SV: Re: [sdiy] OTA VCO bad or not!

[karl dalen]

Hi Rene!

--- René Schmitz <uzs159 at uni-bonn.de> skrev:

> Hi Karl,
> 
> karl dalen wrote:
> > Was OTA's so bad for doing regular triangle VCOs?
> > Is the Electronotes OTA VCO so bad?
> > I cant remember but they cant be that bad or can they?
> > I now the issues about OTA's but so bad cant they be?!
> 
> The main issue is the temperature dependancy of the transconductance.

Yes but was/is that so bad really?
To be honest i havent build a OTA osc for years and my memmory is decomposing!

> You need to set the levels for the OTA to "clip" so that the output 
> maxes out at +-Iabc, with a much reduced thermal coefficient.

The lower the worser the thermal drift i suppose.

> Then also the problem is the OTA is very unsymmetrical internally, one 
> side has one current mirror while the other has three.

Hmm, when doing the electronotes VCO or a  one with fast and accurate
comparator
i se almost perfect symetry but the amplitude start to expand at higher
frequencies,
but it appears to be a slewrate problem of the comparator, still its pretty
much 
very good at 0 to 20Khz! (did this with a Ne5517 and a LM13700).

My problem is that i need a aboslute symetry, and no dent/glitch/slant
overshot/undershot on the tri wave from 35hz to 14Khz. 

> Since the current 
> gain multiplies, you'll have always some difference between the source 
> and sinking currents (unless you're lucky and all gains are exactly 1).

> And you need a PNP expo convertor, which is usually inferior to an NPN 
> based one, because of the poorer carrier mobility in P-Si.

But how much poorer in *cents' is this inreallity?
 
> This all doesn't say its impossible to build an OTA-VCO, but I wouldn't 
> really recommend using them if you have high performance demands.

So what is high performance demands actually?

> Besides, why base something on parts that are (or will be) unobtainable 
> in the near future?

LM13700 obsolete?

KD