How to interpret power supply specifications

[sja]

I am trying to understand the specifications of bench power supplies. Let me take as an example the Siglent SPD4000X series.

The datasheet for these PSU says (CV mode):
(A) Line regulation 0.01%+2 mV
(B) Load regulation (2W mode) 0.01%+10 mV
(C) Voltage programming/readback accuracy +/- (0.03% of reading+10 mV)
(D) Voltage programming/readback resolution 1 mV

To measure potential differences lets assume that I will use a perfectly accurate DMM.

Lets suppose that I set a channel to 5 V with a current limit of 3 A. I now connect a load taking a steady 2 A.

Spec (C) says that the perfect DMM measurement of the channel's voltage will lie in the range 4.98850 - 5.01150 V (neglecting transients).
Lets say it is 5.01100 V. If I repeat the measurement at ANY time afterwards then spec (A) says that the DMM will measure a voltage of between 5.00850 and 5.01350 V (neglecting transients)

Is this the correct interpretation of the datasheet?

[tooki]

No.

(A) and (B) describe the actual regulation performed by the power supply. This is what you’ll measure with an external DMM.
(C) and (D) describe the accuracy and resolution of the power supply’s own display.

[sja]

Thanks for the response.

Yes indeed but when I set the voltage to 5.0000 V I am using the display so a resulting real voltage of 5.0110 V would be in spec. Right?

Having measured the real voltage, then (A) and (B) determine how that real voltage can fluctuate as a result of regulation. Isnt that right?

[jwet]

Tooki's interpretation of the spec's is correct.  Some of this stuff gets almost existential and is why test engineer's make the big bucks and why real source/measure units (SMU's) cost so much more than power supplies.  SMU's account for and minimize all these error sources.

If you set your "imperfect" supply so its display shows 5.000v.  The "actual" voltage error at the output is the sum of the errors in the real output voltage and the error of reading back the voltage.  There are four error sources as you list them, and when summed, give you .05% +-23 mV.   You could make an argument that these things are uncorrelated and may be, so the error is  the RMS sum at least the uncorrelated errors- square root of sum of squares.  However they are probably correlated, by virtue of likely sharing a reference.  As is, you can't separate the errors.  Your perfect voltmeter could eliminate the readback error and give you the true output.  You'd have to track this back through the error sources to quantify them.

The "ANY" time after is also a question.  The line voltage varies significantly- on the order of 10%, you'll see .01% plus 2 mV of this- somewhere around .1%.

I don't know what your ultimate goal is.  Chasing stuff below .01% gets hard (and expensive usually). 

There is a group of very pedantic people called "volt nuts" that play with this kind of stuff.  Its the realm of 8 digit voltmeters and Josephson junctions, etc.  Really interesting stuff.  The other related bunch is "time nuts" that try to get down to "perfect" time.

After 40+ years of analog design and instrumentation development, I have come to the conclusion that its basically an 8 bit world.  If you're really careful, you can get to 10 bits but the real world isn't really that accurate.  There are dynamic range issues but 8 significant bits or best case 3 significant digits is the real world.  Tracking leap seconds is the way to madness.

[MrAl]

I am trying to understand the specifications of bench power supplies. Let me take as an example the Siglent SPD4000X series.

The datasheet for these PSU says (CV mode):
(A) Line regulation 0.01%+2 mV
(B) Load regulation (2W mode) 0.01%+10 mV
(C) Voltage programming/readback accuracy +/- (0.03% of reading+10 mV)
(D) Voltage programming/readback resolution 1 mV

To measure potential differences lets assume that I will use a perfectly accurate DMM.

Lets suppose that I set a channel to 5 V with a current limit of 3 A. I now connect a load taking a steady 2 A.

Spec (C) says that the perfect DMM measurement of the channel's voltage will lie in the range 4.98850 - 5.01150 V (neglecting transients).
Lets say it is 5.01100 V. If I repeat the measurement at ANY time afterwards then spec (A) says that the DMM will measure a voltage of between 5.00850 and 5.01350 V (neglecting transients)

Is this the correct interpretation of the datasheet?

Hi,

Some good replies here already I'll just add a little more to summarize.

It looks like the power supply has three sources of error, A, B, and C, and D is just telling you that you SHOULD be able to step the output in increments of 1mv.
Since D is different than the others and we can assume D is close enough, the main sources of error come from the first three.

If we look at A, B, and C, we see they all have errors in the form:
err=K1*Vset+K2

Since they all have the same form, we can combine them into one lumped expression:
MaxTotalError=A1*Vset+A2

Adding all the K1's and all the K2's to get A1 and A2 we end up with:
MaxTotalError=0.0005*Vset+0.022, where MaxTotalError is in plus or minus volts.

We can then also form a little table with each entry as [Vset, error, percent error] with all errors plus or minus:
1v, 0.0225v, 2.2500 percent
5v, 0.0245v, 0.4900 percent
10v, 0.0270v, 0.2700 percent
15v, 0.0295v, 0.1967 percent
20v, 0.0320v, 0.1600 percent
25v, 0.0345v, 0.1389 percent
30v, 0.0370v, 0.1233 percent

This is following the way power supplies are specified.  If that manufacturer reported something in an unusual way, you'd have to get that information from them.
Feel free to check over the numbers it's always good to do that if this is important.

[sja]

Thank you all very much for your responses and indeed the time you spent on composing them.  I now think I understand how I should interpret those lines in the specifications.

I am certainly not a "volt nut" but I can get distracted down some interesting looking rabbit holes every now again.  I'm an electronics beginner who needs to replace my single output Korad power supply with one having at least 3 outputs but with sensing. That's why I was trying to understand the specifications.

[Smokey]

Also be careful where you are measuring the voltage (and where the supply is measuring the voltage).  Unless you are using sense terminals, the voltage at the load will almost certainly not be what the internal supply is measuring itself, which is probably measuring at the terminals.  Considering a power supply is designed to supply some amount of current the leads will always have some voltage drop at the load.