[time-nuts] Lars GPSDO on EEVblog

[Charles Steinmetz]

Jim wrote:

> I have modified the circuit to use a simple 2 ma current source as
> shown in the attached schematic, with significantly improved linearity.
>
> The diode-connected 2N3904
> has less leakage at a small forward voltage than any small-signal diodes
> I tested

Generally, one uses the B-C junction rather than the B-E junction for 
this, because (i) being a smaller junction (in most small-signal BJTs), 
it has less leakage than the B-E junction, and (ii) it has a much higher 
reverse breakdown voltage than the ~5v breakdown of the B-E junction 
(NB: the B-E junction is degraded with even very brief operation in the 
breakdown region).

According to my tests, the B-C junction of a high-quality 2N3904 has 
about 50pA of leakage at 20vDC reverse voltage.  Mid-high-frequency 
small signal NPNs (eg. MPSH10, 2N5179) have about 5pA of leakage under 
the same conditions, and the very small-geometry 2N4117/PN4117 JFET gate 
diode has about 1pA of leakage.  The HF BJTs and the FET also switch 
faster than the 2N3904, with less transfer of glitch energy.

Some time ago, I posted the results of my leakage tests of small-signal 
diodes, BJT B-C junctions, and JFET gate junctions.  Those results are 
attached.

This has all been known for a very long time.  See, for example, 
Burr-Brown Application Bulletin AB-064 (a/k/a TI SBOA058), 
Diode-Connected FET Protects Op-Amps (1994) regarding the 2N4117A gate 
junction <http://www.ti.com/lit/an/sboa058/sboa058.pdf>.

Best regards,

Charles


-------------- next part --------------
A non-text attachment was scrubbed...
Name: Reverse_leakage_of_diode-connected_BJTs_and_FETs_rev.1b.pdf
Type: application/pdf
Size: 136529 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20180908/22588f1c/attachment.pdf>