I'm designing scheme with using of LOG114.
Task and scheme is rather typical: compression and amplification of photodiode signal.
I'm starting with scheme from LOG114 datasheet (for dual-supply configuration), and calculated my parameters for input from 1nA to 1mA (6 decades), reference current 1uA (2.5MOhm resistor and 2.5V internal Vref) and 0 - 2.5V output. It gives me some reasonable R1, R2, Rg and R values:
R1 = 290k, R2 = 90k, R = 100k, Rg = 110k.
I'm entered scheme to Tina-TI and got very strange results: Vlogout is not where expected -- it starts from ~965mV, not from 1.25V, it crosses zero at ~400nA, not at uA, and scaled and shifted Vout is completely wrong too. Also, it could be seen, that Vlogout and Vout is not strictly logarithmic. When I add "Vref" to outout, I was shocked: Vref was not constant 2.5v, but was around 1,000145v for most of simulation space, and then "rockets" to 1.00017 (which gives non-logarithmic output on this interval).
When I replaced internal Vref with "ideal" voltage and current sources, simulations shows perfect results, as calculated, with perfect logarithmic transfer!
Is it defect in LOG114 model or my scheme completely wrong and real device will work as predicted by simulation?
