4
\$\begingroup\$

I've been working on trying to get this SMPS to work for a while. I'm using an LTC1872, a step up DC controller with current limiting. The circuit used is the typical application provided in the datasheet:

With a couple of slight differences (since I'm not always able to get the correct parts):

  • Capacitor at pin 2 is 200pF
  • Output filter capacitors are a 100nF ceramic, 10uF electrolytic, and 2200uF electrolytic (scattergunning here a little)
  • Inductor is ~3uH (calculated from an LC circuit)

The good news is that the converter does seem to work. I input 3V and was easily able to generate 6V (with no load) out of it:

Oscilloscope, no load

(2 volts/div)

However, when any sort of substantial load (<1kOhm), the output voltage rapidly drops - this is definitely undesirable since I need to be able to provide hundreds of milliamps to an amp for my project. When I used a 100R load, the waveform became this:

Oscilloscope, 100R load

It appears to be pulsing towards target output voltage, but then abruptly terminating and thereafter switching off for the rest of the cycle (resulting in Vout = Vin). To me, this suggests that the current limiting is kicking in - the voltage is rising until the current gets too high, and then falls off as the converter stops boosting.

I've used a jumper cable for the sense resistor, but I have a feeling this has too high a resistance (compared to the one suggested). However, I've also tried shorting pins 5 and 4 with a piece of wire in between (the pins on the IC are less than a millimetre apart, so it's definitely going to be low resistance), and this doesn't seem to have too much of an effect. I'm not sure what else could be the problem. Perhaps the inductor saturating?

UPDATE

I've tried using a 45mOhm sense resistor (repurposed variable resistor) - doesn't seem to work. It's changing something, but I'm definitely not seeing an improvement. I've also tried using two inductors in series, which should give me something like 6uH. Also no luck.

\$\endgroup\$
6
\$\begingroup\$

You made a significant change in the inductor, from 4.5 µH to 3 µH. The smaller inductor will hold less energy at the same current, and will reach the current limit quicker. Also, make sure the 30 mΩ sense resistor really is 30 mΩ. If that is a larger, then obviously the device will limit the inductor current to a lower value.

\$\endgroup\$
2
  • \$\begingroup\$ I wasn't aware this would have a significant effect. I'll try that. \$\endgroup\$ Jan 29 '15 at 14:24
  • \$\begingroup\$ Okay, update. I tried two inductors in series, around about the same inductance. Didn't help. \$\endgroup\$ Jan 29 '15 at 22:32
6
\$\begingroup\$

The LTC1872 is a current-mode controller - that is to say, the current information is used to establish the on-time of the switch.

From the datasheet:

During normal operation, the external N-channel power MOSFET is turned on each cycle by the oscillator and turned off when the current comparator (ICMP) resets the RS latch. The peak inductor current at which ICMP resets the RS latch is controlled by the voltage on the ITH/RUN pin, which is the output of the error amplifier EAMP.

Putting a jumper wire instead of a correctly-computed sense resistor badly corrupts the current information needed for operation of the device. This is "not a good thing". The reference design implies 1A output current, and you're getting into trouble with what should be a 50mA load. Even a few 1 ohm resistors in parallel instead of the jumper wire should show you some different behaviour at that load current.

As to whether or not the inductor is saturating - if it saturates out your MOSFET is essentially shorting your input supply. Is your MOSFET staying on excessively long? Is the input current to the converter going astronomically high, or is the input voltage getting dragged below 3.3V?

Also, is your new inductor suited for the application? Just because it appears to be 3 microhenries (empirically measured from an RC circuit) does not mean it's appropriate for use in a boost converter (i.e. won't saturate out, won't have high core loss at the converter switching frequency, etc.) - I would try to source the original inductor specified or accurately cross-reference it from another supplier.

\$\endgroup\$
1
  • \$\begingroup\$ I have what I believe to be a 45mOhm resistor here - I calculated this using a current-limited power supply to supply 1A through it and measuring the voltage across it (getting 45mV). I placed this across the terminals. It changes it, but it definitely fluctuates like before. \$\endgroup\$ Jan 29 '15 at 22:34
3
\$\begingroup\$

It's sounds like you've got this stuck in a breadboard- that's usually not a good approach with modern SMPS supplies. Also, the use of a jumper wire for a current sense resistor alone could be causing your problem because of the inductance it has. You need an appropriate value of resistance, but the inductance also must be very low or it will 'see' a higher current than is actually present.

It's well worth the time and effort to get an inductor of both the appropriate value and with a known saturation. Random cores (perhaps intended for EMC applications rather than SMPS) tend to lead to problems. You can salvage cores from other SMPS units if you have some idea of the behavior of the material.

\$\endgroup\$
4
  • \$\begingroup\$ I'm not experienced enough with building SMPSs to be confident in permanently building one first. Re jumper wire: if I short out the terminals directly on the IC (which are very close together), surely that would mean very very little resistance or inductance? I was under the impression that the IC would work fine for very very low resistance values (<0.03R). \$\endgroup\$ Jan 29 '15 at 14:56
  • \$\begingroup\$ Also, is there any way to measure saturation? \$\endgroup\$ Jan 29 '15 at 14:57
  • \$\begingroup\$ To measure saturation you have to put a DC current through the inductor and measure the inductance, or do it indirectly- current should increase linearly with time for a fixed voltage (ignoring resistance). If it curves downward, the resistance is not negligible, if it curves strongly upward, it's saturating. \$\endgroup\$ Jan 29 '15 at 16:05
  • \$\begingroup\$ In response to your first question, AFAIUI, the LTC1872 is a current-mode converter so not allowing it to measure current accurately will probably not turn out well for you. \$\endgroup\$ Jan 29 '15 at 16:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.