C-multipliers and simple circuit to measure npn beta.

The ZTX450 are nice, output noise ~1.5 nV/rtHz. For what ever reason the ZTX851's filtered less noise than the 2n4401's.

George H.

Maybe higher C-B capacitance. I often use an RC in the collector as well, like 1 ohm and 10 uF. With a low-V_CEsat transistor, that doesn't cost any dropout voltage.

Cheers

Phil Hobbs

Does ZTX851 have SMD version (SOT-23 prefered)?

Driving home I was thinking I should describe the circuit. From the voltage ref. I've got two series 500 ohm R's and two 100 uF Al. electros to gnd. Between the base and the last C/R junction I've got ten ohms and a ferrite bead, Lower r_bb' can only help down to the ten ohm level. (bead and R_b added to stop oscillations.)

George H.

I use that approach too, though I put in a single 0603 in the base, which starts out as a jumper and gets hacked if necessary. (I've only seen one or maybe two cap multipliers oscillate.) The electros probably have tens to hundreds of milliohms ESR, so a 1-uF MLCC can help.

A bead to prevent oscillation plus another big cap on the emitter will get rid of the residual noise pretty well.

Cheers

Phil Hobbs

Yes, it is not even ripple problem. We have very thin spikes from Traco Pow er's isolated dc-dc converter. And those spikes are really wideband and dif ficult to suppress. They leaks through Murata BLMs and capacitors somehow.. . I'm considering to replace Traco with isolated push-pull converter built around LT1683 (which is slew rate controlled). But it costs a lot of PCB sp ace and some time to understand adum3190 compensation...

I simulated cap mult with single FMMT620 transistor (without PNP) at 500mA load in LTSpice. Behaves very good. Especially in Phil's "5-Pole 2-Stage C apacitance Multiplier" setup. May be FMMT620's model is too optimistic. Hav e to test on bench.