SMPS Bench Supply testing

[EdStraker]

I have an Eventek KPS303DF PSU works great, no issues. I use it for powering and testing low voltage applications from2v-18v and rarely up to 24v or beyond range. As it is a SMPS and as usual is noisy on the output, which in turn makes testing a lot of circuits problematic. I am getting tired of having to remember to put decoupling caps on the test circuit which may or may not have them until I set it up and realize the circuit doesn't respond as it should due to noise.

I would like to know how to "properly" and "simply" test the supply for it's output noise level. Would appreciate so guidance on that.

Secondly, what can be done to reduce noise to a manageable level? I KNOW it can't be as simple as dropping a decouple cap internally across the V+ and GND Terminals. Not shooting for Lab Grade just a usable range. Any thoughts?

 

[danadak]

You could :

1) Put a LC filter on output of supply.
2) Put an adjustable 3 terminal regulator on output of supply, set for each case. They have a
lot of noise rejection, although not at HF.

LM317 :

3) To test noise use Zeitnitz PC sound card scope, its FFT facility. Good to about 20 Khz.

Cautions : Look at post # 2 for cautions - https://www.edaboard.com/threads/typical-beginner-microcontroller-questions.399717/#post-1720496

Regards, Dana.

 

[EdStraker]

I should said in my original post I do have a pocket O-Scope and 100w 8 Ohm Wire Wound Power Resistors (used to use them for tube amp loading) I have looked around for beginner type tutorial info on how do this test but really didn't come up with anything useful in the How and Why department and then what to do about it. All I know is that the you do need a load for the initial test. I have zero knowledge on Inductors.

 

[danadak]

The Zeitnitz software can do a spectrum plot, it can generate a swept sine and plot
the freq response at the load.

You can do a crude test by using a sine generator and hand plotting
some points observing with your pocket scope the pk-pk amplitude
of the waveform at load.

LC filters -

https://www.analog.com/en/technical-articles/switching-regulator-noise-reduction-with-an-lc-filter.html

https://www.analog.com/en/technical-articles/designing-second-stage-output-filters-for-switching-power-supplies.html

Regards, Dana.

 

[Tony Stewart]

Beware of high Q wire resonance effects on pulse loads and SMPS noise amplification.

interactive simulation ( change RLC etc)

 

[EdStraker]

I can see this has already gotten way beyond my pay grade so speak.
Was just wondering if anyone had civilized a cheap PSU for digital circuit work without getting deep in the weeds.

It's just really annoying to have to test a circuit, most of the time designed for battery power on my PSU and have the circuit malfunction due to ripple noise and then having to add a filter cap to smooth the power in order to have the circuit function correctly. I have this issue quite often with anything that is an an oscillation type circuit or MCU's that was originally designed to run from Batt Pwr.
Guess I should have payed more attention and bought a Linear PSU.

 

[Nigel Goodwin]

I can see this has already gotten way beyond my pay grade so speak.
Was just wondering if anyone had civilized a cheap PSU for digital circuit work without getting deep in the weeds.

It's just really annoying to have to test a circuit, most of the time designed for battery power on my PSU and have the circuit malfunction due to ripple noise and then having to add a filter cap to smooth the power in order to have the circuit function correctly. I have this issue quite often with anything that is an an oscillation type circuit or MCU's that was originally designed to run from Batt Pwr.
Guess I should have payed more attention and bought a Linear PSU.

Try posting a complete circuit of a design that didn't work, it sounds like you may just be missing out essential smoothing and decoupling capacitors? - just because something is battery powered you can't leave them out, and in fact a decent sized reservoir capacitor across the power rail is absolutely essential in many battery designs (or the batteries have a very short life).

 

[danadak]

If you spring, for a transformer based linear supply (read isolated), which are quite handy,
I would advise :

1) Dual supply. Some have additional 5V supply which is handy.
2) Each with adjustable V and adjustable I
3) Ability to series the two for higher V, parallel for higher current
4) Tracking between two, for example generating split supplies. You set one supply
and the other becomes a slave of opposite polarity.
5) Digital display of V and I both supplies
6) Range, 0 - 20 to 30V, 1A or more, depends on your design activity

They are worth the investment.

You can buy ones with meters, and add low cost V and I modules to do the display work,
eg. save money overall.

Geekcreit® Mini Digital Voltmeter Ammeter DC 100V 10A Voltmeter Current Meter Tester Blue+Red Dual LED Display

Only US$7.49, buy best geekcreit® mini digital voltmeter ammeter dc 100v 10a voltmeter current meter tester blue+red dual led display sale online store at wholesale price.

www.banggood.com

Quite a few DIY variants on youtube....and using PC supplies.....

Regards, Dana.

 

[EdStraker]

Try posting a complete circuit of a design that didn't work,

Nothing "Didn't work" per se, they worked fine on battery power, but on the PSU power they circuits were flipping out (like and overload condition) untill I dropped a 1uF cap across the pwr input.

1) Dual supply. Some have additional 5V supply which is handy.

Why a dual supply? Something inherently better in design?

I really like the one I've got, has a 5v USB pwr port on the front and does everything I need it to do with a small footprint. Was hoping that someone had over the years came across a quick fix to "civilize" these for sensitive circuits whitout having to overhaul the darn thing.

You can buy ones with meters, and add low cost V and I modules to do the display work,
eg. save money overall.

Thanks for that heads up, Guess I'll have to look into it.

 

[danadak]

Design considerations for logic circuits (including CMOS) :

https://www.ti.com/lit/an/sdya002/sdya002.pdf

Power supply will not fix this unless its very noisy, you always need local
bypassing. Usually a bulk cap line a electrolytic / tantalum, and a ceramic disk...


Regards, Dana.

 

[EdStraker]

Power supply will not fix this unless its very noisy

That's kinda why I wanted test the PSU to see just how noisy it is.

BTW: was just looking around at potential Linear PSU's. Unless you are ready to spend big bucks, I find a LOT of the budget variety stating they are Linear but when you actually look at the specs they are actually SMPS. Buyer Beware.

Guess I'll just have to live with it.

 

[Nigel Goodwin]

Nothing "Didn't work" per se, they worked fine on battery power, but on the PSU power they circuits were flipping out (like and overload condition) untill I dropped a 1uF cap across the pwr input.

That's "didn't work" - post the schematic so we can see if that's the issue.

But fitting a 1uF across the power input strongly suggests your circuit is missing that vital component - which MUST be there regardless of battery on not.

It seems like you 'might' be blaming the PSU when the fault is actually missing components from your circuit.

 

[EdStraker]

It seems like you 'might' be blaming the PSU when the fault is actually missing components from your circuit.

Only one was mine the others were preassembled units which contained SMD PIC MCU's
Was testing them for a friend who suspected they might be damaged. They were not.

 

[Tony Stewart]

For those who have not taken Transmission Line Theory
A wire or a cable is not just a wire, it is a high frequency resonator.

Beware of high Q wire resonance effects on pulse loads and SMPS noise amplification.

interactive simulation ( change RLC etc)

View attachment 141672

Every pair of wires is actually equivalent to lumped inductor and shunt cap or maybe defined as per unit length. The "characteristic impedance" depends on the ratio of sqrt{L/C} while the resonant frequency depends on delay time with harmonics 1f, 3f, 5f, 7f .

Here is a fairly high impedance pair of wires modeled as distributed LC and the same and as the transmission line above it. It is 2.6m and has a resonance at 30 MHz, 90, 150 ,etc.

Here I attempt to demonstrate to you without adding a filter cap or RC filter but just adding a series R or load R or both, how this alone affects the high gain of resonance. This is why we call R's as damping resistors. They don't look like it but act like shock absorbers to attenuate the peaks above 0dB (unity gain)

So when you have any high frequency stimulation or load pulses with energy near these peaks, the noise is amplified. In light loads it is especially high 60 dB peaks is 1000x gain in noise. Each vertical division is 20 dB.

This picture is worth a thousand words.

 

[danadak]

That's kinda why I wanted test the PSU to see just how noisy it is.

BTW: was just looking around at potential Linear PSU's. Unless you are ready to spend big bucks, I find a LOT of the budget variety stating they are Linear but when you actually look at the specs they are actually SMPS. Buyer Beware.

Guess I'll just have to live with it.

You could always do one with an old Microwave oven transformer and
a discrete design off the web. You tubers doing a lot of microwave oven
transformers projects.


Regards, Dana.

 

[ChrisP58]

I have an Eventek KPS303DF PSU works great, no issues. I use it for powering and testing low voltage applications from2v-18v and rarely up to 24v or beyond range. As it is a SMPS and as usual is noisy on the output, which in turn makes testing a lot of circuits problematic. I am getting tired of having to remember to put decoupling caps on the test circuit which may or may not have them until I set it up and realize the circuit doesn't respond as it should due to noise.

I would like to know how to "properly" and "simply" test the supply for it's output noise level. Would appreciate so guidance on that.

Secondly, what can be done to reduce noise to a manageable level? I KNOW it can't be as simple as dropping a decouple cap internally across the V+ and GND Terminals. Not shooting for Lab Grade just a usable range. Any thoughts?

Here is a pretty good video on understanding and measuring power supply noise.

But, as others have said, I expect that power supply noise may not be the cause of the problem. So let's try a couple of tests.

Test one. Your circuit works well with a battery, but not with the bench supply. Question. What is the lead length from the battery to the circuit vs. from the power supply. Test how it works if you put the same length and size of wires between the battery as you have from the power supply.

Test two. The circuit works from the power supply if you add a 1uF capacitor. I'm assuming that the capacitor is added at the input to the circuit. Test how it works if you move the capacitor to the supply output terminals.

These test will give you a better understanding of whether it's really power supply noise or not.

Shoutout to Tony Stewart's post #14 for the deep dive on what might be going on.

 

[Nigel Goodwin]

Only one was mine the others were preassembled units which contained SMD PIC MCU's
Was testing them for a friend who suspected they might be damaged. They were not.

PIC's (assuming they were actually PIC's?) are VERY tolerant of their power supply, so the PSU would have to be horrendous to cause problems.

However, you still don't seem to want to post a schematic of your circuits, so we can potentially diagnose your issue.

 

[danadak]

One test you can do if you have a DSO, often quite revealing, is look
at supply rail(s) with scope set to infinite persistence set on display
mode, and see what pk-pk noise is. Probing techniques :

https://download.tek.com/document/02_ABCs-of-Probes-Primer.pdf

Be careful probing SMPS, you can easily damage scope if you do not
observe ground loops and isolation issues :

https://download.tek.com/document/51W_10640_1.pdf

https://download.tek.com/document/Voltage%20on%20Power%20Supplies_App-Note_51W-60161-3.pdf

Another test is to set triggering for a narrow pulse width, like 50 ns or so,
and probe various inputs to see if it captures any transient.

Another is set triggering for a runt pulse violation on an input. Agilent
and Tektronix have videos discussing triggering detective work.


Regards, Dana.

 

[EdStraker]

Test one. Your circuit works well with a battery, but not with the bench supply. Question. What is the lead length from the battery to the circuit vs. from the power supply. Test how it works if you put the same length and size of wires between the battery as you have from the power supply.

Test two. The circuit works from the power supply if you add a 1uF capacitor. I'm assuming that the capacitor is added at the input to the circuit. Test how it works if you move the capacitor to the supply output terminals.

Test One: Lead length plays an effect but not what I expected. On PSU - no effect 6"-18" leads. On Batt - 12"+ malfunction (no cap) reacts like a weak signal. 6" lead or less (no cap) normal function.

Test Two:

1uF and 10uF Ceramic and Electrolytic tested. On PSU, cap at circuit input = correct function. On PSU Output terminals= no effect. (All lead lengths)

1uF and 10uF Ceramic and Electrolytic tested. On Batt. cap at circuit input= correct function at all lead lengths.

PIC's (assuming they were actually PIC's?) are VERY tolerant of their power supply, so the PSU would have to be horrendous to cause problems.

Checked it under a scope it is indeed a PIC

 

[Nigel Goodwin]

Checked it under a scope it is indeed a PIC

View attachment 141735

Cool - you do find PIC's all over the place

BTW, nice picture.