Post Go back to editing

LT1054 Charge pump component selection

Category: Hardware
Product Number: LT1054

Hi all,

I’m designing an unregulated dual-rail supply to generate ±12 V from a single +9 V source (either battery or regulated DC PSU). The DC-DC stage is based on the LT1054. The topology follows LT1054 datasheet (dual-output voltage doubler/inverter).

Target load current is very low, about 5 mA per rail (10 mA worst-case including margin).

Questions:

1) Capacitor values

Given the light load, is it possible to reduce the 100 µF (input/output) and 10 µF (flying) capacitors from the reference design?

Modern options I am considering:

  • Option 1: 100 µF aluminum-polymer, 10 µF aluminum-polymer (very low ESR)

  • Option 2: 100 µF aluminum-polymer, 10 µF MLCC (noting MLCC bias-derating)

  • Option 3: 100 µF aluminum electrolytic + 10 µF MLCC in parallel (for ESR balance), 3x 10 µF capacitors MLCC 

Will the very low ESR (options 1 and 2) increase peak current stress on the rectifier diodes and LT1054 internal switches? Is this a real concern at ~5 mA load?

2) Diode selection

The reference design uses 1N4001 diodes. Since forward-drop wastes headroom in this application, would a Schottky diode be acceptable (or preferable)?

Given the low current, could the diode be scaled down (e.g., SMD 40–60 V, 0.5 A Schottky)? This would reduce losses and help post-regulation margin.

Any practical guidance or experience with reducing capacitor sizes and using modern low-ESR parts with the LT1054 would be greatly appreciated.

Thanks!

  • Hello,

    This is an extremally old part that was designed before there were many very high value MLCC caps. Using an aluminum polymer provides some ESR which helps reduce switching currents slightly. Low ESR is good to help reduce ripple. Using all 10uF caps should be OK for this. Please look at the ripple voltage to make sure it is acceptable. Trying 1uF where the other caps were 10uF might be worthwhile as well. If this is an S8 package they might all need to be 10uF. 

    Reducing to a Schottky diode might be OK. Please be aware of the leakage currents in the Schottky diode. Some Schottky diodes get very leaky especially at higher temps. 

  •   Thank you for your support.

    I agree that this part is quite outdated. The LTC3265 would indeed be an ideal replacement, especially since it includes integrated LDO regulators. However, for this particular application, it might be somewhat overengineered.

    From what I’ve read in various threads, maintaining a 1:10 ratio between CIN / COUT and CFLY is recommended to ensure proper charge pump startup. Using higher capacitance values can also help support reliable startup behavior.

    I believe that using low ESR capacitors — for example, CIN and COUT at 22 µF and CFLY at 2.2 µF — should work well. Even configurations like CIN/COUT at 10 µF and CFLY at 1 µF with low ESR should be sufficient for proper startup, in my opinion. However, I was hoping to get confirmation from someone with hands-on experience.

    Also, would increasing the switching frequency by connecting a small capacitor between OSC and CAP+ improve or worsen the startup behavior?

    Lastly, can anyone confirm whether the LT1054 architecture is derived from or similar to that of the ICL7660 or MAX1044?

  • Hello,

    Yes using the 10:1 for the caps is recommended. I do not know if 10uf or 22uF will work best for this particular application. Evaluation on the bench is the best way to determine this. 

    The frequency can be increased by adding a cap from C1 to the OSC pin as mentioned in the pin description.

    I actually do not know if increasing the frequency will improve or make startup behavior worst. 

    This part was before my time but I am quite sure it was not derived from the MAX or ICL 7660. I do not even believe it was derived from the LTC660 or LTC1044.