amplification in one package – Tim M May 15 '13 at 13:08

3 Answers3

2

AD8521 single ended input: -

enter image description here

It is a true instrumentation amplifier so it will work from a single-ended source as shown above. The circuit also shows how to connect to provide a differential output but this is irrelevant to your question.

Andy aka
  • 434,556
  • 28
  • 351
  • 777
  • I miss understood the diagram, looks like it will work perfectly. Thanks for all your help. :) – Tim M May 15 '13 at 14:41
1

(Could you link an example PGA you are considering?)

I don't believe that will work - it depends on the internal feedback and coupling of the amplifier, but it won't ignore Vin-.

I have in front of me a datasheet for a TPA2006 amp with differential input. It reccomends the following circuit: enter image description here

pjc50
  • 46,540
  • 4
  • 64
  • 126
  • Hi, thanks for your reply, The PGA I am considering is the AD8251, Datasheet link: http://www.analog.com/static/imported-files/data_sheets/AD8251.pdf Thanks for that link, that circuit looks like what I need. – Tim M May 15 '13 at 13:01
  • One of the diagrams on page 21 does show what you are describing - a voltage source with V- connected to ground. So it looks like it will work with this part. However, because it's single-ended you have to divide all your gain numbers by 2. – pjc50 May 15 '13 at 13:13
  • Ah yes, I had thought, seeing at there were two line coming from the voltage source that it was a differential source, but now I see that the line is indeed tied to ground. Many thanks for your help and time. – Tim M May 15 '13 at 13:31
1

Even using a PGA, say the AD8251 that you have in mind, you have only a limited set of output amplitudes that you have available to you, in this case about 0.6V * (1,2,4,8). This isn't really an 'adjustment' of the amplitude for most sorts of applications. Further, you can't reduce the amplitude unless your PGA also allows fractional gain. If it works for you, though, you should probably keep in mind the following :

  1. The AD9833 output isn't biased to ground. The midpoint of the signal is elsewhere, somewhere around 0.3ish volts. This means that simply putting in the output to IN+ and GND to IN- won't work. EDIT: It will still give you a sine, just one thats at some other bias. If that doesnt matter to you, go ahead.

  2. If you want a signal biased to ground, one option to get around this is to use a high pass filter with a really low cutoff frequency or AC couple the signal (which are usually effectively the same thing). This will give you something that is better biased to ground, but the biasing will not be perfect unless you have a very effective filtering circuit. The 24MHz you're using to the run the 9833 will show up on the output and a frequency this high is likely to get rectified along the chain and produce a small additional DC offset. There are other sources of potential DC offset. What this means is that your PGA output won't be a clean sine wave.

  3. The bias point of the 9833, which, by all rights, should be the IN- to your PGA, isn't accessible in any way. You could try the reverse of the approach of 2, where a low pass filter with a really low cutoff frequency could be used to generate something close to the bias point of the sine wave.

  4. If you use 2. or 3, be careful if you use a PGA using a switched capacitor approach (the AD8521 is not one of those, so you should be fine using that). If you do use something with a switched capacitor input, though, there should be a buffer on whichever signal you filter.

Alternative approaches for achieving this sort of control exist. You can use certain DAC's, with the output of the 9833 going into the REF input of the DAC. Make sure that the DAC supports REF levels over the full range of the 9833 output voltage range (not all do). You could also use a digital potentiometer to scale the signal down to some level (which is essentially how some DACs do it anyway). Since it seems you need a high signal level, this would be followed by a fixed gain amplifier. This should easily be able to give you 8-12 bits of control.

Chintalagiri Shashank
  • 2,241
  • 17
  • 26
  • 1
    I'm slightly confused now. The output from the AD9833 will be a sin wave with Vmin @ 38mV and Vmax @ .65V tying the V- input to ground and setting the gain to 2 (to account for the gain halving due to single ended mode) will, at Vmin produce a Vdiff across the input to the AD8251 of 38mV. Which is what will be observed at the output (due to gain of two being cancelled by a drop of 2 due to single-ended mode). When the input signal reaches Vmax the output of the AD8251 will be 0.65V. This is an exact copy of the input signal. I don't see the problem. . . – Tim M May 15 '13 at 14:00
  • @TimMottram Yes, you are correct. I've edited the answer some. You can probably go ahead and use the diff amp if the flexibility is enough. – Chintalagiri Shashank May 15 '13 at 14:26
  • Just saw your edit. I am not concerned about a non ground bias. If it does become a problem I can alter that elsewhere. Thanks for all your help. :) – Tim M May 15 '13 at 14:26