LM2596-ADJ CC CV Supercapacitor Charger

[UPI]

I am building a circuit that relies on Supercapacitors to run a stepper motor for a couple of hundred milliseconds when power fails. I have designed (mostly plagiarized) the attached LM2596 buck converter set at 14V out that is current limited to ~.5A. C8, C9 and C10 are the 5.4V supercapacitors that will be charged to ~13.5V.

The first half of the LM358DT is used to limit the current to ~500mA. The second half of the LM358 op amp is designed to light an LED when the voltage reaches ~13.5V as the supercapacitors are almost at their fully charged voltage of 13.5V (supercapacitors are after the ~.5V diode drop). 

We have prototyped the circuit using quality components from Mouser and it all appears to work well in our quick bench testing. As of this evening we have not done a full characterization but the output is clean (~5mV ripple), the current is limited, the voltage is relatively stable under load and the fully charged indicator is working.

Main parts used:
Bourns SRR1208-221KL 220uh 1.1A Inductor
Texas Instruments LM2596SX-ADJ/NOPB Adjustable Buck Converter
Vishay VSSB410S-E3/52T 4A Schottky Diode
STMicroelectronics LM358DT Op Amp
Nichicon UHV1V221MPD 220uf 35V Capacitors
TT Electronics LRMAM0805-R02FT5 .020 ohm current sensing resistor
Taiwan Semiconductor BZT55C15 15V .5W Zener
PowerStor / Eaton PHV-5R4H505-R 5F 5.4V Supercapacitors

Does anyone see any glaring issues or have any suggestions for improvement?

[UPI]

See attached Eagle 6.6 file.

[UPI]

We found some major noise in the output voltage during the constant current mode (above 500mA).

We have since changed the design as shown in the attachments. It is stable whether in CC (above ~500mA) or CV (14V) mode.


EDIT: Please see revised schematics attached here and further down the message chain (Thanks zlymex!)

[zlymex]

We found some major noise in the output voltage during the constant current mode (above 500mA).

We have since changed the design as shown in the attachments. It is stable whether in CC (above ~500mA) or CV (14V) mode.

There is no DC path for Ib- of U1B.

[UPI]

Great catch! It was lost in the translation from the bench to the schematic.

Please see the revised schematic.

[UPI]

Please see the attached schematic for the entire cap charging and 5V out section.

Q100 enables the 5V out stage once the caps are charged to ~10.5V (on machine boot) and Q11 allows me to latch this 5V section on even when the cap voltage drops below 10.5V (to keep the 5V until I have finished moving to a safe motor position).

 Any suggestions are much appreciated.

[UPI]

We had to add an extra 220uf cap to the 5V output to stop a large 5kHz ringing under load. A ~470uf through hole cap would be appropriate according to the datasheet. We are limited in capacitor height so we chose to parallel two 220uf caps.

Please see attached Eagle and PDF schematic.

[Mosaic]

Interesting but too complex for my app.
I used a couple BJTs which seem to work well....in the prototype.