[time-nuts] Re: +1/f of transistors
Lux, Jim
jim at luxfamily.com
Fri Apr 15 14:03:33 UTC 2022
On 4/14/22 3:00 PM, ed breya wrote:
> It depends on context and your definition of low noise. I was
> picturing ones commensurate with the 1/f discussion, trying to get
> down toward DC. Many are spec'd for so many nV or uV p-p over 0.1 to
> 10 Hz. The OP-07 is a good example of this class.
>
> A lot of opamps are spec'd ok for their intended use, but you have to
> look fully at the charts and numbers for your actual use. For
> instance, the ADA4817 looks great as a HF/RF amp, with 4 nV/rt-Hz, but
> that's at 100 kHz. Similar for the OPA656. At 10 Hz, it looks like
> it's well over 100 nV/rt-Hz.
>
> The JFET types are fine for high-Z amplifiers and such, but can't
> reach the near-DC region in the few nV/rt-Hz (at 10 Hz, say).
>
> The CFAs may be OK too - they're built for low-Z, so may tend to be
> operating the + inputs at fairly high bias currents, like the low
> noise opamps.
>
> The good part overall, is that to get near the nV/rt-Hz range, the
> source resistance has to be very low anyway, so at least the bias
> currents are manageable, although zero would be nicer.
of course, if your low noise frequencies of interest start at 100kHz,
then whether the 1/f crossover is at 100Hz or 10kHz is less important :)
My recreational complaining gripe is that there's a lot of interesting
low noise parts intended for the microwave market, but they don't even
bother giving data below 50 MHz. So it becomes a matter of tribal lore
- and you have the risk of "we got a really good batch" or "we don't
make those anymore".
Vishay Siliconix *used* to make JFETs (e.g. JANTXV2N4858) with good
noise properties (3 nV/sqrt(Hz) at 1 kHz) - but they're not in the JFET
business any more. They even put the noise on the datasheet, although
with a "not guaranteed JEDEC spec", footnote. And they still sell them,
maybe (It says "contact factory for JFET products" - factory stock, I
suppose). Or you can scrounge around and find 4 or 8 from some dealer.Â
And you can buy plenty of 2n4858s from other people, but *they* don't
say anything about the noise. And how do you find parts like this? Word
of mouth from someone who spent a bunch of time looking at data sheets,
bought a bunch of different kinds, and tested them.
I suppose that's why this is a recreational complaint - I (or anyone)
who needed low noise FETs basically has to suck it up and go through
datasheets, but parts, and test them. There's no "easy fix"
-------------- next part --------------
A non-text attachment was scrubbed...
Name: JANTXV2N4858Vishay70244.pdf
Type: application/pdf
Size: 42523 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20220415/18a27f84/attachment.pdf>
More information about the Time-nuts_lists.febo.com
mailing list