Does one ever need some specific voltages from PSU other than most common ones ?

[unitedatoms]

I have an idea about useability of DC power supply controls. Will typical hobbyist user be satisfied with supply with no digital displays. Just few leds and knobs?

Consider the panel with banana outputs, red led indicating constant current mode, green led indicating constant voltage mode, one knob for 10 fixed voltages: 1.200V, 1.500, 2.500, 3.300, 5.000, 6.300, 9.000, 12.000, 15.000, 24.000V. And one current limit knob for 10 fixed limits: 0.1mA, 0.5mA, 1, 5, 10, 20, 50, 100, 500, 1A.

The numbers are painted on panel for each knob. It will take very little space, will give feel of repeatability and precision. But resolution is coarse.

My reasoning for such UI is that most of the functionality needed from PSU is a constant voltage for some design, where rails are long decided upon. And current limits are also decided and clearly belong to certain damage levels for different circuits. For example LED will never need more than 20 mA. Or relay will never need more than 100mA, etc.

Any fine measurements can be done once with external multimeters. Users need repeatabilty, low noise and clear function more than resolution.
What do you think? Did you ever need some specific voltages for PSU other than listed ?

[unitedatoms]

Agree about ramp requirement and the tolerance testing, but this can be done with any lab supply. Everyone has a good range and resolution on their lab supply.

I am mostly thinking about low noise, low cost, low power, and low care supply with multiple isolated channels, like say 6 channels. So all the rails with all polarities will be available in small package. It looks like the useabilty fails on any multichannel supply when you have more than 1 channel.

[MarkF]

Also, voltage injection.
I sometimes use a power supply to drive a signal input (i.e. not the power rails) for testing a device over various voltage levels.

[MarkF]

Agree about ramp requirement and the tolerance testing, but this can be done with any lab supply. Everyone has a good range and resolution on their lab supply.

I am mostly thinking about low noise, low cost, low power, and low care supply with multiple isolated channels, like say 6 channels. So all the rails with all polarities will be available in small package. It looks like the useabilty fails on any multichannel supply when you have more than 1 channel.

I have built a +/-15V supply from 3 terminal regulators to test analog/op-amp circuits without the need to setup a +/- supply.  (I only have single output supplies.)  I can see a standalone supply with multiple voltages (but without the switches).  I often need multiple 'standard' voltages for circuits.

[not1xor1]

I have an idea about useability of DC power supply controls. Will typical hobbyist user be satisfied with supply with no digital displays. Just few leds and knobs?

Consider the panel with banana outputs, red led indicating constant current mode, green led indicating constant voltage mode, one knob for 10 fixed voltages: 1.200V, 1.500, 2.500, 3.300, 5.000, 6.300, 9.000, 12.000, 15.000, 24.000V. And one current limit knob for 10 fixed limits: 0.1mA, 0.5mA, 1, 5, 10, 20, 50, 100, 500, 1A.

The numbers are painted on panel for each knob. It will take very little space, will give feel of repeatability and precision. But resolution is coarse.

My reasoning for such UI is that most of the functionality needed from PSU is a constant voltage for some design, where rails are long decided upon. And current limits are also decided and clearly belong to certain damage levels for different circuits. For example LED will never need more than 20 mA. Or relay will never need more than 100mA, etc.

Any fine measurements can be done once with external multimeters. Users need repeatabilty, low noise and clear function more than resolution.
What do you think? Did you ever need some specific voltages for PSU other than listed ?

IMHO for general purpose testing/experimenting a PSU with 2 single outputs plus one double output might be more useful:
1) variable CV/CC... something like 0-24V 0-3A (at least)
2) switchable output (or variable one) 3-5V with fixed current limit (1A or more)
3) dual rail symmetrical output with 100mA (at least) current limit, variable voltage from ±1 up to ±15V (or more)

that should cover most needs for a beginner.

[grifftech]

my DIY power supply can output 3.3, 4, 5, 8, 9, 10, and 12 simultaneously.

[rdl]

This design has been around for many years. It doesn't have adjustable current limiting though. It's obvious from the 1.5 volt steps that it's meant as a battery replacement. Still, I'm surprised it's never been updated to voltages more useful with modern devices.

https://www.parts-express.com/power-supply-regulated-3-12-vdc-2a-6-way--120-536

[bd139]

I'm not 100% sure of the value of a current limit other than a failsafe to stop a persistent fire. The issue is that to maintain stability, most power supplies have to utilise an output capacitor as part of the feedback. This capacitor can discharge stupid quick into your DUT (I = dQ / dT) and pop it before the feedback loop has a chance to do anything about it. However a hard limit to keep the power supply inside its SOA is a good thing as that stops it burning your house down.

Linearly variable voltage is essential for me - I regularly sweep the rails on a project to look for things that are supply voltage sensitive.

The absolutely best supply out there for me in genreal use is an HP 6236B or HP 6237B or HP/Agilent/Keysight E3630A really. They have hard limits, foldback, are indestructible and incredibly flexible for most applications. E3630A if you want digital readout. +/-20V @ 500mA each tracking (or 0-40V @ 500mA) and 0-6V @2.5A or 0-18V @ 1A. Literally 99% coverage for my use cases.

[bsudbrink]

I regularly require 8 and 16 volts to power the unregulated side of devices with onboard regulators.