Some questions about a Function Generator Design

[dr pepper]

Sgm yes your right if the signal goes below ground npn's wont work, I assumed the circuit was single supply.
My bad.

 

[MrAl]

The HP analyser at work measured less than 0.005% distortion and noise. People making a switched capacitor filter of a squarewave also got pretty low distortion and noise. I used two National Semi 4th order Butterworth filters in series.

Hi,

Sounds nice, 8th order filter.
It might be hard to detect frequencies that have more distortion though because you have to tune it very slowly and carefully to each and every frequency in the range 0 to max Hertz. That can be hard to do, but if you do encounter a 'bad' one you could always tune a little off and then it should clear up.

It might be interesting to hear about your complete test procedure.

Back in the 1980's i created a sine generator with the old 8038 chip, and also one using 16 steps up and 16 steps down using TTL counter and an exclusive OR gate for inverting the digital pattern. Worked decently i guess but upper frequency was limited by the oscillator divided by 32, so a 32MHz osc only allowed up to 1MHz output. At the time that's all i needed though and it was sort of interesting just that it could be done that way
BTW the original scope i had at the time to view this with was an old Heathkit that only had AC coupling, so i had to use analog switches to make it dual trace and see some DC values (chuckle). Bandwidth was probably 1MHz (chuckle).

 

[audioguru]

I used a variable frequency clock for the digital switched capacitor filter and for the digital stepped "sinewave". The digital divider made certain that they were always at the correct frequencies.