LM5114: lM5114

Hi Robin,

thanks for posting on LM5114 and HV GaN!
it looks like your attachment was lost, can you please re-post it?
PMP40328 uses UCC27712, a 600V gate driver, which would not fit your bus voltage however have you checked out the isolated GaN/MOSFET driver UCC21220A which has a 5 V UVLO and dual isolated channels to more effectively drive HV GaN FETs?

thanks,

GaN_stack.pdfJeff:

I just realized that LM5114 is not isolated. So I have to provide another chip fast enough to not cause problems.

Thinking of NVE isolators: they are ns delay rated...

Now, I notice UCC21220A output sides are from 9V up. But GaNs we are using cannot have drive signals more than 5V. Most do not.

So UCC21220A may not be suitable for any GaN, right?

Here is the earlier pdf file , got lost somehow....here, can you suggest a clever way to minimize drives  ...such as you did using  cascode approach in UCC287xx chip?

thnx for the prompt replay.

-robin

Hi Robin,

TIDA-00173 does use a clever cascode topology for switching very large voltages, but with a unipolar DC bus. With a bipolar AC waveform, this kind of cascode topology does not work very well, since the current into the zener diode gate clamp will be very low when the AC waveform is near 0V and this will distort the gate drive signal.

UCC21220A is suitable for the GS66508T in the attachment. The typical UVLO rising threshold for UCC21220A is 5.5V, and the recommended operating voltage starts at 6V. This is within the gate drive range of GS66508T, as seen on the front page of the FET datasheet: "Easy gate drive requirements (0 V to 6 V)"

Is it possible to rearrange the stacked FETs so that the source nodes meet at the midpoint? Then this circuit can be controlled by 2x UCC21220A (if Q1/Q2, Q3/Q4, Q5/Q6, Q7/Q8 are in parallel) or 4x UCC21220A (if all are driven independently), 1-2 digital isolators such as ISO7720 or ISO7740 (depending on which FETs are driven in parallel), and one 6V isolated supply with a 1.2V or 2.7V series zener for the digital supplies.

If the application will absolutely not tolerate 6V gate drive, you may have more luck with some combination of isolated 5V supplies, digital isolators, and single-channel gate drivers such as UCC27511 (or equivalent in UCC2751x family) or LMG1020.

Out of curiosity, why are SiC MOSFETs not being considered for this application? A 1200V-rated bidirectional switch seems to be within the maximum line voltage ratings.

Regards

Derek:

I absolutely appreciated your curiosity based inquiry: why not use SiC devices! & let us indeed have a few exchanges on this issue- albeit outside normed  Q/A in this forum

A long struggle to accommodate  1200 V devices. Here are the critical reasons why we had to abandon them 

#1 real estate.  GaN devices when not handling extremes of SiC capabilities are almost smaller by a factor of 10 even after considering several GaN in series/parallel condition.

#2   Driver complexity. GaNs are far simpler for this particular case. More on this later. Ours have ZVS/ZCS switching, at fixed 150 kHz( resonant mode converter). There is no DC storage inside the resonant chain.

#3 Power density ( SWAP in aerospace parlance) - we perhaps are offering the smallest most efficient AC-DC possible- all told approx 20kW using fully salient properties of 3 phase aircraft generator.

#4 A specific way to assemble  GaN due to size & thermal plane offers very effective thermal management.

Pl see the layout as it stands today: 8 GaN with local LDOs(all TI), several super compact floating supplies ( using SN6501- of this tiny rigiflex).  Overall size of GaN substrate is approx 40mm by 38 mm...power delivered from this cell will be approx 1000W. We have MOSFET based synchronous rectification. Control is by using  UCD3138. But sync signals will be generated off the controller.

Will appreciate your comments.

I will definitely double check your driver chip suggestions. I may have misread some specs.

thnx much

robinGaN_layout_ti.pdf

Hi Robin,

Would you still like a layout review?

do you think you could also send the schematic so I know exactly what components are on the layout file?

thanks,

Jeff

I have to admit, I did give another try with SiC & drivers Derek suggested.

It just won't work out.

So now we have Gan System GS66508T 8 devices driven by several floating Si8273 LGA drivers.

TI  parts play a vital role all around excepting at this zone. So donno  if you would be interested in a review BUT I would love it. Then got to send it encrypted to an individual email at ti.

thnx

robin

yes will do. thnx much...I will activate the friendship email, indeed appreciate very much

Jeff: sent you a file thru your email.