[sdiy] jfet driver

Uwe Schueler uwe.schueler at uni-tuebingen.de
Wed Feb 27 10:44:53 CET 2019


Declare Update <declareupdate at gmail.com>: schreibt...

> So! this has me wondering, why jfefs instead of mosfets? habit?

most discrete MOSFETs are designed as switches, so low Ron and high  
Vds_max are design goals.
This comes along with huge Cs , 100...1000x the JFETs Cs and leakage  
currents (Ids) up to
microamps, compared to JFETs pA..nA. This makes them almost useless  
for most analog synth
circuits. The situation is different for modern analog CMOS processes.  
They have very small parasitic Cs
and still low Ron, and give us fast precision opamps and analog  
switches. But: small Cs mean tiny silicon structures == low operating  
voltages. All good and cheap stuff ends @ 5V and it's all SMD :-(.
The '3157 (74lvc1g3157,NC7SB3157-D,SGM3157,...) analog switch is one  
of these goodies: for <10 cent you get an ultrafast, low leakage  
switch. I don't want to praise modern CMOS Opamps here - too off  
topic,but this new
analog CMOS stuff made me add +5 or +-2.5V aux. power on my synth PCBS  
and use this stuff in S&H, ring mod,
VCOs etc. Back to JFETs - many obsolete THT JFETs are still in  
production as SMDs: 2N5485 -> MMBF5485.

My main problem with JFETs is the huge variation in process parameters  
as mentioned already
>> ... For 4391, V_GS (cutoff) is minimum -4V, maximum -10V....
means you need a lot of safety margin if you switch signals with a  
JFET. On cure is to keep switched voltages small, e.g. run the  
integrator in a saw tooth VCO between 0 and -2V instead of 0 to -5V, @  
12V supply ,
doubles at least the # of JFET types, that can be used as integrator  
reset switch. But then you could consider
above mentioned CMOS stuff :-)






More information about the Synth-diy mailing list