Short: Op amp is being run at FAR below rated voltage during startup. It acts like a piece of wire. TL499 regulates at ~~~~ 1.25V. Easiest "fix is either a properly specified op-amp (see below) or a cct which clamps FB pin until TL499 output is around say 4.5 V. [eg resistor OA out to TL499. NPN clampimt FB pin. PNP turns NPN off onceT L499 output high enough. 1 x NPN, 1 x PNP, 4 resistors, maybe a cap. Medium: Op amp is specified for Vdd =3D 4.5V min. At Vdd_OA around 1V, Vout_OA is ~=3D Vdd_OA which is about 10x too high. TL499 sees approx its own O/P on its FB pin. SO TL499 IS running but you have a ~=3D 1.25V supply. If/when you manage a Vout_TL499 glitch that is high enough above 1V for the OA to start being an OA and not a piece of wire (or at least have Vout_OA = =3D 0) THEN the TL499 increases in voltage. At Vout_499 >=3D 4.5V the OA works correctly and the system operates. Longer: The design is failing due to extreme naughtiness on the part of the original 'designer'. The LMC6041 is specified for Vddmin =3D 4.5V in single supply operation, bu= t at startup it is required to (ideally) operate correctly, or at least, to maintain Vo low, for all supply voltages above about 1V. It fails. TL499 FB divisor drops from about 11:1 to about 1:1. Cct then bahaves as expected. TL499 Vref is 1.25V nominal LMC6041 op amp datasheet TL499a SMPS controller datasheet Opamps that operate correctly at Vdd < 1.25V (say 1V or less) will allow the circuit to operate correctly. (probably) suitable opamps NCS2001 OA - in stock digikey non-obsolete <- cheapest lowish-risk available. Your oscillographs [tm] 35 and 36 seem to give a clear picture of what is happening. I assume that X scale is 5000 uS/dive (as per top left of screen). The TL499a starts when Vin_rising reaches just over 1 Volt. Vout starts rising immediately. At 2 divisins / 1 mS VoutTL499 reaches ~=3D 1 V and ~~~ so does OA - and = at ~=3D 1.25V 499 starts to regulate Russell __________________ http://www.onsemi.com/pub_link/Collateral/NCS2001-D.PDF http://www.ti.com/lit/ds/symlink/lmc6041.pdf http://www.ti.com/lit/ds/symlink/tl499a.pdf On 4 September 2016 at 01:10, James Cameron wrote: > I've two kit PCBs here, not my design, that use a TL499A to get 5.5V > from a 1.25V NIMH cell. > > The TL499A often fails to get going. It takes convincing; with me > disconnecting and reconnecting the input. > > Antonio Sergio Sena on PIClist twelve years ago had the same problem, > but the thread seemed unresolved and his address bounces. > > Instead of the feedback configuration used in the datasheet, the PCBs > in front of me run an output sense voltage divider into an LMC6041 > configured as a voltage follower. Output of LMC6041 is directly tied > to REF input of TL499A. There's no capacitor between REF and OUT. > The LMC6041 is powered from the 5.5V OUT of the TL499A. > > In my oscilloscope traces; > > - channel 1 (yellow) is the 1.2V input, which has a 470 uF > electrolytic, > > - channel 2 (cyan) is the 5.5V output, which has a 220 uF > electrolytic, a 0.1 uF polyester, and powers the LMC6041 and voltage > divider, > > - channel 3 (purple) is the REF pin of TL499A, > > - channel 4 (blue) is the SW IN pin of TL499A; one side of the > inductor, > > In this capture, the TL499A starts properly. The rising edge of the > input is when the battery was connected; see how the capacitors had not > discharged from previous test. > > http://dev.laptop.org/~quozl/DS1Z_QuickPrint33.png > > In this capture and the next, the TL499A does not start properly. All > capacitors had been carefully discharged; you can see the voltages > start from near zero. > > http://dev.laptop.org/~quozl/DS1Z_QuickPrint35.png > http://dev.laptop.org/~quozl/DS1Z_QuickPrint36.png > > While the traces are very different timescales, when it hasn't started > nothing more happens; output sits around 1V. > > Pin 4 SW CUR CTRL has 330 ohm to ground. > > I'm puzzled; REF seems a bit unpredictable. > > It's late, I may have missed something. > > Can improve traces in a day or so. > > -- > James Cameron > http://quozl.netrek.org/ > -- > http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .