Op amp not working out as planned

[netdudeuk]

The AD9833 has a VOUT Maximum of 0.65V and a VOUT Minimum of 38mV.

So, that's the peak input voltage.

Regarding the peak output voltage, here are the specifications for a branded budget 5MHz arbitrary function generator -

2mVpp~10Vpp (50?, ?10MHz)
4mVpp~20Vpp (high impedance, ?10MHz)

So, maybe something along these lines (Vpp and maximum frequency) looks sensible for my purpose, which is to have a signal generator to experiment with op amps (which I'm now actually doing), filters, etc ?

I'll not be putting the signal down long cables.

I understand that I'm going to have to put in more volts in if I want more volts out.

Thanks

[StillTrying]

The AD9833 has a VOUT Maximum of 0.65V and a VOUT Minimum of 38mV.

Yep, it outputs a constant amplitude sine, triangle or square. But with only 25M samples per second I think any wave over 1MHz is going to be quite rough, and 12.5MHz not really possible other than just a 12.5MHz square, unless I've missed something!
http://www.analog.com/media/en/technical-documentation/data-sheets/AD9833.pdf

Have you scoped the AD933's 38-650mV output directly to see what a ~1MHz+ sine, triangle and square look like?

I wouldn't know where to go from here in producing a variable amplitude, variable offset output, but you could get the AD9833's output to a 0V centered 1.25Vpp with something like this.

[Zero999]

The AD9833 has a VOUT Maximum of 0.65V and a VOUT Minimum of 38mV.

Yep, it outputs a constant amplitude sine, triangle or square. But with only 25M samples per second I think any wave over 1MHz is going to be quite rough, and 12.5MHz not really possible other than just a 12.5MHz square, unless I've missed something!
http://www.analog.com/media/en/technical-documentation/data-sheets/AD9833.pdf

Yes you have missed something: an anti-aliasing filter will theoretically give a pure sine wave, right up to half the sampling frequency. In practise, you'll probably be limited to less than half the sample rate, but it should be possible to get a much higher frequency, than 1MHz and still have a decent sine wave.

The problem will be the triangle wave and other arbitrary wave shapes, assuming this IC can do that, but I haven't chekted the data sheet.

[StillTrying]

"The problem will be the triangle wave and other arbitrary wave shapes"

It only does fixed amplitude 10 bit sine, triangle, and square, with a 28 bit word to set the frequency.
I love the way the data sheet keeps saying "No other components necessary".

[james_s]

At least to start with use a bipolar supply, you can work on making a single rail version after you have it working properly. Even a pair of 9V batteries can work quite well, op-amps typically draw very little current so you should be able to run it for a long time. Building a simple bench supply is another classic beginner project, it's hard to have too many power supplies.

[Wimberleytech]

Yes you have missed something: an anti-aliasing filter will theoretically give a pure sine wave, right up to half the sampling frequency. In practise, you'll probably be limited to less than half the sample rate, but it should be possible to get a much higher frequency, than 1MHz and still have a decent sine wave.

Did you mean "smoothing filter," and more specifically a "brick wall smoothing filter" would be the only way to get a pure sine up to half the sampling frequency?

[StillTrying]

Here's another +/- 12V version. Simply driving the top end R1 in the above schem. with -1V to +8V allows for the 1.2Vpp output to be offset all -Ve, all +Ve or centered on 0V of course, which looks like that.

Amplitude is still fixed at 1.2Vpp, until someone comes up with a good simple idea.

LT .asc if someone wants to play with it.

[Zero999]

Here's another +/- 12V version. Simply driving the top end R1 in the above schem. with -1V to +8V allows for the 1.2Vpp output to be offset all -Ve, all +Ve or centered on 0V of course, which looks like that.

Amplitude is still fixed at 1.2Vpp, until someone comes up with a good simple idea.

LT .asc if someone wants to play with it.

Unfortunately the LT1022 is unsuitable for this application. It's over an order of magnitude too slow. Try a faster amplifier such as the AD818.

The AD9833 has a VOUT Maximum of 0.65V and a VOUT Minimum of 38mV.

So, that's the peak input voltage.

Regarding the peak output voltage, here are the specifications for a branded budget 5MHz arbitrary function generator -

2mVpp~10Vpp (50?, ?10MHz)
4mVpp~20Vpp (high impedance, ?10MHz)

So, maybe something along these lines (Vpp and maximum frequency) looks sensible for my purpose, which is to have a signal generator to experiment with op amps (which I'm now actually doing), filters, etc ?

I'll not be putting the signal down long cables.

I understand that I'm going to have to put in more volts in if I want more volts out.

Thanks

What's your budget?

High speed amplifier ICs aren't cheap, especially ones suitable for running off the higher supply voltages required to give 20Vpp. The AD818 I just mentioned above seems to be the best value for money and meets your specification.

You'll need an anti-aliasing filter. I think a passive, LRC design might be the best option here, as it's tricky to design an active filter at these frequencies.

[netdudeuk]

Here's another +/- 12V version. Simply driving the top end R1 in the above schem. with -1V to +8V allows for the 1.2Vpp output to be offset all -Ve, all +Ve or centered on 0V of course, which looks like that.

Amplitude is still fixed at 1.2Vpp, until someone comes up with a good simple idea.

LT .asc if someone wants to play with it.

Unfortunately the LT1022 is unsuitable for this application. It's over an order of magnitude too slow. Try a faster amplifier such as the AD818.

The AD9833 has a VOUT Maximum of 0.65V and a VOUT Minimum of 38mV.

So, that's the peak input voltage.

Regarding the peak output voltage, here are the specifications for a branded budget 5MHz arbitrary function generator -

2mVpp~10Vpp (50?, ?10MHz)
4mVpp~20Vpp (high impedance, ?10MHz)

So, maybe something along these lines (Vpp and maximum frequency) looks sensible for my purpose, which is to have a signal generator to experiment with op amps (which I'm now actually doing), filters, etc ?

I'll not be putting the signal down long cables.

I understand that I'm going to have to put in more volts in if I want more volts out.

Thanks

What's your budget?

High speed amplifier ICs aren't cheap, especially ones suitable for running off the higher supply voltages required to give 20Vpp. The AD818 I just mentioned above seems to be the best value for money and meets your specification.

You'll need an anti-aliasing filter. I think a passive, LRC design might be the best option here, as it's tricky to design an active filter at these frequencies.

I've ordered a AD818 thanks.

[netdudeuk]

I have the AD818 but TBH, I'm not sure to hook it up for this application.

Here's the data sheet -

http://www.analog.com/media/en/technical-documentation/data-sheets/AD818.pdf

Also, the data sheet states (gain ≥ 2, or gain ≤ –1).  Is this really the limit of how much it can boost the signal from the AD9833 ?

Thanks

[Zero999]

I have the AD818 but TBH, I'm not sure to hook it up for this application.

Here's the data sheet -

http://www.analog.com/media/en/technical-documentation/data-sheets/AD818.pdf

Also, the data sheet states (gain ≥ 2, or gain ≤ –1).  Is this really the limit of how much it can boost the signal from the AD9833 ?

Thanks

No, those are the minimum gains before it can become unstable and oscillate: ≥ means greater than or equal to and ≤ means less than or equal to.

This is very common on faster op-amps. Look at the OP37 datasheet, for example: it says gains >5.
http://www.analog.com/media/en/technical-documentation/data-sheets/OP37.pdf

[jmelson]

Assuming the first scope trace is the output from the DDS chip, that is 600 mV peak to peak.  Your resistors program the op amp to have a gain of 11.  So, you are expecting to get an output of 6.6 V P-P from a 5V power supply.  Clearly, it will HAVE to clip.

Jon

[netdudeuk]

I have the AD818 but TBH, I'm not sure to hook it up for this application.

Here's the data sheet -

http://www.analog.com/media/en/technical-documentation/data-sheets/AD818.pdf

Also, the data sheet states (gain ≥ 2, or gain ≤ –1).  Is this really the limit of how much it can boost the signal from the AD9833 ?

Thanks

No, those are the minimum gains before it can become unstable and oscillate: ≥ means greater than or equal to and ≤ means less than or equal to.

This is very common on faster op-amps. Look at the OP37 datasheet, for example: it says gains >5.
http://www.analog.com/media/en/technical-documentation/data-sheets/OP37.pdf

Thanks.  Can I just use this in an inverting amplifier configuration like I did with the previous ones, using the same two resistor calculation to get the desired gain ?

[Zero999]

Yes you have missed something: an anti-aliasing filter will theoretically give a pure sine wave, right up to half the sampling frequency. In practise, you'll probably be limited to less than half the sample rate, but it should be possible to get a much higher frequency, than 1MHz and still have a decent sine wave.

Did you mean "smoothing filter," and more specifically a "brick wall smoothing filter" would be the only way to get a pure sine up to half the sampling frequency?

Yes, you're right, I was talking about a brick wall filter, which is the only way to get a sine wave, up to half the sample rate. Unfortunately, it's useless for other wave shapes.

I have the AD818 but TBH, I'm not sure to hook it up for this application.

Here's the data sheet -

http://www.analog.com/media/en/technical-documentation/data-sheets/AD818.pdf

Also, the data sheet states (gain ≥ 2, or gain ≤ –1).  Is this really the limit of how much it can boost the signal from the AD9833 ?

Thanks

No, those are the minimum gains before it can become unstable and oscillate: ≥ means greater than or equal to and ≤ means less than or equal to.

This is very common on faster op-amps. Look at the OP37 datasheet, for example: it says gains >5.
http://www.analog.com/media/en/technical-documentation/data-sheets/OP37.pdf

Thanks.  Can I just use this in an inverting amplifier configuration like I did with the previous ones, using the same two resistor calculation to get the desired gain ?

Remind me which resistor values are you planning to use? A few schematics have been posted and I don't know which one you're referring to.

That op-amp should be fine for gains greater than 2, as long as the resistor values aren't too high. The bias current of that op-amp is fairly high, so high resistor values will cause it output DC point to drift significantly. High resistor values will also slow it down, as there it will take longer to charge/discharge the parasitic capacitances.

[netdudeuk]

Yes you have missed something: an anti-aliasing filter will theoretically give a pure sine wave, right up to half the sampling frequency. In practise, you'll probably be limited to less than half the sample rate, but it should be possible to get a much higher frequency, than 1MHz and still have a decent sine wave.

Did you mean "smoothing filter," and more specifically a "brick wall smoothing filter" would be the only way to get a pure sine up to half the sampling frequency?

Yes, you're right, I was talking about a brick wall filter, which is the only way to get a sine wave, up to half the sample rate. Unfortunately, it's useless for other wave shapes.

I have the AD818 but TBH, I'm not sure to hook it up for this application.

Here's the data sheet -

http://www.analog.com/media/en/technical-documentation/data-sheets/AD818.pdf

Also, the data sheet states (gain ≥ 2, or gain ≤ –1).  Is this really the limit of how much it can boost the signal from the AD9833 ?

Thanks

No, those are the minimum gains before it can become unstable and oscillate: ≥ means greater than or equal to and ≤ means less than or equal to.

This is very common on faster op-amps. Look at the OP37 datasheet, for example: it says gains >5.
http://www.analog.com/media/en/technical-documentation/data-sheets/OP37.pdf

Thanks.  Can I just use this in an inverting amplifier configuration like I did with the previous ones, using the same two resistor calculation to get the desired gain ?

Remind me which resistor values are you planning to use? A few schematics have been posted and I don't know which one you're referring to.

That op-amp should be fine for gains greater than 2, as long as the resistor values aren't too high. The bias current of that op-amp is fairly high, so high resistor values will cause it output DC point to drift significantly. High resistor values will also slow it down, as there it will take longer to charge/discharge the parasitic capacitances.

The very first one right at the top of the post.

[Zero999]

Yes you have missed something: an anti-aliasing filter will theoretically give a pure sine wave, right up to half the sampling frequency. In practise, you'll probably be limited to less than half the sample rate, but it should be possible to get a much higher frequency, than 1MHz and still have a decent sine wave.

Did you mean "smoothing filter," and more specifically a "brick wall smoothing filter" would be the only way to get a pure sine up to half the sampling frequency?

Yes, you're right, I was talking about a brick wall filter, which is the only way to get a sine wave, up to half the sample rate. Unfortunately, it's useless for other wave shapes.

I have the AD818 but TBH, I'm not sure to hook it up for this application.

Here's the data sheet -

http://www.analog.com/media/en/technical-documentation/data-sheets/AD818.pdf

Also, the data sheet states (gain ≥ 2, or gain ≤ –1).  Is this really the limit of how much it can boost the signal from the AD9833 ?

Thanks

No, those are the minimum gains before it can become unstable and oscillate: ≥ means greater than or equal to and ≤ means less than or equal to.

This is very common on faster op-amps. Look at the OP37 datasheet, for example: it says gains >5.
http://www.analog.com/media/en/technical-documentation/data-sheets/OP37.pdf

Thanks.  Can I just use this in an inverting amplifier configuration like I did with the previous ones, using the same two resistor calculation to get the desired gain ?

Remind me which resistor values are you planning to use? A few schematics have been posted and I don't know which one you're referring to.

That op-amp should be fine for gains greater than 2, as long as the resistor values aren't too high. The bias current of that op-amp is fairly high, so high resistor values will cause it output DC point to drift significantly. High resistor values will also slow it down, as there it will take longer to charge/discharge the parasitic capacitances.

The very first one right at the top of the post.

So your original schematic?


I advise reducing them by a factor of ten: R_IN = 1k, R_F = 10k.

The advice previously given, regarding a dual power supply and output switch is still relevant: a dual 5V or single 10V power supply is essential with 600mVpp in and a gain of 10.

[netdudeuk]

So, no success, with virtually nothing on the output of the op amp.

I've checked the wiring several times and it looks ok.

RIN = 1k, RF = 10k
Non inverting input is biased to 5v
5v supply for the DDS and Arduino
10v supply for the AD818
Common ground

What could be causing this ?

Thanks

[Audioguru]

You did not use a series input capacitor so its input DC voltage is amplified and causes its output to be close to 0V all the time.

[netdudeuk]

I added a series input capacitor and amended RF to 2K.  So, the gain should be 2.

However, the scope traces still don't look right (especially as it isn't inverting).

Either there is still something wrong with the circuit or I'm measuring it wrong 

[Zero999]

I added a series input capacitor and amended RF to 2K.  So, the gain should be 2.

However, the scope traces still don't look right (especially as it isn't inverting).

Either there is still something wrong with the circuit or I'm measuring it wrong 

That doesn't look right. Which trace is which? Where is the 'scope's 0V connection going to?

Please post a complete schematic and a photograph, showing how it's physically connected up.

[netdudeuk]

I added a series input capacitor and amended RF to 2K.  So, the gain should be 2.

However, the scope traces still don't look right (especially as it isn't inverting).

Either there is still something wrong with the circuit or I'm measuring it wrong 

That doesn't look right. Which trace is which? Where is the 'scope's 0V connection going to?

Please post a complete schematic and a photograph, showing how it's physically connected up.

The yellow trace is connected to the output of the DDS before the series input capacitor.  The blue trace is the output of the op amp.

So, yes, it doesn't look right.

The scope grounds, the ground for the 5 volt supply (Rigol DP832) and the ground for the 10 volt supply (the same DP832) are all connected together.

I'll need to get a photo and a schematic later.

Thanks

[netdudeuk]

Schematic and setup photos.

Thanks

[Zero999]

Schematic and setup photos.

Thanks

That circuit looks fine.

Such a high speed op-amp shouldn't be used on breadboard, but it doesn't appear to be oscillating. If all the connections are good, then I'd suspect it's the op-amp. Can you change the op-amp for the TL071 or TL081?

If not, you could try the old '741 but the frequency would need to be lowered to 10kHz and a larger DC blocking capacitor used: try 1μF.

[StillTrying]

The op amp might have more gain if you connected pin 4 to 0V.

[netdudeuk]

The op amp might have more gain if you connected pin 4 to 0V.

It is working somewhat better now thanks 

I wanted to create an adjustable offset on the output so I connected the non inverting input to another rail on my PSU instead of the 5V.  The ground for this was connected to all the others.

However, if I dialled in anything more than a few hundred mV, I had what I'm guessing we'd describe as unwanted oscillation.  The basic sine wave was still there but rather than a line that traced the rise and fall of the signal, think of it as a many columned bar graph where the numbers followed the sine wave.  I don't have access to the equipment at the moment but hopefully it is clear what I mean.

Why would this be ?

Thanks again.