2-channel parallel 100W GC using LM3886 - diyAudio
Go Back   Home > Forums > Amplifiers > Chip Amps

Chip Amps Amplifiers based on integrated circuits

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 18th November 2004, 05:45 AM   #1
scone is offline scone  United States
diyAudio Member
 
Join Date: Nov 2004
Location: Austin, TX
Default 2-channel parallel 100W GC using LM3886

I'm going to build a 2-channel GC to compliment the speakers I am building this winter break. As you can tell by the impedance graph on John's site, the speakers are 4 ohm nominal, but dip to 3 ohms in several places along the freq range. I've read a number of places that it's a bad idea to drive anything less than 4 ohms with the National chips, so I figure i'll follow the spec found in their application note 1192 and build them using 2 LM3886's in parallel (per channel).

I haven't found too many documented gainclone projects using 2 chips in parallel per channel. Lots of them use the bridged/parallel configuration, but I think that's a little bit too complicated for my first amplifier project.

I modified the circuit diagram on Maarten's site (which looks like an exact copy of the non-inverting PA100 design from the National document) to include the input capacitor and resistor, and the voltage rails for the chips. It shows one channel... i've posted it below. The National spec suggested a voltage of +-37V for the PA100 design, so that is what I'm going with at the moment... so here are my questions:

Does the volume regulating pot go in between the audio in and ground?

Also, would there be any advantage in this case to build the inverting model (also found on Maarten's site) since it has the exact same component count as the non-inverting one?

I haven't done the math on the power consumption, but I'm assuming i'll be able to find a ~37V transformer capable of driving both channels? But I suppose there's no harm in using two if the design dictates it. I have about 5 weeks to build the GC and the speakers... hopefully it will be enough time.
Attached Images
File Type: jpeg chainclone_noninverted_parallel.jpeg (28.7 KB, 3648 views)
  Reply With Quote
Old 18th November 2004, 09:32 PM   #2
diyAudio Member
 
Join Date: Aug 2004
Location: Finland
Here is nice site (matches your request for 2xLM3886):
http://www.shine7.com/audio/pa100.htm
  Reply With Quote
Old 19th November 2004, 07:13 AM   #3
scone is offline scone  United States
diyAudio Member
 
Join Date: Nov 2004
Location: Austin, TX
Quote:
Here is nice site (matches your request for 2xLM3886):
http://www.shine7.com/audio/pa100.htm
thanks savenger, that's the first time i've seen that site.

there are a couple of things about his schematic that deviate from the national spec and which i don't quite understand.

1) there is an extra resistor and capacitor between audio in and the rest of the circuit (R20 and C20).

2) there is a very tiny capacitor between + and - inputs on the LM3886s (220 pF). However, at any normal operating frequencies (20Hz - 20kHz) the capacitor will look like a giant resistance... so I wonder why it's there at all?

3) there is a 4.7 pF capacitor put in parallel with the feedback resistor.

4) there is a 2 ohm resistor and a 0.1 uF capacitor in series connecting the speaker output to ground (thus in parallel with the speaker the amp is driving).

My memory of my circuits course in college is a bit fuzzy... I'm having trouble figuring out why he added the components to the national design. Assuming the minimal-is-better approach to gainclone design, is there any reason the amp wouldn't work if I removed the modifications and stuck to the national spec? I can always add them later if I think there is a need. Now if they are necessary for the amp to function correctly (and they are simply missing from the an-1192 doc to make those schematics easier to understand), then that is another matter entirely.

PS - There is some additional modifications to the GND and Mute pins on the LM3886 and a few added capacitors connecting the rails to ground... Not sure what those are either. Perhaps that is a question for another day.
  Reply With Quote
Old 19th November 2004, 11:36 AM   #4
diyAudio Member
 
Join Date: Aug 2004
Location: Finland
Datasheet example circuit also includes some of those extra components:
http://www.national.com/ds/LM/LM3886.pdf

Mostly these extra components prevent amp from oscillation.

1. Prevents high freguency entering in amp.
2. Prevents oscillation (i guess)
3. As above
4. Zobel http://www.trueaudio.com/st_zobel.htm

None of these components are mandatory if amp is stable and do not oscillate.

I may also be wrong , so i suggest you wait for another board member gives his impression.
  Reply With Quote
Old 19th November 2004, 02:04 PM   #5
diyAudio Member
 
Join Date: Mar 2001
Location: Brazil
Quote:
Originally posted by scone

there are a couple of things about his schematic that deviate from the national spec and which i don't quite understand.


All those components basically control bandwidth and filter what is not essential for audio.

Item 4 is a zobel and there was a specific thread over it. This should be implemented.

The other three are optional and should be tried and listened as to whether they improve or detract on what you listen or like to hear.

The feedback capacitors, C2 and C6, are also optional and should be used only if your amp has a large DC offset.

The chips used on the regulators (LT1083) are also expensive and hard to find. Use LM338 and do the supply Carlosfm suggests.


Carlos
  Reply With Quote
Old 23rd November 2004, 08:14 AM   #6
scone is offline scone  United States
diyAudio Member
 
Join Date: Nov 2004
Location: Austin, TX
Quote:
Use LM338 and do the supply Carlosfm suggests.
I assume you are referring to the schematic found in this thread ? I've copied the schematic i'm talking about below so no one needs to hunt for it

Also, I found an inconsistency in the 2x100W LM3886 GC schematics (http://www.shine7.com/audio/pa100.htm ). If you look closely, the values for C2/C6 and C5/C9 on the diagram are both shown as 100uF polarized caps. However, on the PCB layout, the board area for C2 and C6 is much bigger than the board area for C5 and C9. Also, on the pictures the C5/C9 caps look tiny compared to the C2/C9 pair. I assume that the value for C5 and C9 must be different than stated, but I have no idea what it should be.
Attached Images
File Type: gif regulated_psu.gif (13.0 KB, 2738 views)
  Reply With Quote
Old 23rd November 2004, 01:31 PM   #7
diyAudio Member
 
Join Date: Mar 2001
Location: Brazil
Quote:
Originally posted by scone


Also, I found an inconsistency in the 2x100W LM3886 GC schematics (http://www.shine7.com/audio/pa100.htm ). If you look closely, the values for C2/C6 and C5/C9 on the diagram are both shown as 100uF polarized caps. However, on the PCB layout, the board area for C2 and C6 is much bigger than the board area for C5 and C9. Also, on the pictures the C5/C9 caps look tiny compared to the C2/C9 pair. I assume that the value for C5 and C9 must be different than stated, but I have no idea what it should be.
What's the inconsistency? C2/C6 can be polarized or non-polar, but C5/C9 shuld be polarized.

The pin spacing on all 4 caps is the same, and my guess is the drawing on the feedback is probably larger. In fact that should be fine for using larger caps and improve on that.

The values are fine for both, but as I said you can eliminate C2/C6 if DC offset is low.


Carlos
  Reply With Quote
Old 23rd November 2004, 02:23 PM   #8
alexw88 is offline alexw88  Hong Kong
diyAudio Member
 
Join Date: Mar 2004
Location: HK
I think I can explain a bit about my PCB...

Basically the answer given by savenger and Carlos are correct.
I added those extra caps because before this project I did a LM3875 GC, it worked very well except one day I put my DIY DAC on top of it and the noise from the DAC coupled to the GC, and I can hear some high frequency noise from the speaker whenever the DAC is turned on. After adding those extra HF filters the noise is gone.

Also, C2 and C6 is bigger because I used Blackgate for them and used Panasonic for C5,C9. C2 and C6 have a bigger impact on sound so I use better quality caps. It would be best if you use non-polar BG for C2 and C6, but they are even bigger.
__________________
My Audio Page
  Reply With Quote
Old 23rd November 2004, 05:51 PM   #9
scone is offline scone  United States
diyAudio Member
 
Join Date: Nov 2004
Location: Austin, TX
Quote:
Also, C2 and C6 is bigger because I used Blackgate for them and used Panasonic for C5,C9. C2 and C6 have a bigger impact on sound so I use better quality caps. It would be best if you use non-polar BG for C2 and C6, but they are even bigger.
Thanks for the explanation. I had incorrectly assumed that both C2/C6 and C5/C9 were Blackgate capacitors, since on the pictures they look very similar.

Quote:
The values are fine for both, but as I said you can eliminate C2/C6 if DC offset is low.
Is DC offset purely a function of the differences between the values of the resistors and capacitors connected to the opamp (i.e. caused by the fact that perfect symmetry in the amplifier design can never be achieved), or can it be affected by the load or the source (audio in)? I'm asking because I don't know whether the DC offset might change if I swap a different audio source and a different set of speakers in the future.
  Reply With Quote
Old 23rd November 2004, 11:17 PM   #10
diyAudio Member
 
Join Date: Mar 2001
Location: Brazil
Quote:
Originally posted by scone

I'm asking because I don't know whether the DC offset might change if I swap a different audio source and a different set of speakers in the future.
There are two DCs that may come out of your GC: one produced by the GC itself, which you may cut with the feedback cap or with a servo; another one coming through the input, which you may cut with an input cap. A DC servo can take care of both if you don't want a cap in the signal path, but the servo uses a cap too and adds some lag of its own.

It's up to you what you use.

Different speakers won't affect the DC offset. Some audio sources might input DC offset.



Carlos
  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
100w Bridged lm3886 Ted205 Chip Amps 2 3rd April 2009 08:29 AM
Bridged or Parallel 100w LM3886? rockgardenlove Chip Amps 4 2nd March 2009 07:26 PM
Will this design work for a LM3886 100w BR amp?? Minion Chip Amps 4 19th October 2007 05:16 PM
FS: 5 channel parallel LM3886 bjefferys Swap Meet 2 13th March 2006 03:03 AM
Are two separate lm3886 = 2*lm3886 in parallel or bridge mode rs1026 Chip Amps 11 21st October 2004 11:24 AM


New To Site? Need Help?

All times are GMT. The time now is 02:40 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