Electronic load and ripple measurements

[Darkover]

And what do you think happens when you connect these two?
This happens when the current drawn by your load is higher than the power supply’s set value.
The two will fight for control.

I agree by 101% and I have developed this by myself. 
You can not connect two control loops where a least one is unknown and expect that it is always stable. It is hard work that it is stable often. Especially when the power supply is not well tested because it is still in the developing process. And it becomes harder when the supply and the load should faster because you like it faster, (for example for modulation) And if a problem happens, who is guilty? The power supply or the load? At the end a resistor is more predictable.

Olaf

[Hydron]

Did you mean a ripple in the water?

[Kleinstein]

Sometimes there is an instability caused by an interaction between the load, and what it is connected to. An this even happens every now and then by expensive / commercial loads.

Take for example a power supply with a current limit at it's output. What will happen if you increase the load until the power supply trips into current limit mode?

Some loads can be switched between either a constant current mode, or a variable resistance mode. This may help with preventing such problems.

There are 2 possible issues with an electronic load:
One is from the impedance of the load: With controling the current and the BW limit in the control loops one gets a kind of capacitive impedance. Depending on the details of the load control loop, the phase angle can be rather close to 90 degrees (like an ideal capacitor with 0 ESR) or even slightly negative. Constant voltage regulators may not like a large capacitive load with low ESR and especially not more than 90 deg. phase shift (corresponds to negative ESR) and this can than cause oscillation.
A way to work around this issue is to have some normal resistance in series to the electronic load. In good electronic loads the internal fuse can do that job.

The other possible problem is when the supply changes to constant current or similar protection. Than the voltage at the load can get too small and the regulator in the load can get some wind up. Once voltage comes back the current overshoots. A good load should be save with that, e.g. by changing to a constant resistance more when the voltage is too low (e.g. less than the minimal drop plus a little reserve).

[Darkover]

I’ve created several designs for DC loads, and I’ve designed most of them to be suitable for AC testing as well, since that was important to me at the time.

Just because I am curious. What was the maximum bandwith and the idea of your design? I think somewhere in the audio range?

When I did it, I made a bandwith somewhere between 100k and 1Mhz (from my memory), because my idea was to put fast loadchange
to switching power supply. But the problem was that it is impossible to assign a bad behavior to my switching supply or the load or the combination. So later I went to switching a resistor....
And even for testing only battery, a short (for example 1ms) loadswitch and calculate Ri was easier than do a slow discharge.

Olaf

[blackdog]

Hi,

Where should I start...

I use various pieces of equipment to measure noise in analog power supplies.
In no particular order, some of the devices:

Audio Precision: Portable One Plus
Homemade 40dB/60dB low-noise preamplifiers for my scoops.
EULER: LFLNA-80
PRP1: Power Rail Probe V2.0
Quant Asylum: QU403
Battery Scoop: OWON and Micsig
Battery Function Generator: OWON
High-quality shielded cables
Shielded isolation transformer

Interference-free environment.
Many electronics workbenches are a nightmare when it comes to EMC interference.

LED lighting, a lot of equipment powered by switching power supplies are sources of broadband interference.
Since not all of my equipment can run on battery power, isolation transformers are sometimes needed to keep the noise level low enough.

When I want to measure really small noise level, P.A.R.D. signals, I grab the measuring equipment I need
and then sit in another room with a camping table, away from other sources of interference.

You'll also need some cookie jars to store your test device in.
Keep your cell phone away from your D.U.T. and your measuring instruments, including those belonging to your coworkers and/or family. :-)

I hope you didn't think that accurate low-level measurements were easy.
God can still hear the echoes of all the swearing caused by the nearly insurmountable EMC problems I encountered during my measurements.   

DC loads, whether modulated or unmodulated, when connected to the power grid, almost always inject a noise signal into your D.U.T.
If the DC Load uses a switching power supply, it is almost certain that you will experience this issue.
If the DC Load’s architecture uses PWM instead of a good DAC, that PWM will also inject interference into your D.U.T.

For power supplies that deliver no more than about 0.5 amps, use resistors as the load, and no, don't use spaghetti wiring!
You can then test with, say, two resistors—one for a 5% load—and then switch on a second resistor that provides an 85 to 95% load on the power supply.
That second resistor doesn’t need to be high-power, since you’ll be turning it on and off with a 10% duty cycle.
You can do this with an LMC555 and a small power transistor like the BD139.

And one more thing: power this circuit from a 9V battery and... don't use spaghetti wiring!
Every piece of wire acts as an antenna, a resistor, and an inductor.
I’ve learned a lot about this over the years; I gained a lot of practical knowledge while building illegal transmitters—yes, I’ve been a bad boy, too.

I was pulling my hair out when the transmitter started interfering with the audio inputs again.
It taught me to really think about the ins and outs of decoupling, RF-shielded enclosures, and how thin metal provides very little shielding against the 50 or 60 Hz transformer field.

Most manufacturers of DC loads do not produce them for currents ranging from 10 to 200 mA,
which would allow you to accurately measure the noise level from a linear power supply.
That is not the target audience for the affordable devices from Rigol, Siglent, Korad, and GW-Instek.

Here's an example of a small DC load powered by a 9V alkaline battery.
If you need a non-modulable DC load that doesn't cause interference, maybe the circuit below could serve as inspiration?
For example, build it into a Bimbox with some extra cooling.
Don’t fall for the misconception that it can also must handle 10 amps and 150 V, etc.
That’s not what it’s designed for!



The LM10 is an OpAmp and a 200 mV reference in a single IC, which also allows for a compact design and low power.
The way the LM10 is used here in the schematic is rated for a maximum of 5 amps, but at 24V that’s naturally 120 watts,
which won’t work with passive cooling in a Bimbox!
So replace R9 for a 1 Ohm power resistor so the max current wil be 500mA.

The red LED is a low current type,
I scale de 200mV reference i nde LM to about 454mV, de switch R5 and de 10K 10Turn potmeter P1 set the current.
Trimpot TR-1 trims the ofset of the opamp, use this at a low current setting.
As always use GOOD Cooling for the MOSFet, for the 500mA version a IRF540 is OK, he has an DC SOA profile that is good enough.

The 60-degree Clikson disables the gate drive when the temperature is too high,
and R8 ensures that the gate is then connected to the source so that no drain-source current can flow.

The fuse for the 500mA version can also be smaller, a 1-Ampere, fast-acting fuse will work.
R6 (4K7) and C3 (4n7) determine the frequency compensation.
The cutoff frequency has been set slightly lower here than I normally would, specifically to accommodate the IRFP064 used in the schematic,
which has large internal capacitances.
However, since this load is intended solely for DC behavior, the values of these components can remain the same for both MOSFET types.

For anyone who is concerned that the LM10 might inject noise into the D.U.T. due to its own noise:
I won’t stop you from choosing a different type of op-amp.
The OPA140, for example, is a good candidate; in that case, the offset trimpot can/must be omitted.

Happy building/thinking about EMC, Current Sources and measuring setups.

Kind regards,
Bram

PS
And one more thing... I'm a dyslexic Monkey who uses Deepl to translate my texts, so they might not always be clear.
If I make any silly mistakes, please let me know, and I'll fix them right away.

[blackdog]

Hi Olaf,

I have a day where I can do whatever I feel like—so you're in luck, because I love talking about my little projects. 
For high-bandwidth testing, I use a Jim Williams DC/Active load. The bandwidth is a few MHz.

The Schematic how i build de Jim Williams design.
The op-amp and the MOSFET must be connected to each other in an HF configuration.
Otherwise, it is not really possible to generate a clean current pulse in your DUT.

.

Connecting the OpAmp to the measuring resistor and the MOSFet, here you go, the shortest possible wiring.

.

A test setup, notice how short the connection is between the DC load and the linear power supply being tested.
Also note where the oscilloscope probe is connected, it's on the back of the banana plug terminals.
That’s the point where the sense wiring is also connected, and therefore the right place to measure how the control loops behave.
And not on the front panel, where the DC load is plugged in.
Those three to four centimeter long connections act as excellent inductors, what are you going to measure there, when the DC load is generating fast current pulses?
A lot of overshoot! Which the control loop of the power supply under test can’t do anything about.

.

Below is a link to Jim Williams' PDF, which covers a lot of ground, and he explains it much better than I do.

https://www.bramcam.nl/AL/an104f.pdf

Time for Lunch!

Kind regards,
Bram

[blackdog]

Hi,

It always makes sense to think about what you want to measure and how...
Why would you test a power supply with very fast pulses if those pulses don’t normally occur in the power supply you’re testing?

Testing with very fast pulses requires a carefully selected measurement setup.
Even if you don’t have to use Jim Williams’ fast pulse DC Load, you can still easily go wrong if you position your measurement cables incorrectly.

With a “normal” linear power supply, I temporarily attach a BNC connector to the point where the sensor wires are connected to the output.
I position the coaxial cable I connect to the BNC connector at a 90-degree angle to the current wiring.
This is usually necessary for pulse currents starting at about 1-Ampere.
If you don't maintain about that 90-degree angle, you'll induce interference in your test lead.

And then there’s this: you’ve tested a power supply for its dynamic behavior at the point where the sense wiring connects, everything there is OK, just as you want it to be.
Now think of a lab bench power supply; connect it to the device you’re testing.
What is the DC and AC Ri behavior after the cable reaches the device being powered?

The DC Ri is almost always higher due to the connecting cables and the contact resistance of the connections used.
When it comes to AC behavior, it gets even trickier, from what I’ve seen on YouTube, testers often use loose cables rather than twisted pairs when performing AC dynamic tests.
And when they use sense wires for compensation, these are sometimes connected to the current wiring, Yeah, that's easy, but it's not right.

And by that I mean mechanically attached, that’s not the right technique.
This wil works for DC, but it’s not a good idea for Dynamic AC testing.

For static DC loads with the lowest interference level, use short, twisted-pair wiring to connect to power resistors that are rated for the power required to perform these tests.
Mounting a 50-watt power resistor on a small piece of metal won’t work during longer endurance tests.
Those resistors always fail; it’s best to avoid the cheap Chinese knockoffs with only two mounting holes on heat sinks.
Again, use high-quality resistors mounted with thermal paste on a heat sink suitable for the application.

There are so many things you need to keep in mind when conducting proper tests, so you don't fool yourself. 

Kind regards,
Bram