XR2206 : strange behaviour !

[ricothetroll]

Hi !

Doing some tests with that chip (http://www.exar.com/Common/Content/ProductDetails.aspx?ID=XR2206), I niticed a strange thing. My schematic is here :

http://www.wuala.com/ricothetroll/public/test.pdf

That's the basic test circuit, with a voltage divider at the output. What's strange is that with that voltage divider the triangle wave has rounded edges, as if there's a hi-cut on it ! Before the voltage divider the waveform is perfect, and then after it's highly modified. I really don't understand why !

Replacing the 680k-510k by 68k-51k, it now works perfect !

The same phonomenon occured with the 820k-220k I put at the square wave output.

I would like to understand what happened, so if anyone has a clue...

Best regards !

Eric

[Thomeeque]

Looks like input capacitance of your oscilloscope (probe) forms LPF with the output resistor.. T.

[PRR]

> with that voltage divider the triangle wave has rounded edges, as if there's a hi-cut on it!

What Tomas says.

It IS a hi-cut.

There's hi-cuts everywhere.

> Replacing the 680k-510k by 68k-51k, it now works perfect !
> The same phonomenon occured with the 820k-220k I put at the square wave output.

The output of 680K+510K is 291K; of 680K+510K is 29K.

Say your 'scope is 50pFd input capacitance to ground.



50pFd against 291K is 11KHz. 50pFd against 29K is 109KHz.

If the bandwidth is not 10 to 100 times the fundamental of the triangle wave, it will show (roughly) 10% to 1% tip-rounding.

So an 11KHz hi-cut is visible on a 1KHz triangle, very obvious on a 10KHz triangle.

Same for the square. 820K||220K is 173K, against 50pFd is 18KHz hi-cut.

The 68K+51K case "looks better" because the 109KHz hi-cut is far above audio frequencies. Crank the XR and 'scope to 500KHz, it will be real roundy.

WHY are you using such high divider values??

Dividers should be a compromise between the source and the load.

Take a simple case. Source is 1K, load is 100K. Divider can be around 10K total. For 2:1 loss, say 4K7+4K7.

The XR2206 tri-sine output is 600 ohms. I do not know what your real load is. Your test load seems to be 1Meg+50pFd 'scope. This is 1Meg up to 3KHz, 159K at 20Khz, 32K at 100KHz, 3K at the XR2206's 1MHz ceiling.

For good 20KHz performance you need divider equivalent impedance well below 159K.

Since XR2206 will comfortably drive 4K7+4K7 (or 10K+2k2), which work out well below 159K, that's the direction you should be going.

i.e.: use RED-stripe resistors, NOT yellow-stripe.

There is a BETTER way to control XR2206's output. A resistor from pin 3 to half-supply reduces the output voltage while preserving the low 600 ohm buffer impedance.

The square is a different case. Stock, it is like 10 ohms low and 10K high. The easy way to reduce it is to put like 3K3 to ground: this prevents the output from rising to full supply voltage.

[MetalGuy]

What would you suggest to control square wave's amplitude?
You say there's a buffer inside the chip but I see another one added here:

http://bp2.blogger.com/_BQac7gcHPK0/RjyjRHbcE0I/AAAAAAAAAO0/2UirYWgOAG4/s1600-h/circuit_XR2206+Function+Generator-1.jpg

[PRR]

> You say there's a buffer inside the chip

Not on the square-wave. It is low-Z pull-down but hi-Z pull-up.

Given 50pFd loading, the rise-time will be horrible at 300KHz, visibly slow at 30KHz.

> but I see another one added here

That one has a small glitch but will mostly be very-fast to ~~~1MHz.

[Thomeeque]

What would you suggest to control square wave's amplitude?

E.g. use the buffer from the circuit_XR2206+Function+Generator-1.jpg, just replace R13 by the 2k2 pot and connect square wave signal output (left leg of R14) to it's wiper..

[ricothetroll]

Hi,

Thanx a lot for your answers ! I must admit that I didn't know there was such a cap to ground on a scope !  Now my observations make sense.

I'd like to include that oscillator in an AD 633 ring mod. My problem is that the 3 waveforms have different levels and also the 6V offset is a problem because I don't want it at the input of the AD 633. I first though of capacitive coupling with voltage dividers but for Fc low enough to use the circuit as a tremolo I needed pretty big capacitors and the circuit took several seconds to establish, especially for the square wave.

I though then of direct coupling : with a summing circuit, I add -6V and then control the gain of the opamp to have a constant output level. Here is my schematic :

http://www.wuala.com/ricothetroll/public/Oscillator.pdf

The simulation of it works pretty good, but as I'm not familiar with direct coupling, maybe there are some other things to consider. What do you think ?

The 633 audio circuit is here simplified.

Best regards.

Eric

[MetalGuy]

Given 50pFd loading, the rise-time will be horrible at 300KHz, visibly slow at 30KHz.

That one has a small glitch but will mostly be very-fast to ~~~1MHz.

I'm interested only in the 20Hz-20kHz range for audio amps testing so I guess it will be OK.

[tubelectron]

Hi MetalGuy,

I'm interested only in the 20Hz-20kHz range for audio amps testing so I guess it will be OK

If you do not search any precision in frequency or low distortion operation, the XR2206 will be OK, that is to say : for making a simple tone generator/oscillator. Otherwise you may be disappointed !

A+!

[MetalGuy]

If you do not search any precision in frequency or low distortion operation, the XR2206 will be OK, that is to say : for making a simple tone generator/oscillator.

It's not for Hi-Fi stuff so I guess it will do the job.

[ricothetroll]

Hi,

Thanx a lot for your answers ! I must admit that I didn't know there was such a cap to ground on a scope !  icon_redface Now my observations make sense.

I'd like to include that oscillator in an AD 633 ring mod. My problem is that the 3 waveforms have different levels and also the 6V offset is a problem because I don't want it at the input of the AD 633. I first though of capacitive coupling with voltage dividers but for Fc low enough to use the circuit as a tremolo I needed pretty big capacitors and the circuit took several seconds to establish, especially for the square wave.

I though then of direct coupling : with a summing circuit, I add -6V and then control the gain of the opamp to have a constant output level. Here is my schematic :

http://www.wuala.com/ricothetroll/public/Oscillator.pdf

The simulation of it works pretty good, but as I'm not familiar with direct coupling, maybe there are some other things to consider. What do you think ?

The 633 audio circuit is here simplified.

Best regards.

Eric

Up !