3
\$\begingroup\$

Trying to repair a non-working Fiber to Ethernet Converter. When connected to power, the unit made a hissing/sizzling/scratching sound and the all LEDs glowed dimly. I isolated a small surface-mount choke on the input power side, filtering 12VDC input from wall-wart to input of main 3.3V voltage regulator. (Schematic below.) Removed this choke and replaced with a straight wire shunt. The unit powers up and works perfectly.

I don't have an inductance meter, but the choke measures about 0.1 Ohms, so it's not burnt out.

The voltage regulator is a LM2591HVT-3.3 (a 150kHz Buck switching regulator). I see a clear 150kHz signal at the input after replacing the choke with shunt. Before replacing, there was a very noisy signal at this point, almost impossible to see with my scope.

Three questions: 1) What actually failed inside this choke? Why does it "sizzle" when powered? 2) What possibly caused this failure?
3) Is it OK to replace this choke with a shunt? What problems may arise from this modification?

Output of power circuitry is now stable at ~3.4VDC and unit powers up normally.

Thanks!

Schematic:

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
  • 1
    \$\begingroup\$ Could be C1, C2 bad and they can't supply enough current, as you removed the choke it takes from adapter cap until it will overheat. \$\endgroup\$ – Marko Buršič Feb 1 '16 at 6:01
  • \$\begingroup\$ Here are scope measurements at M1 and M2. M1: is.am/ez and M2: is.am/ey . M3 is a flat line 3.4VDC. \$\endgroup\$ – Ryan Griggs Feb 1 '16 at 16:09
2
\$\begingroup\$

There are many options.

First of all, as said in the comments it's possible it wasn't the choke, but the caps that have gone bad. It's possible it was the choke after all.

The choke is there to get out the buck converter's switching noise spikes, so that:

  1. The adapter isn't overstrained (adapters are cheap and designed on the edge of well)
  2. To conform to EMI/EMC standards.

The second may not be an issue for you, the adapter will probably filter that out before it gets to the AC and possibly it will not couple from the DC wire onto anything else.

If it was the choke its inductance may have increased due to a mechanical default in the core. But you should also make sure the caps aren't dry by looking at the ripple if you put in a small extra resistance in the supply path.

Or just replace them anyway with a couple fresh ones.

\$\endgroup\$
  • \$\begingroup\$ I can definitely see the 150kHz spikes being injected back into the supply. See my comment above with scope screenshots. Also, at M2, I'm seeing a 12Vpp square wave, which is being filtered by L2 to produce a 3.4VDC flat line at M3. The regulator gets quite warm to the touch. So it's likely C1 is bad, because C2 is a tiny ceramic SMD cap? \$\endgroup\$ – Ryan Griggs Feb 1 '16 at 16:12
  • 1
    \$\begingroup\$ It's a large deal more likely to be C1, yes. Even if the ceramic was the only one going bad, it'd have less of an influence on the audible whineyness of filter components. \$\endgroup\$ – Asmyldof Feb 1 '16 at 16:21
  • 2
    \$\begingroup\$ It's nearly always the electrolytic caps in failing home network equipment. The ceramic cap is quite robust and only there to damp frequencies in the MHz range. When you replace C1, make sure the replacement is Low-ESR and matches or exceeds the ripple current rating of the failed capacitor. You can find the ratings in the data sheets. \$\endgroup\$ – Michael Karcher Feb 1 '16 at 20:37
7
\$\begingroup\$

Quite likely, the choke isn't broken. It is normal for chokes to act as speakers, especially under high momentary current (think of transformer hum on loaded transformers). So your choke got noisy, because there are irregular current peaks across it. You observed the irregularities on your scope.

The most likely reason for erratic voltage at the choke (causing erratic current) is instability of the regulator. The idea of a switch-mode buck regulator is to draw current spikes at a high voltage and provide continuous current at low voltage. In your design, the current spikes should be delivered from C1/C2, and the capacitors should get smoothly recharged through the choke. If C1/C2 fail to source the required current, the input voltage of the regulator breaks down, up to the point where the regulator stops working. The external supply can not source the required current spike, because of the coil impedance at high frequencies.

If you replace the coil by a shunt, you lower the impedance if the external power supply, so the output filter cap can substitute C1/C2, at the cost of high-frequency components on the supply cable, so the modified device is likely to violate EMI regulations.

The proper fix in that case is to replace the input caps by new, proper low-ESR (important!) caps. To check the health of those caps, a capacitance measurement is insufficient; you need an impedance or ESR test.

\$\endgroup\$
1
\$\begingroup\$

I'm not sure about the questions 1 and 2. About Q3, in general yes, you can replace choke with the shunt. The choke is there to filter out voltage spikes that occur when you plug the unit/turn on the switch. It is recommended to use choke here, so C1 and C2 don't get (large) voltage spikes. You can put in a random choke with inductance of few 100uH.

\$\endgroup\$
  • \$\begingroup\$ I can't find any chokes around the shop :( . Without an inductance meter, can you suggest any way of testing the choke to determine if it has failed? C1 looks fine, but I could also desolder it to test if necessary, but I don't have an ESR tester. Any suggestions for narrowing this to the actual bad component? \$\endgroup\$ – Ryan Griggs Feb 1 '16 at 16:16
  • \$\begingroup\$ not really. You can get in to detail here, but without instruments to measure this it will be hard job to do. I would recommend try finding a choke on farnell, digikey, mouser, rs-components, etc. One of those should have what you need. Change the caps to, just to be sure. This should be all quite cheap. If you really can't find it either leave 0R resistor on or make a small coil from a wire yourself, as many turns as you can get. And don't forget to take wire cross-section area into account (thick enough to be able to pass current you need) \$\endgroup\$ – ursusd8 Feb 1 '16 at 17:03
  • \$\begingroup\$ I tried winding my own choke out of 30 AWG wire, getting about 100 turns on a small steel nail. I did three layers, and the entire device was about 1.5cm in length. Total resistance about 0.4 Ohms. This choke got VERY hot, and did produce some squealing like the previous one. So i am going to assume it's C1 causing the problem. Don't have any caps to replace it with... so we'll call this job "done" and toss the device in the scrap bin. :/ Thanks to everyone for your advice. \$\endgroup\$ – Ryan Griggs Feb 5 '16 at 1:26
  • \$\begingroup\$ I think it is the best "solution". If you (we) can't identify the issue it's safer to get rid of the device :/ \$\endgroup\$ – ursusd8 Feb 5 '16 at 9:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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