1
\$\begingroup\$

I used an LM350 adjustable voltage regulator https://www.onsemi.com/pub/Collateral/LM350-D.PDF and to keep the wattage dissipated by the device to a minimum I have opted to make the output voltage 24VDC. Max current drawn is around 500mA. The input voltage can vary from 19VDC to 32VDC but will be 24VDC under normal conditions (24V Vehicle Supply). When the input to the circuit "Vin" is below 25.1VDC the regulator Input-Output differential will be too small and the regulator will not regulate the voltage. I cannot find any information if this will have any negative consequences eg increased noise or limiting of current flow? The circuit is intended only to limit the voltage when it goes above the normal 24V and some noise filtering as a by-product. Is there any reason not to use the regulator in the "non-regulating" / "drop-out" state most of the time?

\$\endgroup\$
  • \$\begingroup\$ If it is in the non-regulating state then the input voltage is in the range of what you need so the regulator has nothing to do. \$\endgroup\$ – Solar Mike Jul 17 '17 at 10:08
  • \$\begingroup\$ Indeed. I only need it when the input goes above 25V (during charging of the vehicle battery) but if I force the output voltage to say 18VDC then the regulator will sit and generate heat with wasted energy 90% of the time when the vehicle is not charging the battery. Hence my question \$\endgroup\$ – Christo Candiotes Jul 17 '17 at 10:10
  • \$\begingroup\$ Have you considered to use a buck-boost instead of a linear regulator? \$\endgroup\$ – Simus994 Jul 17 '17 at 10:18
  • \$\begingroup\$ Good comment, thank you Simus994. I have used a switching regulator previously and it is an option but where possible we are trying to stay away from anything that could create electrical noise. \$\endgroup\$ – Christo Candiotes Jul 17 '17 at 10:44
2
\$\begingroup\$

Let's have a look at datasheet Figure 10 "Dropout Voltage". At 500mA it will be a bit above 1.5V.

When operating in this mode, the output pass transistor will be fully on (saturated in this case, since it is a NPN bipolar).

If this was a PNP pass device LDO, you would expect excess ground current, as the regulator attemps to saturate its output device. However, this one uses a NPN, so the excess base current will simply go into the output. No problem here.

The regulator will not regulate anything though, this means it will act either as a resistor or as a couple of diodes in series, so output voltage may vary depending on current draw. Also, output voltage will follow input voltage, so input noise will not be suppressed.

If your load works on 19V, and you have 24V input, and the load can tolerate the regulator not rejecting input noise, then you're fine. In this case it would simply act as a voltage limiter.

If you also want to filter noise, then something like a capacitance multiplier with an output voltage limit would be more suitable.

EDIT: Example

schematic

simulate this circuit – Schematic created using CircuitLab

This is a simple capacitance multiplier. It lowpass-filters the input (RC network) to filter out noise. Zener limits the voltage. I put in a CFP (double transistor) for lower output impedance, so you can say that's the "luxury" version!

\$\endgroup\$
  • \$\begingroup\$ Thank you Peufeu. I did miss the fact that it will not provide any noise filtering if the voltage is below the standard vehicle charging voltage. To be safe I will drop the output to force the regulator into the working region or consider the switching alternative as suggested by Simus994. \$\endgroup\$ – Christo Candiotes Jul 17 '17 at 10:51
  • \$\begingroup\$ Yeah, buck-boost will give you stable voltage. I added a simpler solution in my edit, which will remove noise, but not boost voltage. What is the load ? \$\endgroup\$ – peufeu Jul 17 '17 at 11:39
  • \$\begingroup\$ The load is a modem. I have never heard of a capacitance multiplier until today. Tnx for sharing. \$\endgroup\$ – Christo Candiotes Jul 17 '17 at 12:03
  • \$\begingroup\$ I'm gonna bet the modem has a DC-DC at the input to create its internal voltages and doesn't give a damn about variations on the "24V" supply... \$\endgroup\$ – peufeu Jul 18 '17 at 9:53
1
\$\begingroup\$

I'd definitely be considering a buck-boost regulator like the one below: -

enter image description here

They're not particularly cheap but they do what they say on the tin and with everyone trying to stop the planet heating, this is a fairly green solution in the longer run. There may be cheaper and similar solutions from TI of course.

I cannot find any information if this will have any negative consequences eg increased noise or limiting of current flow?

This would be my concern and it will happen on most LDO regulators if taken too close to the limit i.e. there will be progressively more output noise as you approach the regulation voltage from a higher voltage. Whether your circuit might cope I cannot say. My philosophy is that if there is a more expensive solution that offers peace of mind then it is worth considering.

\$\endgroup\$
  • \$\begingroup\$ Agree Andy aka. Thank you for the suggestion. I will go the switching route. \$\endgroup\$ – Christo Candiotes Jul 17 '17 at 10:54
0
\$\begingroup\$

If noise is important, then consider the PSRR of LDOs versus switchers.

At high frequencies (and "high" may be 100Hz and above for some 1uA Iddq LDOs), the input noise comes through un-attenuated. Examine the datasheet.

Switchers have their own added synchronous noise, with lumped Ls and Cs to filter; input noise lower than the switcher's frequency will be somewhat attenuated; examine the datasheet.

Part of your tradeoff will be the standby current. You can purchase 1uA LDOs that use Pchannel FET regulation. You cannot purchase 1uA switchers.

\$\endgroup\$
  • \$\begingroup\$ Thank you @analogsystemsrf. I cannot quickly find comparable results for load regulation between the linear and switcher devices but I believe the switcher comes out on top in that regard? This, and the efficiency is more important in my design than the added noise of the switcher, which is a known constant that I can eliminate mostly with LC filters. \$\endgroup\$ – Christo Candiotes Jul 18 '17 at 7:25

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.