Why go down to 0V on a bench PSU

[Cicada]

Hi

I have being looking at the posts on the forum about people building their own bench power supplies. I have being wondering why it is important that a bench power supply should be able to go down to 0V.

I have quickly done a test with my TTi QL564TP at work. It can go down to 50 mV and supply 1 amp. It can even do better. It was just a quick check. I have never had to go down lower than 3V if I remember correctly.

Now my questions are:
1. What is the voltage (1.25V??) below which it usually becomes more difficult to design a bench power supply. I.e. where you need to start add extra circuitry to be able to go below that voltage. For instance a simple regulator or a switching regulator have a lowest voltage that it can output.
 
2. For what kind of lab work (i.e. what kind of circuits) would a person regularly go down below this easily achievable lowest voltage. Maybe you kind forum users can list the cases where you use low voltages, with and without high currents, in your daily work and home projects. It would be interesting to see what applications use such low voltages.

3. How often (% of your projects) do you have to go down to really low voltages.

Thank you
Cicada

[abraxa]

Some food for thought: what is current limitation in reality? How is it usually realized in a power supply circuit? What happens when the output terminals (i.e. the device being powered) see a short circuit and the output voltage can't go below a certain voltage?

[kjr18]

Not only current but also voltage is necessary for a magic smoke to escape, so in case of a short component or other fault there will be not enough power to make any damage. Some time ago I connected cheap step up converter to non current regulated power supply. Angry pixies in that MT3608 module were not happy with that and took like 2 amps at 5V and decided to burn a hole through ic. But when I connected another faulty module to a current limited power supply nothing bad happened even because voltage dropped to almost 0V. That 50mV i think it's due to resistance of connectors/cables.

[Ian.M]

The problem is highly integrated regulator chips that use a feedback loop to compare part of the output voltage with an inaccessible fixed internal reference voltage.  Getting them to deliver an output voltage less than that reference voltage without compromising stability or accuracy is *DIFFICULT*.

Why would you want to be able to go below Vref, which is commonly in the 1V to 2.5V range?
Well, here's a few examples:

* Testing any circuit that is to be powered by a single Alkaline cell as the fully discharged voltage is commonly accepted as 0.8V.

* Setting up a known current through a low resistance current sensor + an ammeter to check function and accuracy.

* Investigating brownout and power on reset behaviour of low voltage MCUs.

* Powering coils to provide small magnetic fields

[David Hess]

I have being looking at the posts on the forum about people building their own bench power supplies. I have being wondering why it is important that a bench power supply should be able to go down to 0V.

Usually it is not important.

I have quickly done a test with my TTi QL564TP at work. It can go down to 50 mV and supply 1 amp. It can even do better. It was just a quick check. I have never had to go down lower than 3V if I remember correctly.

My ancient Tektronix PS503As actually go slightly negative because of an internal current pull down to make sure they can reach zero volts without trimming.  The operational amplifiers used as error amplifiers are deliberately imbalanced so that their worst case input offset voltage can never prevent the power supply output from reaching zero volts.

Now my questions are:
1. What is the voltage (1.25V??) below which it usually becomes more difficult to design a bench power supply. I.e. where you need to start add extra circuitry to be able to go below that voltage. For instance a simple regulator or a switching regulator have a lowest voltage that it can output.

The 1.25 volt limitation comes from easy to use 317/337 style adjustable floating regulators where without an auxiliary bias supply, they can only produce a minimum output voltage of 1.25 volts.  If you build a power supply based on the common 723 voltage regulator, then operation down to zero volts is not a problem.
 

2. For what kind of lab work (i.e. what kind of circuits) would a person regularly go down below this easily achievable lowest voltage. Maybe you kind forum users can list the cases where you use low voltages, with and without high currents, in your daily work and home projects. It would be interesting to see what applications use such low voltages.

Modern low voltage logic requires a power supply which goes down to zero output if you want it to shut off.  Some low voltage circuits can be damaged with improper power supply sequencing unless the supply voltages drop to zero.

3. How often (% of your projects) do you have to go down to really low voltages.

If you mean with a power supply, I would say none because if I am doing this, then I likely need more performance than a power supply will provide.  Sourcemeters which combine a multimeter and precision power supply would be more suitable for these applications but I usually cobble something together as needed.

[Tomorokoshi]

Part 1: Not sure.

Part 2: Assuming the supply is isolated, not all uses are to simply power a board. It can also be used as a signal source. Applications for going to low (with precision!) voltages:

1. Sweep a part from 0 to V that would otherwise be troublesome to do with a resistor divider.

2. Provide a small DC offset or differential voltage to a sensor circuit.

3. Provide an adjustable DC offset to the main supply in series for checking tolerances.

4. Ability to turn off the device without a switch in case of some event, such as controlling the output of a programmable supply.

5. While running the supply in current-limit mode, it allows the limit to be achieved in case the load is shorted.

Along with other variations on this theme.

Part 3: Let's say I need that functionality 5% of the time. That's about 10x the usage frequency where it becomes a requirement for the equipment.

[Cicada]

Some food for thought: what is current limitation in reality? How is it usually realized in a power supply circuit? What happens when the output terminals (i.e. the device being powered) see a short circuit and the output voltage can't go below a certain voltage?

This link will answer these questions to some extend.  Constant Current, Constant Voltage and Current Limiting.
https://community.keysight.com/community/keysight-blogs/general-electronics-measurement/blog/2016/11/09/the-difference-between-constant-current-and-current-limit-in-dc-power-supplies

[Cicada]

* Testing any circuit that is to be powered by a single Alkaline cell as the fully discharged voltage is commonly accepted as 0.8V.
* Investigating brownout and power on reset behaviour of low voltage MCUs.

Very good examples of the need for low voltage.

* Setting up a known current through a low resistance current sensor + an ammeter to check function and accuracy.
* Powering coils to provide small magnetic fields

True. This is probably where you would use a "constant current" "constant voltage" type PSU in the  "constant current" mode.

[mikerj]

Some food for thought: what is current limitation in reality? How is it usually realized in a power supply circuit? What happens when the output terminals (i.e. the device being powered) see a short circuit and the output voltage can't go below a certain voltage?

This link will answer these questions to some extend.  Constant Current, Constant Voltage and Current Limiting.
https://community.keysight.com/community/keysight-blogs/general-electronics-measurement/blog/2016/11/09/the-difference-between-constant-current-and-current-limit-in-dc-power-supplies

Those were questions to get you thinking about your own question, not a request for knowledge!

[Cicada]

Modern low voltage logic requires a power supply which goes down to zero output if you want it to shut off.  Some low voltage circuits can be damaged with improper power supply sequencing unless the supply voltages drop to zero.

I was not aware of that.

If you mean with a power supply, I would say none because if I am doing this, then I likely need more performance than a power supply will provide.  Sourcemeters which combine a multimeter and precision power supply would be more suitable for these applications but I usually cobble something together as needed.

Very interesting observation.

[Cicada]

Some food for thought: what is current limitation in reality? How is it usually realized in a power supply circuit? What happens when the output terminals (i.e. the device being powered) see a short circuit and the output voltage can't go below a certain voltage?

This link will answer these questions to some extend.  Constant Current, Constant Voltage and Current Limiting.
https://community.keysight.com/community/keysight-blogs/general-electronics-measurement/blog/2016/11/09/the-difference-between-constant-current-and-current-limit-in-dc-power-supplies

Those were questions to get you thinking about your own question, not a request for knowledge!

Hi mikerj

I am aware of that. I have posted the link so that someone else reading the thread can very easily navigate to the information they need to understand the issue discussed.

[TheUnnamedNewbie]

Some food for thought: what is current limitation in reality? How is it usually realized in a power supply circuit? What happens when the output terminals (i.e. the device being powered) see a short circuit and the output voltage can't go below a certain voltage?

Often voltage regulation and current regulation is done seperatly. It could be that the supply cannot regulate a voltage under a certain level, but is capable of regulating a current which results in a voltage lower that that level.

1. What is the voltage (1.25V??) below which it usually becomes more difficult to design a bench power supply. I.e. where you need to start add extra circuitry to be able to go below that voltage. For instance a simple regulator or a switching regulator have a lowest voltage that it can output.
 
2. For what kind of lab work (i.e. what kind of circuits) would a person regularly go down below this easily achievable lowest voltage. Maybe you kind forum users can list the cases where you use low voltages, with and without high currents, in your daily work and home projects. It would be interesting to see what applications use such low voltages.

3. How often (% of your projects) do you have to go down to really low voltages.

Thank you
Cicada

1] Depends on the topology. The main issue is that if you want to work with a single supply rail, you need rail-to-rail capable feedback networks. Otherwise, you need negative supplies for your control circuit so that a zero volt output is not against the rail of the feedback network.

2] I often go very close to zero when I am doing things like playing with precision measurements. I might want to use a few hundred millivolts as a reference voltage in a feedback circuit, and use a powersupply to generate it.
At work, I do research on ICs and we might have bias voltages below 100 mV, which we want to control with very high precision (to a few tens of microvolts). If I were to apply 1.25 V on most of the chips we work with they would die within seconds. (\$V_{DD}\$ below 800 mV is not uncommon. The low power guys at my research group do ARM cores that operate on a \$V_{DD}\$ of 300 mV, if I remember that correctly).

3] At work, most of the time. At home, not that commonly, although I am in the process of making a device to provide precision voltages precisely for biasing and calibration purposes.

If you mean with a power supply, I would say none because if I am doing this, then I likely need more performance than a power supply will provide.  Sourcemeters which combine a multimeter and precision power supply would be more suitable for these applications but I usually cobble something together as needed.

Interesting point, though I would like to add that source meters are far more expensive. If you are working something where you need 7-8 bias voltages in the few hundreds of mV, using a source meter for each will quickly become very expensive (compared to using precision supplies)

[bd139]

Another application: I was recently using my (TTi PL310) power supply in mV territory to characterise varactors which were going in a PLL loop. The PLL phase comparator and loop filter goes pretty low Vout so you want to make sure that the oscillator won't drop dead while it's hunting around for a lock worst case.

Also typically if you have battery powered systems with DC input i.e. sensor etc, then you want the common mode input range of things to bottom out as close to ground as possible to give you more voltage swing so it makes sense to use a power supply which can hit zero volts easily as an input source.

I can think of a thousand more use cases given time.

[David Hess]

If you mean with a power supply, I would say none because if I am doing this, then I likely need more performance than a power supply will provide.  Sourcemeters which combine a multimeter and precision power supply would be more suitable for these applications but I usually cobble something together as needed.

Interesting point, though I would like to add that source meters are far more expensive. If you are working something where you need 7-8 bias voltages in the few hundreds of mV, using a source meter for each will quickly become very expensive (compared to using precision supplies)

I have thought about designing and making a floating low current low noise precision power supply for these types of applications but so far alternative methods have been easier.  Trying to use a power supply as a bias generator at low voltages in a precision application will reveal that most of them are not suitable for this task due to lack of stability, lack of precision, and too much noise.  They are not variable voltage references.