Reliable High side drive in Offline Bridge SMPS's

[Faringdon]

Hi,
Bootstrapping chips  give a nice and small solution, but  can blow up due to negative voltages on the switching node pin.
Gate drive transformers  have the problem of resonating LC's when duty cycle change is sudden.

So what do you say to putting the two together, and having the best of both worlds?...ie, Gate drive transformer just to provide high side power, and bootstrap chip just to provide the actual gate drive signal....(into a hi side fet driver)........then a big value resistor can be put into the switching node pin of the bootstrap chip (to protect against the big negative voltages) ...and you dont have to worry about it slowing down the gate drive, because its just the "signal" into the hi side fet driver....ie not the actual gate drive "power" pulse.

[Miyuki]

Isn't something like EiceDRIVER from Infineon a solution to this problem?
When the driver is totally isolated from the logic side, and the power can be provided by bootstrap with protective filtering.

[Faringdon]

Thanks...
Eicedriver.
Yes looks good, the best single chip i've seen yet for mitigating this issue, but their datasheets have a CMTI value of 300kV/us
(common mode transient immunity)
The very fact that it states a CMTI figure, tells us that the Eicedriver is not immune to this problem of negative voltage on the switching node connected pin.
So i am afraid no, with 400V offline SMPS's, you must use a separate gate drive transformer to act as hi side power provider...then a bootstrap chip
with a series high value resistor into its switching-node-connected pin.
In truth, also, the Eicedriver is the only one of its kind on the market, so leaves you open to nil-stock issues.

In truth, the Eicedriver also needs a hi side supply anyway.

https://www.infineon.com/dgdl/Infineon-1ED314xMU12F-DataSheet-v01_00-EN.pdf?fileId=8ac78c8c85c5e5aa0185e8c38e856d7e

Isn't something like EiceDRIVER from Infineon a solution to this problem?
When the driver is totally isolated from the logic side, and the power can be provided by bootstrap with protective filtering.

Thanks its interesting ,the datasheet doesnt recomend hi side bootstrap supply for EiceDriver.

[jbb]

300 kV/us is really fast. That’s 0-400V in 1.3ns; are you going to get anywhere near that?

[Faringdon]

Thanks, no, but the fact of it "being there", declares that there is this "connection" between the isolated and non-isolated sides...and thus, that a big negative voltage on the isolated side's switching node, may screw up the chip.

[jbb]

I think the ‘connection’ is through the stray parameters of the transformer eg winding capacitance.

Whatever gate drive approach you use will be subjected to common mode transients. The amount of CM transient any given solution can cope with varies from ‘poor’ for a cheap LED optocoupler all the way through to ‘amazing’ for a fiber optic link (note you’d still need an isolated supply to drive the gate).

[uer166]

Thanks, no, but the fact of it "being there", declares that there is this "connection" between the isolated and non-isolated sides...and thus, that a big negative voltage on the isolated side's switching node, may screw up the chip.

So.. you expect to have a isolated gate drive chip to have infinite CMTI? Does your gate drive transformer solution have infinite CMTI?? Do you think galvanically isolated optos have infinite CMTI???

[Faringdon]

Thanks, i think the solution involving separate gate drive transformer as isolated supply...and separate  "digital isolater" chip,  with big value resistor into the switching_node_connected_pin is best here.
Whatever its CMTI is...it is the best way to solve the problem of the top post.

I'd say, with 400V inputs (eg PFC output), its the only reliable way.

Having said that, Phase shift full bridge never changes the duty cycle and so PSFB with gate drive transformer alone is fine, as long as you dont suddenly take duty cycle to zero in eg some response to a fault.

[Zero999]

There are isolated gate drivers with a built-in DC:DC converter for the output. I've seen the ADuM5230 used for this purpose and it works quite well.
https://www.analog.com/media/en/technical-documentation/data-sheets/adum5230.pdf

[Wolfram]

Isolated gate driver chips are superior to GDTs when it comes to CMTI, managing more than 50 V/ns with a gate drive transformer can get tricky due to injected current into the driver stage through interwinding capacitance, especially if the transformer is wound for low leakage inductance.

Midpoint voltage undershoot and CMTI-induced latchup are two completely separate issues. Fully isolated gate drive chips are immune to negative voltage on the upper FET source, within the device isolation ratings. Bootstrapping is perfectly fine, and the common solution in cases where you don't need to reach 100 % duty cycle, and don't need negative gate turn-off voltage.

https://www.infineon.com/dgdl/Infineon-EiceDRIVER_Gate_Driver_ICs-ProductSelectionGuide-v02_00-EN.pdf?fileId=8ac78c8c80027ecd018094fa56806ee1

Note that there are a ton of parts available from different manufacturers, as low as 20 cents per channel in moderate quantity.

I have used off-the-shelf GDTs to provide drive power to isolated drive chips in the past, and it works fine of course, but these days you can get dedicated 2W gate drive DC/DC converters with very low isolation capacitance and characterized CMTI and TDDB behavior for around 2.50, so it hardly seems worth it.

[JohnG]

Thanks, no, but the fact of it "being there", declares that there is this "connection" between the isolated and non-isolated sides...and thus, that a big negative voltage on the isolated side's switching node, may screw up the chip.

The fact that they quote the spec means they bother to spec it. If the number doesn't exist on a datasheet, it is probably much worse (more susceptible) to CM transients, because either they measured it and it was not good, or they didn't measure it and it probably is not good because they don't understand the importance of it. I would personally not use a part for a high side drive if the manufacturer did not specify the CMTI.

50 kV/us used to be considered good, but the days of SiC and GaN are here and they can reach >100 kV/us easily. In fact, so can superjunction Si MOSFETs during turn-off due to the crazy Coss step in the Coss vs VDS curve.

Many GD transformers have 5-10 pF CM capacitance or more, sometimes much more. This would give ~3 A (!) of CM current from secondary to primary. That current will travel into the circuit driving the transformer. Unless the transformer drive circuit has perfectly balanced impedance up to very high frequency (100s of MHz), much of this 3 A will be converted to differential mode and thus cause problems because it will completely overwhelm the logic level signals. If you think otherwise, I suggest doing some measurements on transformers.

John

[Faringdon]

Many GD transformers have 5-10 pF CM capacitance or more, sometimes much more. This would give ~3 A (!) of CM current from secondary to primary.

Thanks for bringing this up. I know what you mean.
But if i put in values of
Stray C (interwinding) =10pF
dV = 400V
dt = 20ns

...then it gives i = C.dv/dt = 0.2A.

I wasnt sure where the 3A figure came from?

GDTs do have problems as you discuss, i agree, and they always have significant stray interwinding capacitance because they usually must be wound for low leakage inductance.
However, scores of lierature  speaks of the severe problem of bootstrap chips when the input voltage is at say 400V (PFC output).
Dr Basso mentions bootstrap problems in his book.
Dr Ridley also speaks of the dangers of the bootstrap negative going voltage.

Pg 16 of this..
https://www.onsemi.com/pdf/datasheet/ncp5106-d.pdf
...speaks of the terrors of the bootstrapper.

Dr Ridley speaks of the terrors of the bootstrap chips here...
https://ridleyengineering.com/design-center-ridley-engineering/49-circuit-designs/192-a11-gate-drive-design-tips.html
...and in this document, Ridley states how the GDT offers the most rugged hi side drive solution when voltage is around 400V or so.

We cannot point to interwinding capacitance of GDTs as a massive problem, since all GDTs will be   either interleave wound or bifilar wound (to reduce leakage inductance), and so will have very significant stray interwinding capacitance, yet Ridley still says its the most rugged hi side drive method when voltages are ~400V.

Thanks Zero999,  the ADum5230 is  indeed a  superb solution to this, but pricey and a nil-stock-worry.



 

[JohnG]

Thanks for bringing this up. I know what you mean.
But if i put in values of
Stray C (interwinding) =10pF
dV = 400V
dt = 20ns

.

The 3 A came from the 300 kV/us figure you gave earlier. You gave no other numbers earlier, so I worked with that. You never mentioned anything about the actual numbers you were expecting, but those only yield 20 kV/ us. Why are you even bringing up CMTI if your real worry is something else?

By the way, many engineers seem to be able to use high-side driver ICs successfully, though they do have their gotchas. IR211x has been around for decades, and there are many more choices today. Careful attention to layout prevents a lot of negative voltage issues.

FWIW, it's been a few years since I worked above 200 V, but I had good luck with SiLabs (now Skyworks) Si827x series drivers at the time, switching GaN FETs at ~350 V in about 3-4 ns.

John

[Mr John]

Isolated gate drivers are pretty standard in HV isolated DC-DC converters nowadays. Bootstrap for the HS driver is fine as long as you have a consistent switching frequency. Switch node under shoot during hard switching causing bootstrap supply overshoot can be addressed with a resistor in series with the bootstrap diode, and if necessary, a clamping Zener. Usually the Zener isn't needed because you'll be operating in soft switching mode or burst mode most of the time.

Just make sure the bootstrap diode is fast reverse recovery. Anything over 75ns TRR will heat up pretty fast at the higher switching frequencies of today's topologies.

[mzzj]

Dr Ridley speaks of the terrors of the bootstrap chips here...
https://ridleyengineering.com/design-center-ridley-engineering/49-circuit-designs/192-a11-gate-drive-design-tips.html
...and in this document, Ridley states how the GDT offers the most rugged hi side drive solution when voltage is around 400V or so.

Note that the Ridley article is nearly 20 years old. Today the high side driver selection is far more comprehensive than the dreaded ir2110.