Single switch flyback for 350W - diyAudio
Go Back   Home > Forums > Amplifiers > Power Supplies

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 13th December 2011, 01:49 PM   #1
Banned
 
Join Date: Jun 2011
Default Single switch flyback for 350W

Hello,


We're doing a single switch 350W flyback (CCM) from a 390VDC input.

The dissipation due to the leakage inductance at max power is 26W.

So basically i need two 120V TVS's in series in the primary clamp.......they need to be rated to obviuosly 15W.

...so i am looking for a TO220 TVS with a TAB so i can connect it to a heatsink.

However, all the TVS's i see are axial.

Do you know of a TO220 TVS? (or similar)
  Reply With Quote
Old 13th December 2011, 04:53 PM   #2
mag is offline mag  Switzerland
diyAudio Member
 
Join Date: Jan 2008
Location: Chiasso
Why not using an RCD clamp instad of a TVS clamp?
TVS clamp is a nightmare for EMI.

You can also make a combination of both: set your RCD clamp at around 200V and parallel the RC with 2 120V TVS.

The TVS will normally not conduct but they can save your mosfet during transient conditions at startup, turnoff or short circuit.
Your 26W will be dissipared in the RCD resistor, you can use a bunch of them
in parallel to spread the dissipation.

I am not aware of TVS is TO220 or other high power packages a have also seen only axial or SMD types.

26W to be dissipated in the clamp is huge.
Consider the following:

1) Reduce the leakage inducatance as much as possible
2) Reduce your switching frequency
3) Let the drain swing to higher voltage and use a 800-900V rated mosfet.
The dissipation in the clamp is inverse proportional to the ratio of
Vclamp/Vor where Vclamp is the maximum clamp voltage, Vor is the
reflected voltage. The higher you let the clamp voltage go, the lower the
losses but the higher the voltage stress on the mosfet.

4) consider using 2 switch (rated 500-600V): no more leakage inductance
problems and increased efficiency. Single switch flyback at 350W
is not so easy to do....

ciao
-marco
  Reply With Quote
Old 13th December 2011, 05:12 PM   #3
Banned
 
Join Date: Jun 2011
Thanks..........though i am thinking of doing an RCDZ clamp for the primary.

This only dissipates the 26W when its on peak power of 350W.

(Its the same as an RCD clamp but there is a TVS in series with the resistor)

This 26W of dissipation is only for those "stray" transformers which somehow manage to get through the manufacturer's inspection process even though they have a horrible 2% leakage inductance.

-we expect most of our transformers to have 1% leakage, so thats just 15W dissipation in the clamp at max load.

-call me a hypocrit...... but i dont expect this smps to be doing more than 1/8 peak power, so i am not bothered about the high clamp dissipation at peak power.

So, its 350W, but only 350/8 = 45W in truth....and i just can't justify a high side drive and a second fet for 45W.

I have found that 800V FETs these days with <1R of rdson are quite cheap (<1$ for 5000 pces).

I would really like to do this with no PFC, but in that case the clamp dissipation will get a bit too much.

This smps is certainly going to look silly with an RCDZ clamp comprising six 30V TVS's in series.

But as i mentioned, the average power is very low.

-Or at least, i am assured that it will be by those here.
  Reply With Quote
Old 13th December 2011, 06:45 PM   #4
mag is offline mag  Switzerland
diyAudio Member
 
Join Date: Jan 2008
Location: Chiasso
2% leakage is not that bad.
Which core are you using? Using long cores you can reduce the leakage somewhat.
Independently on how you make your clamp the power to be dissipated is always the same;
I was suggesting RCD(Z) just because it is easier to burn big power on resistors rather than tvs.
If you are working at fixed switchig frequency you can also thing about resonant clamp.
It is a kind of clamp based on a LCD combination that can partially recycle the energy.
I don't know it in details but if I remember well there was an old Unitrode seminar explaining this. Take a look at the TI website.

What is your output voltage?
what is your switching frequency?
Does the smps need to be isolated? If not a sepic can be an alternative but you will need a huge coupling capacitor.

Infinieon has a lot of 800V mosfets in their coolmos C3 series that are reasonably priced.
ST has a mosfet called stw6n120k3 rated at 1200v with 1.6ohm priced around 1.2$ for 5k
I have used it for a 160w single switch flyback working from 187 to 550vac, the efficiency was something around 86% at 187vac full load.

You can try a very high voltage mosfet and check on paper how much do you gain in clamp dissipation but I don't expect miracles from it.

Now you see why I like two switch flybacks? Imho it is the clamp the real bottleneck of a flyback smps.

Ciao

-marco
  Reply With Quote
Old 13th December 2011, 07:36 PM   #5
Banned
 
Join Date: Jun 2011
output is +/-40V, 4A plus there is a 5V, 1A processor supply.

F(sw) = 67KHz.

I will use ETD type core.

I see what you mean about the clamp, but it only dissipates big power at peak load............and that wont happen often...........also, i will put an integrating current sensor on the output which turns the smps off if it is delivering current levels which are too much for normal music....(ie the user is abusing the euipment).

Im sorry but i just cant justify a high side drive and fet for this applicaion which is 350W peak, but just 45W on average.......ill have an ugly RCDZ clamp network instead.

2% leakage may be good but im hoping to find a winder that can do 1% for all trafo's.
  Reply With Quote
Old 13th December 2011, 08:11 PM   #6
diyAudio Member
 
Join Date: Apr 2006
Location: Minnesota
I noticed you are using CCM. As I recall, the inductance for DCM mode can be made quite a bit small than for CCM. This should help reduce the leakage inductance. On the other hand I think the peak primary current is higher in DCM.
  Reply With Quote
Old 13th December 2011, 08:17 PM   #7
Banned
 
Join Date: Jun 2011
yes i believe youre right.....but ill just use a bigger core and hence less turns if leakage is too high with this CCM design.
-also loss due to leakage is proportional to 1/2LI^2.f so you are right , it probably wouldnt make that much difference.
  Reply With Quote
Old 13th December 2011, 10:24 PM   #8
Banned
 
Join Date: Jun 2011
Actually i have changed my mind about this single switch 350W flyback, -on simulation i am getting enormous switching losses......switching losses alone are about 37W in the FET.......The peak drain voltage is clamped at 640V.

The conduction losses are just 2W in the FET.

So i think i have found out why people dont do single switch flybacks at 350W from 390VDC input...switching losses.

A simulation of a two transistor flyback of the same power gives switching losses in each fet of about 3.8W......much less than with the single transistor flyback.

Last edited by eem2am; 13th December 2011 at 10:25 PM. Reason: spell
  Reply With Quote
Old 14th December 2011, 10:04 AM   #9
Banned
 
Join Date: Jun 2011
woops.....i changed the FET from SPA11N60C3 to STW11NM80 and the switching losses have gone down to just 6W now.
*
These two transistors have equal Rdson and equal Qg......but the latter is rated to 800V...the former only to 650V.
*
So i am assuming that the simulator must have been modelling avalanche fet breakdown or something.
-or maybe the SPA11N60C3 has a high Cds.

-i
  Reply With Quote
Old 14th December 2011, 10:15 AM   #10
mag is offline mag  Switzerland
diyAudio Member
 
Join Date: Jan 2008
Location: Chiasso
Strange...

Infineon CoolMos C3 and ST MDMesh II (NMxxxN in the part number) have more or less the same characteristics.

The higher Cds (Coss) for the Infineon part is an advantage. The number stated in the datasheet is measured at 20V or something about, it is practically useless.

Coss is voltage dependent in a highly non linear way, decreasing as the VDS increases. In coolMOS the Coss is very high at low voltage and decreases to very small values above 100V. This created a kind of snubbing effect to the VDS rise at turn off that contribute to decrease the switching losses.

Be careful with coolMOS, these parts are so fast that your dv/dt can reach dangerously high values (I have seen more than 50V/ns) if turned off too hard. In my experience I have had the best results using not less that 10R for gate turn off resistor.

What is your peak drain voltage? do you really need a 800V part or you can stay on 600/650V?

Using a 800V part will be safer but for the same price you have higher RDS_on and higher swiching times.

These things have to be tried on a prototype, forget the simulator. Even if you have model for your mosfet parts the results you get are often completely wrong and don't represent the reality of thing.
Simulating an SMPS with Spice is hard when it is related to switching events or other highly non linear phenomenon.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Switching dual supply rails with a single switch SirNickity Construction Tips 20 23rd August 2012 05:31 AM
A switch for 2 power source for a single amp abcbeck Power Supplies 4 7th August 2011 10:04 AM
Is this off-line flyback? B+ switch, Startup troubleshooting steps? gmunee Power Supplies 3 21st December 2010 08:34 AM
poweramp as good as gets for 350w prorms Solid State 21 20th September 2006 08:22 PM
How to switch single pair speakers across multiple amps? ransom peek Multi-Way 0 21st February 2005 03:30 PM


New To Site? Need Help?

All times are GMT. The time now is 10:54 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2