ZVS and ZVT converter techniques

[Winfield Hill]

Would anyone care to elaborate, contrast, compare, and
expound on the relationship between the ZVS and ZVT
dc-dc converter concepts, approaches and technologies?
Etc., etc. [end, boki-style question.]

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Harry Dellamano]

Would anyone care to elaborate, contrast, compare, and
expound on the relationship between the ZVS and ZVT
dc-dc converter concepts, approaches and technologies?
Etc., etc. [end, boki-style question.]

Thanks,
- Win

whill_at_picovolt-dot-com

Hi Win,
ZVT is normally a fixed frequency switching PWM, with rise and fall times
controlled by LC networks to allow the switches to be changed with zero
volts across them. ZVS is normally variable frequency switching in which
one state controls input energy charging and the other state controls the
stored energy discharging to the output.
In HS physics class they told us that if you try to move energy from a
charged input cap to an uncharged output cap with a switch you would loose
half the energy, conserve charge and generate a lot of EMI. If you add an
inductor in series with the switch and open the switch at zero current, all
the energy will be transferred to the output cap with no EMI. Now this
switch is opened and closed at zero current to allow low losses but ZVS
normally employs ZCS.

boki style answer
harry

 

[Winfield Hill]

Harry Dellamano wrote...

Winfield Hill wrote ...

Would anyone care to elaborate, contrast, compare, and
expound on the relationship between the ZVS and ZVT
dc-dc converter concepts, approaches and technologies?
Etc., etc. [end, boki-style question.]

Hi Win,
ZVT is normally a fixed frequency switching PWM, with rise and fall times
controlled by LC networks to allow the switches to be changed with zero
volts across them. ZVS is normally variable frequency switching in which
one state controls input energy charging and the other state controls the
stored energy discharging to the output.
In HS physics class they told us that if you try to move energy from a
charged input cap to an uncharged output cap with a switch you would loose
half the energy, conserve charge and generate a lot of EMI. If you add an
inductor in series with the switch and open the switch at zero current,
all the energy will be transferred to the output cap with no EMI. Now this
switch is opened and closed at zero current to allow low losses but ZVS
normally employs ZCS.

boki style answer
harry

Thanks very much Harry, for your boki-style answer! I should have said
contrast and compare phase-shift ZVS with ZVT; both are fixed-frequency
approaches. Actually, Unitrode/TI in discussing the UCC3895 phase-shift
controller IC calls it ZVT switching, whereas LTC with their new LTC3722
calls it ZVS switching (or say ZCT and ZCS, as appropriate, natch).

In one IEEE paper* covering both ZVS and ZVT (few articles use both terms),
there was a distinction made with "soft switching," another widely-used,
under-defined term. Elsewhere a suggestion was made that FETs associated
with ZVS can and should have low-gate-current turnon (given that they're
really in ZCS mode), yet employ high-current hard-switching turnoff (given
that full current may be flowing at turnoff). Both these sources showed
capacitors added across the FETs, not as snubbers (no series resistors,
etc.), but as circuit elements. Are capacitors key soft-switching parts?

I'm interested in high-frequencies, e.g. 600kHz, and have been using the
FET's intrinsic Coss capacitance; although nonlinear, it is natural. :>)
Another issue is that at high currents we're not really talking resonance
with 1/4-cycle waveforms, but rather dV/dt = I/C, which results in faster
rail-rail transition times at high currents and pushes one toward adaptive
turnon delays. (It should be pointed out that in my present project design
an over 100:1 power-delivery range is involved. Design for full-power soft
switching and let the chips fall where they may at low power?)

More about role of soft-switching ... boki-style responses please!

* Maria D. Bellar, et. al., "A Review of Soft-Switched DC-AC Converters,"
IEEE Trans on Ind. Apps, 34 - 4, pg 847-860, July 1998.

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Joe Legris]

More about role of soft-switching ... boki-style responses please!

* Maria D. Bellar, et. al., "A Review of Soft-Switched DC-AC Converters,"
IEEE Trans on Ind. Apps, 34 - 4, pg 847-860, July 1998.

Thanks,
- Win

whill_at_picovolt-dot-com

What is this thing boki-style?

 

[Harry Dellamano]

Harry Dellamano wrote...

Winfield Hill wrote ...

Would anyone care to elaborate, contrast, compare, and
expound on the relationship between the ZVS and ZVT
dc-dc converter concepts, approaches and technologies?
Etc., etc. [end, boki-style question.]

Hi Win,
ZVT is normally a fixed frequency switching PWM, with rise and fall times
controlled by LC networks to allow the switches to be changed with zero
volts across them. ZVS is normally variable frequency switching in which
one state controls input energy charging and the other state controls the
stored energy discharging to the output.
In HS physics class they told us that if you try to move energy from a
charged input cap to an uncharged output cap with a switch you would loose
half the energy, conserve charge and generate a lot of EMI. If you add an
inductor in series with the switch and open the switch at zero current,
all the energy will be transferred to the output cap with no EMI. Now this
switch is opened and closed at zero current to allow low losses but ZVS
normally employs ZCS.

boki style answer
harry

Thanks very much Harry, for your boki-style answer! I should have said
contrast and compare phase-shift ZVS with ZVT; both are fixed-frequency
approaches. Actually, Unitrode/TI in discussing the UCC3895 phase-shift
controller IC calls it ZVT switching, whereas LTC with their new LTC3722
calls it ZVS switching (or say ZCT and ZCS, as appropriate, natch).

In one IEEE paper* covering both ZVS and ZVT (few articles use both terms),
there was a distinction made with "soft switching," another widely-used,
under-defined term. Elsewhere a suggestion was made that FETs associated
with ZVS can and should have low-gate-current turnon (given that they're
really in ZCS mode), yet employ high-current hard-switching turnoff (given
that full current may be flowing at turnoff). Both these sources showed
capacitors added across the FETs, not as snubbers (no series resistors,
etc.), but as circuit elements. Are capacitors key soft-switching parts?

I'm interested in high-frequencies, e.g. 600kHz, and have been using the
FET's intrinsic Coss capacitance; although nonlinear, it is natural. :>)
Another issue is that at high currents we're not really talking resonance
with 1/4-cycle waveforms, but rather dV/dt = I/C, which results in faster
rail-rail transition times at high currents and pushes one toward adaptive
turnon delays. (It should be pointed out that in my present project design
an over 100:1 power-delivery range is involved. Design for full-power soft
switching and let the chips fall where they may at low power?)

More about role of soft-switching ... boki-style responses please!

* Maria D. Bellar, et. al., "A Review of Soft-Switched DC-AC Converters,"
IEEE Trans on Ind. Apps, 34 - 4, pg 847-860, July 1998.

Thanks,
- Win

I'm sticking with ZVS uses frequency modulation for power control and ZVT
uses fixed frequency PWM but maybe Maria knows best, but only boki knows.
Look at adding Cds and increasing the circulating current to force this
added capacity to the other rail thus increasing rise and fall times,
putting less stress on your driver and generating less EMI. This will
greatly reduce your FET's turn-off loses with some added conductive losses.
Since the newer FET's lower Rds_on, the overall efficiency will be improved.
Nothing like those controlled rise and fall times and the room radio coming
in crystal clear.
regards
harry

 

[Yzordderex]

I've been playing with the older UC3875 part. My plan is to use it to
modulate a class E resonant ZVS AM transmitter. I have similar
problem - Have to go from zero current to about 3 amps at 5-10 kHz on
the output. I'm also playing with the adaptive delay time. I now
have current sources on the delay set pins which are slaved to the
input voltage command - messy. I spoke to Unitrode and they agreed
that this may be one of the applications where curent mode control
isn't the better way to go. I think the newer 3895 part looks like it
has an enhanced adaptive delay scheme which should get you down to low
current switching. I think it looks at peak current and adjust delay.
The idea of a small capacitor sounds counter productive, but my gut
is agreeing with you. I'll have to mull that one over.

regards,
Bob

Harry Dellamano wrote...

Winfield Hill wrote ...
Would anyone care to elaborate, contrast, compare, and
expound on the relationship between the ZVS and ZVT
dc-dc converter concepts, approaches and technologies?
Etc., etc. [end, boki-style question.]

Hi Win,
ZVT is normally a fixed frequency switching PWM, with rise and fall times
controlled by LC networks to allow the switches to be changed with zero
volts across them. ZVS is normally variable frequency switching in which
one state controls input energy charging and the other state controls the
stored energy discharging to the output.
In HS physics class they told us that if you try to move energy from a
charged input cap to an uncharged output cap with a switch you would loose
half the energy, conserve charge and generate a lot of EMI. If you add an
inductor in series with the switch and open the switch at zero current,
all the energy will be transferred to the output cap with no EMI. Now this
switch is opened and closed at zero current to allow low losses but ZVS
normally employs ZCS.

boki style answer
harry

Thanks very much Harry, for your boki-style answer! I should have said
contrast and compare phase-shift ZVS with ZVT; both are fixed-frequency
approaches. Actually, Unitrode/TI in discussing the UCC3895 phase-shift
controller IC calls it ZVT switching, whereas LTC with their new LTC3722
calls it ZVS switching (or say ZCT and ZCS, as appropriate, natch).

In one IEEE paper* covering both ZVS and ZVT (few articles use both terms),
there was a distinction made with "soft switching," another widely-used,
under-defined term. Elsewhere a suggestion was made that FETs associated
with ZVS can and should have low-gate-current turnon (given that they're
really in ZCS mode), yet employ high-current hard-switching turnoff (given
that full current may be flowing at turnoff). Both these sources showed
capacitors added across the FETs, not as snubbers (no series resistors,
etc.), but as circuit elements. Are capacitors key soft-switching parts?

I'm interested in high-frequencies, e.g. 600kHz, and have been using the
FET's intrinsic Coss capacitance; although nonlinear, it is natural. :>)
Another issue is that at high currents we're not really talking resonance
with 1/4-cycle waveforms, but rather dV/dt = I/C, which results in faster
rail-rail transition times at high currents and pushes one toward adaptive
turnon delays. (It should be pointed out that in my present project design
an over 100:1 power-delivery range is involved. Design for full-power soft
switching and let the chips fall where they may at low power?)

More about role of soft-switching ... boki-style responses please!

* Maria D. Bellar, et. al., "A Review of Soft-Switched DC-AC Converters,"
IEEE Trans on Ind. Apps, 34 - 4, pg 847-860, July 1998.

Thanks,
- Win

I'm sticking with ZVS uses frequency modulation for power control and ZVT
uses fixed frequency PWM but maybe Maria knows best, but only boki knows.
Look at adding Cds and increasing the circulating current to force this
added capacity to the other rail thus increasing rise and fall times,
putting less stress on your driver and generating less EMI. This will
greatly reduce your FET's turn-off loses with some added conductive losses.
Since the newer FET's lower Rds_on, the overall efficiency will be improved.
Nothing like those controlled rise and fall times and the room radio coming
in crystal clear.
regards
harry

 

[Peter O. Brackett]

Joe:

[snip]

What is this thing boki-style?

[snip]

"boki" is a "well known" USENET poster.

If you hang on USENET engough and the boki posts you will have your answer.

 

[Fred Bloggs]

Would anyone care to elaborate, contrast, compare, and
expound on the relationship between the ZVS and ZVT
dc-dc converter concepts, approaches and technologies?
Etc., etc. [end, boki-style question.]

Thanks,
- Win

whill_at_picovolt-dot-com

See Robert W. Erickson's FoPE- I know you own it-)

 

[Genome]

Would anyone care to elaborate, contrast, compare, and
expound on the relationship between the ZVS and ZVT
dc-dc converter concepts, approaches and technologies?
Etc., etc. [end, boki-style question.]

Thanks,
- Win

whill_at_picovolt-dot-com

You are looking for Definitions of Acronyms.

ZVS sounds like zero voltage switching
ZVT sounds like zero voltage transistions

Here's another one

MVS, minimum voltage switching.

And you might try

AMVS, adaptive minimum voltage switching.
Also known as AdMVS.

While I'm being marketing man I'll also include AdZVS... It just rolls of
the tongue.

Just made them up and they are now copyright and trademark me. Genome in
conjunction with the sign off DNA are also copyright and trademark me, as is
anything that me does.

OK

ZVT is a guess as to when things might give you ZVS. That's your phase
modulated bridge thing. That's why your latest version of the IC tries to
cope with some idea of the available energy that is used to stick the switch
voltage at nuffink.

It's a bloody hard subject.

ZVS uses something to look at when the switch has nuffink volts across it
and turn it on.

AdZVS knows it can't catch the sweet spot and adjusts itself accordingly.

PredZVS predicts it on the basis of something else.

MVS deals with converters that can't reach zero volts on the switch but
tries to minimise the voltage across the switch when it turns on.

AdMVS adapts the turn on time to coincide with its idea of when it should
happen.

PredMVS adapts the turn on time according to other parameters.

Lovely, all makes sense now.

Blarg Blarg, Off to buy some spudsz.

DNA

 

[R.Legg]

In one IEEE paper* covering both ZVS and ZVT (few articles use both terms),

there was a distinction made with "soft switching," another widely-used,
under-defined term.

This must have been an interesting exercise in semantics, but the
terms are used interchangeably, depending mostly on the power level,
application and school.

Elsewhere a suggestion was made that FETs associated
with ZVS can and should have low-gate-current turnon (given that they're
really in ZCS mode), yet employ high-current hard-switching turnoff (given
that full current may be flowing at turnoff). Both these sources showed
capacitors added across the FETs, not as snubbers (no series resistors,
etc.), but as circuit elements. Are capacitors key soft-switching parts?

The unavoidable existence of the capacitance makes ZV turn-on a
desirable feature at higher frequency. When this capacitance becomes a
required feature for predictable operation, however, it tends to be
desirable to swamp the strays with known, controlled quantities.

Current need not be low at the node being switched - in resonant
converters it may even be at a maximum in the reverse direction, when
fet gate bias is applied . If reverse transfer capacitance is not
discharged by the driver, then drive power is reduced. If low dV/dT
gate drive waveforms are acceptible then resonant gate drive may be
acceptible - with drive energy being recovered in each cycle.

I'm interested in high-frequencies, e.g. 600kHz, and have been using the
FET's intrinsic Coss capacitance; although nonlinear, it is natural. :>)
Another issue is that at high currents we're not really talking resonance
with 1/4-cycle waveforms, but rather dV/dt = I/C, which results in faster
rail-rail transition times at high currents and pushes one toward adaptive
turnon delays. (It should be pointed out that in my present project design
an over 100:1 power-delivery range is involved. Design for full-power soft
switching and let the chips fall where they may at low power?)

Very few practical loads exhibit an operating consumption range that
approaches 100:1. In critical applications, eating 5% of the rating at
lighter loads will not affect the thermal profile of the total
package. Choosing unrealistic or unnecessary goals can be a major
stumbling block in getting anything constructive done.

The whole idea behind the design of application-specific power sources
is to address the application's needs only, realizing a pre-calculated
commercial benefit.

RL

 

[Winfield Hill]

Genome wrote ...

Winfield wrote ...

Would anyone care to elaborate, contrast, compare, and
expound on the relationship between the ZVS and ZVT
dc-dc converter concepts, approaches and technologies?
Etc., etc. [end, boki-style question.]

You are looking for Definitions of Acronyms.

ZVS sounds like zero voltage switching
ZVT sounds like zero voltage transistions

Here's another one

MVS, minimum voltage switching.

And you might try

AMVS, adaptive minimum voltage switching.
Also known as AdMVS.

While I'm being marketing man I'll also include AdZVS... It just rolls of
the tongue.

Just made them up and they are now copyright and trademark me. Genome in
conjunction with the sign off DNA are also copyright and trademark me, as is
anything that me does.

OK

ZVT is a guess as to when things might give you ZVS. That's your phase
modulated bridge thing. That's why your latest version of the IC tries to
cope with some idea of the available energy that is used to stick the switch
voltage at nuffink.

It's a bloody hard subject.

ZVS uses something to look at when the switch has nuffink volts across it
and turn it on.

AdZVS knows it can't catch the sweet spot and adjusts itself accordingly.

PredZVS predicts it on the basis of something else.

MVS deals with converters that can't reach zero volts on the switch but
tries to minimise the voltage across the switch when it turns on.

AdMVS adapts the turn on time to coincide with its idea of when it should
happen.

PredMVS adapts the turn on time according to other parameters.

Lovely, all makes sense now.

Blarg Blarg, Off to buy some spudsz.

DNA

Thanks DNA, there's gold in that answer, someplace. :>)

- Win