XR2206 Sine distortion problem

[jmur]

Has anyone any experience with the**broken link removed** Function generator IC, Im prototyping a function generator design based on the XR2206, but Im having trouble with distortion on the positive peaks of sine o/p. The distortion is present for the full range of frequencies and is not affected by DC offset or amplitude... .Adjusting the symmetry (22k pot) makes no difference either.

+VCC = 5v and -VCC = -5V
**broken link removed**

Sine @ Pin 2
**broken link removed**

Zoomed in on the peak...
**broken link removed**

 

[ericgibbs]

hi,
Do you still get that glitch when the CD40106 is disconnected.??

 

[jmur]

Hi,
Forgot to mention that neither the CD40106 or the TL-072 are connected at the moment, so to answer your question... yes.

 

[ericgibbs]

Hi,
Forgot to mention that neither the CD40106 or the TL-072 are connected at the moment, so to answer your question... yes.

hi,
These images are from a LTspice simulation I have just done, the spike is also on the Sim, it must be a design problem.????
Very strange....

 

[jmur]

Hmm interesting and yes "very strange" any ideas on whats causing it?

 

[jmur]

Iv messing around in LTspice with component values but nothing helps...

 

[jmur]

it must be a design problem.????

Sorry to rebumb this thread... eric did you mean a design problem in the XR2206, as thats what its starting to like! After searching a bit on the net it appears that Im not the first to have this problem, I cant find any solutions though. This is really frustrating!!

 

[crutschow]

I believe the XR2206 generates a sine wave by diode shaping of the triangle wave. There is apparently a small glitch that is generated at the peak of the triangle wave, which is also the transition point of the square-wave output.

You might try a high frequency low-pass filter, with a break frequency slightly above the highest frequency you plan on generating, on the output to suppress the glitch. The oscilloscope picture shows a glitch rise-time of about 200ns which translates to a frequency of about 1.75MHz. Thus a low-pass filter of 1MHz or lower would provide some suppression of the glitch. Configuring the output op amp as a 2nd-order low-pass active filter would improve the suppression over a simple 1-pole RC passive filter. Active Filter Design Application - FILTERPRO - TI Software Folder will help with the active filter design.

 

[Mikebits]

According to the data sheet the resistor between pins 13 and 14 should be a 500 ohm pot and adjusted for lowest distortion. You might try swapping out the 200r resistor with a pot.

**broken link removed**

 

[jmur]

I believe the XR2206 generates a sine wave by diode shaping of the triangle wave. There is apparently a small glitch that is generated at the peak of the triangle wave, which is also the transition point of the square-wave output.

You might try a high frequency low-pass filter, with a break frequency slightly above the highest frequency you plan on generating, on the output to suppress the glitch. The oscilloscope picture shows a glitch rise-time of about 200ns which translates to a frequency of about 1.75MHz. Thus a low-pass filter of 1MHz or lower would provide some suppression of the glitch. Configuring the output op amp as a 2nd-order low-pass active filter would improve the suppression over a simple 1-pole RC passive filter. Active Filter Design Application - FILTERPRO - TI Software Folder will help with the active filter design.

Thanks I think your right about the shaping method. Iv been playing around with the TI app, but for the filter it generates it says that I need an op-amp with a minimum of about 82MHz GBW, could that be right, I haven't formally studied Active Filters yet but that figure seems rather high, The fastest Op-Amp I have is an NE5534(10MHz GBW) If I do need a faster opamp, which would you recommend?

**broken link removed**

According to the data sheet the resistor between pins 13 and 14 should be a 500 ohm pot and adjusted for lowest distortion. You might try swapping out the 200r resistor with a pot.

**broken link removed**

You can use a 200R (fig.11 in data sheet) the 500R pot is to adjust THD, though I have already tried varying the value in LTspice, Unfortunately It doesnt remove the glitch though, it just alters the shape of the wave. Thanks

 

[crutschow]

Thanks I think your right about the shaping method. Iv been playing around with the TI app, but for the filter it generates it says that I need an op-amp with a minimum of about 82MHz GBW, could that be right, I haven't formally studied Active Filters yet but that figure seems rather high, The fastest Op-Amp I have is an NE5534(10MHz GBW) If I do need a faster opamp, which would you recommend?

Try a Salen-Key rather than the MFB topology. I believe it requires a lower GBW for a given corner frequency.

 

[jmur]

Try a Salen-Key rather than the MFB topology. I believe it requires a lower GBW for a given corner frequency.

I just tried but It didn't make any difference

 

[crutschow]

I'm not sure why they suggest such a high GBW. It would seem that 10MHz would be sufficient for a 1MHz filter with a passband gain of 1.

 

[Frosty_47]

Thanks I think your right about the shaping method. Iv been playing around with the TI app, but for the filter it generates it says that I need an op-amp with a minimum of about 82MHz GBW, could that be right, I haven't formally studied Active Filters yet but that figure seems rather high, The fastest Op-Amp I have is an NE5534(10MHz GBW) If I do need a faster opamp, which would you recommend?

**broken link removed**

You can use a 200R (fig.11 in data sheet) the 500R pot is to adjust THD, though I have already tried varying the value in LTspice, Unfortunately It doesnt remove the glitch though, it just alters the shape of the wave. Thanks

I recommend you use OPA633 it is an amazing op amp with 2500V/uS Slew rate and 260Mhz Bandwidth.
Also, please download Filter Pro from Texas Instruments. It is simply the best Filter Design Program there is, and it is 100% FREE!!!!!
Here is the link, Active Filter Design Application - FILTERPRO - TI Software Folder
P.S. you will need to register in order to download it...
I love Ti, I received over $400 of parts from them for absolutely free, aside from that, their parts always work to their specs....

 

[jmur]

I'm not sure why they suggest such a high GBW. It would seem that 10MHz would be sufficient for a 1MHz filter with a passband gain of 1.

Yes, I think the TI software is wrong I tried another filter design app and the required GBW was quoted as 1.818MHz which seems more reasonble. However I went ahead and simulated the filter design produced with the TI app, but it only barely rounded the tip of the glitch so I experimented with the values of the components and eventually settled on changing C3 from 10pF to 200pF which seems to have sorted it(see attachment) . What do you think?

I recommend you use OPA633 it is an amazing op amp with 2500V/uS Slew rate and 260Mhz Bandwidth.
Also, please download Filter Pro from Texas Instruments. It is simply the best Filter Design Program there is, and it is 100% FREE!!!!!
Here is the link, Active Filter Design Application - FILTERPRO - TI Software Folder
P.S. you will need to register in order to download it...
I love Ti, I received over $400 of parts from them for absolutely free, aside from that, their parts always work to their specs....

Thanks, that does look like an amazing opamp! but it comes at a cost $$$$. I dont think I need such a high spec for this application -The NE5534 seems to be doing the trick!

 

[ericgibbs]

hi jmur,
That looks reasonable, what the application use for the function generator.?

 

[jmur]

hi jmur,
That looks reasonable, what the application use for the function generator.?

Thanks eric,
Im building the function generator for fun and to save myself the cost of buying one for my home lab. I will be using it for testing and prototyping -so basically it could be used for anything, but If at some point if I find that its not up to scratch I can always upgrade...

 

[jmur]

It seems I spoke too soon, naively I only simulated the filter with an input of 10KHz, realizing this I altered the values of timing capacitor and resistor to get a 1MHz O/P but the XR2206 model doesn't seem to want to go past 90KHz so I settled for testing at that frequency. The results were disappointing -300mV attenuation.

I set C3 back to the value calculated to by the TI app and sure enough the attenuation was almost completely reduced at the same time filtering out the glitch.

It looks to me like I may need to use a different filter topology in order to effectively filter out the glitch for the full range of input frequencies while limiting the amount of attenuation.

As Iv said though my knowledge in the area of active filter design is lacking.
Anyone, any suggestions?

 

[audioguru]

A Sallen and Key second order Butterworth lowpass filter is non-inverting and has C2 double the value of C3.
If it cuts 1MHz by -3dB then it will have very little effect on the distortion.

 

[Frosty_47]

Thanks, that does look like an amazing opamp! but it comes at a cost $$$$. I dont think I need such a high spec for this application -The NE5534 seems to be doing the trick!

What about Free samples form Ti ?