Miniature LED driver based on a BPS BP9021 chip

[ym58]

I have recently purchased in CN (ebay object #231497809833) a pair of 1 .10USD led drivers made with a BP9021 chip (BPS).


Literature on the chip or the 2 sq.cm (0.3 sq.in) PCB does not abound on the internet but at least I got hold of a circuit diagram :


Anyone here could comment on how this chip works ?

[sleemanj]

[ym58]

yes, of course I had seen this video ... I made a few researches before posting here 

this guy "bigclivedotcom" is certainly knowledgeable generally speaking but lacks confidence on this very specific PCB (and chip) !
looks like he's discovering the thing while shooting his video  ...
---
I am sure one or two guys here on EEV have already tackled this circuit, technically speaking.

[miguelvp]

Datasheet of similar chip:

http://www.bpsemi.com/pdf/BP3122/BP3122_EN_DS_Rev%201.1.pdf

[Yansi]

That is not a filter circuit, thats voltage spike protection circuit. The leakage inductance of the xformer makes voltage to overshoot on the primary side.

The circuit is nothing more than a simple voltage clamp. The RC time constant is something about 5 times the switching period (to dissipate the overshoot energy quickly enough not to charge the cap up too much), designed by selecting the resistor first according to the primary reflected voltage and power to be dissipated on the resistor. Then tha capacitance is calculated for desired time constant as mentioned above.

The DRC circuit can be replace better by unipolar transil with a diode in series. For price reasons, the chinese manufacturer probably didn't bother with these components (they cost money) and the power msofet switch is usually avalanche rated (reacts as zener diode when overvoltage). It is relatively fine omitting the DRC circuit, unless there is too much power dissipation in the power switch device. And it also really does not help the service life of the thing.

And just note about efficiency: The RC circuit is the least effective - it wastes power even if there ar almost/no spikes. The solution with TVS diode is more costly (few cents) than the DRC, doesn't waste much power (dissipates only the spike energy) if properly selected TVS voltage and the solution with no overvolt protector is the most efficient of them, because it clamps only the spikes that are over the absolute max of the power switch. (But I generaly don't suggest you to design flybacks like that).

[mmagin]

I wonder what the primary/secondary isolation on that is good to!

I've thought it would be a fun project to build my own hi-pot tester that ramps from 0 to 5kv with a preset leakage current to shutoff at (and record the voltage it reached).  Sure I could buy one, but it's not something I really 'need', and it's a fun high voltage project.  And then I could test all this cheap stuff around and see how good it is.

[ym58]

Datasheet of similar chip:
http://www.bpsemi.com/pdf/BP3122/BP3122_EN_DS_Rev%201.1.pdf

Finally, a datasheet in English, I am a lucky man ... thanks for the link 

[Yansi]

Have you read my post?

[ym58]

That is not a filter circuit, thats voltage spike protection circuit. The leakage inductance of the xformer makes voltage to overshoot on the primary side.

The circuit is nothing more than a simple voltage clamp. The RC time constant is something about 5 times the switching period (to dissipate the overshoot energy quickly enough not to charge the cap up too much), designed by selecting the resistor first according to the primary reflected voltage and power to be dissipated on the resistor. Then tha capacitance is calculated for desired time constant as mentioned above.

The DRC circuit can be replace better by unipolar transil with a diode in series. For price reasons, the chinese manufacturer probably didn't bother with these components (they cost money) and the power msofet switch is usually avalanche rated (reacts as zener diode when overvoltage). It is relatively fine omitting the DRC circuit, unless there is too much power dissipation in the power switch device. And it also really does not help the service life of the thing.

And just note about efficiency: The RC circuit is the least effective - it wastes power even if there ar almost/no spikes. The solution with TVS diode is more costly (few cents) than the DRC, doesn't waste much power (dissipates only the spike energy) if properly selected TVS voltage and the solution with no overvolt protector is the most efficient of them, because it clamps only the spikes that are over the absolute max of the power switch. (But I generaly don't suggest you to design flybacks like that).

I am not quite sure the DRC circuit has been totally omitted on my PCB.
You may be right when you say that a TVS can also do the trick ... I'll go & get my best magnifier but I know for sure that one or two SMDs are hidden under the transformer and one of those could well be a TVS !

But besides this spike protection circuitry (be it present or not on the PCB), how does the chip REALLY works  ?

I can see a RC cell (not the spike protection but the one connected to VCC - pin#3). Do you think this cell is designed to provide a RC time constant to the chip or has it nothing to do with this and simply provides VCC to the chip thru a resistor and a filtering cap ?

Then how do the two DRAIN pins (5 and 6) behave in respect of the primary coil ?
How such a circuitry can possibly provide a constant current on the secondary ?

( needless to say I have very poor expertise in SMPS ... reason why I am asking  )
 

[ym58]

Have you read my post?

NO ...  (just kidding)
It took me a little time to answer to it, that's why (see above)

[Kjelt]

How such a circuitry can possibly provide a constant current on the secondary ?

My guess:
The ic measures the current in the primary winding of the transformer, it knows the voltage on the primary of the transformer,
then with an accurate transformer the losses are known and the ratio between primary and secondary is known so you know the secondary voltage (at least a good guess within 5% as claimed in the datasheet) hence you know (or again a good guess) the current in the secondary.
Still a lousy shitty circuit if you ask me, most circuits have at least some form of feedback with an optocoupler from the secondary, this is just fire and forget, if something is wrong (different transformer, or leds short circuit) it keeps on going, quality! But then $1 can't beat that.
I myself like to work with DC voltage and Maxim led drivers that measure the current through the leds, I don't trust HV led drivers, i like my house too much to let it burn down by a $1 hunglo circuit.

[ym58]

I wonder what the primary/secondary isolation on that is good to!

I would not even take a chance to measure it ... looking at the transformer, the PCB layout and the cabling of this USD1.10 driver speaks for itself !

[ym58]

My guess:
The ic measures the current in the primary winding of the transformer, it knows the voltage on the primary of the transformer,
then with an accurate transformer the losses are known and the ratio between primary and secondary is known so you know the secondary voltage (at least a good guess within 5% as claimed in the datasheet) hence you know (or again a good guess) the current in the secondary.
Still a lousy shitty circuit if you ask me, most circuits have at least some form of feedback with an optocoupler from the secondary, this is just fire and forget, if something is wrong (different transformer, or leds short circuit) it keeps on going, quality! But then $1 can't beat that.
I myself like to work with DC voltage and Maxim led drivers that measure the current through the leds, I don't trust HV led drivers, i like my house too much to let it burn down by a $1 hunglo circuit.

So, if I understood you well, the VCC value is the key since it implies definite primary current/voltage and, hence, a secondary voltage and thus a constant current thru the LEDs ...
Then, how the PCB manufacturer can possibly claim that this "house-burner" can be supplied within a 85V-250VAC range ?

[Kjelt]

So, if I understood you well, the VCC value is the key since it implies definite primary current/voltage and, hence, a secondary voltage and thus a constant current thru the LEDs ...
Then, how the PCB manufacturer can possibly claim that this "house-burner" can be supplied within a 85V-250VAC range ?

No the primary current is the key, the primary voltage is also used (to switch off when it exceeds the threshold)  it is all explained on the above datasheet page 6 Constant Current Control 

[Kjelt]

The chip of the datasheet does have some protections which is nice. Unsure if your chip has the same.

[ym58]

So, if I understood you well, the VCC value is the key since it implies definite primary current/voltage and, hence, a secondary voltage and thus a constant current thru the LEDs ...
Then, how the PCB manufacturer can possibly claim that this "house-burner" can be supplied within a 85V-250VAC range ?

No the primary current is the key, the primary voltage is also used (to switch off when it exceeds the threshold)  it is all explained on the above datasheet page 6 Constant Current Control 

Brilliant ... I had not (yet) gone that far in reading what miguelvp had kindly posted  earlier (the english datasheet).

The more I look around at those new electronic circuits, the more astounded I am : those guys can make very sophisticated & dedicated chips whose cost is probably less than a dime (Maxim chips is another good example, a bit more expensive though) !

The reverse of the medal being that these circuits can probably burn a Netherlands house to ashes in less than a minute ... 

[mmagin]

I wonder what the primary/secondary isolation on that is good to!

I would not even take a chance to measure it ... looking at the transformer, the PCB layout and the cabling of this USD1.10 driver speaks for itself !

Yes, but, for science!

[Yansi]

Think a bit before asking.

The RC on VCC pin is not "RC" as you probably think. The C is the supply decoupling. The R from the top provides probably startup curent for the IC.

The VCC exact value doesn't matter. The constant current output is met bcs the primary current is known, so is the secondary one.

By controlling primary peak current*, secondary peak current is set. The secondary filter cap makes it filtered, DC average out of the secondary waveform.

*the primary current looks liek a ramp.  The IC switches on, current rises (determined by i(t) = 1/L* integral of u(t)dt) and after reaching a specific level, the switch is off, energy starts to transfer to the secondary. The secondary current then looks like similar ramp, but inverted. Starts at high current, decreases to zero (as the magnetick energy in the core discharges). Then the whole cycle starts again.

Sorry, this is only very unaccurate description of the basic working of a flyback converter. Use your favourite search tool to get more info on flyback converters. But beware, some complex math can follow. 

[ym58]

Yes, but, for science!

No, not even for science coz by the time I will have pointed and documented where the isolation problem is and how much it could be a safety concern, the manufacturer will have long time retired, and gone fishing with the billion dimes earned from selling each of these unsafe drivers !

[ym58]

Think a bit before asking.
The RC on VCC pin is not "RC" as you probably think. The C is the supply decoupling. The R from the top provides probably startup curent for the IC.
(...)
Sorry, this is only very unaccurate description of the basic working of a flyback converter. Use your favourite search tool to get more info on flyback converters. But beware, some complex math can follow. 

Thanks for enlightening the ignorant that I am (or the ignorants that we are)
I will now take a little time for digestion and will come back to you when I am sure I understood it all, so don't take offence if I don't answer you within a minute ...

[miguelvp]

Let me clarify that the datasheet I linked is a similar device, not the same one, but it should work based on the same principles

[ym58]

Let me clarify that the datasheet I linked is a similar device, not the same one, but it should work based on the same principles

Yes, but that's fine ... as we are all lacking English literature on BP9021, we'll have to assume that the BP3122 functional diagram is "somehow" similar.
This is the only BP9021 datasheet that I've been able to find so far :


I believe that the very relevant explanations & comments from Kjelt will also apply to this chip.
Thanks again for the finding

[spyking]

Quick question, would it be possible to modify/adapt this driver to accept PWM dimming?