3
\$\begingroup\$

I've been using a USB voltage/current measurement device to evaluate power consumption of USB devices and performance of USB chargers and batteries. Most of the USB power supplies I've used are switched-mode power supplies which have a no-load voltage just over 5V and sag by varying amounts under load; however, I've noticed that one of my USB chargers, a Samsung "Travel Adapter" (5V, 0.7A rating) supplied with a feature phone, exhibits a significant voltage increase under load, until the supply's limit is reached.

Here's what I'm getting:

Amps    Volts   Error
0.00    5.11    0.02
0.03    5.14    0.02
0.06    5.17    0.02
0.12    5.20    0.02
0.17    5.23    0.02
0.25    5.28    0.02
0.31    5.32    0.02
0.35    5.35    0.02
0.41    5.39    0.02
0.46    5.41    0.02
0.50    5.42    0.02
0.56    5.45    0.02
0.61    5.49    0.02
0.64    5.51    0.02
0.67    5.53    0.02
0.70    5.54    0.02
0.73    5.55    0.02
0.82    5.60    0.02
0.83    5.59    0.02
0.85    5.07    0.05
0.89    3.90    0.15
0.93    3.70    0.25

Chart depicting output voltage against amperage

Notice that the voltage rises as the load increases up to 0.82A. Trying to pull more power than that causes the voltage to fall off a cliff as the limit of the power supply is exceeded.

I find this behavior to be bizarre because no other SMPS I've worked with outputs increased voltage under load. What kind of SMPS design would cause the voltage to increase under load, and how would be it different from more typical designs? What advantages, if any, would this design carry?

\$\endgroup\$
  • 1
    \$\begingroup\$ Maybe there is a load-dependent error term in the feedback circuit. \$\endgroup\$ – dext0rb Jul 25 '14 at 20:37
3
\$\begingroup\$

Some cheap chargers don't use a regulator on the output side, monitoring the output voltage, and feeding back this information to the primary via an optocoupler. Instead, they regulate based on what the "see" on the auxiliary winding on the primary side, saving the cost for the optocoupler. On the aux winding of a flyback converter, you get a very approximate information about the voltage on the secondary winding, and sometimes, ringing and spikes dominate the picture instead of the theoretically ideal reflection of the output voltage. It could be that under mid- to high-load conditions, the ringing on the aux winding becomes less and the regulator increases the power just because it receives a decreasing ratio of the output voltage.

If this is the case, the error actually would occur at light loads (because this is where the ringing might be severe), but is compensated for by decreasing the overall regulation such that the voltage is somewhat within the specification over the entire load range. The details would depend on what the snubbers are optimized for, for example.

Cheapo USB chargers are sometimes reduced to an absolute minimum of components because the cost pressure is dramatic. Muntzing is anything but an old-fashioned trick of the trade.

Searching for "primary side regulation in flyback converters" might lead to some background info like this article.

\$\endgroup\$
3
\$\begingroup\$

Sometimes a negative load regulation term is introduced on purpose in order to compensate for IR drop in the cable between the regulator and load. This is common in automotive or industrial systems where the cable is long and bulky, thick cabling is undesirable.

If this regulator was designed for a specific phone, it is also possible that it's making up for internal drops in phone's power circuitry. For example, if an LDO is used to generate a 5V internal rail, it's dropout voltage will go up as it's load current increases. The phone designer could compensate for that and keep the LDO out of dropout by increasing the input voltage as a function of load...

\$\endgroup\$
1
\$\begingroup\$

That's a really cool feature. The 0.5V/A rise compensates for a 0.5 ohm cable drop. Sure, 0.5 ohms sounds like a lot until you consider #28 wires, which are 65 milliohms per foot. If you use a 3' USB cable made with #28 power wiring, this compensation is just about right. Or, a 10' cable with #24.

\$\endgroup\$
  • \$\begingroup\$ Not really an answer to the question. You can comment on the question when you have enough reputation. \$\endgroup\$ – pipe Mar 16 '17 at 21:50
  • 2
    \$\begingroup\$ @pipe: It's an answer to a perfectly reasonable interpretation of the title question. The other answers all address how it could happen, this one mentions why you would design it that way. The title asks why and the body of the question asks both (what kind of design, what advantages). \$\endgroup\$ – Ben Voigt Mar 16 '17 at 21:56
0
\$\begingroup\$

The ratings never mean maximum value. Ratings are based on some standard test conditions. Also if you see the V - I curve, this seems more or less similar to a battery charging cycle except that V & I are reversed. Usually in a typical battery charging cycle, the voltage rises up initially during which the current remains a constant. After the battery's OCV (Open Circuit voltage) reaches a set point, the voltage remains a constant but the current reduces. After some amount of time, the charging terminates. But in your case, it is exactly opposite to a typical charging cycle. Assuming you have reversed the order of the data, then there isn't anything wrong with the charger. It is working fine.

You may find more details to the battery charging process here

\$\endgroup\$
  • \$\begingroup\$ A USB charger is supposed to be a regulated 5V power source. The conversion to the correct charging voltage is handled by the device to which the power source is connected; the power isn't fed directly into the battery. \$\endgroup\$ – bwDraco Jul 29 '14 at 0:11
  • \$\begingroup\$ Ah now I get it. I was thinking that these measurements were from the output of the USB Charger. In that case, I believe what Zebonaut said is right. \$\endgroup\$ – LJanardhan Jul 30 '14 at 20:02

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.