Tempcos

[gstopp at fibermux.com]

     I have a couple of these "tempermental tempco's" here, and I haven't 
     had the chance to figure out what the darn deal is with them. They 
     look good on an ohmmeter (2K exactly) no matter how much fingering 
     they get. Sure enough, put them in a VCO and it goes to pot.
     
     Does anybody have the manufacturer's data sheet for these weirdos?
     
     Based on recent experience I know that if you touch 2N3904s or 2N3906s 
     with your finger, in an exponential generator, the VCO/VCF will drift 
     quite readily. If you touch a MAT-03 (matched pair in a can) the drift 
     will be negligible. Although I haven't tried it lately, I'm pretty 
     sure that if you glob a bunch of heat sink compound on the 3904/3906 
     pairs and wrap them tightly together with a strip of copper (like ARP 
     used to do) they will stay closer to the same temperature and drift 
     less when touched. Usually I just keep my finger out of the guts when 
     playing.
     
     I've got lots of tempco's installed in a lot of VCO's, and they work 
     fine. Most of them are in ARPs and Moogs so I didn't install them 
     myself in the first place. The rest of them are Q81's in circuits that 
     I built back in the 70's which work fine. Unfortunately I can't easily 
     remove these for testing or comparisons or else I would.
     
     In my latest stuff I use metal film 5% 2K resistors. These seem to 
     work fine for normal operation. Since I've been experimenting in 
     fairly constant and pleasant environments I can't claim that they 
     won't drift in other circumstances. Keep in mind that the matched 
     pair/tempco combinations are designed to exhibit the same behavior 
     under all conditions, not just a constant 75' F room temperature.
     
     So - apparently my old Q81s are "good" tempcos, and the new crop that 
     are being supplied these days are "bad"? Maybe we forgot to specify 
     something when we ordered them? How complicated can it be? I'm 
     stumped. Okay here's an idea - I'll dig through all my Electronotes 
     and try to find some disseration on tempco's. If anybody can find any 
     engineering data on tempco's, please post!
     
     - Gene
     gstopp at fibermux.com


______________________________ Reply Separator _________________________________
Subject: Tempcos
Author:  Roland U K Ltd <101574.2042 at CompuServe.COM> at ccrelayout
Date:    5/7/96 5:26 AM


FROM:  101574,2042 at compuserve.com
TO: synth-diy
DATE: 5/7/96 9:29 AM
     
Re: Tempcos
     
Hi list,
     
Has anybody actually tried using tempco's in the Electronotes VCO circuits?   I 
put a couple of 1k's ( all I could get over here ) in place of the 2k metal film
in the EN76 tri-square and it made things worse.  The osc definitely became less
stable and started to drift .  Now, the tempco was the correct type with 3300ppm
rating so what could be going on?   It's a feedback system that seemed to add 
instability, and taking them out resulted in the osc. going back to its normal 
behaviour.   I'm using an Elantec quad PNP dil package  ( only L2.60 ) so I 
wonder if it is just not so sensitive to temperature changes.   You can 
certainly touch it for quite a while without hearing any drift whereas , with a 
discrete transistor, touching it instantly affects it.  ( As I learnt while 
matching up transistors for the Moog type VCF ).
So maybe the tempco is reacting much quicker to slight fluctuations in 
temperature whereas the Elantec PNP's do not due to their construction.  This 
would cause errors would it not?
Any ideas?  It's not really a problem as the OSc is very stable as it is ( ie no
tempco ).  But it worries me that the tempco made things worse.
     
Thanks
     
Joe Farler