6
\$\begingroup\$

So here I have a standard 19V noname laptop power brick rated at say 4A output. Inside is a typical SMPS regulated by a TL431 and an optocoupler.

I added in a potentiometer to the reference resistors on the TL431 so am able to vary the output voltage (say 10-19V now). I am curious what the limiting factor is regarding the possible voltage window that can be achieved. Is there something stopping the SMPS outputting 3V or 50V, if the TL431 commands it so (apart from the voltage ratings of the output caps and diodes)? Is the transformer/coil optimized for just a narrow optimum output voltage?

Also, if the rating was originally 19V@4A=76W, how does that rating change in practice as I go lower in voltage, can I consider 76W to be my guiding max number, or the 4A to be a maximum regardless of voltage (I suppose overheating the rectifying diode would be my first concern with higher amp flow at a lower voltage?)? Is a 24V power brick for example almost identical to a 12V one, with just different resistors on the TL431, or are there more design considerations to adapt to a different voltage?

Cheers!

\$\endgroup\$
5
  • \$\begingroup\$ I wouldn't imagine that a "typical" laptop power brick has a TL431 in it. Most converter ICs have their own integrated voltage reference. \$\endgroup\$ Mar 27 at 2:58
  • 1
    \$\begingroup\$ I just tore apart a few power bricks out of the trash I am using as power supplies for some lighting, and most of them had the TL431 and optocoupler control system, I suppose they are of the 10 years old vintage maybe that explains it... \$\endgroup\$
    – parkside
    Mar 27 at 3:01
  • 1
    \$\begingroup\$ Ah, you are right. These are mostly isolated flyback, and the voltage reference is on the other side of the transformer from the controller chip. \$\endgroup\$ Mar 27 at 3:14
  • 1
    \$\begingroup\$ After the voltage/current limitations and considerations, the next issue you're going to run into is the stability - changing the output voltage is likely also changing the gain of the feedback network. So now, you have to consider loop compensation. \$\endgroup\$
    – MOSFET
    Mar 27 at 4:41
  • 1
    \$\begingroup\$ Most is covered by the answers below. To fully answer this for yours, please either measure both Vds of your primary switch to see how much margins you have to go up in output voltage as well as Vaux on the primary side which will probably be the limiting factor for going down in voltage, or calculate it from reverse engineering and update your question with the info. \$\endgroup\$
    – winny
    Mar 27 at 9:12

3 Answers 3

4
\$\begingroup\$

In a TL431-based regulator, the REF pin of the TL431 will be kept at the IC's internal reference voltage: ~2.5V for TL431/432 or ~1.24V for TLV431. The former is more common.

Is there something stopping the SMPS outputting 3V or 50V, if the TL431 commands it so (apart from the voltage ratings of the output caps and diodes)? Is the transformer/coil optimized for just a narrow optimum output voltage?

Without knowing anything about the circuit, it's safe to say that the theoretical lower bound for adjustment is the REF voltage. For your example, you can't expect an output voltage lower than 2.5V. In practice, at such low voltages, the ripple (percentage-wise) can be really high, so this could be a limiting factor. For example, if the design is optimised for 1% ripple at 19V then you may end up with, say, 5% ripple (or maybe even higher) if you adjust the output to 5V.

As for the upper bound, it all depends on the rest of the circuit. Without knowing the internals we can't give a satisfactory enough but there are some limiting factors:

  • Duty Cycle Limitation: Most laptop chargers are isolated flyback converters where the duty cycle, which determines the power delivery and the output voltage, is somewhat internally limited. Assuming there's no active power factor correction (APFC) circuit inside, we can expect the transformer to be designed for a considerably wide input voltage range (e.g. European: 180..264 Vac or universal: 85..264 Vac). Now if you try to adjust the output to a relatively high voltage (e.g. 50V) then you may not get the desired output when the input voltage is relatively lower due to the aforementioned duty cycle limitation.

  • Component Ratings & De-ratings: You are ignoring the exceeding of capacitor & diode ratings but higher output voltage means higher reflected output to the primary (assuming for a flyback) and this will increase the stress of the primary-side components. Although the voltages are clamped, there's still a risk of having higher stress which may result in failure. If the design is a two-switch flyback then this may not be a problem but it's unlikely to see a two-switch design due to the aggressive market and cost targets.

NOTE: The placement of the adjustment pot is important. As you might already be aware, the TL431 takes the output voltage sample from a resistive divider. If you put the pot to the high side then you'll change the dynamics of the circuit as the high side of the divider network appears in the transfer function. So, you may end up with an unstable system by adjusting the output from a pot put on the high side. The pot should be a part of the low side of the divider.

Also, if the rating was originally 19V@4A=76W, how does that rating change in practice as I go lower in voltage, can I consider 76W to be my guiding max number, or the 4A to be a maximum regardless of voltage (I suppose overheating the rectifying diode would be my first concern with higher amp flow at a lower voltage?)?

Again, it's difficult to answer without knowing the internals but we are sure that the transformer and the output section (PCB/layout) are designed/optimised for 4 Amps output current. Lowering the output voltage doesn't mean the charger can deliver a higher output current. But if you adjust the output to a higher voltage, the internal power limitation may not allow you to draw 4 Amps. So, at 10V you can possibly draw 40W but at 30V you may not draw 120W.

Is a 24V power brick for example almost identical to a 12V one, with just different resistors on the TL431, or are there more design considerations to adapt to a different voltage?

The transformer design is different, even if we can expect the rest to be the same. If the same transformer is used for 12V and 24V output, assuming the input voltage is the same, the losses will quite possibly be higher for 24V even at the same power level.

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

In a typical offline flyback converter, the voltage can't be lower than the aux (primary side) minimum operating voltage, nor higher than the respective limits of all parts involved: the controller, transistor, capacitors, rectifiers, etc.

By "typical", we refer to something like this:

enter image description here

Source: Designing Isolated Flyback Converter Circuits | Techweb | Rohm.com

Common controllers include UC3842, L6565, NCP1342, etc. Note that, after startup (here, provided by the VH pin; others use a startup circuit, or for integrated switch e.g. TOPSwitch etc. parts, it may be derived from the drain/SW pin), control power is obtained from the aux winding. Brownout of the controller causes it to shut down, until supply voltage charges back up to the rising threshold; this hysteresis is how hiccup-mode current limiting is typically implemented. Thus, an overloaded or browned-out supply tends to pulse or "hiccup" rather than operating continuously in a steady current-limit mode.

The transformer doesn't actually care very much; core loss will be higher at higher peak voltages, but you're probably not going for maximum efficiency in an adjustable unit anyway. Current is still limited by magnetic saturation, and by frequency (mostly in DCM).

Power supplies made for wide output variable range, typically use an independent aux supply (perhaps one of the integrated-switch regulators), and current limiting can be made more precise by adding output DC current measurement, or using architectures better suited for output current regulation, like forward. The aux supply will typically be to both sides (primary and secondary) so that both primary/power and secondary/interface circuits can operate regardless of output setting -- that is, such a supply might have an LED or LCD display, MCU, and ADC/DAC to program, monitor and control the output, and these need to be operational regardless of setting.

Basically, to adapt an off-the-shelf supply for general-purpose use, you're putting in as much effort as building one from scratch anyway.

The comment suggesting to use a secondary stage is an attractive one: the PSU has already solved all the hard problems for you (mains isolation, protection, EMC), and provides steady output at ratings. A buck converter is much easier to design or use than a full offline converter, and easier to solve EMC issues on (it's common-ground, so the input and output can be filtered with respect to a single ground point, avoiding spooky common-mode issues). And modules are available, though do mind they are typically of poor quality (reliability, counterfeit components used, poor efficiency, noisy, etc.). You can even make a boost, or combination (buck-boost), or SEPIC topology to afford both higher and lower output voltages (say if you wanted a 0-30V range for example).

\$\endgroup\$
0
\$\begingroup\$

Respect both the current and voltage limits simultaneously; adding capability in either costs money and no company is going to populate 50V caps when 25V ones will work or add in traces that can carry 10A when the max design load is 4A (yes, sometimes extra capability will be built in because parts or boards are used across different designs but you can't count on it). There might be other issues - undervoltage detection, efficiency loss, etc - but it's impossible to tell without a complete schematic. Basically, anything that's powered off the output of the converter circuit can't be assumed to operate correctly or at all.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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