How Did This Circuit Go So Wrong?

Thread Starter

mriksman

Joined Aug 31, 2010
113
I built this circuit based on the examples in the datasheet. It worked.
1598578819209.png

Transistor Q3_1 is BUJ100LR,412Bipolar Transistors - BJT Bipolar Transistors - BJT Trans GP BJT NPN 400V 1A 3-Pin
https://www.mouser.ie/datasheet/2/848/BUJ100LR-1846655.pdf
C2_1 is 4.7uF Tantalum Capacitors 20V
https://www.mouser.ie/datasheet/2/40/avx_taj_e-1158783.pdf

Voltage at Vs is 16.3V (created by an internal zener).
Voltage at Vdd is 5V.


I thought perhaps I'll shrink those two devices, and replaced them with

Transistor Q3_1 with a NEXPERIA PMBTA45,215 Bipolar (BJT) Single Transistor, NPN, 500 V, 35 MHz, 300 mW, 150 mA, 100 hFE
http://www.farnell.com/datasheets/812952.pdf
C2_1 with 10uF Murata 0805 (2012M) MLCC 25V dc ±10% SMD GRM21BB31E106KA73L
https://docs.rs-online.com/671d/0900766b8145fba0.pdf


However, things didn't work.
Voltage at Vdd was 5V as expected.

However
* with the transistor installed, the Vs pin would oscillate between 13V and 15V. I don't have an oscilloscope, but the period was visible on a DMM. I removed the transistor (and thus, the L_Pwr feature), and then measured a stable 15V on Vs. This should be 16.3V as before.
* I replaced the 10uF with the original 4.7uF, and it worked - it was showing 16.3V, and the dimmer worked.

So how were my choices so very wrong!?



FL5150 examples/datasheet https://www.mouser.ec/datasheet/2/149/FL5150-1008825.pdf
 
Last edited by a moderator:

ci139

Joined Jul 11, 2016
1,898
https://www.mouser.com/datasheet/2/302/BUJ100LR-353035.pdf -- hFE≈17.3 , IEBO≈1mA , tf≈80ns , PMAX≈2.1W , see Fig.6
https://assets.nexperia.com/documents/data-sheet/PMBTA45.pdf -- hFE≈100 , IEBO≈100nA , tf≈800ns , PMAX≈300mW , sww Fig.8 , Fig.10

suppose the mosfets are driven by ? 20mA.RMS (÷√(50%)≈30mA ?) ~> BUJ100 (Fig.6) VCE≈70mV /// PMBTA45 () VCE≈70mV
/!\ but since the tf difference one will be dissipating 10x more power at closing /!\ or not close at all /!\ since the lower IEBO . . . etc. . . .

it takes detailed analysis to see what everything forms (is involved in formation of) the 16.3V and how it all comes together
basically (incase of no thermal failure of your PMBTA45) it should be longer ON (get above 16.3V) ...
... about Pg.17 (as much as it's described there) https://www.onsemi.com/pub/Collateral/FL5150-D.pdf + Pg.4 Fig.6
-- shows an internal zener in parallel to VS,GND - as i suspected there would be . . . . . . . .
??? if the PMBTA45 fuses to constant ON (or is const.-ly ON due it's IEBO) the 33k sees (120-17)V.RMS 320mW <-- but that will pulse due half period ~AC in . . . maybe ???
103V.RMS/30kΩ≈3.4mA -> 6.9mA if the MOSFETS are switched at 50% (Pg.6 IS.MAX 25mA) . . . 20mA if the FET's are switched at 14% duty
 
Last edited:

Thread Starter

mriksman

Joined Aug 31, 2010
113
Just an update.

I have an oscilloscope, and I have removed the transistors for now. The circuit is now this;
1599325019739.png

For C2_1,
10uF MLCC 17.2V peak, dropping down to 11.4V
10uF Electrolytic 17.2V peak, dropping down to 15.7V

I'll test the 4.7uF Tantalum tomorrow.

How can the ripple be so much worse on an MLCC? Shouldn't they discharge at the same rate?
 

BobaMosfet

Joined Jul 1, 2009
2,110
I built this circuit based on the examples in the datasheet. It worked.
View attachment 215795

Transistor Q3_1 is BUJ100LR,412Bipolar Transistors - BJT Bipolar Transistors - BJT Trans GP BJT NPN 400V 1A 3-Pin
https://www.mouser.ie/datasheet/2/848/BUJ100LR-1846655.pdf
C2_1 is 4.7uF Tantalum Capacitors 20V
https://www.mouser.ie/datasheet/2/40/avx_taj_e-1158783.pdf

Voltage at Vs is 16.3V (created by an internal zener).
Voltage at Vdd is 5V.


I thought perhaps I'll shrink those two devices, and replaced them with

Transistor Q3_1 with a NEXPERIA PMBTA45,215 Bipolar (BJT) Single Transistor, NPN, 500 V, 35 MHz, 300 mW, 150 mA, 100 hFE
http://www.farnell.com/datasheets/812952.pdf
C2_1 with 10uF Murata 0805 (2012M) MLCC 25V dc ±10% SMD GRM21BB31E106KA73L
https://docs.rs-online.com/671d/0900766b8145fba0.pdf


However, things didn't work.
Voltage at Vdd was 5V as expected.

However
* with the transistor installed, the Vs pin would oscillate between 13V and 15V. I don't have an oscilloscope, but the period was visible on a DMM. I removed the transistor (and thus, the L_Pwr feature), and then measured a stable 15V on Vs. This should be 16.3V as before.
* I replaced the 10uF with the original 4.7uF, and it worked - it was showing 16.3V, and the dimmer worked.

So how were my choices so very wrong!?



FL5150 examples/datasheet https://www.mouser.ec/datasheet/2/149/FL5150-1008825.pdf
You do realize the 4.7uF and 10uF are very different right? In terms of how C2 is used- that is a high-frequency (aka noise) filter. By changing the capacitor you alter the filter. Q3 is there solely as a current control (which alters voltage, as voltage and current do not work independently of one another).
 

MisterBill2

Joined Jan 23, 2018
18,179
The first problem is that we have no hint as to what the circuit is supposed to actually do. That adds a lot of mystery to deciding if it is working or not.
 

Thread Starter

mriksman

Joined Aug 31, 2010
113
@MisterBill2 It's a dimming circuit. The datasheet for FL5150 explains it all.

@BobaMosfet I wouldn't have thought 50Hz AC half-rectified to be 'high frequency'. C2 is used to smooth the half rectified AC. Internally, the FL5150 has a 17V shunt zener diode. That's why the voltage ripples between 17.2V (fully charged during AC on cycle) to 11.4V-15V during the discharge.

I have found the answer to my first question.

DC Bias.

I never knew that capacitance of a MLCC drops SIGNIFICANTLY after a few volts DC.
For example, a 4.7 uF 25 V, 0805, X5R capacitor operating at 25 volts DC would actually be about 15 percent of its rated value or 0.705 uF. A 4.7 uF 50 V, 1210 X5R capacitor at 25 volts DC would be about 50 percent of its rated value or 2.35 uF. A 4.7 uF 100 V, 1210, X5R capacitor would be 70 percent of its rated value or 3.3 uF.
So I don't actually have a 10uF at all when it's running at 17V (close to rated voltage of 20V). The capacitance is more like 1.5uF. Which is why it appears to discharge quicker than a 4.7uF Tantalum.
 

MisterBill2

Joined Jan 23, 2018
18,179
The variation of capacitance with applied voltage is driven by the movement of charge carriers. That is why tuning diodes work. The variation is different with different materials, which is a fact that circuit designers must be aware of. Interesting and educational for the rest of us, isn't it?
 

BobaMosfet

Joined Jul 1, 2009
2,110
@MisterBill2 It's a dimming circuit. The datasheet for FL5150 explains it all.

@BobaMosfet I wouldn't have thought 50Hz AC half-rectified to be 'high frequency'. C2 is used to smooth the half rectified AC. Internally, the FL5150 has a 17V shunt zener diode. That's why the voltage ripples between 17.2V (fully charged during AC on cycle) to 11.4V-15V during the discharge.

I have found the answer to my first question.

DC Bias.

I never knew that capacitance of a MLCC drops SIGNIFICANTLY after a few volts DC.


So I don't actually have a 10uF at all when it's running at 17V (close to rated voltage of 20V). The capacitance is more like 1.5uF. Which is why it appears to discharge quicker than a 4.7uF Tantalum.
It's not there to filter the 50Hz AC, it's their to filter higher frequency, spurious noise.
 

Thread Starter

mriksman

Joined Aug 31, 2010
113
C2 is to smooth the half-rectified AC.

The FL5150
1599469187873.png

Because of the 17V Zener, C2 has a DC bias of ~17V. This is pretty close to the rated 20V voltage of the MLCC. Because of this, C2 is NOT operating like a 10uF capacitor; more like 1.5uF. Hence it discharges quicker than the 4.7uF Tantalum. At face value, you would think the 4.7uF would discharge quicker than a 10uF.


The second issue with the transistor/Low Power Mode.

R2 was chosen from the datasheet. However, I didn't include the additional op-amp I have hanging of Vdd. So when it went into Low Power Mode, there was not enough current flowing through R2 to keep C2 charged. It eventually depletes, hits 7.2V, and the chip causes an Under Voltage Reset. The cycle repeats, causing me to see this oscillating voltage when Dim Control = 0V.

I just need to reduce R2 to allow more current to flow.


1599469140034.png
 

ci139

Joined Jul 11, 2016
1,898
as for before your last entry (#10)
i speculate your 33KΩ is just a bit too high value (the datasheet suggest also different "self-powering" options ... if this one in particular fails . . .)
 
Top