[time-nuts] 4046 variations (was EPE GPS....)

[Lux, Jim (337C)]

On 9/27/09 9:54 AM, "Javier Herrero" <jherrero at hvsistemas.es> wrote:

> Yes, of course, some applications needs the parts that are in the better
> side of the distribution curve for a particular parameter and for a
> particularly stringent application, as you mention :), but I think that
> a 4046 seems a part that should not have to be so critical that even in
> the same lot number some work, some don't, and it is hard to believe
> that no alternatives existed on that time that could have avoided the
> high costly selection process.
> 
> Regards,
> 
> Javier
> 
> Lux, Jim (337C) escribió:
>>
>> 
>> Not always marginal or poor design.  Sometimes, you can't get the
>> performance needed with (any) part that simply meets the datasheet specs, so
>> hand selection is needed.  Picking matched pairs is a time honored method
>> for instance.. In some cases, the mfr does the picking for you (RF power
>> transistors). In another case, there's an extremely low noise FET used in
>> some charge amplifiers that not only has to come from one mfr
>> (notwithstanding the 2N JEDEC number) but you want to buy a bunch and pick
>> the quietest one.  High value resistors and leakage currents also come to
>> mind.



That was an aircraft radio, and while I don't know a whole bunch about what
certification requirements they have, but for spaceflight, we often use
older (design-wise) parts which have a "flight heritage", even though the
performance is not what you could get with a more modern part (even if the
new part is available as Class S). (anyone know of a flight qualified opamp
to replace the venerable OP27.. Not that the OP27 isn't good, but a lower
power, lower noise, wider band device is always nice to find)


I could see the radio having been designed years and years ago, when the
4046 was all you could get, and even with the cost of hand selecting, it was
still better than the alternative (dozens of carefully selected crystals and
multideck rotary switches).  Move forward 10-15 years, and rather than go
through the considerable effort and cost to "certify" a new design, they
just stick with what they have.  Then, move forward another 10 years, and
someone needs that radio repaired (I had an early 70s Piper Cherokee in the
80s with the original radios, and I'll bet the next owner kept them for a
while, too.