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 27th March 2005, 06:09 PM   #21
SSassen is offline SSassen  Netherlands
diyAudio Member
 
SSassen's Avatar
 
Join Date: Mar 2005
Location: Edam, The Netherlands
MikeB,

Good work, just one remark that might help:

Quote:
I had some sucess with adding a 3nf to the Re of predriver/levelshifter.
How about you give the predriver/levelshifter a little bias voltage, so that the voltage swing needed for it to switch levels is lower, hence it'll be more like a schmitt-trigger?

Best regards,

Sander Sassen
http://www.hardwareanalysis.com
  Reply With Quote
Old 27th March 2005, 07:06 PM   #22
MikeB is offline MikeB  Germany
diyAudio Member
 
MikeB's Avatar
 
Join Date: Apr 2004
Location: Gütersloh
Hi Ssassen !

The problem is, that i create the necessary deadtime by not having
biased the predriver, giving a delay turning on these transistors...
So i use these 3nf to create some nearly overshooting, resharpening
the edges of the signal... The effect of this was very visible on scope,
the signal went from nearly sinusoidal to nearly squarewave.
Looking to the pchannel-vgs explains why i need that much deadtime...

Some small sidenote...

I still can't believe how superior the sound of ClassD is compared to
analog amplifiers... It plays details and dynamics that effortless !
And turning up the volume just makes the music more intense, not louder !
Even this far from perfect simple ClassD plays at a qualitylevel... Holy shid !

Mike
  Reply With Quote
Old 27th March 2005, 07:52 PM   #23
SSassen is offline SSassen  Netherlands
diyAudio Member
 
SSassen's Avatar
 
Join Date: Mar 2005
Location: Edam, The Netherlands
Thanks MikeB,

Alright, how about the following modification? I devided the 3nF cap up in two 1.5nF caps and attached their center to the Vcc, this should help speed up the rise/fall times and it sure is easy to try. See below quoted schematic, I edited in the changes.

Click the image to open in full size.

Best regards,

Sander Sassen
http://www.hardwareanalysis.com
  Reply With Quote
Old 27th March 2005, 07:58 PM   #24
Bgt is offline Bgt  Netherlands
diyAudio Member
 
Join Date: Jan 2005
Location: NL, Voorschoten
Is'nt it possible to put the feedback after the output coil like the UCD's?
  Reply With Quote
Old 27th March 2005, 08:16 PM   #25
Account Disabled
 
Join Date: Feb 2004
Send a message via AIM to classd4sure Send a message via MSN to classd4sure
Hell yah Mike well said.

It is impressive how even crappy class d can sound is it not? Makes it worth experimenting on your own even if only in P2P. Since I'm contemplating making another more complex version (than I've done before) using veroboard I'd very much to see the bottom side of your yours, would you be so kind as to provide a pic of it please?

I have to point out one thing that really sucks though about experimenting like this, eventually it gets to be really very good, but there's still certain "things" that will rub you the wrong way... and how to get rid of those right? Dual layer PCB.... SMT.. sniff.

It's hard to hit that point where you just have to admit to the limitations of the way it was built, and walk away from it leaving it "unperfected". Still lots of fun though.

You mentioned using the heatsinks as some sort of shielding.... they might be making better transmitters, which your coil might pick up nicely. It seems well worth bolting a small HF bypass cap to each heat sink, espcially given where you take your feedback from.

Seriously I'd like to see the bottom side of that board. Excellent work as well.

Best Regards,
Chris
  Reply With Quote
Old 27th March 2005, 09:06 PM   #26
ilimzn is offline ilimzn  Croatia
diyAudio Member
 
Join Date: Feb 2005
Location: Zagreb
Default Re: Update

Quote:
Originally posted by MikeB
Okay, here is the "exact" calc for the Vgs:
The gain of the predriver/levelshifter is ~14, given a virtual additional Re of 20ohms you get 1000/(50+20) = 14. The currents delivered by "vas" are ~8ma, this gives voltageswing of 1.2volts because of the 150ohms to ground.
Yes, my bad, I just looked at my post and noticed I put gain of predriver 200 instead of 20. Sorry for that.
Interesting that you actually limited the drive voltages by limiting the gain on the Vas via the 150 ohm resistor. If you have the time I would appreciate if you could write why you decided for that approach?

Quote:
Originally posted by MikeB
But, it all seems to be the problem with too slow switching off...
Theoretically both mosfets are IRF, but you never know what you really get... Both mosfets have the same labeling, using the IRF-brand, the IR with the diode between. But they costed only 60cents...
That would be about their usual price. I'm just asking because there are differences between manufacturers, speciffically in the threshold voltage. I only have IRFP140 and IRFP140N datasheets at hand at the moment and there is a good 1V of difference, lower on the N part. But looking at your diagrams and oscillograms I don't think this is the case for you.

Quote:
Originally posted by MikeB
I will try to improve the driverstage, as the pchannel-vgs shows some "sticking" when vgs drops below 4volts... Also there is some ringing in the vgs.
Yes, looks like that may actually be the problem. 4V puts it right in the middle of the 'linear' area. Also, it seems like you could be having some source inductance (or stray inductance) issues re ringing? The sticking part seems to last about as long as the ringin part, and the sticking part shows a ringing crest.
It's difficult to accurately overlay the P and N gate drive waveforms, maybe that would reveal cross-conduction during the sticking part on the P side?
  Reply With Quote
Old 27th March 2005, 09:06 PM   #27
MikeB is offline MikeB  Germany
diyAudio Member
 
MikeB's Avatar
 
Join Date: Apr 2004
Location: Gütersloh
Hi !

SSassen, there would be one small problem with your modification...
If you look at the scopeshot of the output, you see that Vcc is not
stable, it wings with some volts. Or is this the idea to create some
feedback ? But, i must admit, i don't understand your change...

Bgt, changing the feedback isn't that easy, i already checked, it would
need heavy changes, because of the thresholds of the schmitt-trigger
beeing to big, i would not get above 100khz oscillation...

Chris, i attached a shot of the first prototype bottomside...
(This one used folded cascode for levelshifter)
It misses the 3 paralelled 1nf-caps...
I added one 1nf after the other until i reached near overshooting !
Yes, a PCB would help improve the signals, and should greatly
reduce ringing. But my intention with this amp was to make a simple
amp, easy to build (this excludes SMD) using standard components,
so that everyone can build this. Just to get back to the DIY-spirit like
it is with solidstate-amps. Just build and listen... No exotic parts...
It took me ~4 hours to construct the whole thing... (and 2 weeks to design...)
I think, if i can reduce deadtimes, it should evolve from "crappy" ClassD
to serious amplifier.
At least it does not sound crappy... Really !
But i consider to switch to n-channel only, seeing how cold the
nchannel keeps, and looking to the IRF-page seeing the newest
mosfets specifically developed for ClassD...
Looking at the datasheet just makes my mouth wet !
Have you looked at the irfb4212 ? Ouch ! 40ns recoverytime!


Another thing... I always observed temperatures of transistors, but
when listening for longer period, i wondered what was stinking...
The outputcoil heated up ! How is this possible ? I would expect
several amperes to heat up a 1mm-copperwire !
Is this normal or is there another problem ? The coil was too hot to
touch for longer than a second, burning my "fingometer"...
I would estimate above 80°C ... An aircoil !!!
At least the caps were cold, seems that wima-mks2 are real serious quality...

Mike
Attached Images
File Type: jpg img_0037.jpg (74.3 KB, 1167 views)
  Reply With Quote
Old 27th March 2005, 09:21 PM   #28
MikeB is offline MikeB  Germany
diyAudio Member
 
MikeB's Avatar
 
Join Date: Apr 2004
Location: Gütersloh
Hi ilimzn !

The mosfets where matched to Vgs-threshold of 3.6v...
But even with the same manufacturer i had one irf9540n having
a vgs-treshold above 5volts...

I choosed the 150ohm in the vas to get very low zout, because of
the highly reactive load from the bases in the levelshifter.
Even with this 150ohm these are visible on the scope, but only some minor spikes.
Also i wanted to have 1.2v swing, to be able to cut the signal at
it's half, to create all the deadtime i want.
Also, a that small resistor speeds up the vas like hell, but with the 1nf...

I believe this sticking part in the Vgs-waveform of the p-channel is the
reason why i need that much deadtime !
Very interesting is, that this stucking seems to occur exactly at the
vgs-threshold, when the mosfet finally closes...
In sims this effect is not visible. Impossible to design ClassD without scope !
I made some checks in sims, without any crossconduction the losses
in the pchannel are ~650mw, enough to heat up that small heatsink !
I think, right now i have no crossconduction, just way too high rds-on...

Mike
  Reply With Quote
Old 27th March 2005, 10:18 PM   #29
Account Disabled
 
Join Date: Feb 2004
Send a message via AIM to classd4sure Send a message via MSN to classd4sure
Hi,

Well Mike thanks alot now I'm ashamed of my work after seeing that. I'm not shy to say this either, the bottom of your board is cleaner than the top of mine. I was keeping my "good perboard" for the second project though, this one has traces, and by using them I see now I've compromised my layout, which I'm paying for now because it's noiser than it was on a protoboard. Before I give up on it though I'm going to try re-routing a few things (is it there yet?).

Hey you know when I said "crappy class d" I didn't at all mean "your crappy class d" but rather the non store bought, tuned to perfection with many hours of listening tests along side of an AP2, using the very best construction techniques going ..

Basically class d the DIY way (they said it couldnt' be done!).
I fully agree with you anyway, the dynamics from them put your average amp to shame, and I find you can tell that right away, well before any optimizing has been done.

I placed my PNP turn off transistor right on the mosfet pins on the underside of the board. The turn on transistor is about an inch away on the top side, would have been closer but I had to place it around the heatsink.

The IR device you mentioned really does look good, I hadn't seen it before. I'm using FDP3682 N-channels from Fairchild. They stay perfectly cold now.

I'm also using an aircoil, though mine is much smaller. At no point in time along with all the changes and mods I've made to the circuit has that coil ever even gotten warm. I tried a toroid once with a mystery iron powder core and that got a little warm. It also didn't sound as good, but it lowered the noise floor notably. You have about twice the rail voltage I do though.

Maybe you need a bigger gauge of wire/fewer turns, lower the DC resistance? As others have said though, a good toroid would be best anyway, but you did some kind of job making that up and keeping it one piece. I "painted" mine with 5 minute epoxy, it only has about 18 turns, holds it together well though.


Anyway I really admire your true DIY approach and your goal with this project. If it helps you to know you've inspired me already to continue with my next project using the same sort of board you are, as I now see it's much cleaner than my previous approach, and just this morning I was thinking I wouldn't bother with it because it would likely be worse. So thanks for that picture. I'm looking forward to following your progress with this.

Best Regards
Chris
  Reply With Quote
Old 27th March 2005, 10:57 PM   #30
subwo1 is offline subwo1  United States
diyAudio Member
 
subwo1's Avatar
 
Join Date: Jan 2002
Location: North American Continent
Quote:
Originally posted by MikeB
...outputcoil heated up ! How is this possible ? I would expect
several amperes to heat up a 1mm-copperwire !
Is this normal or is there another problem ? The coil was too hot to
touch for longer than a second, burning my "fingometer"...
I would estimate above 80°C ... An aircoil !!!
At least the caps were cold, seems that wima-mks2 are real serious
Hi Mike, I attribute this wire heating to the skin effect. It causes the current to flow on the surface of a wire at high frequencies. On one of my early switching power supplies, the enamel insulation on an air core choke actually developed cracks from the heat. Consider too, that the coil conducts more current than is output to the load.

Also, you may be able to lower radiated interference from your coil on the circuit by winding it in two sections, half in one direction, the other half physically oriented at a 180deg angle to the first.
__________________
USMPS
http://groups.yahoo.com/group/switchmode/
  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
PLaying with ClassD again TroelsM Class D 12 20th February 2008 08:18 AM
New to ClassD DarkOne Class D 2 24th September 2006 04:11 PM
Discrete fully differential amps and no CMFB rtarbell Solid State 1 24th August 2006 06:22 AM
Fully Discrete Switched PS tguzella^TuX64 Power Supplies 16 27th July 2005 12:29 PM
GC VS classD etalon90 Chip Amps 0 3rd November 2004 05:29 PM


New To Site? Need Help?

All times are GMT. The time now is 04:22 PM.


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