hf ringing after inductor on positive halfwave. - Page 3 - diyAudio
Go Back   Home > Forums > Amplifiers > Class D

Class D Switching Power Amplifiers and Power D/A conversion

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 10th October 2012, 03:25 PM   #21
diyAudio Member
 
Join Date: Oct 2010
Location: Canada
Hey terry I posted gate driver schematic in one of my first posts on page one. I can post comparator/level shift schematic later if need be... OK will add more caps and vias after work. Also I only use irf530n to initially turn on a circuit to make sure I don't blow up a good fet! I have some fdp2575 s to throw in..

Last edited by switchleg; 10th October 2012 at 03:32 PM.
  Reply With Quote
Old 10th October 2012, 03:38 PM   #22
diyAudio Member
 
Join Date: Mar 2008
Switchleg, I saw that. I just cant figure out how the first stage of the gatedrive works. take the high-side gatedrive. D10/R33/Q17/R32 is straightforward, providing a low-impedance pull-down.

and the soft-started diode-coupled power supply is likewise straightforward.

but R31/D9/Q16 looks wrong - shouldnt Q16 be npn, emitter to junction of D9-A & R32, forming a low-impedance pull-up? ditto for Q18.

So I figured there must be something interesting going on with Ghi & Glow that I cant see. hence the question.
  Reply With Quote
Old 10th October 2012, 07:41 PM   #23
diyAudio Member
 
Join Date: Oct 2010
Location: Canada
Ghi and Glo are tired to the collectors of npn's in the discrete comparator/xor standard design from um10155 except they are both cascoded with both bases tied to a 5.6v supply... I'll post schematic later...
  Reply With Quote
Old 11th October 2012, 03:23 AM   #24
diyAudio Member
 
Join Date: Mar 2008
Oh ok, thanks. then D9 is a Baker clamp for pnp pullup Q16. that means the drive must be current limited.
  Reply With Quote
Old 11th October 2012, 04:39 PM   #25
diyAudio Member
 
Join Date: Oct 2010
Location: Canada
Sorry didn't get around to getting the schematics, but yes it aids turn off. Added two more 100nf in supply to FETs. Changed FETs to fdp2572. I have to adjust dead time as I have a bunch of shoot through but the ring frequency has not changed one bit. Granted the Coss's are similar but it should be different. Could the bootstrap diode excite this ring?
  Reply With Quote
Old 12th October 2012, 07:01 AM   #26
diyAudio Member
 
Join Date: Mar 2008
Switchleg,
firstly, the Baker clamp - it'll work a lot better if you use a schottky diode. as it is, the Baker Clamp will only start to work as Vce approaches zero, clamping it to Vce = (Vbe - Vf_clamp). For a silicon diode Vf_clamp will be similar to Vbe. Using a schottky changes this to Vce = (Vbe - Vschottky), which will be several hundred millivolts.

I downloaded a few UcD schematics, most used schottky diodes but one of them they did a proper Baker clamp, using a dual diode - putting one diode in series with the Base, and the Baker Clamp diode from the Collector to the Cathode of the first diode. This means that the baker clamp gives Vce = [(Vbe + Vf_base) - Vf_clamp] and will be about 0.6V

As far as your ringing is concerned:
start with a low-inductance probe connection. Linear Tech application note AN47 has an appendix devoted to exactly this problem.

the reason is so you can see just where your ringing actually is - its possible your probe is picking it up from elsewhere, even though its connected across the LC filter cap.

It may also be your probe setup oscillating (probe capacitance & ground lead inductance). If you have one, a switchable 10:1/1:1 probe can be helpful here - when you change it from 10:1 to 1:1 the probe capacitance goes up quite a bit (probe datasheet will tell you just how much). if the ringing frequency changes significantly, its the probe setup. Assuming the 1:1 probe bandwidth is still > 35MHz....

run some L,C numbers at 35MHz - it'll give you an idea of the sorts of parasitics you are looking for. Have you measured the resonant frequency of your output inductor? your output cap? at 35MHz LC = 20.7*10^(-18). so if its 20nH of cap ESL, it'll be ringing with 1nF. rinse & repeat until you find some numbers that make sense.

a T106-2 has a permeability of 10 - so 90% of the energy is stored in the core, 10% in free space. placing the wound core flat on the PCB will give you fairly high capacitance from winding to PCB. does the ringing frequency change if you stand the core up on its edge? what about adding a small cap in parallel with the output inductor?

Eva's post #8 is a very good description of whats going on. but until you sort out a decent measurement, I predict male pattern baldness as you tear your hair out in frustration - without a reliable measurement, you'll have to fix it by deducing the cause & eliminating it (this is often the hard way).

If you post decent pics of your layout (both sides) it'll help with figuring out actual current paths - current flows in loop; minimise them. and that nasty "ground" symbol on a schematic does a beautiful job of hiding the current return path. draw the actual current paths on your layout (eg in CAD, using a non-copper layer), remembering that the current starts and ends with the DC bus filter caps - out the cap, thru the FET, the filter inductor and then the filter cap, along the "ground" plane and back into the DC bus cap.

Once you've drawn the physical loop(s), the solution will probably leap out at you.
  Reply With Quote
Old 12th October 2012, 04:38 PM   #27
diyAudio Member
 
Join Date: Oct 2010
Location: Canada
After removing shoot through the ringing is almost gone (10mvp-p) but seems to remain at 35 MHz... I guess I need to increase my snubber resistance to match frequency. Also found away to match charge times and dead time!

Btw a bat54 is a schottky. Terry sounds like a plan, I'll have much more time this weekend to look into it

Last edited by switchleg; 12th October 2012 at 04:42 PM.
  Reply With Quote
Old 13th October 2012, 05:42 AM   #28
diyAudio Member
 
Join Date: Mar 2008
Switchleg,
you get -1 points for using the wrong symbol. I get -2 points for poor reading comprehension

If you dont already know, heres how to figure out your snubber values:
The Well Regulated Power Supply

That uses a rough approximation, and gives good results. It turns out that the optimum Rsnubber depends on the ratio of Csnubber to Cring. You have no control over Cring, and Csnubber comes in discrete values, but should be between 2 and 3 times larger than Cring. Here's how to calculate the optimal Rsnubber, once you've selected Csnubber:
The Well Regulated Power Supply

HTH
  Reply With Quote
Old 14th October 2012, 01:48 AM   #29
diyAudio Member
 
Join Date: Oct 2010
Location: Canada
Terry i apologize but i had replied while at work so i didn't have time to read over your post. I have calculated snubber values for SMPS before and i will go through the procedure on this board at a later time. I was concerned about capacitance from my toroid to ground plane but i also read on the ncore site where they mention to lay it on the ground plane to minimize EMI. Though i am always looking to experiment so it will be something to try, thanks.

But after thinking about the issues with the discrete gate drivers timing something eva said in another thread sometime ago gave me an idea... One cap and one resistor later i now have much closer matched gate driver dead times and charge times (lower gate driver was much slower then the upper driver before). I do not have any noticeable ringing at this time but i will gradually lower my dead time till i get cross conduction. Pretty sure i will get some ringing at that point and will try the suggestions you have mentioned.

Thanks for all the help btw.
Attached Images
File Type: jpg bottom1.jpg (406.2 KB, 84 views)
File Type: jpg bottom2.jpg (317.5 KB, 78 views)
File Type: jpg top1.jpg (300.2 KB, 72 views)
File Type: jpg top2.jpg (395.6 KB, 69 views)
File Type: jpg ADS00002.jpg (61.8 KB, 71 views)
File Type: jpg ADS00001.jpg (62.2 KB, 21 views)
  Reply With Quote
Old 15th October 2012, 04:55 AM   #30
diyAudio Member
 
Join Date: Oct 2010
Location: Canada
I tried my best to make my probe less inductive (image 1). I would be willing to hack it apart but after i find a supplier that sells probes that will work with my ATTEN.
I couldn't find any info on the probe it self but the scope manual claims the input of the unit is 1M 17pf +-3pf. (60mhz unit).

I tried switching btw 1x and 10x but it did not see a noticeable difference.

Then i started placing a 1nf ceramic cap across: (image 2 is without)
-Lower Fet drain source, no change. (image 3 change in amplitude is due to different output volume)
-Output filter Cap, little change. (image 4)
-Output filter Inductor, massive resonances at multiple frequencies (image 5,6,7)
Attached Images
File Type: jpg IMG_20121014_211549.jpg (160.6 KB, 19 views)
File Type: jpg ADS00012.jpg (57.3 KB, 17 views)
File Type: jpg ADS00009.jpg (59.1 KB, 11 views)
File Type: jpg ADS00003.jpg (61.4 KB, 12 views)
File Type: jpg ADS00004.jpg (66.0 KB, 19 views)
File Type: jpg ADS00005.jpg (63.3 KB, 16 views)
File Type: jpg ADS00006.jpg (65.9 KB, 19 views)
  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
Tapped HF inductor: where to get? Stabby Parts 0 21st April 2011 12:38 AM
Ringing on class d output inductor. nigelwright7557 Class D 36 25th February 2011 08:49 PM
Separate DACs for each halfwave ? Bernhard Digital Source 94 20th December 2006 07:06 AM
Cure for HF oscillation on positive cycles only RichardJones Pass Labs 4 27th June 2002 01:20 AM


New To Site? Need Help?

All times are GMT. The time now is 01:03 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