What to change in LTSpice BJT to account for Vce change ?

Thread Starter

DarthVolta

Joined Jan 27, 2015
521
I made a current mirror with a common collector stage in real life and in LTSPice, for a discrete diff-amp lab, and it works well and more or less matches what I calulate if I choose Ibase1 or measure something. And that matches LTspice too, for GND to+8 or 10V rails, w/ around 320uA.

But when I sweep Vce3 to near 16V-GND on the breadboard , or in a diff-pair circuit w/ +8/-8V rails, I get 785uA sinking into the Imirror in real life. (I never built output section). I need to chart it more tommorw to see where the Vce of the current mirror really puts the current up so high.

The problem is, LTSpice still only predicts 365uA for a Vce of 16V. So something about my 2N3904 is what even LTspice is missing. If I put in a current soucre of 785uA (in the attached diff-amp), then my calulations match LTspice and the actual transistors pretty spot on w/ n=1.03312480007.....how did I memorize that so easy, but not my passwords.

I used a Wheatstone bridge diff-amp circuit (not below) to try matching the 3 BJTs to within a few mV. I can't wait for my SMD3065X to get here and make some curve tracer projects.
 

Attachments

Last edited:

Thread Starter

DarthVolta

Joined Jan 27, 2015
521
I used 300 and so does LTpice, never tried to measure/calulate it. For the RC1 resistor current, I said it's current equals [Beta +3/(beta+1)]*Ib1 (for Q1/4/7), and Vce1 of the super transistor , must be
Vce1=nVt*ln[3*(Beta*Ib)^2/(Beta+1)Is^2]
and that works for some n=1.03312480007 I got from comparing LTspice to calulations on my calculator of whatever chips it has(from last weeks calulations on other stuff, comparing the equations to LTSspice predictions, to get an n....and that matches very close to the 3 BJTs and LTSpice, until it must ramp right up. w/ Vce3=10.7V , I had still pretty close agreement between breadboard and LTspice.
 

Attachments

Last edited:

crutschow

Joined Mar 14, 2008
34,280
You might try a Wilson current-mirror which uses negative feedback to maintain the output current nearly constant with a change in output voltage.
That greatly reduces the effect of the finite collector resistance (as defined by the Early voltage) on the variation of current with voltage.

LTspice simulation below:
The simulation shows only about a half microamp change in Q2's collector current for a collector voltage change of 1V to 16V.

1601063560186.png
 
Last edited:
Top