SCR vs MOSFET and FLYBACK vs FORWARD vs RESONANT HALF BRIDGE

[SSShocked]

hi guys
i have a few questions
what is better to use in high power SMPSs-SCR or MOSFET and how to drive SCR with SG3525
the other is what topology should i use-flyback,forward or resonant half bridge
also how to make a MOS controlled thyristor

 

[WimRFP]

All depends on what you want (power, efficiency, money available, supply voltage, etc).

Most SCRs are slow devices and are used at very high power where mosfets can't operate. If you want to use a SCR, you need a zero current resonant topology the get the SCR in the off state, or you need to divert to topologies where the current is temporary carried by another SCR.

Forward topology is mostly used in high power circuits (as magnetic usage is better, less stringent on leakage inductance, less peak current). The main advantage of flyback is the low component count (one switch only) and the large control range. For flyback you need a low leakage inductance transformer and a snubber. If you use energy recycling snubbers, leakage inductance can be higher.

Full bridge forward has low ripple current, can be used in ZVS mode (where leakage inductance is part of the quasi resonant circuit).

Regarding the switch in applications fed from low voltage mains (say up to 400V) I would go for mosfet or IGBT.

You may use (quasi) resonance to reduce switching loss and EMC problems.

 

[SSShocked]

then i will use ncp1382
it's much simpler with it and is quasi resonant
but, can't i use MOS controlled triac
it is powerfull like a normal triac and is fast switching like mosfet
(All depends on what you want (power, efficiency, money available, supply voltage, etc).) -i have it all

 

[mtwieg]

SCRs are very rarely used in SMPS nowadays, since few converter topologies can guarantee that the switch current will fall to zero every cycle under all line/load conditions. Plus they are inherently slow, which goes against recent trends in SMPS design. They're really only applicable to AC-DC and DC-AC converters, and aren't useable for DC-DC converters (as far as I know. I'm fairly sure there are some arcane topologies from decades back which can use SCRs).

Even with the development of GTO thyristors, they're not really usable for SMPS, except maybe at very high power levels (hundreds of kW).

So for DC-DC converters, you have to use FETs or BJTs.

 

[SSShocked]

DAMNIT
what about
MOS controlled triac
it is powerfull like a normal triac and is fast switching like mosfet

 

[mtwieg]

Can you post the datasheet for what you had in mind?

 

[SSShocked]

http://www.onsemi.com/pub_link/Collateral/NCP1381-D.PDF
i think this is the best and lowest noise pwm
http://www.datasheetcatalog.com/datasheets_pdf/M/C/T/A/MCTA75P60E1.shtml
the MCT

 

[mtwieg]

If it's a true MOS controlled thyristor, then it's theoretically possible to use it in DC-DC topologies, but again, thyristors are really only suited to very high power levels and soft switching conditions. Even MOS controlled thyristors are much slower than typical MOSFETs, and they will perform poorly in hard switched high frequency converters.

Also you'll need specialized a gate driver to use it with normal PWM controllers. Also since it's a P channel device the gate drive will likely have to be fully isolated.

 

[SSShocked]

my is 60khz 100v 25a soft switch
i think it's ideal for it

 

[mtwieg]

my is 60khz 100v 25a soft switch
i think it's ideal for it

Quasiresonant converters are not completely soft switched (no matter how much the part manufacturers say it is). They have to hard switch off (under peak switch current and voltage conditions), which is the one thing that GTO and MOS controlled thyristors do very poorly.

I mean, just look at the datasheet. The current fall time is 1.15us typical (a factor of ten slower than typical IGBTs or FETs). The datasheet actually describes explicitly why the device would work poorly at high frequency. Just look at figure 8, and plug in your numbers. At 100V, 25A, at 60KHz you're looking at over 100W off dissipation. And figure 9 actually shows theoretical max switching frequencies, and they never go over ~25KHz. I bet in practice it's impossible to go over 10KHz in practical scenarios.

Thyristors (of any kind) aren't suitable for this application. A properly selected IGBT or FET will work far better.

 

[SSShocked]

so for high eff. and freq. i must sue IGBT

 

[WimRFP]

my is 60khz 100v 25a soft switch
i think it's ideal for it

Given the voltage and current, I would go for mosfet. Depending on (quasi) resonant topology, the switch may experience higher voltage and higher current.

A resonant topology will reduce switching loss, but may increase conduction loss and is more complex. As mtwieg also mentioned, drive circuitry for mct is more complex.

What is the maximum current and voltage in your quasi resonant topology (that your switches will experience)?

 

[SSShocked]

400v 7a
coming from the PFC

 

[WimRFP]

Depending on topology you have in mind, the maximum voltage and current as experienced by the switches can be higher then the DC link voltage (think of a flyback converter).

If the voltage experienced by the switch equals the DC link voltage (400V in your case), my first choice would be mosfet. I would also look to (partly) soft switching in combination with IGBT.

Actual topology and components will depend on available components, experience of the designer, development time available, etc.

 

[SSShocked]

forward topology

 

[mtwieg]

2.5Kw is pretty high for single-ended converters like forward and flyback (even quasiresonant flyback), so I would look more at half or full bridge converters. Maybe a two switch forward converter. At high power levels you want to try to reduce voltage stresses on the switches, so that you can use lower voltage rated devices.

 

[SSShocked]

700v 96a- hope it is enough
also i will not have the advantages of NCP1381/2
but if you say so...
any suggestions
lm2524
ncp1034
tell me what to use
can i use NCP1381/2 like that to controll two mosfets

this circuit is for LED driver, but with a few changes...
look at the second