Signal Generator for DIY Audio Electronics

[Amphiprion]

I just wanted to pass along some info that may be useful for those here that dabble in audio electronics and active crossovers.

I came across an old IC made by Exar called the XR2206 that still happens to be in production. It is a 16 pin IC that is a monolithic function generator with many additional features. It is a $2 chip that can create low distortion sine, square, and triangle waves of constant amplitude with variable frequency using a single potentiometer. It also has built in AM and FM modulation.

While a modern PC is perfectly fine for generating sine waves in the audio band, the XR2206 can function from 0.1Hz up to 1MHz (although I have found that above 100KHz the 1% sine wave distortion spec is not accurate).

If anyone needs a frequency generator (for instance it might be of use in the H@LF project posted here on HTGuide) this is a very easy to use (and cheap!) solution. I was able to rig up an amplitude modulation system using two chips and parts from my grab box of electronics (nothing that you can't find at Radio Shack).

Back in the day these types of IC's were common, but the Exar is the only one that I have found that is still in production.

[Hdale85]

Hmm interesting.

[JonP]

Wow, a blast from the past...

I kept meaning to build something like this from a chip I can't even remember the manufacturer of now... sounds very similar, but went from audio to 20Mhz...

Who's selling these nowadays? Think this would be a worthy tool to have...

[Amphiprion]

The MAX038 was the 20MHz chip, it was the better chip but it's not in production anymore. Some places may still have new old stock laying around. The 2206 seems to be fine for distortion at audio frequencies, and I might have just chosen a bad RC combo for the higher frequencies. By 300KHz I was seeing 2nd harmonic at -20dB on my scope's FFT and 3rd at -30dB.

I bought my XR2206's from Future. If you go to www.findchips.com it and search for XR2206 it will pull up all the suppliers that stock it. It's a really useful site for hard to find items.

[Johnloudb]

The MAX038 was the 20MHz chip, it was the better chip but it's not in production anymore. Some places may still have new old stock laying around. The 2206 seems to be fine for distortion at audio frequencies, and I might have just chosen a bad RC combo for the higher frequencies. By 300KHz I was seeing 2nd harmonic at -20dB on my scope's FFT and 3rd at -30dB.

I bought my XR2206's from Future. If you go to www.findchips.com it and search for XR2206 it will pull up all the suppliers that stock it. It's a really useful site for hard to find items.

Very interesting find. Are you sure the MAX038 is not in production? Maxim makes no note of this on their website. They still let you request quotes. If you're an engineer or student Maxim is also very good about sampling parts. I've had a lot a parts sampled to me for free in the past. These parts might be more difficult to get sampled though.

[Amphiprion]

I found this on Maxim's site:

This product was manufactured for Maxim by an outside wafer foundry using a process that is no longer available. It is not recommended for new designs. For further information, contact us. The data sheet remains available for existing users.

Findchips doesn't turn up any sources for it, and I've read that it's no longer in production on several electronics forums.

[Johnloudb]

I see it now ... I missed that. :Z

[JoshK]

Do you have an example schematic for how you are using it? Or is it in the tech notes?

If it doesn't have heaters I am not that familiar with working with it.

[Johnloudb]

Do you have an example schematic for how you are using it? Or is it in the tech notes?

If it doesn't have heaters I am not that familiar with working with it.

On page 6 of the data sheet there is a simple functional test circuit. It shows pin outs of the XR2206 integrated circuit with component values.

Also, Figure 12 looks like a good circuit to try: "Circuit for Sine Wave Generation with Minimum Harmonic Distortion. (R3 Determines Output Swing - See Figure 3)"

EXAR XR2206 data sheets

I've never built anything that has glass or heaters. Not that I wouldn't like to ... I should probably try hybrid tube/solidstate amp sometime.

[Amphiprion]

I have the first chip on my board hooked up like Figure 11 on Page 8. I am running Vcc at 24V.

The sine wave output is biased to 1/2 Vcc at the output of the chip by nature of how the chip works (the sine wave fluctuates around 12VDC). The amplitude of the sine wave is about 3Vpk-pk for 75% modulation of the next stage (see figure 6 page 7). I set that amplitude using R3.

I take that output and AC couple it to the next circuit using an electrolytic capacitor and a pot whose wiper is set at 14V. The reason for this can also be seen in Figure 6 on Page 7. I could have also biased it at 10V and gotten the same result. Now the signal is the same sinusoid, but fluctuates around 14VDC.

The second chip is configured exactly as the first chip (Figure 11 page 8 ), EXCEPT that the above 14V + sine wave signal is connected to the amplitude modulating signal input (AMSI) pin instead of grounding the AMSI pin. I also used a smaller timing cap so the low frequency from the first chip would modulate a higher frequency from the second chip.

That's what generated the graph above. It's very much a cookbook circuit, you just have to read the data sheet carefully.

As a side note, if you were to run the circuit at 100% modulation and make the frequency ratio 1:10, you would be making the same test signal that Linkwitz uses for distortion testing.

[Davey]

Mark,

The XR2206 is ubiquitous. It's the chip you'll find in just about every cheap, audio signal generator on the market. Unfortunately, the lowest distortion levels are rather high. Even with tweaking nominal values of 0.5% THD+N are about the best that can be achieved.

If audio signals only are the requirement it's best to let your computer to the work..IMHO. Use the Expression Evaluator in Goldwave to generate the waveforms you want and then burn them to a CD.

As an example this formula will 100% AM modulate an existing waveform with a cosine envelope (just like Linkwitz, as you mentioned) and yield a very low distortion result.

wave(n) * (0.5 + 0.5 * sin(2*pi*f*t))

Some changing of the field values in the Goldwave setup box will allow 1:10, 1:5, or other modulations...or whatever you'd like. The sky's the limit.

Building circuits is a lot of fun, but it gets tedious and time-consuming...at least for me.

Hope that helps.

http://www.goldwave.com

Cheers,

Dave.

[Dennis H]

Speaking of cosine envelopes, I was looking at the ARTA manual and it can calculate what he calls a burst decay graph from the impulse response. He says it's mathematically equivalent to the linear distortion graphs SL does but it does it over a whole range of frequencies with either 1/3 or 1/6 octave resolution and it plots a waterfall-style graph based on periods rather than msec. It looks a lot more useful than just testing a few frequencies with shaped tones and hoping you happen to hit a problem area.

[Davey]

Dennis,

I think the likelihood of hitting a problem area is much better than you might expect because the total spectrum of the 10:1 test signal can highlight problems above (and outside) the nominal range of the fundamental test frequencies. It doesn't seem intutive that a 400Hz test tone with sidebands at 360 and 440 Hz would highlight some driver problem at 1760 Hz, but it actually can. However, you're correct that this method has the potential to miss something that a more "continuous" type of analysis might catch.

http://www.linkwitzlab.com/mid_dist.htm

I think part of SL's recommendation of this type of test signal is for folks who have no test equipment......as it can highlight problems with just audible (ears only) evaluation.

Good info on ARTA. I need to fiddle even more with that program.

Cheers,

Dave.

[Amphiprion]

Davey,

I agree, the more I have used it the less I have been impressed with the THD performance of the XR2206. The peaks of triangle waves are even a bit off, and the square wave rise/fall times are asymmetrical by a fair amount (400 and 40nS respectively as I measured it). 0.4us is pretty poor, I wish I could find a MAX038 somewhere.

At the same time, the project I am using it in is something that I wouldn't want connected to anything remotely expensive, or turned on without me being a few feet away. Everything is a socketed DIP in case it blows up. I would use the computer for a frequency generator, but laptops are more expensive than a 2206

[mazurek]

I've been getting progressively more agressive in hooking crazy stuff up to my sound card. I have an EMU 1616M, and have found it to be excellent for crazy testing (especially after they finally fixed the drivers this year).

The bandwidth up to 192kHz is usable for some electronics measurements.

For the crossover circuit I just did, I poked around the circuit with the + wire of some xlr adapter probe I made, and the measurements were surprisingly good (surprising because the reference was circuitously connected). It was good enough that in a different circuit I could diagnose where harmonic distortion was entering the signal chain, and I found I had poorly modified a trace.

I also used the sound card to check out the signal from my amplifiers. I always check DC and AC levels with a multimeter (and spare driver) before hooking up something real. But after that I fed the output of the amp (with the gain reduced) into my sound card to measure performance up to a volt or so. I was just checking out low level performance and linearity so that was good enough.

ARTA and TrueRTA have ok signal generation and time record/oscilloscope, the sound card has a pretty good noise floor.