Stabilising SMPS simulation for Amp supply

[eem2am]

Hello, please could you help to stabilise my LLC SMPS simulation.

Its got a split output so it can be for a half-bridge class d amplifier. (330W)

Its in LTspice.

I have made it have a minimum frequency by adding a current source I1 to the LTC6908.

I then implemented variable frequency by adding a PNP transistor to the frequency changing parts of the LTC6908.

It doesn’t seem possible to stabilise it, so I am hoping that you may assist me.

Here is the LTspice simulation in .txt form, and if you make it to .asc form, then it will be runnable in LTspice

2shared - download LLC.txt


Here also is simulation

View attachment LLC.txt

 

[gmarsh]

Any particular reason you're using a pure oscillator IC instead of a purpose-built LLC controller?

 

[eem2am]

yes, linear.com dont have a model for an LLC controller

 

[dtproff]

I will be honest ant tell you I have never been able to stabilize the LLC in either symetrix or ocad. From what I understand from talking to the resident PHD here, this issue is in the small signal model for the LLC's is not well described for a spice simulation because of the way Lp/Lr add in over the power range. Not sure I am in total agreement with him but he does have the PHD in his tittle and he heads up the advance R&D group.

If you just want help stabilizing the loop for the LLC using a standard TL431/Opto I can certainly provide you some values that will get you pretty close the first time up.

Tony

 

[dtproff]

Alternatively, if you just need the open loop waveforms yuo could set it to resonance and see what you get.

Tony

 

[chas1]

I would like the values you offered for TL431, I just started a design using L6599AD or more than likely UCC25600 (fewer parts count and lower startup current required) and have read many of Christophe Basso's appnotes and shared results of his DC-DC converters feedback and control presentation but using type 3 with TL431 doesn't simulate using LTspice and I am using the LLC spice model's he sent me, I would like to simulate your solution using the model's I have.

Thanks in advance

 

[dtproff]

I would like the values you offered for TL431, I just started a design using L6599AD or more than likely UCC25600 (fewer parts count and lower startup current required) and have read many of Christophe Basso's appnotes and shared results of his DC-DC converters feedback and control presentation but using type 3 with TL431 doesn't simulate using LTspice and I am using the LLC spice model's he sent me, I would like to simulate your solution using the model's I have.

Thanks in advance

Sorry my friend but that is 10 years ago. Not even sure I kept my notebooks for the design. The design will depend a lot on how narrow a frequency range you plan on using (see ST design notes on the Mv transfer function). A relatively high Lr/Lm ratio gives you good control range (Fmin to Fmax being narrow) but a more difficult loop to stabilize because of the high loop gain needed in the TL431. A lo Lr/Lm ratio gives you a relatively large range for frequency and a lower loop gain. One more thing, the innerwinding C on transformer will determine you burst frequency entry/exit point as you will transfer a discreet charge to the output every cycle through that C. That C will tend to drive the output up at light loads so you will need a burst mode action.

 

[chas1]

Thanks for quick reply, I have simulated a TL431 circuit that I got from a couple of appnote's for the NCP1395 & 97 and after a few changes it should do the job. I am at a crossroads with the magnetic's, I am using EER4220 core set and the gap required is very small and difficult to shim for testing I might end up with discrete inductor using RM core so I can bench test the over all design. I have a pcb designed that will support the L6955, UCC25600 as well as NCP1395 chip set's with external supply inputs. I should have pcb's in a couple of weeks and I hope to have transformer ready by then.

Again Thanks

 

[dtproff]

Core size is fine. Be sure the core is wound symmetric. Meaning the leakage to the primary to each leg on the secondary. This is measured by shorting one leg of the secondary then the other leg and measuring the primary side impedance (one side conducting and then the other is what you are measuring). Asymmetry can lead to poor current share between the two legs. I like my transformer to run <50KHz so that the third harmonic is less than the start of the conducted emissions band and your switching losses are a smaller percentage of your overall losses. Your core should be goor for about 800W or so.

Good luck.

Tony

 

[chas1]

Thanks Tony, Fr for my design is 90kHz,Cr 100nf (2 x 47n across switches reduce current and cap size) operating frequency below 50kHz? up to 800 Watts is the goal. thanks for input , all help appreciated. Handmade dual chamber bobbin using white silicon tape 3mm wide.Trying to find Chinese house to wind at least 5 samples of transformer but it looks as if none use EER core sets larger than EER 28.In states most are ETD for my purpose large footprint for pcb. EER4220 Ae 240, Al ~5200nH small footprint.