[sdiy] AN-299 VCO troubles [was: SDIY list troubles]

[Aaron Lanterman]

On Apr 28, 2008, at 11:42 PM, Paul Schreiber wrote:

> a) never believe National Semi app notes written before 2000. Most  
> of them
> were authored by co-op students, especially the 1980s ones. Linear  
> Tech app notes are wonderful as are Maxim Semis. But a lot of  
> National's are a bit....odd.

Good bit of advice.

I admit that I've somewhat been using students as guinea pigs to try  
out things I don't have time to try out. I pay attention to how much  
time they're putting in (I'm in lab almost all the time when they're  
working on their projects) and take that into account, along with the  
"risk" of a given project, when grading. I expect people working on  
"proven" designs to get further than people who are working on less- 
proven thing (say random data sheet examples) or more experimental,  
original designs.

> b) Is the substrate of the array tied through a 10K to -15V?  
> National likes to leave this little tidbit off most app notes. It's  
> the 'Q5' of the *array*, not in the schematic (notice how 'Q5' is  
> skipped in the LM3046, but there is no reference to connecting the  
> substrate to anything.). If this pin is let floating (which I'm  
> betting it is) then you can get all sorts of interesting things,  
> especially if you use old-fashioned ProtoBoards (the white things  
> you can stick stuff in and not use solder).

Ohhhhh... you know... I'm pretty sure that's currently floating. I  
will pass all this on to Adam.

That substrate pin has caused my students any number of problems in  
the past. I try to remember to tell them about it but sometimes I  
forget.

> c) If you see *any* circuit using a LF412 *RUN LIKE HELL*. Easily  
> the worse op amp ever made (second to the National LM833 total POS  
> audio part...ack!). I worked for a major central office telecom  
> supplier, and we almost went bankrupt over LF412s in circuits doing  
> all sorts of interesting things like oscillating at 10Mhz as the  
> *humidity* changed, Q of filters moving depending on how much  
> *light* the IC package was exposed to. I designed fancy multi-band  
> filter (a delay equalizer for 2400 baud modems over leased-lines)  
> using NE5532s and the factory "upgraded" to LF412s and I spent many  
> hours sitting in telecom closets in elevator shafts getting it to  
> work, we finally had VP of Engineering literally throw out every  
> LF412 in the system, the factory was solder-sucking parts off  
> finished goods for weeks.

Wow.

Adam's using TL08x/TL07x's (they're sort of my standard op amp to hand  
students unless there's particular reason to do otherwise).

> It would not surprise me if this particualr circuit used a 'quirk'  
> of the LF412 to "start up", or worse the designer 'just got lucky'.  
> This is like Buchla paralleling CA3160 op amp *outputs* to exploit a  
> 'quirk' in the output stage (plugging in any other op amp causes bad  
> things to happen).

Oh - if that's the case that's bad news... I can pick up some LF412s  
from Ack tomorrow (I doubt they have them in our stockroom) and see if  
that makes a difference.

Re: the Buchla - good grief! What circuit did he do that in?

> d) LF412s are *VERY VERY VERY* sensitive to power supply noise and  
> decoupling. Silly observation: is the op amp powered off true +-15V?  
> You need to be sure pins 4 and 8 ahave at lrast a 0.1mf ceramic  
> bypass, perferred with a 10uf electrolytic in parallel. Even if  
> using a TL072 here, is it bypassed?

Probably not... and that's my fault. I have been negligent in  
instructing my students in proper bypass hygene.

- Aaron