issues with a TD352

[fab04]

Hi Everybody,

In the case of a LEDs tubes dimmer, I'm using a TD352.

It works very well. But in some case (and for the moment, I don't know why), my IGBT and the resistor wich controls the gate of the IGBT burn.

I just replace these two components and after it works well. It happens very rarely, but it happens.

IGBT can drive up to 80A (@ 25°C), and Vce is ok up to 650VAC, and the maximum inrush current that I measure is 35A, and the peak voltage, it's around 450VAC.

The value of the resistor is 51R in 0805 case.

The IGBT that I use is an IK40N65WR5.

If you have some suggestions to try to understand what happens, I'll be very happy.

[AScha.3]

Hi,

show full circuit, not only a part.

btw

LED dimmer, at 450V + 80A -> 36kW LED ??

[fab04]

Hi,

Maybe my explanations were not enough clear.

There are two channels on the product, and the maximum consumption per channel is 864W (each LEDs tubes consumes 18W, and 48 tubes per channel).

The system is supplied with 230VAC.

80A is the maximum that the IGBT can support (I choose this one, because the inrush current at starting can be very high - 35A), but in normal functionning, the current is low (around 3/4 A).

[AScha.3]

Hi,

ok - i see. You have a lot of protective diodes... so overvoltage should (!) be no problem.

Burned gate resistors are collateral damage from strange IGBT suicide . :)

So my ideas : change gate circuit...to just a 20 ohm (1206) resistor, as ds shows , is used for tests:

no extra diodes, no clamp connection. (IGBT switching anyway at its own speed - its not a mosfet.)

+

where is VCC for driver coming from ? 0/15V ? should be close to driver , or some small (5 ohm or so) in ground and vcc connection, to avoid any spikes coupling in.

+

Gate circuit :  driver-20r-IGBT gate - - source - driver : has to be small/close to IGBT , small included area.

+

What frequency you drive the dim pwm ? (and : any short "ON" states , like 100ns ON, IGBTs dont like. You avoid this short on-times ?)

[fab04]

Thank you very much for your quick answer.

OK, I'll change my resistor.

You suggest me to remove all the diodes (D15, TVS4, TVS5 and D16) and to remove connection between clamp pin and the gate ?

Concerning the supply for driver, I use a power supply module from RECOM (https://recom-power.com/en/products/ac-dc-power-supplies/ac-dc-pcb-mounted/rec-s-RAC03E-K!s277.html?0)

I think that all components are closed (you can see my layout in attached picture).

Concerning the frequency of PWM, it's 100Hz.

[AScha.3]

20r ok, and just remove D15, tvs4 . And 20r at gate 1206 size - minimum.

tvs5 and D16 keep, might be useful to increase c32 . Did you check with a scope, what voltage at c32 is coming?

+

100Hz pwm is so low - why not use a thyristor or triac ? with a triac you need even no diode bridge and no gate driver+supply, just a optocoupler-trigger, like MOC3052 (random control) and BTA16-600 : 

+

btw

Your violent PWM "dimmer" might work on some LED tubes - on some just destroy them, because they have switching supply inside,  with PFC SMPS . And even Zigbee remote control for dimming - we have these in company. 

https://www.ledvance.de/professional/produkte/produkt-stories/led-roehren-online-special/osram-substitube-t8-connected

[fab04]

Hi,

Thank you for answers and advices.

I'll try with a 20R resistor, and I'll increase this size.

Concerning C32, I can try with 220nF & 470nF, I've seen in this application note (https://www.infineon.com/dgdl/Infineon-Snubber_Considerations_for_IGBT_Applications-Article-v01_00-EN.pdf?fileId=5546d462533600a401535748b5103fe8) that they use 0.5 & 1µF.

You can see in attached file, the voltage on C32.

Concerning the LEDs tubes, we use only a kind of tube and they didn't contain PFC.