XR2206 Sine distortion problem

[jmur]

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.

So what are you saying Its not possible to both filter out the glitch and to limit attenuation for the full range of frequencies?

What about Free samples form Ti ?

Sure if its easy to get them then Id give it a shot. Whats there criteria? surely if they were that easy to get then they'd be out of bussiness (obviously they wouldn't as company's would want bulk for production... but I think you get my point)

 

[crutschow]

So what are you saying Its not possible to both filter out the glitch and to limit attenuation for the full range of frequencies?

A higher order filter, such as two op amps to give a 4th-order filter would give more attenuation of the glitch. You could also switch in a lower frequency filter when you are generating low frequencies to better filter the glitch. Some sort of tracking filter would be ideal, but they are difficult to build in analog form with more than one pole.

 

[jmur]

A higher order filter, such as two op amps to give a 4th-order filter would give more attenuation of the glitch. You could also switch in a lower frequency filter when you are generating low frequencies to better filter the glitch. Some sort of tracking filter would be ideal, but they are difficult to build in analog form with more than one pole.

I just tried a 4th order filter there wasn't much of an improvement. What I will do (as you have suggested) is to use a different filter for each range.

Im using a double pole 6-way switch to switch between timing capacitors for the different ranges, only one of the poles is used so I could use to the other to switch between fiters, or simply switch between different C3's.. should be quite easy.

Now to figure out the values for the caps...

 

[audioguru]

You need to switch both resistors or both capacitors so that the filter remains as a sharp Butterworth second-order filter insted of switching only C3 then the filter becomes a droopy Bessel first-order filter.

 

[jmur]

You need to switch both resistors or both capacitors so that the filter remains as a sharp Butterworth second-order filter insted of switching only C3 then the filter becomes a droopy Bessel first-order filter.

I though that was a little too easy to be true! still though it shouldn't be too much of a problem
Edit: ie to switch both the caps or...
Thanks

 

[MRCecil]

Your problem stems from the "current switches" of the XR2206, not unlike the crossover distortion of some op amps. You only noted the issue at the predominate point, but if one looks at the signal at the opposite polar transition one can see the same issue at a somewhat lesser degree. The real issue is with the deviations from the manufacturer's recommended configurations.

I'm thinking that what one should be more concerned with is first, following the datasheet recommendations, and second the noise superimposed on the sine output as displayed.

Is the MO input (pin 3) bypassed properly? There should be a 1uf-10uf tantalum cap. there as well as the proper configuration as outlined in the datasheet.

You say the symmetry pot across pins 15 & 16 has no effect. Have you tried a 25k 20 turn cermet pot to get the desired results?

In conjunction with the symmetry pot, have you tried the recommended 500-ohm with at least a 10 turn pot for the THD adjust between pins 13 & 14? One may have to make adjustments to the amplitude also to ameliorate the effects of the crossover spikes, which should only be apparent in the upper frequencies.

Another note of departure from the recommendations is the 680pf cap used for the upper end frequency output. That would indicate to me one has constructed the circuit on a breadboard. That is right around the value needed to get up the hill to 1Mhz on a breadboard owing to the stray capacitance inherent with that construction. One cannot trust all the results observed as credible when implementing that device on a breadboard.

Before I forget it, get rid of that 100nf cap between pins 4 & 12 and put one on pin 4 to ground and another from pin 12 to ground, and you just might get a cleaner sine wave sans noise if the other items are dealt with.

If the device is constructed correctly, there will be no need to for add-on circuitry, for which to make corrections to meet the THD limits specified. Been there and done that with the 2206 many times with my students waaay back when.

 

[jmur]

Is the MO input (pin 3) bypassed properly? There should be a 1uf-10uf tantalum cap. there as well as the proper configuration as outlined in the datasheet.

I forgot for to put that in for this round of tests, thanks for reminding me.

You say the symmetry pot across pins 15 & 16 has no effect. Have you tried a 25k 20 turn cermet pot to get the desired results?

No only a regular pot, I need to get some...

In conjunction with the symmetry pot, have you tried the recommended 500-ohm with at least a 10 turn pot for the THD adjust between pins 13 & 14? One may have to make adjustments to the amplitude also to ameliorate the effects of the crossover spikes, which should only be apparent in the upper frequencies.

I only tested this in spice as I dont have any 500-ohm pots at them moment..

Another note of departure from the recommendations is the 680pf cap used for the upper end frequency output.

Yes your quite right, I thought I had changed that to a 1nF which I will do.

That would indicate to me one has constructed the circuit on a breadboard. That is right around the value needed to get up the hill to 1Mhz on a breadboard owing to the stray capacitance inherent with that construction. One cannot trust all the results observed as credible when implementing that device on a breadboard.

I gave up on protyping on a breadboard a while back and switched to stripboard.

Before I forget it, get rid of that 100nf cap between pins 4 & 12 and put one on pin 4 to ground and another from pin 12 to ground, and you just might get a cleaner sine wave sans noise if the other items are dealt with.

will do!

If the device is constructed correctly, there will be no need to for add-on circuitry, for which to make corrections to meet the THD limits specified. Been there and done that with the 2206 many times with my students waaay back when.

Glad to hear it, I was starting to get really frustrated at the lengths I had to go to get a reasonable quality signal(and still not obtaining one)

It maybe a couple of weeks before I can implement all of the changes as I need to order parts, +Im taking a weeks holiday next Wednesday. Once I have implemented... Ill post back

Thanks again.

 

[audioguru]

I have not seen an XR2206 but many years ago I used the ICL8038 that works the same. It also had pretty bad-looking distortion.
The datasheets show that it is normal for the distortion to get really bad at high frequencies.

 

[jmur]

I have not seen an XR2206 but many years ago I used the ICL8038 that works the same. It also had pretty bad-looking distortion.
The datasheets show that it is normal for the distortion to get really bad at high frequencies.

Yes Iv noticed the sine gets distorted an attenuated at higher frequencies which can be calibrated a little by adjusting the symmetry pot, but it doesn't seem possible to find a one size fits all value so instead I think I will only run it up to 200-300KHz so that it will have a consistent output throughout the range

 

[zesla]

Hi,

Just want to use XR2206 as a just sine generator at frequencies below 100kHz, Do I have the problem of distortion at the output of the chip?
Is there any better function generator chip to be used instead of XR2206?

And my last question, Is it possible to have a sine wave with a duty cycle of upper/lower 50%? Is the sine wave generated by XR2206 fixed at 50% of duty cycle at all its operating frequencies?

Thanks

 

[audioguru]

The IC is the only function generator IC that is still made. I used the similar ICL8038 IC about 35 years ago.
It takes its triangle-wave and "breaks it" into a sine-wave looking shape. Its typical distortion at frequencies up to 100kHz is 0.5% if it is trimmed properly and if its duty-cycle is set to 50:50. You can hear the distortion and see it on a 'scope.

 

[MrAl]

Hi,

audioguru:
OH yeah, i remember that chip. I built a function generator out of that chip a long long time ago, and used a huge knob on the front to get good adjustment resolution. It was pretty nice until i accidentally connected the 9v battery up backwards when plugging it in. That did it in. I used a 9v battery snap, which if you align the battery backwards while trying to plug it in the two contacts touch in reverse for a brief but destructive moment. That's a lesson hard learned.

Also, i noticed the distortion on the 8038 chip was better than with the XR2206 part chip. The XR2206 part chip isnt very good as the frequency increases. One way around this is to build a filter, which probably has already been talked about, and it ends up being a little frequency selective so it may need adjustment with frequency. On the other hand, the high frequencies are the worse so maybe it could be designed to kick in after a certain frequency has been reached (ie cutoff frequency could be somewhat higher).

 

[bountyhunter]

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....

I use a 2206 in my sine wave generator and I don't see any noticable distortion until relatively high frequencies.

 

[atferrari]

Is it possible to have a sine wave with a duty cycle of upper/lower 50%?

Talking of sinewaves "duty cycle" means nothing.

 

[zesla]

Talking of sinewaves "duty cycle" means nothing.

Really? So please can you tell me the waveforms which the duty cycle means there?
Thanks

 

[audioguru]

Duty cycle applies to "square" (rectangular) waves and to "triangular" (sawtooth) waves.

 

[Neil Groves]

Hi Guys....

Well i just joined this what looks to be an amazing site, having stumbled across it in the bing search engine whilst looking for the xr2206 function generator chip, i have just built from a kit using this I.C a function generator, i am experiencing thesame distortion "blip" on the waveform, however my question is a little more complex, having looked at my waveforms on the scope i notice they all have a D.C bias of a couple of volts, i thought A.C (audio) signals bounced around 0 Volts and had no D.C offset? i intend to use the generator as an audio source but am a little concerned that my output has a D.C offset, can anyone explain this to me please?

I used to work as an electronics technician 11 years ago, and am very rusty since i have only just got around to setting up the 'Lab' again.

thanks in anticipation

Neil.

 

[audioguru]

If the power supply for the XR2206 has a single polarity then of course the output will have a DC voltage that the signal swings around. If you add an output coupling capacitor then the capacitor blocks the DC but passes the signal.

 

[Neil Groves]

If the power supply for the XR2206 has a single polarity then of course the output will have a DC voltage that the signal swings around. If you add an output coupling capacitor then the capacitor blocks the DC but passes the signal.

Of course....now i stop to think about it, how can a signal go outside the supply rails....DOH!

Thanks

Neil.

 

[wkrug]

Why you dont use a DDS Function Generator IC like the AD9833 or AD9834?
This Parts Generate Sine Waves by an Sine Table and a D/A converter.
The Output Filter can be made passive an must be configured for the maximum needed frequency.
The dissortion increase as higher the frequency. Maximum is 12,5MHz for the AD9833, but then the Sine Wave has harmonics. I would end at 4MHz.
The generated frequencies are quite exact ( about 0,3Hz - appends from the Clock Source ) and append only from an external Crystal Clock Generator an a frequency tuning word.

It have 2 Outputs Output 1 Sine/Triangle. Output 2 Rectangular.
Output 1 has 2 Ports with an differential output that offers 180° deg phase shift.

The Chips can be steered by an microcontroller via an SPI Interface.