Componenets to Replace to Eliminate LZ Whine

[NotFitForPurpose]

Hi,
ohh yes most definitely failed !!
That thermal is the before fix image.

The key points being:
The unit appeared to function normally, powered on, screen worked.

In this case, I grabbed replacements from a donor unit to obviate compatibility issues. Measured those parts prior and post fitting, also comparing to parts in boards - to make sure my techniques sound.

Detection
Detection is straight forward, measure the current draw in sleep mode. As the SMD device was low resistance it continuously drew current of ~20 mA. (Hence why it was heating, I couldn't actually feel anything on the board. I used the thermal camera to rapidly locate what might have been consuming the energy).

Diagnosis
Straight forward enough, the SMD (I suspect BCX18 - TR11) had same resistance forward and backwards ~28 Ohms (query Base - Collector). The Inductor, more resistor than inductor - circa 1.8 Ohms & 58 uH @ 1 kHz. The inductor needs to be 1mH @ 1 kHz, Q @ ~1.8 and around 4.5 Ohms The SMD should at least 'test' a bit like a Diode, even if it's a transistor (I will confirm the actual part later, as my boards are not exactly as the circuit diagram - It seems close enough).

Repair
The repair process was straight forward, remove the Inductor then the SMD. Replace SMD, then Inductor.

Validate
Measure the quiescent current in sleep mode (running current may mask the issue, though it was ~20 mA higher).

Root Cause
As previously indicated, I am still ruminating on true root cause. Is it distinct issues or correlated, I lean to the latter. Diode (heat) - Coil - Ripple current - Capacitor - Noise ... Needs careful thought and perhaps a Fault Tree.

I have seen similar wound inductors with the same appearance under heat / high voltage stress in similar boost convertor circuits. The order might change slightly, though not the failed parts. The inductor DC resistance can only fall if the winding is shorted somewhere and that needs most likely heat (or electrical stress) to cause the magnet wire insulation to fail. If the failure rate not insignificant, design / specifcation related causes should not be ignored - needs more measurements.

Comment
The short (Query Base - Collector short on TR11) would clearly cause problems when trying to program some EPROM's

I posted, as it was something that might have gone un-noticed without care (apart from flat batteries).
I hope this helps?

[amenjet]

Interesting, this is L1 and either TR11 or TR12 on Martin's circuit diagram? I'd have thought EPROM programming would have been adversely affected?

[NotFitForPurpose]

Hi,
I agree, hence the need for a fault tree. Perhaps the device spent excessive periods at lower voltages leading to an increase in 5 volt ripple current. It must have eaten batteries at quite a rate, surprised there wasn't evidence of a leaked battery, perhaps it was used with a power supply more than a battery.

I hope I was just 'unfortunate', having two (possibly un-related) faults.
The check for the condition and component tests straight forward, so it can quickly be checked for.

For future testing (and fun), I need a test jig and code for the EPROM voltage pump..

Long weekend, so cheery by for a bit.

[NotFitForPurpose]

Hi,
purely for confirmation, the bubble residue is magnet wire insulation that has ''cooked" off.

The wire was carefully removed and tested for resistance at various points. Normal magnet wire may have occasional pin holes but should never offer a low resistance. In this case the material is no longer present on the wire and the copper is exposed allowing a short circuit to form as hypothesised.

It was difficult to capture in the image, the central section where the bubbled medium was is a lighter copper colour than the insulated wires near the extremities of the bobbin. Un-winding carefully allowed the magnet wire resistance to be checked along its length. You can see how little had to be removed until continuity between the start and that point just off the coil, revealed a low resistance to be measured (no insulation).

20250422_135151.jpg