Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Class D Switching Power Amplifiers and Power D/A conversion

SystemD LiteAmp
SystemD LiteAmp
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 2nd September 2014, 12:43 PM   #121
basreflex is offline basreflex  Spain
diyAudio Member
 
Join Date: Oct 2007
great results.

what output transistors did you finally use? have you considered the IPD320N20 of infineon ? (22nC, 32mohm,34A,200V)
  Reply With Quote
Old 2nd September 2014, 04:26 PM   #122
Capt Grogg is offline Capt Grogg  United States
diyAudio Member
 
Join Date: Aug 2009
also Infineon IPP120N20NFD 65nc 12mO 84a with fast diode for higher power amps
  Reply With Quote
Old 2nd September 2014, 08:44 PM   #123
ChocoHolic is offline ChocoHolic  Germany
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
Some insights regarding the switching stage.
Please refer to the attached schematic and PCB.
The MosFets are two IRFI4212H.
Their specific beauty is the package, which allows low parasitic inductances, combined with an easy to handle fully isolated TO220-5.
It allows to keep small the critical half bridge loops:
Q6, C17, C18 and in the same way Q7, C13, C14
This keeps the intrinsic ringing frequency and damping high.
==> You get perfect signal integrity free of ringing already with very light snubbering.
Low Qrr of the IRFI types also keeps the situation nice during heavy load with hard switching conditions.
Downside of course is the very restricted portfolio of MosFets in this package.

Gate drive:
The shown values are the softest, but still reliable gate drive which at the same time allows to skip D7 & D8 for high end spectrum shaping.
Particularly in the mid power range (i.e. 10W into 4R) with D7 & D8 you will get some more high order harmonics, vs the posted spectrums.
However D7 and D8 are strongly advisable for heavy duty party operation into 2R, because D7 and D8 substantially reduce the switching losses at heavy loads.
During idle operation the losses are identical for both variations, no shoot through, no forced transitions, but nice ZVS.
Overall the idle losses of the main board are 3W, whereof 2W happen in the auxiliary supplies and small signal area.
Approx. 1W idle losses sum up in the MosFets and inductor.
Attached Images
File Type: png LiteAmpPWR.png (35.4 KB, 1043 views)
File Type: png LiteAmpPCB.png (103.8 KB, 1040 views)
  Reply With Quote
Old 2nd September 2014, 10:08 PM   #124
gommer is offline gommer  Belgium
diyAudio Member
 
Join Date: Sep 2012
Very nice design. Do you have any idea what causes the extra harmonics with the extra diodes? Is it the extra asymmetry of switching timing (faster off with diodes)? If so, can it be tweaked with a different dead time setting?

Another question, not for implementing in this design, but have you ever tried a separate npn-pnp push-pull for gate driving? Especially if you parallel FET's this is very beneficial in my experience (in power electronics, remember, not class D designs).

Cheers, Marc
  Reply With Quote
Old 3rd September 2014, 06:23 PM   #125
ChocoHolic is offline ChocoHolic  Germany
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
Quote:
Originally Posted by gommer View Post
... any idea what causes the extra harmonics with the extra diodes?
It is a combination of dead time distortion and dv/dt shaping.
IMHO the dominant portion is the dead time.
Dead time has to defined more detailed.
There is the dead time of the driver IC and there is the time during which upper and lower MosFet are both non conductive, which I call effective dead time.
In order to avoid heavy brute force during removing of Qrr and avoiding exzessive dv/dt afterwards it is helpful to chose moderate high drive impedances for turning ON. In order to achieve low losses during turn OFF and (ensuring that the reverse transfer capacitance cannot cause unintended turn ON effects) it is often helpful or even necessary to turn OFF with lower impedances. In many cases this leads to effective dead times, which are longer than the dead time of the driver IC. Unpleasantly the effective dead time is also depending on the load.
A multiparameter optimization which becomes even more difficult with the restrictions in dead time setting the IRS....
Besides dead time setting of the IRS also the level of gate driver supply can be used to tune a little bit.
I would not dare to state that the shown solution is the ultimate, but it is one of the good performing options, so in the moment I tend to stay with it.



Quote:
Originally Posted by gommer View Post
... ever tried a separate npn-pnp push-pull for gate driving? Especially if you parallel FET's this is very beneficial in my experience (in power electronics, remember, not class D designs).
Yes, that's fine in class D as well.
In first step already the PNP can be of great help and for real heavy loads NPN and PNP.
Just have a look to the 2kW thread...
  Reply With Quote
Old 4th September 2014, 06:44 AM   #126
gommer is offline gommer  Belgium
diyAudio Member
 
Join Date: Sep 2012
Thanks for the effort of elaborating. Very interesting. Every single detail you mention is relevant in power electronics too. I remember a design where I added an extra boost switcher in order to be able to meet minimum VCC for a gate driver IC and at the same time keep the rail for gate driving at a lower voltage. And then it all changes again when you start using SIC FET's. An interesting hobby/profession, isn't it

One thing that you might easily try: I'm used to very different gate on and off ratio's.
You could try 2E2 off and 2E2+22E on. That's usually a good starting point in my experience. I'm also used to snubbering with a much lower resistive part (sub 10E).

Last edited by gommer; 4th September 2014 at 06:47 AM.
  Reply With Quote
Old 4th September 2014, 01:28 PM   #127
Reactance is offline Reactance  South Africa
diyAudio Member
 
Reactance's Avatar
 
Join Date: Jun 2005
Location: Cape Town
Quote:
Originally Posted by ChocoHolic View Post
Some insights regarding the switching stage.
The MosFets are two IRFI4212H.
Their specific beauty is the package, which allows low parasitic inductances, combined with an easy to handle fully isolated TO220-5.
Choco

Have you looked at these ?
http://www.irf.com/product-info/data...4019h-117p.pdf

Won an cheap Chinese class-d last night off ebay and its fitted with one of those, to lazy to check do parasitic at the moment looks interesting as both fets are bonded into one package.
1pc IRS2092 Class D Mono 200W Amp Board | eBay
__________________
There’s nothing worse than the smell of burned output transistors!!
  Reply With Quote
Old 4th September 2014, 07:50 PM   #128
ChocoHolic is offline ChocoHolic  Germany
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
Quote:
Originally Posted by gommer View Post
Every single detail you mention is relevant in power electronics too. I remember a design where I added an extra boost switcher in order to be able to meet minimum VCC for a gate driver IC and at the same time keep the rail for gate driving at a lower voltage. And then it all changes again when you start using SIC FET's. An interesting hobby/profession, isn't it

One thing that you might easily try: I'm used to very different gate on and off ratio's.
You could try 2E2 off and 2E2+22E on. That's usually a good starting point in my experience. I'm also used to snubbering with a much lower resistive part (sub 10E).
...I see - you are sort of addicted as I am..
Power electronics: ClassD amps are power electronics.
More precise in SMPS terminology most classD amps would be called high speed adjustable precision synchronous rectifying four quadrant buck converters.

2E2 off and 2E2+22E on:
Fine in terms of power electronics, but for low distortion this would need a driver IC which enables negative dead time at its output in order to achieve a very small effective dead time at the MosFets.
And if the driver would provide this - I would clearly prefer your proposal over the slowish turn OFF, which I am using here.
If you do it with the restricted options of the IRS your proposal will work reliably, but dead time distortion in the mid power range would be of average level. Please also note that 2E2 OFF will shift much more heat into the IRS driver.
But true the DIL package can take pretty some losses and most likely I will need this capability for the +/-70V...+/-80V version.

snubbering with a much lower resistive part (sub 10E)
This snubber is not intended to limit dv/dt sloping with a large cap + small series resistor as you often find it in half bridge SMPS with inductive loading for ZVS.
Here the snubber is only intended to dampen parasitic HF ringing during hard switching, which is massively triggered by the snappy di/dt of the body diodes right after Qrr is removed.
At the ringing frequency this snubber should behave mostly resistive.
The parasitic ringing frequency is defined by the inductances of the MosFet + PCB + caps + PCB tracks ( this loop should be small to avoid high parasitic energy) in combination the Coss of the MosFet.
In case of the LiteAmp the parasitic main mode is at 150MHz .
100pF at 150MHz show a Xc of 10 Ohms ==> in order to achieve an overal resistive snubber behavior the series resistor must be at double of that, most common rule of thumb is even triple ==> roughly 30 Ohms.
In fact not only the Xc of the cap are relevant but also the series inductive impedance of the snubber. As long as the XL is not larger than Xc it is helpful.
A resonant adjustment of the snubber (go for XL=Xc) allows a very efficient snubbering with very small capacitors and very small snubber losses.
Well... in the lite amp this is close to XL=Xc and is intended to allow lower resistors for damping more heavy MosFets without the need of much larger snubber caps. ==> Keeping the IRFI4020 in mind for the +/-70V...+/-80V version.
...the +/-40V would most likely also be fine with 47pF, but I did not check in detail, because already with 100pF the snubber losses are completely trouble free for +/-40V and 400kHz. There was simply no need to tune the low voltage version for small snubber losses.


Quote:
Originally Posted by Reactance View Post
The PCB is intended to suit the IRFI4212, IRFI4019 & IRFI4020.
However particularly the IRFI4019 is not in my focus so far, because the body diodes are already much worse than in IRFI4212 but the step in usable amp power or drive higher speaker impedances is not really convincing, IMHO. Here I clearly prefer to step directly to IRFI4020 for the +/-70V...+/-80V version of the Lite Amp.

P.S.
You are now the third guy who is pointing to IRFI4019.
Is there somebody in the backyard strongly promoting and pushing particularly the IRFI4019? If yes, why?
  Reply With Quote
Old 4th September 2014, 08:02 PM   #129
Reactance is offline Reactance  South Africa
diyAudio Member
 
Reactance's Avatar
 
Join Date: Jun 2005
Location: Cape Town
Quote:
Originally Posted by ChocoHolic View Post
You are now the third guy who is pointing to IRFI4019.
Is there somebody in the backyard strongly promoting and pushing particularly the IRFI4019? If yes, why?
not that I'm aware, I think it was the marketing paragraph in the datasheet.
__________________
There’s nothing worse than the smell of burned output transistors!!
  Reply With Quote
Old 5th September 2014, 07:56 PM   #130
alfsch is offline alfsch  Germany
diyAudio Member
 
Join Date: Jan 2002
Location: augsburg
>pointing to IRFI4019 < is probably just because all (?) the cheap amps on ebay use this...
ex.:

IRS2092 200W Class D Mono Amplifier Board 50V High Efficiency | eBay

Last edited by alfsch; 5th September 2014 at 07:58 PM.
  Reply With Quote

Reply


SystemD LiteAmpHide 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

Forum Jump


New To Site? Need Help?

All times are GMT. The time now is 10:03 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.79%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio
Wiki