[sdiy] Voltage references in VCO

[René Schmitz]

René Schmitz wrote:
> Hi Benjamín,
> 
> Benjamín Velasco wrote:
> 
>>> Thanks Rob
>>>
>>> So what precision regulators exist for voltages near +-12-15V??
> 
> 
> If you're thinking of a three terminal regulator then the answer is: 
> (Next to) None. In that case you would start with a so called voltage 
> reference, e.g. a LM336 which gets you a stable 2.5 or 5v and you have 
> to add a little amplifier to bring that up to the desired voltage, then 
> you can use that as your power source.
> 
> But what gets confused here is _references_ with _regulators_.
> The actual regulator is only the amplifier which regulates by amplifying 
> the difference of the output voltage and one derived from a reference. 
> What is commonly refered to as regulator contains a reference together 
> with an amplifier, called the error amplifier. (Have a look at the LM723 
> datasheet, this chip is a little kit containing several parts which the 
> user has to configure.)
> 
> It makes sense to derive the critical voltages within a VCO from a 
> reference (voltages for the pots, reference current of the expo 
> convertor, trip points of the relaxation circuit), but if you would want 
> to patch that to your VCOs you're in for a serious hack. Best done at 
> design time.
> 
> Your next option would be to use a LM336 (or something even more fancy 
> than that), and use an opamp and a pass transistor. You're designing a 
> power supply then. This can get tricky to get it stable and fast (for 
> good transient behaviour).
> 
> An LM317 will have far better performance than what is indicated in the 
> datasheet, if you only draw little current. Its the heat that is 
> produced in the pass transistor which disturbs the reference which 
> happens to be build into the same chip (the typical problem of 
> integrated regulators). It will at least be a lot better than a wobbly 
> supply.
> 
> One thing you also might want to consider is adding a three terminal 
> regulator to the offending LFO, to avoid putting that noise on the power 
> supply in the first place. This improves the effectiveness of any 
> decoupling thats local to the LFO as well.

I should add that you also would need to keep GND clean then. If the 
current going to GND is below 20mA then a voltage follower with input to 
GND, you connect the offending modules GND to the output of the 
follower, this makes a so called dustbin. Or you run an extra wire 
straight back to the PSUs GND. You must connect the GND of jacks or the 
panel to the not dustbinned GND then. In other words don't allow the 
output of that follower to become shorted to GND.

Cheers,
  René

-- 
uzs159 at uni-bonn.de
http://www.uni-bonn.de/~uzs159