diyAudio

diyAudio (http://www.diyaudio.com/forums/index.php)
-   Class D (http://www.diyaudio.com/forums/class-d/)
-   -   Disabling MOSFET body diode... (http://www.diyaudio.com/forums/class-d/122797-disabling-mosfet-body-diode.html)

peranders 14th May 2008 02:21 PM

Quote:

Originally posted by Pafi
peranders!

What is "475 A"?

It is four hundred and seventy five amperes :D

http://www.diyaudio.com/forums/showt...97#post1484597

ChocoHolic 22nd May 2008 11:06 PM

Quote:

Originally posted by RiskCord
Thanks Workhorse for your comments. The amplifier I am designing will work in +- 75. Then the impact of freewheeling currents will be quite smaller than in your case, but it would be quite interesting to compare both performances with and without series schottkys. Sadly I dont have the time to build both prototypes so I guess I'll have to pick one in advance. I'll run simulations as accurate as possible and decide on this basis...

..as accurate as possible?
How about trying as accurate as necessary?
It is allowed to neglect quite a lot of things, but I would like to highlight two crucial points that are most often not simulated, but tend to show their dragon face in reality.
- The parasitic layout inductances and the parasitic inductances of the devices.
Nice start for TO-220 is to simulate roughly 5nH in every leg in order to include the unavoidable geometric properties of that package. Also put about 0.5nH inductance for every mm copper trace in your high current paths (MosFets, free wheeling diodes, rail caps....).
- Your diode models should incoporate the reverse recovery characteristic.

And then simulate hard switching conditions, means go for output currents above the filter ripple current... Force PSPICE to go for small time steps (<10ns), otherwise you might miss the clue...

RiskCord 23rd May 2008 04:39 AM

As accurate as necessary... I like that better.

Yes, I included the parasitic inductances, the non-linear Gate-Drain and Drain-Source capacitances and I made a model representing the reverse recovery characteristics of the diode. The waveforms looks a bit ugly, so I guess the simulation is working just fine :D

Thanks for the advice on the inductance of PCB traces. I didnt know how to estimate them so I just doubled the values of the MOSFETs inductances. Anyway, I'm just about to start the PCB design and I guess I'll upload my design here so you can have some fun criticizing it :D

By the way, I'm not using PSpice. Instead, NL5 is my choice. It works on piece wise linear approximation which is quite suitable for non-linear applications as it is extremely fast and doesn't have (usually) convergence issues.

ChocoHolic 23rd May 2008 07:21 AM

Quote:

Originally posted by RiskCord

The waveforms looks a bit ugly, so I guess the simulation is working just fine :D


...resonances in the frequency range between 10MHz-100MHz ?
You seem to be on the right track.
;)

fernando_g 23rd May 2008 05:44 PM

Quote:

Originally posted by Pabo
Tekko and Dave

Just consider that the current has to leave the channel and enter the parallell diode in a matter of nano seconds if it is going to give an improvement on reverse recovery. If the parallell diode has 0,2V lower drop and the stray inductance and internal inductance inside the MOSFET/diode is 10nH you can easily calculate that it takes 20A of current 1 microsecond to "move" to the other diode. Has anyone here seen an improvement on EMI, efficiency or THD by doing this?

I remember attending a technical presentation by one of the major MOSFET manufacturers (IR, Fairchild? don't remember) that explained this very well. They made the point that the only way to achieve the intended results was to co-package the Fet and the schotkky.
Of course, they also had in the market Fet-schottky devices they were trying to sell...


;) ;) ;) ;) ;)

mias 26th May 2008 04:46 AM

2 Attachment(s)
Hello,

I am new here in this forum and I am very interested in this topic.

An interesting alternative to the shottky diode is a
- voltage control
- of the output voltage of the freewheel FET
- during the transition between the
- freewheel phase and the
- switching phase

If you
- control the output voltage to a value, e.g. 0.6V
- the intrinsic diode of the freewheel FET
- will not become conductive.

I attached some waveforms. If anybody is interested, I will give much more explanations.

Greetings,

Matthias

P.S.: I was posting something before. Please consider this as a draft. I was pushing the Submit Reply Bottom unintentionally.

lumanauw 26th May 2008 04:56 AM

Hi, Fernando,

Quote:

I remember attending a technical presentation by one of the major MOSFET manufacturers (IR, Fairchild? don't remember) that explained this very well. They made the point that the only way to achieve the intended results was to co-package the Fet and the schotkky.
What part #/manufacturer is that FET/Schottky? The IXYS one is very difficult to get.

lumanauw 26th May 2008 04:58 AM

Hi, MIAS,

Quote:

An interesting alternative to the shottky diode is a.......
Very interesting. Could you tell more?

In your right drawing, there is additional C at the base of totem pole and R+dioda from base of totempole driver to drain of mosfet.

What are these for?

classdphile 26th May 2008 09:32 AM

It's just a snubber, common technique to slow the switching transition, long since demonstrated on this forum as well.

Pafi 26th May 2008 11:41 AM

Not really. It's there to control on-state Vds. It works very similar to schottky transistors.

The problem is it is useful only if Iout*Rdson<0.5 V, wich is not usual in high-power ClassD amps.


All times are GMT. The time now is 04:11 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright ©1999-2017 diyAudio

Wiki