FET driver IC with adjustable dead time for inverting buckboost SMPS?

[treez]

Hello,
We are designing a +/-48V , 400W Power supply for a Class D, Bass guitar amplifier supply.

The +48V rail will simply be supplied by the output of an offline, 48V PSU bought off-the-shelf.
The –48V rail will be provided from the same +48V rail by putting it through an inverting buckboost converter as in the diagram and LTspice simulation below.

The problem is that we need it to be synchronous. (Use a FET across the output diode).
Normal synchronous controller IC’s are for synchronous buck or synchronous boost converters and can’t be used with inverting buckboost converters like we have here.

Therefore, we need an IC which will drive the power FETs and have a variable dead time, then we can add the gate drivers and gate_drive_isolators ourselves. Most of the relevant driver ICs have dead time which is far too short, and can’t be used with the external gate driver ICs that we will be using, since gate driver ICs have relatively long turn-on delay times which would result in overlap and “shoot-through”.

So do you know of eg a synchronous buck driver IC which has adjustable dead time? Such an IC could be used by us for this inverting buckboost.

 

[D.A.(Tony)Stewart]

Mismatched Power Supplies for output impedance and load currents is only asking for trouble with lag on startup and brown-out conditions with large DC outputs. woof...

Are these not isolated that you can invert polarity directly?

What is your power budget 200W per rail?
What is your cost budget $0.10/W? and budget for startup, overload synchronous protection and EMI control?

Deadtime is easily controlled with nonlinear series impedance. Fast turn off , slower turn on.

Also take a look at my https://www.edaboard.com/threads/327654/#post1400634 LC filter.

 

[treez]

yes they are isolated ,but we need +/-48v, 400w.

 

[D.A.(Tony)Stewart]

yes they are isolated ,but we need +/-48v, 400w.

So connect two 400w floating supplies in series with centre grounded then you have 800W +\-40V? Is that too obvious?

 

[treez]

thanks, but that's using two offline power supplies........which are big things...we don't need two offline psu's. We just need a +/-48v, 400w output......and the way to do that, which takes up less room, is to just put a +48v to -48v smps at the output of the single +48v psu, and then we have +/-48v.

The offline psu's are good but they are big as they have internal pfc etc etc...we only have room for one OTS offline psu.

 

[D.A.(Tony)Stewart]

Mismatched Power Supplies for output impedance and load currents is only asking for trouble with lag on startup and brown-out conditions with large DC outputs. woof...

Are these not isolated that you can invert polarity directly?

What is your power budget 200W per rail?
What is your cost budget $0.10/W? and budget for startup, overload synchronous protection and EMI control?

Deadtime is easily controlled with nonlinear series impedance. Fast turn off , slower turn on.

Also take a look at my https://www.edaboard.com/threads/327654/#post1400634 LC filter.

You are still asking for problems with your suggested solution without addition protection frm imbalance.

Can you answer my questions?
What about two 200W 48V supplies in series, which can be quarter bricks, but cost more.

 

[treez]

What about two 200W 48V supplies in series, which can be quarter bricks, but cost more.

One 400w OTS PSU, with us adding a buckboost at the output would be cheaper, and smaller.

You are still asking for problems with your suggested solution without addition protection frm imbalance.

This will not be a problem, since we account for this, we have already done it with a flyback at the output to provide the -48v, and its fine, but now we wish to use buckboost instead of flyback.

I cannot say cost budget exactly, we want as cheap and small as we can get.

Power budget is 200w per rail

 

[D.A.(Tony)Stewart]

One 400w OTS PSU, with us adding a buckboost at the output would be cheaper, and smaller.


This will not be a problem, since we account for this, we have already done it with a flyback at the output to provide the -48v, and its fine, but now we wish to use buckboost instead of flyback.

I cannot say cost budget exactly, we want as cheap and small as we can get.

Power budget is 200w per rail

I can get PSU for $0.10/W or $0.50/W or $2/W.

Each has tradeoffs for quality, size and performance.

Since you cannot specify any of these, it cannot be solved.

IMHO if you cannot specify cost, size and performance that avoids imbalance issues, you cannot design it.

 

[treez]

We have already got round any imbalance issue.
We can find the OTS ,offline, +48V PSU no problem. We have that done.
We are already using a flyback to get -48V, but want to do it with a buckboost instead...and we wish the buckboost to be an synchronous inverting buckboost.

I appreciate you are discussing costs, but just in case, we must say that we simply cannot afford to employ external PSU consultancies or contractors.
We had a consultancy come to use and ask £100K for an offline +/-48V, 400w psu, and then they wanted 30% of any profits that we made from it. We just had to stop thinking about consultancies.

 

[D.A.(Tony)Stewart]

Can you at least spec
Max Size and
Not to exceed cost and
Noise, transient tolerance ( worst case) ( to power line quality issues ... brown out step input change, 10k power cycles etc)

bad consultants 'n doctors and lawyers exist everywhere. skill comes from recognizing the attributes of a good one.

 

[treez]

Well, Here is our OTS 500W, 48v psu
**broken link removed**

 

[treez]

The attached SMPS is a Cuk converter and can give us 48V to minus 48V at 200w conversion, and does not need a transformer, however, it has a resonating problem.
That is , if the output load is pulsating at approx. 7600Hz (which is very likely with a class D audio amplifier load), then it will go into damaging oscillatory behaviour due to the resonant frequency of the L,L,C of the Cuk power stage.

LTspice simulation also attached.

- - - Updated - - -

 

[D.A.(Tony)Stewart]

The attached SMPS is a Cuk converter and can give us 48V to minus 48V at 200w conversion, and does not need a transformer, however, it has a resonating problem.
That is , if the output load is pulsating at approx. 7600Hz (which is very likely with a class D audio amplifier load), then it will go into damaging oscillatory behaviour due to the resonant frequency of the L,L,C of the Cuk power stage.

LTspice simulation also attached.

- - - Updated - - -

Yes if course it will oscillate with a tuned circuit in the inverter in the audio band.

Consider changing the design parameters to increase f by x3-x5 and reduce L-L to <5uH which has impedance of 1 Ohm and ensure both supply ripple and audio signal is suppressed with sufficient band stop above 20KHz with suitable filters.

N.B. Series caps MUST be able to handle the current ripple .


Otherwise change control to PWM only with carefully designed PID feedback on output voltage to eliminate the chaotic behaviour and at least 10db gain margin under worst case input noise.

- - - Updated - - -

Otherwise considering your cost budget is around $300, reconsidering the case design for 2 independent 1U high 80$ 200W~250W isolated PFC 50V supplies.

 

[mtwieg]

I have to agree with what Sunny is getting at. Trying to make a bipolar supply by cascading the negative output from the positive output leads to all sort of stability and balancing concerns, not to mention that now the positive regulator needs to have twice the capacity. Also the standard buckboost converter is not very suitable for hundreds of watts, much like the flyback.

I would also suggest not cascading any supplies, and using two identical 200W isolated 48V supplies in series.

 

[D.A.(Tony)Stewart]

I have to agree with what Sunny is getting at. Trying to make a bipolar supply by cascading the negative output from the positive output leads to all sort of stability and balancing concerns, not to mention that now the positive regulator needs to have twice the capacity. Also the standard buckboost converter is not very suitable for hundreds of watts, much like the flyback.

I would also suggest not cascading any supplies, and using two identical 200W isolated 48V supplies in series.

Yes a 96V supply or a +/-48 supply with a differential load , can add balance problems and coupling instabilities with intermodulation of any residual noise but if half bridge driver uses single supply drain, then a good braided central ground ought to stablize it.

I can suggest 1u high PSU's if I have a dynamic load spec.

also, unless the Class D has some negative feedbak with compensation, poor PSSR will show as audible noise.

dual bipolar supplies need to befined and designed from scratch for high audio performance with good efficiency, stability possible but never by willy nilly cascading converters. Good control, soft start, stable slope compensation over a wide load range requires control theory and accurate design skill on pulse power flow thru each component from CDSS to RdsOn, ESL, ESR and EMI management.

 

[mtwieg]

Also out of curiosity, why would an isolated class D amp need bipolar supplies? The output wouldn't be steering current from the midpoint, right? Unless it's some sort of multilevel class D?

 

[D.A.(Tony)Stewart]

Also out of curiosity, why would an isolated class D amp need bipolar supplies? The output wouldn't be steering current from the midpoint, right? Unless it's some sort of multilevel class D?

For DC coupling, I guess they either use a full bridge with single supply and each speaker lead has 50V dc on it which is OK for Car speakers that are permanently mounted.

I think they prefer bipolar supplies with a half or full bridge for 4x the power, so each speaker connection is at 0V mean with DCP ( DC offset protection) . Perhaps it depends if the the AMP is built-in or external with cables so the DC voltage is not live on standard speaker wires or plugs.

Not sure how state-of-the art boom boxes or stadium Amps work, but IF I designed one it would be a 2kW single supply system @ 2 Ohm with 2 level supply voltage and cascode drivers or Class F ? ( I think)