0
\$\begingroup\$

I'm trying to debug very strange behavior I'm seeing while I'm trying to breadboard out a circuit to test the passive values for an LED driver I'm designing. I am designing for a 28 V input and a 25.4 V / 0.36 A output.

The issue I am encountering is that the breadboarded circuit only outputs ~18 V at 0.01 A. However, and this is the strangest part, as I was probing around the board with my oscilloscope, I accidentally hooked the probe up backwards, with the ground clip on Vin and the probe tip on the SW output of the LM3402 IC, and the circuit suddenly started outputting the correct power.

Some additional observations: I don't actually need to connect the probe at all, just touching the ground clip of the probe (even if the oscilloscope is powered off) to the SW pin causes the driver to output 360 mA. However, if I just ground the SW pin, the driver emits magic smoke. So it has something to do with earth referencing the SW pin.

[edit] Additional info: connecting the earth ground of my power supply to the SW pin has the same effect. I have tried connecting the earth ground to the circuit ground, which had no effect. It's literally just connecting the SW pin to earth ground.

This the schematic of my circuit:

enter image description here

This is what I see if I probe the SW pin with the ground clip on the power input ground:

enter image description here

And this is what I see if I reverse the oscilloscope probe, which results in the driver suddenly working as designed:

enter image description here

What is it about the oscilloscope probe that suddenly makes this circuit work?

\$\endgroup\$
7
  • 1
    \$\begingroup\$ looks like the trigger level is not set correctly in the first example \$\endgroup\$
    – jsotola
    Feb 4, 2022 at 0:47
  • \$\begingroup\$ It's actually set in the same place for both samples. There's just a tremendous amount of noise and ringing in the first sample. \$\endgroup\$
    – flimsy
    Feb 4, 2022 at 0:56
  • \$\begingroup\$ Stray or parasitic reactance from loose non-twisted pair wires getting creating positive feedback can make any design unstable. This is especially true when for DCDC converters with isolation and no common mode EMI suppression. Using the scope earth grounded probe lowers the common-mode impedance and will alter the leakage paths of common-mode noise. Although normal one usually connects the probe ground to 0V and not SW. Use short connections , observe orientation of components or wires with switched currents and used twisted pairs where possible. for shielding. \$\endgroup\$ Feb 4, 2022 at 7:38
  • 2
    \$\begingroup\$ These kind of problems are always caused by a missing ground in your circuit. My best tip for fixing breadboard problems is: throw them in the garbage. Then pick up soldering. \$\endgroup\$
    – Lundin
    Feb 4, 2022 at 7:40
  • 1
    \$\begingroup\$ ”the breadboarded circuit” Sorry to be the bringer of bad news, but you can’t successfully build an SMPS on a breadboard. You need to design a proper PCB with ground plane. \$\endgroup\$
    – winny
    Feb 4, 2022 at 14:06

3 Answers 3

4
\$\begingroup\$

Breadboards, especially cheap ones, are notorious to have bad connections, might be something that doesn't have good contact on the breadboard that is fixed by the physical touch of the probe.

Also it is not trivial to get decent performance out of a switching regulator on a breadboard due to high inductance in every connection, especially if using the standard 10cm jumper wires to do all the connections. Extra care must be taken to add some extra decoupling (close to the regulator pins), as short leads as possible etc.

Not like the typical mess one often sees on breadboards. messy breadboard

By the way, why would anyone draw a circuit diagram like that? Off-page connectors are to be used when going to another schematic page (very occasionally within the same page). This schematic is so much easier to read (could be tidied up even more, but I can't bother in MS Paint): Improved schematic Now that I can actually read the schematic, I can see that the only decoupling capacitor you have on the supply voltage is a large (I assume) electrolytic, 470µF. You should add some smaller ceramic capacitor that can handle high frequency better, something like 0.1-1 µF to the regulator, as close as possible to the VIN and GND pins.

\$\endgroup\$
1
  • \$\begingroup\$ I have actually tried a number of different capacitor combinations on the input. I've tried a 1µF, a 4.7µF and the 470µF I have on the schematic, plus various combinations of the above. Adding input capacitors actually results in the current output decreasing, which is also really puzzling. I also know that breadboards are super noisy, though I have managed to breadboard out working supplies with other ICs. They weren't perfect but they worked well enough to verify that my design would work. \$\endgroup\$
    – flimsy
    Feb 4, 2022 at 19:18
1
\$\begingroup\$

The clip on the scope is connected to AC power ground. It is also probable that you have other equipment such as the power supply grounded as well.

You may be completing a power path required by the circuit.

\$\endgroup\$
3
  • \$\begingroup\$ But what specifically about the oscilloscope probe makes it work? Simply connecting SW to GND would cause a short (I actually tried this and it killed the IC). Is it capacitance? Resistance? Some sort of RC filtering effect? \$\endgroup\$
    – flimsy
    Feb 4, 2022 at 0:55
  • \$\begingroup\$ @flimsy - Not Sw to ground. I would first just connect the ground clip to Vin (don't touch the probe anywhere) and see what happens. It may be you are completing the power path through the scope. \$\endgroup\$ Feb 4, 2022 at 1:03
  • \$\begingroup\$ So this is very strange. Here's what I'm seeing: clip to GND: 20mA boost to output current clip to VIN: current jumps by 5mA clip to SW: current jumps to the intended level (360mA) probe to SW: current falls by ~50mA probe to GND: current jumps by 5mA prove to VIN: no change \$\endgroup\$
    – flimsy
    Feb 4, 2022 at 1:26
1
\$\begingroup\$

Only guesses are possible because we do not have your setup in our hands. Here's one:

At first your scope obviously doesn't have galvanic contact to the GND of your +28V DC supply at the same time. If it had, the +28V would have shorted. Connecting the ground of the scope to pin Vin of the IC inserts only some stray capacitance.

That stray capacitance probably is just what's needed to compensate too long wire or poor contacts or too high inductance of of your capacitor Cin1 (=470uF).

Test that guess by inserting a 1uF ceramic capacitor with short (=less than half inch each) wires straight between pins GND and Vin of the regulator IC. Your Cin1 should be as near the regulator IC.

BTW using breadboard for switching circuits is a gamble even in case there's no poor contacts. The stray capacitances are equivalent of having unwanted capacitors between the circuit nodes. The crisscrossing wires do the same and they also work as inductors. As already said by others, a proper soldered circuit with minimal wire lengths and ground plane is the right way to make something predictable. IC manufacturers often give layout drawings for a good circuit board. It should be taken seriously and your IC is not an exception. See pages 29...30 in your linked datasheet!

\$\endgroup\$
2
  • \$\begingroup\$ This work is all in preparation for getting a PCB printed. I'm breadboarding this out because the design math for this IC is really annoying, and I can't actually tell if my input/output power is possible. As I noted above, I've tried a few combos of input capacitors, but adding capacitors on the input actually results in a lower current out. I'm current running it with no input filtering. I'm also still super confused as to why connecting the SW pin to earth ground would have the effect it is having. If anything it should fry the IC. \$\endgroup\$
    – flimsy
    Feb 4, 2022 at 19:20
  • \$\begingroup\$ Frying the IC: it happens due overcurrent. There can be a high capacitive loading route scope GND -> scope power supply -> mains AC -> your 24VDC power supply -> GND of your +24VDC. Feeding capacitance with voltage pulses causes a current rush. The capacitances inside devices are intentional and they are there to reduce unwanted radio noise emissions. The makers of the devices would say: It's not our fault that someone makes harmful connections which in addition are 100% useless. \$\endgroup\$
    – user136077
    Feb 4, 2022 at 20:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.