uA726 replacement

[Paul R. Higgins]

Maybe I'm wrong, but I am almost POSITIVE that the 3340 uses some kind of 
heater; I remember talking to a tech at PAIA years ago because I was concerned 
about how warm the thing got in my Hyperflange unit .  That's what the tech 
said--there's a substrate heater in it.  And wasn't there an article some time 
ago in Elektor (???) about running the negative rail of the 3340 at around -11V 
instead of -15 to lower the thermal dissipation?  The thing definitely gets 
warm.

I wish that I still had a data sheet for the 3340 so that I could confirm 
whether I'm wrong.  But then again, 3340s aren't really a feasible part for new 
designs these days, so I'm not really knocking myself out to get the 
documentation.  (I've got only one left in my parts box, aside from the one in 
the Hyperflange).

Anyway, what's so bad about cancelling the thermal drift with a heater?  As long
as all expo converter transistors are on the same substrate, you are cancelling 
out the temperature term for them.  As regards the integrator capacitor, perhaps
this is a problem as well, but I've never heard of this problem until I 
subscribed to this list...note that I'm NOT saying here that your claims are BS.
(No flames, please!).  Obviously this is a part that can't be fixed with a 
heater.    But isn't this why a high-grade cap is specified for this part (e.g. 
in Electronotes designs, etc.)?  How bad is the cap drift problem?

Maybe Tom G can pitch in on the stability of his heated 3046 VCO designs.  
They're certainly simple but elegant designs, essentially improvements on Moog 
circuits.

Regards,
PRH


> >In fact, the much venerated CEM 3340 VCO chip used this heated substrate
> trick
> >with great success.  Warm-up time was < 30 sec, much better than (dare I
> say it)
> >tubes
> 
> This is an incorrect statement.
> 
> The 3340 does NOT heat the chip up. Rather, it *senses* the die temperature
> and feeds a second-order correction current into the integrator.
> 
> The exponential converter is 70% of the temperature dependence. You also
> have other
> factors, such as the temperature coefficient of the integration capacitor.
> 
> The MOTM-300 Ultra VCO has *5* seperate temperature servo loops, does NOT
> use a heater,
> and can track within 0.05HZ over 6 octaves and within 0.2Hz over 10.
> 
> Paul Schreiber
> Synthesis Technology
> www.synthtech.com/motm



 
>> The only drawback to the heated chip approach is that additional current
>> is required for the heater transistor.  I don't consider this a big price to
>> pay for the stability it affords.

> Paul, please check the archives about that.


> In fact, the much venerated CEM 3340 VCO chip used this heated substrate
> trick

Definetly not.

JH.



Paul Higgins
email: higg0008 at tc.umn.edu
University College, University of Minnesota