expo-converter and other ramblings

[Rene Schmitz]

Hi Tony and all!

At 17:27 08.10.99 +0100, Tony Allgood wrote:
>This sounds very good indeed. But has anyone verified just how good the
>gm linearity and absolut gm is over temperature for the LM13700. My HF
>VCO, appears OK at room temperature using 2220 pnp pair and 1k TC as the
>expo, but I haven't done any proper temperature tests yet. 

gm of any diff pair drifts at the same 3300ppm. Same for the OTAs gm, but in 
the tri/square VCO it should not play a significant role, since the input 
is at either positive or negative saturation, and the output current of 
the OTA is equal in magnitude than the input current then (apart from 
inacurracies of the mirrors) regardless of gm.

(A note for Juergen: If you used the gm of the OTA for the compensation,
the expo trannies would have to be at the same die/temperature than the
expo pair! 
Selfheating due to powerconsumption inside the OTA would be an issue.)

>My other point, other than for the sheer fascination of the subject, why
>do we need a non-temp co approach? The temp co's that Paul S and Farnell
>are selling seem to do the trick. Paul's MOTM uses a TC, albeit with
>some other stuff as well, and this is 'ultra-stable'. I have to say that
>I have not seen the schem for this, but he does have two sets of pairs
>in there as well as a TC. Also, most people seem to be happy with the
>ASM-1 VCO.

Haveing done a solid state compensation, I'm now "converted" to the tempco. 
(Well sort of.) Thus I felt I should comment here. Tempcos are a very
effective 
way of compensation, easy, and very noise free and linear. I see my solid
state 
scheme now as an exercise. There are other mostly overlooked methods apart
from wirewound tempcos. A few of which I explored, and you can find them at
my site. 
(Look at the expo of VCO566 & VCO6 for NTC and VCO3 for KTY-81)

>This is not a dig at semiconductor temperature stabilised designs, no
>sir; more of a 'trying to find out whether my new VCOs really do need
>something else apart from a TC' sort of enquiry.

The thing is (in theory) that you make the temperature compensation on 
the same die than the convertor, thus when the temperature drifts the 
compensation is quicker and more precise. (theoretically all that is,
infact my 
TB3030 is quicker back in tune after turning it on than my solid-state
compensated 
VCO due to the opamps offsets.)
There are circuits in which it makes sense, e.g. if you have the space, and 
if you can't get the TCs. (Which was the reason why I did mine back then.)
Or if you 
happen to design a VCO chip (see CEM3340).

Now I use NTCs and (cheap, nonlinear, Si) KTY-81 PTCs, with success. The
wirewound 
resistors are a bit better theoretically, but considering that even the
3000ppm 
tempco in my TB3030 does its job great, there is probably only a minor
difference.
>From my practical experiences with either (TC/PTC/NTC) circuits I cannot tell 
which one is superior.(But they haven't been in a thermo chamber yet.)

Oh, and BTW, I'm partly finished with an introduction tutorial on expos and
temperature compensation.

Bye
 René





error:asciiartisnotfully  \ uzs159 at uni-bonn.de
supportedwithyourcurrent   \ http://www.uni-bonn.de/~uzs159
displaydriversettings...    \ http://members.xoom.com/Rene_Schmitz