A source of error?

[René Schmitz]

Hi Jörgen!

At 09:25 26.07.00 +0200, jorgen.bergfors at idg.se wrote:
>Hi all.
>I just built the sample and hold from René Schmitz page
(http://www.uni-bonn.de/~uzs159/sah.html).
>When doing the board layout, I realized that the circuit diagram symbol 
>for JFETs doesn't tell which is source and which is drain.
>Is there a convention for this? I tried it both ways, and it didn't seem 
>to matter. Does it make a difference? If not, why do they have different
names?

Many JFETs (like BF245) are noted for symmetrical operation in the
datasheets, so it really doesn't make a difference. The drain is just
whichever of the two is more positive.

>Another problem with this circuit was that the response was very non-linear. 
>But when I lowered the 1M resistor between gate and source (or is it drain?) 
>to 100k, it became perfectly linear. Comments? René?

It was a while ago that I built this, looking at it now, it makes perfect
sense to lower this resistor, since the gate gets charged only thru that
resistor when the trigger is near the positive rail. Lowering this makes
the FET open faster, so the cap can more precisely settle in the small
aperture time.

>Also it doesn't seem to hold voltages below -8 volts. The voltage just jumps 
>back to around -8 volts if the sampling rate isn't fairly high. Any ideas?

I have at the same effect at (roughly) -10V. Vgs(off) is delimiting the
lower end. The potential of the gate must be by Vgs(off) more negative than
the source (A1 output) to prevent conduction. (Vgs(off) is spec'ed for 10nA
drain current in the datasheet, which is too much here, it must be more
negative than that.) You could try a BF245A for a lower Vp, but conversely
they have more Rds(on). 
Also the gate resistor R1 forms a voltage divider with the 4.7k resistor in
the driving circuit, when the diode conducts. So it can't go all the way to
-15V.  (I wish they'd make CMOS with 30V swing anyway ;-)

> P.S. René, can you please write 4k7 instead of 4.7k? In the MS20 VCF
schematic it is impossible to tell wether resistors are 1.8k or 18k.

Yes, I'll try to keep that in mind. They're 1k8 BTW. 

Bye,
 René

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