UCC29950: The output driver GD1 and GD2 are not working if they are clamped

Hello Robert

Thanks for getting in contact. I've had a look at your schematics and one thin I see is that the output capacitance for the LLC stage is very low - C4015 is only 470nF. Normally at this power level one would need several mF to absorb the ripple current, for example if you look at the EVM for this device we have used 4300uF. You don't show a value for L4003 but assuming it's a common mode choke, it will still have a significant differential mode inductance. This will force most of the secondary ripple current to flow in C4015 and will generate a very large voltage ripple across it. I'm not certain how this would affect the controller, perhaps the LLC_CS will trip but I'd advise you to move the bulk of the output capacitance to the left hand side of L4003.

The other thing you should check is that the MD_SEL/PS_ON pin is reaching its turn-on level of 10.5V (VCC_ab_uvlo(start)). There is some margin here but if Vcc is 12V you will lose 700mV in the pass transistor and the resistors form a potential divider of about 0.8 leaving 9V at the pin - this is rough calculation but you can check the pin voltage. It's possible that the 12V rail drops a little when the LLC starts and this could cause the MD_DEL/PS_ON pin to go below its 'on' threshold.

Please let me know how you get on.

Regards

Colin

Hello, Colin,

Thank you so much for your evaluation.

The first point is very interesting for me and I will try it in the next days directly. I didn't think about the differential part of the Common Mode Choke, but it sounds conclusive. This is the 744824622 choke from Wuerth. 
It always looks to me as if the LLC level starts briefly at the beginning, but then remains inactive. The switch-off after the description in the data sheet would support your thesis with LLC_CS. I have chosen the size of the electrolytic capacitors so, because in the end I will have an extreme capacity by batteries. In your opinion, should 4300uF still be dimensioned independently of the load on the output stage? This would mean that an idle operation would cause problems, which I can't imagine, because the capacity is dimensioned to the nominal ohmic load. As I said, in my case it is rather extremely capacitive.
Apart from that I don't have room for a 4300uF/450V capacitor bank.

The second point is not completely correct. Especially since in the application TIDA_00704 exactly this circuit is used to control the LLC stage. I'll check your point again at this point. But it's not the problem itself at the moment. Especially since the PFC stage can be switched on and off separately via MD_SEL/PS_ON.

Hello Robert

I was wrong about the capacitance you need - I forgot your output voltage is 260Vdc and not 12Vdc.

Common mode chokes always have an amount of leakage inductance and this is the inductance that appears as a differential inductor. Sometimes, indeed usually, CM chokes are deliberately wound so as to have a significant amount of leakage inductance. Anyhow, your best option here is to talk to Wurth.

The excel calculator for the UCC29950 gives a ripple current of about 2.14A in the output circuit of the LLC at 260V and 500W so you will need enough capacitance to be able to carry this current and keep the ripple voltage to reasonable levels. Assume you want to have 5Vrms ripple when running at resonance of 120kHz then you will need about 55uF or so of capacitance. The output inductor then forms a second stage filter so that the ripple current into your battery load is acceptable. Please double check these numbers using the device excel calculator - you can find it on the device web page.

Please let me know how you get on.

Regards

Colin

Hi Robert,

Since there has been no reply to this post for over a week i will close this post.

Please open a new post if you have more questions.

Thanks

Peter

I have still this problem to drive the resonance tank.

I have followed exactly the application skills. But it does not work. In this case, I start another solution with another concept.

Thank you for your service.

Hi Robert,

Sorry to hear that this did not work out for you on this occasion. Best of luck with your new solution.

Regards

Peter