CoB LED filament analysis

Looks to me that the four diodes make a full-wave bridge, driving a single series string of LEDs, with two series resistors.

A CT without a burden resistor becomes a differentiator. Big cheap iron-core CTs are very nonlinear at low currents anyhow.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

"And, by the way, there's yet another fundamental misunderstanding"

There sure is.

Yep.

Danke,

No, the phaseshift was caused by the current transformer being unloaded. The perfect load of a CT is a short (e.g. a transimpedance amplifier). The breadboard would not cause a measuremable shift. The scope probe alone is more capacitive.

Arie

It still doesn't make sense. The Fluke current transformer has its shunt built in. There is no need to -indeed you should not- add another. It's designed to be loaded by 1MOhm//47pF.

The manual for the probe is easy to find on line:

Jeroen Belleman

I don't understand the flat top. You are presumably scoping the 240 volt AC line.

A 240-volt led lamp will be different from 120. It might have half-wave rectification.

And I'd expect lots of different light bulbs. Some have electronics in the base.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

What you call two half wave rectifiers are a full wave bridge, or if you prefer the inventors name a Graetz bridge.

Most mains is now slightly flat topped from the huge number of SMPSUs in use and utility transformers running into mild saturation near peaks.

Thank you for your kind, thoughtful followup, Arie. AI adds additional food for thought:

There is no publicly available comprehensive teardown video or detailed technical internal diagram specifically for the Fluke i1000s AC current probe. Because the device is a ruggedized, safety-rated tool designed for industrial environments, it is typically serviced as a sealed unit rather than opened by users.

### Internal Architecture Overview

While a detailed schematic is not publicly provided, the mechanical and electrical construction follows the standard design for a high-quality, passive clamp-on current transformer.

• **Split-Core Transformer:** The "jaws" of the clamp contain a laminated magnetic core assembly that pivots. When closed around a conductor, this core forms a closed loop, concentrating the magnetic flux induced by the AC current in the conductor * **Secondary Winding:** Wrapped around this core is a secondary coil with a high number of turns. This coil converts the primary current into a proportional, lower-level current signal. * **Passive Signal Conditioning:** Inside the probe's handle, the secondary signal passes through a network of resistors and potentially capacitors. This passive network provides the selectable output ranges (1 mV/A to 100 mV/A) and includes filtering components to minimize high-frequency noise and ringing, which is crucial for accurately capturing distorted waveforms on an oscilloscope. * **Shielding and Construction:** The internal circuitry is housed within a flame-retardant, high-impact plastic enclosure. The output is fed through a high-quality coaxial cable, which is shielded to prevent external electromagnetic interference (EMI) from corrupting the low-level signal before it reaches the measurement instrument.

It will take me a while to digest it. Here's an excerpt from Fluke's user manual:

Operation

Use the following procedure to make a measurement:

1. Connect the i1000s Current Probe to the desired input channel on the oscilloscope. 2. On the Current Probe, select the least sensitive range (1 mV/A). See Figure 1 for selector switch location.
2. On your oscilloscope, select an appropriate range. Use a 1:1 probe setting.
3. Clamp the probe around the conductor to be measured, and observe the current waveform on your oscilloscope display.
4. Calculate the Amps-per-division on your oscilloscope by dividing the vertical scale (mV or V per division) by the i1000s switch position (1, 10, or 100 mV/A). For example:

It's time to end my participation in this thread. In closing, I hope this Chinese aphorism is correctly displayed:

不诱于誉，不恐于诽，率道而行，端然正己。

Not tempted by praise, not intimidated by criticism; walking the path of principle, one remains upright and true to oneself. Danke,

Why?

Any "light bulb" that relies on LEDs includes lots of electronics. There's no particular reason to put it all in the base.

My flat came with lots of Philips tungsten halogen spot lights. One of them ended close to a temperature sensor for the buildings revised fire control system, so I replaced the conical plug-in bulb with a LED based plug-in equivalent, which ran a lot cooler. The base of the replacement wasn't transparent but it wouldn't have offered much space for electronics.

My AC source is not straight mains. I value my life. Instead, I used a toy train variac transformer feeding a little 15VA mains transformer in reverse. Plenty of opportunities to distort the waveform.

The LED strings I tested do indeed conduct only one way. There are some electronics in the base which I haven't yet looked at.

Jeroen Belleman

Lots of people are afraid of electricity. Strange.

A 120 volt LED lamp can apparently work with just resistive current limiting. At 240, the economics is different, although one could just put a bunch of the 120 v COB things in series.

I have a bulb here with no electronics and four COBs. It starts to light up around 80 volts. The COBs may be in series.

What's interesting is the bulbs with a long twisty string of LEDs. How can they do that?

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

The economic aren't going to be wildly different. It's going to pay off to put in twice as many LEDs. LED lamps dissipate a whol;e lot less power than their tungsten filament predecessors so the extra power dissippation isn't gong to be a problem, but it may make sense to go in for smaller LEDS. It's not an elegant solution - some sort of reactive current current switching with fast current switches could give the same amount of light with half the power dissipation, but it going to take more expensive components, if fewer of them.

Who knows.

We all know about flexible printed circuits. The image suggests that there might be bits of rigid support for the strips of flexible material that carry the LEDs and their connecting conductor. It's an exercise in visual design rather than any kind of serious engineering.

A real lunatic might put LEDs onto strips of glass and rely on transparent tin oxide for the conductor.

I'm not afraid, but I'm not stupid either. 240V straight from the mains can kill. For sure it bites.

That's what they did. There were two series pairs. The PCB in the base has a full-wave rectifier, an MT7606 LED driver and a bunch of resistors. I didn't reverse-engineer it; the copper traces are hidden under a thick layer of white laquer, but I can sort-of guess how it's done.

I'd expect a roughly constant amount of light for a wide range of voltage. I didn't try that.

Jeroen Belleman

Here's the AC line here:

This is the AC hot, relative to the grounded scope. But there isn't much on the neutral wire, so the waveform is about right.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Now is the time to re-engage with this thread.

A cheap CoB LED filament probably uses a Graetz bridge, constructed with aluminum wire bonding:

COB technology is widely used in LED designs, providing higher lumen density and improved thermal management. Wire bonding enables compact LED arrays with efficient heat dissipation, leading to brighter, longer-lasting lighting solutions in automotive, industrial, and consumer applications. ... Wire bonding remains a crucial technology in modern electronics, offering flexibility and cost-efficiency in a variety of applications, including 3D ICs, power electronics, and COB LEDs. While material and manufacturing costs can vary, especially for high-volume production, the cost advantages of wire bonding become evident as production scales. [1]

In regards to Arie's remarks, the Fluke i1000s is an AC current probe. In other words, a capacitor couples its current transformer to its transimpedance amplifier. Perhaps that capacitor is culpable? The Fluke i1000s 100 mV/A Users Manual [2] response curve shown in Figure 2 for a 100 mV/A range indicates a large phase shift is plausible. Yet all phase shift disappears when a CoB LED bulb is substituted for a bare metal filament. The breadboard's parasitic capacitance of 31 pF is negligible.

EUREKA by USA Poet Laureate Edgar Allan Poe, argues that scientific inquiry begins with an intuitive leap of imagination and more-or-less ends with measurement. In other words, a hypothesis provides a roadmap to give direction to measurement.

Here's the first image shared by me in this thread:

It shows a color mismatch between the blue tinted voltage trace and the green tinged current curve. That's a crucial clue. My mind mulled this clue during a bicycle spin up a nearby mountain with Bach playing through the earbuds. Then an intuitive leap of imagination took place. What if neither the breadboard nor the probe's intrinsic parasitic capacitance causes the phase shift? Is it possible that an idling transformer mounted near the breadboard causes the anomaly? Yes, the idling transformer creates the problem. And a power switch added to the transformer to power it off when not in use restores the current curve to zero phase shift.

Note.

Danke,

A cheap iron-core current transformer will behave very badly at low current.

You are trying to measure mA with a 1000A CT. Parts per million.

Instrument it better.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

I should have thought of that possibility. As a youngster, I almost got a scope for free because it had an unstable trace with lots of hum. "This must be a defective supply in an old scope, nobody wanted to repair." When moving the scope to get at the backside cabling the trace changed a lot. Pulling it away from the isolation transformer next to it cured the problem. No free scope for me that time...

Arie

Old Eico-grade scopes had low voltage CRTs and usually no magnetic shielding, so were very sensitive to mag fields.

A clamp-on split-core current transformer will have some sensitivity to local magnetic fields. He's trying to resolve milliamps using a

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics