Types of diode-connected BJTs
in [Design]
|
From: John Larkin on 9 Jul 2010 22:35 On Fri, 9 Jul 2010 18:32:35 -0700 (PDT), George Herold <gherold(a)teachspin.com> wrote: > > >John Larkin wrote: >> On Fri, 9 Jul 2010 10:00:08 -0700 (PDT), George Herold >> <gherold(a)teachspin.com> wrote: >> >> >On Jul 9, 11:20?am, John Larkin >> ><jjlar...(a)highNOTlandTHIStechnologyPART.com> wrote: >> >> On Fri, 9 Jul 2010 06:48:53 -0700 (PDT), George Herold >> >> >> >> >> >> >> >> >> >> >> >> <ggher...(a)gmail.com> wrote: >> >> >On Jul 8, 8:35?pm, John Larkin >> >> ><jjlar...(a)highNOTlandTHIStechnologyPART.com> wrote: >> >> >> On Thu, 8 Jul 2010 00:53:14 -0700 (PDT), C Egernet >> >> >> >> >> <eger...(a)hushmail.com> wrote: >> >> >> >What are the relative merits of diode-connecting a BJT by shorting >> >> >> >base and collector (using B-E junction) versus shorting base and >> >> >> >emitter (C-B junction)? >> >> >> >> >> >Na?vely, I have assumed that one always uses the B-E junction but then >> >> >> >I saw this post from s.e.d by Phil Hobbs: >> >> >> >> >> >> BFT25A C-B junctions are at least as good as 2N4117As as diodes. >> >> >> >> >> >> Cheers >> >> >> >> >> >> Phil Hobbs >> >> >> >> >> >What gives? >> >> >> >> >> >Chris >> >> >> >> >> Most diodes-sold-as-diodes, like BAV99 and 1N4148 and such, leak >> >> >> nanoamps, and the glass ones are photosensitive. >> >> >> >> >> I took data on using BFT25A C-B junctions as diodes. They are >> >> >> fantastic. I measured about 20 fA reverse leakage at a few volts, log >> >> >> linearity from 1 pA to 10s of mA, and about half a pF. I didn't try >> >> >> the B-E junction, because it will zener at a few volts so isn't as >> >> >> generally useful. The measurements are tedious. >> >> >> >> >That's great! ?It would be useful as a over-voltage protection 'diode' >> >> >on a low noise front end. ?Any idea how much current it can handle. >> >> >Seems like if used as over-voltage protection you might want to short >> >> >the E and B and get a bit of current gain. >> >> >> >> Reverse beta lowering the voltage drop? Probably so. C would go up. >> > >> >I was just thinking that if E and B were shorted then not all the >> >current would have to flow through the base. There must be some small >> >amount of current gain in this 'backwards' transistor. >> > >> >George H. >> >> >> >> I'd always assumed that "microwave" transistors would be leaky for >> >> some reason. As Phil pointed out, they make good low-leakage diodes >> >> because the junctions are so small. >> >> >> >> We created PADS schematic and PCB symbols for the BFT25 as a diode. >> >> Schematics get weird and ugly when you use a bunch of transistor >> >> symbols as diodes. >> >> >> >> I did the testing for a couple of projects. One is a photodiode amp >> >> where we want to prevent windups and inject some test currents, and >> >> the other is an FTMS preamp where we have a kilovolt of transmit RF >> >> millimeters away from a nanovolt receive antenna, and we need to >> >> recover quickly but add minimal leakage and capacitance. The resulting >> >> circuit is cute but un/fortunately too good to publish in the open. I >> >> *did* Spice it because I *didn't* entirely understand how it would >> >> work; too damned nonlinear, too diode dependent, no hard definition of >> >> "best." >> >> >> >> John- Hide quoted text - >> >> >> >> - Show quoted text - >> >> My Fluke seems to output 0.6 mA on the "diode" range. A BFT25A reads >> 0.843 for the C-B diode, down to 0.771 if I short the base to emitter, >> so there is some advantage from reverse beta. Capacitance is 0.55 and >> 0.83 pF respectively. Those are the easy measurements; maybe I'll get >> around to doing leakage, too. >> >> The high voltage drops at low current suggest a very small chip, no >> surprise. >> >> Hmmm, delta-V is about 70 mV. Does that imply a reverse beta around >> 10? >> >> >> John > >That's Great! I'll try and remember to measure a 2N3904/6 on Monday. >How did you measure the capacitance? We've got an SRS RCL >'meter' (box) but I've never tried it on an active device. I used my AADE capmeter, with the surface-mount adapter. I'm not sure what the drive volage is, so maybe I should check it with one of my old pale green Boonton analog c-meters; I know it runs about 0.1 volts. John
From: whit3rd on 10 Jul 2010 21:00
On Jul 9, 6:32 pm, George Herold <gher...(a)teachspin.com> wrote: > [John Larkin wrote:] > > My Fluke seems to output 0.6 mA on the "diode" range. A BFT25A reads > > 0.843 for the C-B diode, down to 0.771 if I short the base to emitter, > > so there is some advantage from reverse beta. Capacitance is 0.55 and > > 0.83 pF respectively. > That's Great! I'll try and remember to measure a 2N3904/6 on Monday. That '2N3904' only means a JEDEC specified part, it could vary widely from one manufacturer to another (and from one year to another) because any transistor that meets the loose specifications can be so labelled. If you want a meaningful result for future guidance, it's better to pick a manufacturer-specific part, and maybe even one with some kind of premium specifications, like low noise or high bandwidth. Hopefully, if NXP or Zetex redesigns, the 'newer' parts don't have the same ID number as the ones they replace. Unless the part is a JEDEC standard and they want to bid on the big jobs that use those standard parts. BFT25A is Philips/NXP part with 5 GHz bandwidth, only available in surface mount SOT-23 (low stray capacitance due to leads). I'd expect its character to be more consistent than that of 2N3904. |