New programmable bench supply, minimum step time?

[Bosray]

Hi, im looking to purchase a new bench power supply. Look for something with a minimum of 3 channels and a programmable output curve that can be configured on a windows PC.

So far ive narrowed it down to UNI-T UDP3305S-E or Siglent SPD3303X-E but im open to other suggestions.
From these two it seems to be the Siglent SPD3303X-E thats the best choice due to fulle isolated outputs.

But before i make my choice i need to know whats the minimum time for changing the output voltage, i need to be able to go from 24v down to around 5v then up to 24v again after 200ms. The manual says nothing about a minimum time just a max of 99999s on the timers, has anybody tried too see what the minimum is on either of these two models?

[tautech]

Welcome to the forum.

You're right, programable step times are not listed for the SPD3303X/E but in the 4ch SDP4000X range they are better than 50ms.
List time accuracy < 50 ms
SPD4000X also have webserver control which will work with any PC with a browser.
https://int.siglent.com/u_file/download/25_08_06/SPD4000X_DataSheet_EN01C.pdf

[AVGresponding]

@Bosray I'd test that for you but mine's an older pre-facelift version.

@tautech Is the 3303X-E still flashable to a 3303X or did they patch that?

[2N3055]

Hi, im looking to purchase a new bench power supply. Look for something with a minimum of 3 channels and a programmable output curve that can be configured on a windows PC.

So far ive narrowed it down to UNI-T UDP3305S-E or Siglent SPD3303X-E but im open to other suggestions.
From these two it seems to be the Siglent SPD3303X-E thats the best choice due to fulle isolated outputs.

But before i make my choice i need to know whats the minimum time for changing the output voltage, i need to be able to go from 24v down to around 5v then up to 24v again after 200ms. The manual says nothing about a minimum time just a max of 99999s on the timers, has anybody tried too see what the minimum is on either of these two models?

Well, if you think about it there will be no single number there.

It will depend on load type and power consumption, voltage difference, current limits set, and it will be different for programming up, or down.

For instance, if your DUT has 10000µF capacitor on input, expect slower ramp up and down than if it has 1000µF or 100µF. Depending on current limit, current consumption and capacity, down programming might be slower or faster than up.

And that is after PSU received command and interpreted it and actually started to change voltage on output.
There will be a delay (latency) from sending SCPI command from PC to PSU to actual start of change of voltage on output. That is internal processing time for CPU to process command and reconfiguring analog parts of PSU for new config.

A piece of advice? This is backwards.

First you need to analyse your requirements, and then write down your required (or wished for) timings. You need to draw a testing protocol and verify worst case scenario (fastest) transitions, up and down.

If your requirement is 200 ms (up or down), I think Siglent SPD3303X-E can do it.
But a load (current consumption of DUT and capacitance onboard) will have large influence on down programming.
For up programming, as long as load capacitance is not large it should cope well.

[nctnico]

Hi, im looking to purchase a new bench power supply. Look for something with a minimum of 3 channels and a programmable output curve that can be configured on a windows PC.

So far ive narrowed it down to UNI-T UDP3305S-E or Siglent SPD3303X-E but im open to other suggestions.
From these two it seems to be the Siglent SPD3303X-E thats the best choice due to fulle isolated outputs.

But before i make my choice i need to know whats the minimum time for changing the output voltage, i need to be able to go from 24v down to around 5v then up to 24v again after 200ms. The manual says nothing about a minimum time just a max of 99999s on the timers, has anybody tried too see what the minimum is on either of these two models?

The specification you are looking for is settling time. Technically, what is needed to have a specified fall time is an active load circuit which discharges the output capacitors to lower the voltage towards the new (lower) setpoint. The Uni-T UDP3305S-E has a specified settling time in the datasheet. The Siglent SPD3303X-E does not.

[2N3055]

Hi, im looking to purchase a new bench power supply. Look for something with a minimum of 3 channels and a programmable output curve that can be configured on a windows PC.

So far ive narrowed it down to UNI-T UDP3305S-E or Siglent SPD3303X-E but im open to other suggestions.
From these two it seems to be the Siglent SPD3303X-E thats the best choice due to fulle isolated outputs.

But before i make my choice i need to know whats the minimum time for changing the output voltage, i need to be able to go from 24v down to around 5v then up to 24v again after 200ms. The manual says nothing about a minimum time just a max of 99999s on the timers, has anybody tried too see what the minimum is on either of these two models?

The specification you are looking for is settling time. Technically, what is needed to have a specified fall time is an active load circuit which discharges the output capacitors to lower the voltage towards the new (lower) setpoint. The Uni-T UDP3305S-E has a specified settling time in the datasheet. The Siglent SPD3303X-E does not.

While that specification in UNI-T might look like they have better specified datasheet, it is just a poster number.
Useless in practice, unless you are measuring PSU loaded with active load at max current or unloaded PSU.
Also they don't specify what they mean, we can presume that they meant from 0 to Vmax to 1% accurate set voltage.

From their specification you can see exactly why I said that DUT is important. If DUT is pulling small current,  UNI-T UDP3305S-E is clearly too slow for OP with 400ms down programming.
If DUT has it's own capacity on power input those numbers fall apart. These numbers are all about charging output capacitor in PSU. They do not mention how that changes with current limit as it will..I don't understand why it takes different time to charge up unloaded (max current available) than to discharge (where nominal current to load and down programmer together should discharge faster), here it is reverse.

Dynamically, those both have 50 µs load change regulation. They should both behave similarly in that regard.

Only thing that I see is that UNI-T UDP3305S-E has command processing time specified (at quite slow 100ms) and SPD3303X-E does not. But I think it won't be more than 100 ms.

So for the OP neither of these two can be recommended on the basis of the datasheet alone and without specifying DUT and protocol better. That is my opinion.

Many do not understand that most of these modern PSU, despite all the screens, software and such, are still 1-quadrant devices and they will have limited dynamic programming behaviour that will vary significantly depending on circumstances.
Only guaranteed spec will be load regulation response time.
Everything else needs to be characterised with actual DUT or realistic analog.

[nctnico]

Disagree. The fact there is a spec for maximum down programming time for the UNI-T PSU means there is some hardware in place to discharge the output capacitance. And ofcourse the actual time depends on how much load capacitance is connected externally. But realistically, most loads will also draw current so it is very likely that the capacitance in the load is discharged quicker compared to power supply's no-load conditions.

[tautech]

@Bosray I'd test that for you but mine's an older pre-facelift version.

@tautech Is the 3303X-E still flashable to a 3303X or did they patch that?[/color][/b]

AFAIK, never tried it but there is info in threads here.

I don't currently have any PSU's unpacked to do checks for the OP.

[2N3055]

Disagree. The fact there is a spec for maximum down programming time for the UNI-T PSU means there is some hardware in place to discharge the output capacitance. And ofcourse the actual time depends on how much load capacitance is connected externally. But realistically, most loads will also draw current so it is very likely that the capacitance in the load is discharged quicker compared to power supply's no-load conditions.

Siglent also have down programmer. Fact is that its specifications for rise/fall time on programming change are  not specified in data sheet. On this type of PSU those specs are highly dependable on what kind of load is connected.

My SPS5000X not only has down programmer, but you can also chose to use it or not. Also you can program slew rate so you get exact risetime's when changing voltage. You can also specify output impedance. So it can act as a voltage source with specified internal resistance.
But that is not 400 € power supply.

SPS4000 has more detailed specifications also.

As I said, I am not saying anything specifically against UNI-T PSU in question. Did not test it myself. I only know that with UNI-T what says in datasheet is not necessarily transferring to real life. I would need to test it myself or to have someone I trust do it. It looks interesting on paper. At least it has normal keyboard.

Just few days ago I recommended a good friend of mine to buy a UTP3313TFL-II.
It is small, no nonsense linear PSU that fits perfectly his needs. It is working well and was really cheap.
I liked it for what it is and recommended it to someone I care about.

[Martin72]

Many do not understand that most of these modern PSU, despite all the screens, software and such, are still 1-quadrant devices and they will have limited dynamic programming behaviour that will vary significantly depending on circumstances.
Only guaranteed spec will be load regulation response time.

This.
In order for a power supply to quickly follow a programmed curve, it must be able to absorb load.Otherwise, its behavior is highly dependent on load, which makes sense when you think about it.

[tautech]

Many do not understand that most of these modern PSU, despite all the screens, software and such, are still 1-quadrant devices and they will have limited dynamic programming behaviour that will vary significantly depending on circumstances.
Only guaranteed spec will be load regulation response time.

This.
In order for a power supply to quickly follow a programmed curve, it must be able to absorb load.Otherwise, its behavior is highly dependent on load, which makes sense when you think about it.

Also output current limit settings.
https://www.eevblog.com/forum/testgear/lab-power-supply-turn-on-and-off-characteristics/

[nctnico]

Many do not understand that most of these modern PSU, despite all the screens, software and such, are still 1-quadrant devices and they will have limited dynamic programming behaviour that will vary significantly depending on circumstances.
Only guaranteed spec will be load regulation response time.

This.
In order for a power supply to quickly follow a programmed curve, it must be able to absorb load.Otherwise, its behavior is highly dependent on load, which makes sense when you think about it.

That is why a small DC load to sink current has been a standard on many higher end power supplies for a long time. Keysight PSUs have had this feature since the HP days. This DC load gets turned on whenever the output voltage is higher than the set voltage. In some cases for a limited amount of time to prevent overheating when external power is fed into the PSU. Some PSUs are even designed to have this DC load active when power is lost to make sure the output voltage always goes to zero.

[Smokey]

Here is another discussion about Rigol supplies:

https://www.eevblog.com/forum/testgear/rigol-dp832-remote-command-update-rate/msg4593808/#msg4593808



As noted, these can't discharge so don't expect great settle times when going to a lower voltage with light load.

[Bosray]

Unfortunantly i dont have anything to measure the input capacitace of the device but the load is between 20mA at idle and 120mA at full load. But i have plenty of load resistors i can attach in parallell with the device so i can experiment to get the down programming to acceptable times i think. The command processing time i must admit i did not think so much about since we have so many powerful MCU these days so i just assumed it was a non issue, but from what you guys are saying the Siglent SPD3303X-E is the better choice. Up programming with such a small load should also be acceptable then. But can anyone confirm if the Easypower allows me to set steps at 0.1s?

[2N3055]

Many do not understand that most of these modern PSU, despite all the screens, software and such, are still 1-quadrant devices and they will have limited dynamic programming behaviour that will vary significantly depending on circumstances.
Only guaranteed spec will be load regulation response time.

This.
In order for a power supply to quickly follow a programmed curve, it must be able to absorb load.Otherwise, its behavior is highly dependent on load, which makes sense when you think about it.

That is why a small DC load to sink current has been a standard on many higher end power supplies for a long time. Keysight PSUs have had this feature since the HP days. This DC load gets turned on whenever the output voltage is higher than the set voltage. In some cases for a limited amount of time to prevent overheating when external power is fed into the PSU. Some PSUs are even designed to have this DC load active when power is lost to make sure the output voltage always goes to zero.

I am going to repeat:

Rigol DP831 has down programer. (that is how it is called, not load).
Siglent has it also.
Many cheaper designs also nowadays.

But those down programmers, even on Keysight (check the schematics available), are not symmetrical to main power regulator.  They just provide some bleed to discharge output capacitance to some degree.
What you need for a 2 -quadrant device is proper push-pull output stage, like on a audio amplifier.  With proper bidirectional current limiting and such.
Then you can have symmetric charge-discharge.

Funny enough, there are no devices like that on market (if we disregard some used devices and specialistic devices).
It is either PSU like we are discussing here or full blown 4-quadrant SMU.

Also, sometimes it is not even good to have down programming capacity, because you might have battery on the output that you don't want to discharge.

That is why we need to speak about how things work so people can make proper choice.

[nctnico]

Where it comes to batteries: you never connect those to a PSU without a series diode. In case of a failure, the result will be catastrophic if the battery can push power into a PSU. To make matters worse, some PSUs have a crowbar circuit inside which kicks in when the output voltage remains above the set voltage in a fault condition. I've seen the end result of what happens when the crowbar kicks in with a battery attached directly. Rather messy.

[2N3055]

Where it comes to batteries: you never connect those to a PSU without a series diode. In case of a failure, the result will be catastrophic if the battery can push power into a PSU. To make matters worse, some PSUs have a crowbar circuit inside which kicks in when the output voltage remains above the set voltage in a fault condition. I've seen the end result of what happens when the crowbar kicks in with a battery attached directly. Rather messy.

Sometimes DUT has badly designed battery backup architecture. I've seen those.
And yes, you are right, crowbar is another VERY fun way to explode things.

That is why both you and I always warning people not to charge batteries with lab PSU or to at least put serial diode at output if you really need to do it. You can compensate 0.5 V or whatever is diode drop by simply raising voltage of PSU.
If you want more precise charging, then use proper charger.

[2N3055]

Unfortunantly i dont have anything to measure the input capacitace of the device but the load is between 20mA at idle and 120mA at full load. But i have plenty of load resistors i can attach in parallell with the device so i can experiment to get the down programming to acceptable times i think. The command processing time i must admit i did not think so much about since we have so many powerful MCU these days so i just assumed it was a non issue, but from what you guys are saying the Siglent SPD3303X-E is the better choice. Up programming with such a small load should also be acceptable then. But can anyone confirm if the Easypower allows me to set steps at 0.1s?

If you don't care about directly measuring current consumption with PSU, additional loading might help you achieve dynamics you need.
I mean you can still measure consumption, you will just need to subtract additional load. Which can be easily calculated from resistance and voltage.

I am sorry, I don't have PSU in question so cannot test it for you.

[mawyatt]

Believe the GW Instek GPP-4323 allows 2 independent channels to be configured as a load rather than a source. Since it has 4 independent channels total, this might prove useful altho we've never used such.

Edit: Just for fun set it up like shown. Load can be configured as CV, CC or CR. CH1 as 100Ω Load, CH2 as 20Ω Load, CH3 as 2V Source  and CH4 as 10V Source. CH1 Load connected to CH4 Source and CH2 Load connected to CH3 Source. Seems to work OK

Best

[2N3055]

Believe the GW Instek GPP-4323 allows 2 independent channels to be configured as a load rather than a source. Since it has 4 independent channels total, this might prove useful altho we've never used such.

Best

Yes, but it can be load or PSU. Not a 2-quadrant source, IIRC..

[mawyatt]

Believe the GW Instek GPP-4323 allows 2 independent channels to be configured as a load rather than a source. Since it has 4 independent channels total, this might prove useful altho we've never used such.

Best

Yes, but it can be load or PSU. Not a 2-quadrant source, IIRC..

See edit above. If you parallel the Source and Load then the result can Source or Sink levels set independently.

So you can get 2 independent Source/Load combos (2-Quadrant) within the supply. Sure these are limited to low current/voltage/power usage but might prove useful sometimes!! 

Best

[2N3055]

Believe the GW Instek GPP-4323 allows 2 independent channels to be configured as a load rather than a source. Since it has 4 independent channels total, this might prove useful altho we've never used such.

Best

Yes, but it can be load or PSU. Not a 2-quadrant source, IIRC..

See edit above. If you parallel the Source and Load then the result can Source or Sink levels set independently.

Best

Yeah, sure, but good luck doing that dynamically.
OP needs to sweep/step through multiple power source voltages. I presume he/she is looking into DUT behaviour through some ranges of supply voltages.  He needs steps to transition through in 100 ms.

I guess you could send separate command to raise voltage limit on load and then raise voltage on source. And lower voltage on source and then follow up with with lowering load voltage.

But there is better idea to get much better dynamic behaviour. A parallel regulation.

You need current limited voltage source set to max needed voltage. You connect a load in parallel set to work in constant voltage (CV) mode.
And then you program the load CV voltage.
Load will serve as fast clamp (they react in 50s of µs ranges), and will keep constant voltage. Parallel regulation.

Down programming will be much faster than what PSU can do (if you use nice load that can sink 30 A), and ramp up will be as fast as PSU can do unloaded output regulation.

If you need DUT current consumption you will need to get data from both PSU and load and subtract or measure it separately, inline with DUT.

I will try on my DP321 and Maynuo M9812.

Best,
 Siniša

[mawyatt]

Yeah, sure, but good luck doing that dynamically.
OP needs to sweep/step through multiple power source voltages. I presume he/she is looking into DUT behaviour through some ranges of supply voltages.  He needs steps to transition through in 100 ms.

I guess you could send separate command to raise voltage limit on load and then raise voltage on source. And lower voltage on source and then follow up with with lowering load voltage.

But there is better idea to get much better dynamic behaviour. A parallel regulation.

You need current limited voltage source set to max needed voltage. You connect a load in parallel set to work in constant voltage (CV) mode.
And then you program the load CV voltage.
Load will serve as fast clamp (they react in 50s of µs ranges), and will keep constant voltage. Parallel regulation.

Down programming will be much faster than what PSU can do (if you use nice load that can sink 30 A), and ramp up will be as fast as PSU can do unloaded output regulation.

If you need DUT current consumption you will need to get data from both PSU and load and subtract or measure it separately, inline with DUT.

I will try on my DP321 and Maynuo M9812.

Best,
 Siniša

You can program CH1 and CH2 for dynamic behavior, so in this case the Load can be Dynamic. Altho we've never used any of this, seems quite a bit of flexibility built in the GPP-4323.

Edit: Here's what the dynamic menu looks like, it's available for CH1 and CH2 only.

Best

[bingo600]

But those down programmers, even on Keysight (check the schematics available), are not symmetrical to main power regulator.  They just provide some bleed to discharge output capacitance to some degree.
What you need for a 2 -quadrant device is proper push-pull output stage, like on a audio amplifier.  With proper bidirectional current limiting and such.
Then you can have symmetric charge-discharge.

I think my My HP/Ailent 6632B'is are real 2 qudrant.
I can use them as a load, if i ie. set current tto -2A.
But they also warn about excess capacitance on the dut, could cause damage to the psu if pressing off.

Edit: Written on a tablet , running Debian 13 - Gnome & FF
The OSK is  not in my fingers yet 🙄️

[2N3055]

I think my My HP/Ailent 6632B'is are real 2 qudrant.
I can use them as a load, if i ie. set current tto -2A.
But they also warn about excess capacitance on the dut, could cause damage to the psu if pressing off.

Edit: Written on a tablet , running Debian 13 - Gnome & FF
The OSK is  not in my fingers yet 🙄️

Yes 6632B is a 2-quadrant source measure unit. If you look into service manual you can see it has symmetrical push-pull output stage. Looks like output of audio amp.

But there is not something like that in current mass of PSU being sold as new for some normal price.
If you have one, take good care of it.