Does anyone know of any old-school function generator ICs (similar to the ICL8038 or XR2206) that are still on the market? (i.e. not obsolete / NFFD - I'm not interested in scouring the globe for old inventory)

I'm aware of digitally-programmable parts that use I2C / SPI, but am looking for something that won't involve adding a micro to a legacy design.

I'm looking primarily for an IC that will output a triangular waveform at low frequency (640Hz). The XR2206 was the original part in the application which is now obsolete. The frequency can be adjusted a small amount from an external input (+/- 50Hz or so).

Exar has obsoleted most (all?) of their function generator parts:

enter image description here

The product has passed into obsolescence and the assigned engineer is no longer with the company. The question is now moot.


closed as off-topic by PeterJ, Ricardo, Dave Tweed Mar 20 '15 at 14:00

This question appears to be off-topic. The users who voted to close gave this specific reason:

If this question can be reworded to fit the rules in the help center, please edit the question.

  • 3
    \$\begingroup\$ What features do you need? Sine/Square/PWM/Saw/Triangle, frequency, etc. \$\endgroup\$ – Kevin Vermeer Apr 1 '11 at 19:38
  • \$\begingroup\$ Updated the question. \$\endgroup\$ – Adam Lawrence Apr 1 '11 at 20:03
  • 1
    \$\begingroup\$ What properties do you want from this triangle wave that can't be provided by something simple like a 555/556 IC? Frequency stability or wave shape is critical? Consistent amplitude at different frequencies? \$\endgroup\$ – endolith Apr 4 '11 at 16:37
  • \$\begingroup\$ The legacy design has an XR2206. The behaviour of the triangle wave should be 'compatible' with the XR2206 - fixed amplitude and clean egdes. The engineer working on this activity isn't comfortable going with a discrete design. \$\endgroup\$ – Adam Lawrence Apr 4 '11 at 20:43
  • \$\begingroup\$ eBay shows many sellers having the XR2206 available at prices ranging from less than $2.00 on up. Although eBay should not be used as a supplier for components that are intended for use in either production or life-safety, the parts available from there are usually adequate for hobbyists or for internal use within your company. I'm reasonably confident that XR2206 chips that you purchase from eBay will work correctly. \$\endgroup\$ – Dwayne Reid Mar 20 '15 at 16:27

This would seem to be similar to the XR2206 :)

The XR2209 is a monolithic voltage-controlled oscillator (VCO) integrated circuit featuring excellent frequency stability and a wide tuning range. The circuit provides simultaneous triangle and squarewave outputs over a frequency range of 0.01Hz to 1MHz. It is ideally suited for FM, FSK and sweep or tone generation, as well as for phase-locked loop applications.


  • Waveform Generation Triangle, Sawtooth, Pulse, Squarewave
  • 2
    \$\begingroup\$ That's what I get for listening to my colleague whine, "We've scoured the globe and can't find anything equivalent!" Well done sir. \$\endgroup\$ – Adam Lawrence Apr 2 '11 at 15:15
  • 1
    \$\begingroup\$ Sorry, I had to toggle this answer off - my colleague contacted Exar and was told that this part was also discontinued, as per PDN 07-003-01 (which obsoleted the XR2209, coincidentally). \$\endgroup\$ – Adam Lawrence Apr 4 '11 at 14:48
  • \$\begingroup\$ Weird. It doesn't mention that anywhere on the website? It's still listed in parametric searches and is still for sale. \$\endgroup\$ – endolith Apr 4 '11 at 15:34
  • \$\begingroup\$ Agreed. I updated the question with the PCN they sent us. \$\endgroup\$ – Adam Lawrence Apr 4 '11 at 16:20

If you can't find a proper part in your area, then you might end up with a micro and software based DDS like this. Small AVR and resistor ladder is all it takes.

  • 1
    \$\begingroup\$ How does the cost and accuracy of a resistor ladder compare with that of a DAC? \$\endgroup\$ – supercat Dec 13 '11 at 19:56
  • \$\begingroup\$ @supercat, wow, didn't know that until you made me aware. Thanks. \$\endgroup\$ – abdullah kahraman Jul 8 '12 at 20:30
  • 1
    \$\begingroup\$ @abdullahkahraman: I should have been a bit less brief in my comment. I've built and used R/2R DACs in the past, back in the days when the only EEPROM-based Microchip part was the 16C84. They are certainly cheap in mass-production (sixteen 1% resistors would cost less than $0.02 as discrete components, or $0.06 as four arrays of four), but the accuracy isn't great. Further, for hand-built projects, soldering a DIP8 may be a lot more convenient than soldering sixteen resistors (and of course, using a processor with built-in DAC may be more convenient yet). \$\endgroup\$ – supercat Jul 9 '12 at 15:34
  • 2
    \$\begingroup\$ That's a better price for those R/2R packages than I remember the last time I looked at them. On the other hand, the accuracy of an R/2R DAC will be limited by the cleanliness of VDD and the ability of the processor's pins to pull cleanly to that level, while using a monolithic DAC will generally allow one to use a filtered supply as a source. \$\endgroup\$ – supercat Jul 9 '12 at 18:20
  • 1
    \$\begingroup\$ @abdullahkahraman: Both approaches definitely have their places. Actually, one nice advantage of an R/2R DAC is that it's easy to add extra input whose weight equals one lsb, and drive that with a PWM. Using e.g. a 6-bit R/2R DAC and a 4-bit PWM, one can get 10-bit resolution a a much faster PWM rate than one could output 10-bit PWM. The resulting output pattern will have good differential non-linearity characteristics, but the integral non-linearity will be limited to that of the R/2R DAC. \$\endgroup\$ – supercat Jul 9 '12 at 18:24

I encounted this issue about one year ago trying to source an analogue function generator and noticed that they are generally obsolete or not for new design.

I also had a restriction that prevented the use of a microcontroller for the design. In the end I settled on 555 timer with supporting components and was able to obtain a reasonably precise sawtooth generator. It worked well, and took a few attempts to get my calculations correct, in the end the result was ideal.

With a 555 timer you can use it to generate various waveforms and it does not require any digital components.


Why not the simple two op-amp circuit that's been known for decades, show for example here with explanation of how it works. There's also the well-known book, IC Op-Amp Cookbook by Walter Jung. I can't find my copy right now, but I'm fairly sure the same circuit is described in that book.

If you're keen on having physically one chip, there are inexpensive dual op-amp chips such as the LM358.

I'd better include a simple schematic here in case someday that link goes bad (though there are plenty of place on the web to find the same circuit)

What the 8038 or XR2206 have inside: a constant current source charging a capacitor (not inside) which switches polarity when the capacitor's voltage reaches certain limits. The op amp circuit has a resistor not a constant current source, but by the principle of virtual ground and the first op amp acting as a comparitor putting out either voltage rail (or close), for all practical purposes the charging current is nearly constant. Unless you need super-good linearity, temperature stability, or some other quality undescribed in your question, the two op-amp circuit is doing the same thing.

Simple two op-amp circuit to generate square and triangle waves

  • \$\begingroup\$ Please note the comment on the question itself: "The engineer working on this activity isn't comfortable going with a discrete design." \$\endgroup\$ – Anindo Ghosh Jul 18 '13 at 8:27

Not the answer you're looking for? Browse other questions tagged or ask your own question.