My fully discrete ClassD - Page 4 - 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 28th March 2005, 01:17 AM   #31
MikeB is offline MikeB  Germany
diyAudio Member
 
MikeB's Avatar
 
Join Date: Apr 2004
Location: GŁtersloh
Hi !

Chris, point taken... I did not understand the "crappy" like mine beeing bad...
Of course it is not optimized yet, it's even not cold yet !
But, it works !
And yes, i wanted to proove that you can build ClassD the old fashioned way !
My coil has 60 turns, giving in theory 55uH, made of 2.5meters wire !
What kind of noise are you talking about ? I did not hear anything...
I like these boards, they are very good for just building a prototype !
But when you start changing, they become a pain...
What oscillationfreq do you use ? How cold is your amplifier ?
Is yours fully discrete, or do you use the ir2011 ?
I forgot to mention.. I even "forgot" to isolate the mosfets from
the heatsinks ! So, the 300khz squarewave is fed into the heatsinks...

About the coil... When i was ready winding it, i found it very flexible,
just like a spring, very unhandy. So i just attached these stripes, making
the thing rock-hard !
Is it possible that mechanical resonance is heating up the coil ?

subwo1, if the skineffect causes this, the "dc"-resistance gets very high ?
Isn't there a specific winding-technology, like using twisted wires ?
Sims say that i have ~400ma current through the coil. (idle)
If i put 2 coils in reverse direction, don't they cancel out each other ?
I think Mr.Evil's suggestion, to bend the coil to a toroid could make
a perfect job to contain the field within the choke...

Mike
  Reply With Quote
Old 28th March 2005, 05:03 AM   #32
subwo1 is offline subwo1  United States
diyAudio Member
 
subwo1's Avatar
 
Join Date: Jan 2002
Location: North American Continent
Hi MikeB, the fields would cancel if the windings were not separate and oppositely oriented (side by side) but concentrically wound in opposite directions (one winding around the other). So twisted wires wound into a coil would give very little inductance too. I think the minimum external field would result if two of the toroidal shaped air cores wound in opposite directions were placed one on top of the other.

Quote:
...if the skineffect causes this, the "dc"-resistance gets very high...
I think it does if for no other reason than the temperature of the whole wire becomes elevated. Even though the low frequency audio component may be able to pass through the inside of the wire, the wire would still heat from the outside to the inside.
__________________
USMPS
http://groups.yahoo.com/group/switchmode/
  Reply With Quote
Old 28th March 2005, 08:01 AM   #33
Account Disabled
 
Join Date: Feb 2004
Send a message via AIM to classd4sure Send a message via MSN to classd4sure
Hi Mike,

Well while I had the comparator/input portion on a cheap breadboard all I had for noise was a slight buzz and a faint hiss, which largely vanished as soon as I put some audio through it.

Now I have that same noise but louder, and if I try to turn it up even a fifth of what I could before it crackles. It seems to have gotten a bit better just from relocating my ground wire but there's more to be done before I call it a day.... at the same time, it plays audio and there's no smoke..... running _COLD_... so I'm ready to leave it as it is, call it a done deal and move onto the next one. This is mostly because of not having a scope to tweak it with.

That's the funny thing see I can't really tell what the frequency is... It should be ~300kHz and it does a great job with the high notes.

How cold is it? It's cold Mike. Even at idle or low volume. I'm now using my upper lip'ometer on the heatsinks and they stay at room temp, +-18V rails/4ohm speaker load, from idle to full volume, doesn't matter. I tried putting my finger on the mosfet itself and it's only a bit warmer than the heatsink, still cool.

I'm using 100 ohm turn on resistor with the FDP3682 mosfets.

Originally these mosfets couldnt' last ten seconds in it and I had to use IRF640's which ran hot as all hell but lasted for a good while.

I'd like to put it on a scope one day soon, working on getting a cheap one. If/when that happens I'll optimize it further.

It is a most basic version of the fully discrete UCD. Next one will be fully differential and use better parts, I'm also going to go a bit higher on the rail voltage, but not by much.

Mechanical resonance heating the coil? Sure I'd imagine it would to some extent but I highly doubt anywhere near the heat you're seeing, in fact I would guess you'd have a hard time measuring any heating caused by that.

I think skin effect forces the AC signal, so your ripple frequency, towards the outer portion of the wire, and a bigger diameter wire would only help by providing a larger circumferance.

Ways to improve it all seem to have one thing in common, give it a thicker skin, or more of it.

That means you could try anything from winding your own air core out of foil (probably the best way I think) or using multiple paralleled strands of wire, or just a thicker gauge.

I was tempted to try the foil method but haven't yet done so yet.

How is your DC offset anyway? Maybe that could have something to do with it.

Also had you considered doing your level shifting via capacitive coupling?

Best Regards,
Chris
  Reply With Quote
Old 28th March 2005, 04:16 PM   #34
MikeB is offline MikeB  Germany
diyAudio Member
 
MikeB's Avatar
 
Join Date: Apr 2004
Location: GŁtersloh
Hi !

As long as the coil does not smoke, i can live with that for the moment...
But a foilcoil, that would be great ! But can't make a toroidal winding
with foil...

I checked datasheet on the FDP3682, not bad, but this new IRF seems
to be slightly better for ClassD ?

Considering that the irf540n still has 77mohm Rds-on, given double the
supplyvoltage it is already impressingly cold. But on +/36v i would not
risk using a lip'o meter, imagine getting 80volts onto the tongue...
Why are you using only 18volts ? Did you already explode too many mosfets ?

I measured 7mv DC-offset, this should really cause nothing !

I think i have to get rid of this "stucking" in the Vgs-waveform,
This "nearly closed" state for ~100ns can't be good, this might be
the reason why the p-channel heats up.
Also, i could reduce deadtime by 100ns if i get rid of this !

Michael
  Reply With Quote
Old 28th March 2005, 06:59 PM   #35
Account Disabled
 
Join Date: Feb 2004
Send a message via AIM to classd4sure Send a message via MSN to classd4sure
Hi,

I had a 48VA 24V CT transformer which was used in a +-12V supply regulated to 1amp for a microcontroller, so I used it to make an unregulated supply. It's just what I had.

Not a bad voltage for experimenting with and with the speaker I have it's actually enough to annoy the neighbors with.

For the next one I'm planning on a slightly bigger supply though, 18Vac, 110VA.

BTW you said you forgot to isolate the mosfets from the heatsink and so the square wave is being fed into the heatsinks right. Even if you had isolated them I think it would be getting fed into the heatsinks anyway via capacitive coupling. Tying them to ground with a small cap seems to make a big difference.

Best Regards,
Chris
  Reply With Quote
Old 28th March 2005, 07:44 PM   #36
MikeB is offline MikeB  Germany
diyAudio Member
 
MikeB's Avatar
 
Join Date: Apr 2004
Location: GŁtersloh
Yeah, i know what you mean... A long time i worked with a 15v
transformer, giving 21v DC... This was enough to torture my speakers
AND my neighbours !

I have isolated my problem... This stucking at exactly the Vgs-threshold ...
Very simple ! When the mosfet closes, the voltage at the drain jumps
to -Vcc because of the coil. But this transient "inducts" a lot of voltage
into the gate of the p-channel because of its D-G-capacitance, and
the driver for closing is not capable to deliver more current.
Now i could improve the driverstage or just do not use mosfets with
such big capacity... This excludes the use of p-channels...
Sims do not show this behaviour correct as the mosfets models are
simply BAD. It seems that realworld-mosfets are a LOT of faster than
sims say. Or the D->S capacitance is much smaller than in model,
giving a much faster transient ?
As the p-channel is stuck at this Vgs, it still conducts with quite
big current, but the drainvoltage falls to -vcc. So it conducts with
S->D voltage of 70volts, nicely heating up...
Good to know that this new IRF has only 66pf outputcapacity,
compared to the 400pf of the irf9540n.

Mike
  Reply With Quote
Old 28th March 2005, 10:09 PM   #37
diyAudio Member
 
destroyer X's Avatar
 
Join Date: Feb 2004
Location: Recife - Brasil Northeast
Default When already satisfied with the sonics, please Mike, made some recording and let me..

Hear some comparison.

Oh!....you have lost your DAT recorder!!!!..... will be not possible!

regards,

Carlos
__________________
Brazilian Airways Christmas Message to all of you!
https://www.youtube.com/watch?v=HFGM5K-WWnY#t=135; Unbeatable!; https://www.youtube.com/watch?v=u4tGBiqMnzQ
  Reply With Quote
Old 28th March 2005, 10:25 PM   #38
MikeB is offline MikeB  Germany
diyAudio Member
 
MikeB's Avatar
 
Join Date: Apr 2004
Location: GŁtersloh
Hi Carlos,
No problem that the DAT-recorder is stolen, i never used it for these recordings....

Will make some samples... But, sonics is not all with an amp, some
reliability and coolness is also important !

Just found out that sims are making bad games with me according
mosfet-models. Having scoped a lot i understand now much more
about these devices. Dropping Vgs below threshold instantly turn
off/reduce the current through these devices, not exactly 51ns later like some
value says in datasheet. Do you believe this, sim is just delaying the
response abut 51ns ? Means i have to do it the old fashioned way,
change parts and measure with scope... Damn it !

Mike
  Reply With Quote
Old 28th March 2005, 11:31 PM   #39
Account Disabled
 
Join Date: Feb 2004
Send a message via AIM to classd4sure Send a message via MSN to classd4sure
Hi Mike,

The small signal characteristics below Vth are basically unmodeled and should be seen as totally unreliable, as well as their variation with temperature or possibly even voltage. Things like base spreading resistance aren't accounted for at all.

Turn off behavior of the models/body diode recovery, is also said to be rather poor and unrealistic.

I think spice can be used to get a very basic running start at some ballpark values to start off with but it will def. need further optimizing in circuit.

Best Regards,
Chris
  Reply With Quote
Old 29th March 2005, 05:53 AM   #40
subwo1 is offline subwo1  United States
diyAudio Member
 
subwo1's Avatar
 
Join Date: Jan 2002
Location: North American Continent
Hi MikeB, I just got an idea you might like to try. You could try reducing the emitter resistor in the level-shifter stage to something like 30 ohms and the collector resistors to 620R.
__________________
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 09:18 AM
New to ClassD DarkOne Class D 2 24th September 2006 05:11 PM
Discrete fully differential amps and no CMFB rtarbell Solid State 1 24th August 2006 07:22 AM
Fully Discrete Switched PS tguzella^TuX64 Power Supplies 16 27th July 2005 01:29 PM
GC VS classD etalon90 Chip Amps 0 3rd November 2004 06:29 PM


New To Site? Need Help?

All times are GMT. The time now is 05:48 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