The Perils of Paralleling - 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 14th March 2014, 08:27 PM   #1
diyAudio Member
 
Robert Kesh's Avatar
 
Join Date: Oct 2012
Default The Perils of Paralleling

About to build my first parallel chip amp, because I need lotsa watts into 4 ohms.

I'd like to use this thread to gather the issues that paralleling chip amps brings up. Some searching on diyaudio has revealed some that I will list here, but I'd be grateful for any other input.

The schematic I will be using as a basis is the one in the LM4780 datasheet attached, though I will be using LM3886s.

1. Match feedback network resistors to 0.1%. No problem there.

2. Match NFB caps. This might prove difficult. I have a cap metre on my DMM, but it's not particularly accurate. I've heard a solution is to enormously oversize the NFB caps, up to 1000u. Presumably this low, no frequencies that are affected by any mismatch will be entering the amplifer due to input cap. I wondered if matching ESR might be necessary too, and wondered if it varied a lot from cap to cap.

3. Increase output resistors. 0.1 isn't reckoned enough, and 0.22 is preferred it seems, at the expense of damping factor, whatever that is.

Offset is an issue, I imagine, as if the two chips differ in offset you will have current flowing between them. At low volumes this may dominate. The high power app note (I forget its number) mentions using DC servo. I'd like to avoid this complication.

Any thoughts? Anything else I should be aware of?
Attached Images
File Type: jpg Parallel chip.jpg (50.9 KB, 188 views)

Last edited by Robert Kesh; 14th March 2014 at 08:37 PM.
  Reply With Quote
Old 15th March 2014, 07:48 PM   #2
diyAudio Member
 
Robert Kesh's Avatar
 
Join Date: Oct 2012
No one? I must have got everything right then.
  Reply With Quote
Old 15th March 2014, 08:23 PM   #3
tomchr is offline tomchr  United States
diyAudio Member
 
tomchr's Avatar
 
Join Date: Feb 2009
Location: Greater Seattle Area
Quote:
Originally Posted by Robert Kesh View Post
No one? I must have got everything right then.
Um... Dude. Some of us were asleep when you wrote...

Quote:
Originally Posted by Robert Kesh View Post
About to build my first parallel chip amp, because I need lotsa watts into 4 ohms.
Use two chips rather than a dual chip-amp then. I also suggest reading National's app note, AN-1192 (commonly referred to as BPA200) on the thermal challenges involved.

Quote:
Originally Posted by Robert Kesh View Post
I'd like to use this thread to gather the issues that paralleling chip amps brings up.
The issues are mostly related to thermal management. If you don't know how to calculate the thermal resistance of the needed heat sink, you will need to learn. The good news is that it's pretty simple. It's basically ohm's law with the following analogies:

Voltage (V, electrical) = Temperature (C, thermal)
Current (A, electrical) = Power (W, thermal)
Resistance (Ω, electrical) = Thermal resistance (C/W = K/W, thermal)

A 1 increment in Kelvin (K) is equivalent to a 1 increment in Celsius (C), hence, C/W and K/W are used interchangeably.

How to do the math is explained both in the LM3886 data sheet and in AN-1192.

Quote:
Originally Posted by Robert Kesh View Post
The schematic I will be using as a basis is the one in the LM4780 datasheet attached, though I will be using LM3886s.
Good man!

Quote:
Originally Posted by Robert Kesh View Post
1. Match feedback network resistors to 0.1%. No problem there.
0.1 % resistors are not that expensive anymore. 0.05 % or 0.02 % are available for a (relatively small) premium.

Quote:
Originally Posted by Robert Kesh View Post
2. Match NFB caps.
Or use a DC servo. I'd prefer the servo as it's much easier to get good matching between channels that way.

Quote:
Originally Posted by Robert Kesh View Post
3. Increase output resistors. 0.1 isn't reckoned enough, and 0.22 is preferred it seems, at the expense of damping factor, whatever that is.
You'll burn an ungodly amount of power in the output resistor. I suggest trying with 0.1 Ω first and only increasing to 0.15 or 0.22 Ω if necessary. You'll probably need a 7 W or 10 W type if you go with a 0.22 Ω. Do the math...

Quote:
Originally Posted by Robert Kesh View Post
Offset is an issue, I imagine, as if the two chips differ in offset you will have current flowing between them.
True. So the amp burns a bit more power at idle. Big whoop... None of that current flows through the speaker, so I fail to see how this is an issue for sound quality.

~Tom
__________________
Modulus-86: Composite amp achieving 0.00018 % THD. Damn Good 300B, Novar Spud, 21st Century Maida Reg., Filament Reg., etc.
Neurochrome : : Audio - http://www.neurochrome.com/audio - Engineering : : Done : : Right
  Reply With Quote
Old 15th March 2014, 10:33 PM   #4
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Quote:
Originally Posted by tomchr View Post
You'll burn an ungodly amount of power in the output resistor. I suggest trying with 0.1 Ω first and only increasing to 0.15 or 0.22 Ω if necessary. You'll probably need a 7 W or 10 W type if you go with a 0.22 Ω. Do the math...
Well, I did, and with the typical 68 W into 8 ohms that the '3886 is good for I'm getting 1.8 W of dissipation.

Everyone and their dog is using .22 ohm 5 W emitter resistors in regular 100-150 W amps. Guess why?

.22 ohms of output R are good for a nominal damping factor of 36 with an 8 ohm load. (Ditto for .11 ohms and 4 ohms.) Not great but workable - 0.11 ohms is the equivalent of a good 6 meters of AWG14 twin lead, or 8 m of 2.5 mm. If you know the minimum and maximum speaker impedance in the audible range, you can calculate the resulting frequency response deviation.

Last edited by sgrossklass; 15th March 2014 at 10:40 PM.
  Reply With Quote
Old 16th March 2014, 12:35 AM   #5
diyAudio Member
 
nigelwright7557's Avatar
 
Join Date: Apr 2008
Location: Carlisle, England
I prefer to bridge rather than parallel amplifiers.
  Reply With Quote
Old 16th March 2014, 03:49 AM   #6
tomchr is offline tomchr  United States
diyAudio Member
 
tomchr's Avatar
 
Join Date: Feb 2009
Location: Greater Seattle Area
Quote:
Originally Posted by sgrossklass View Post
Well, I did, and with the typical 68 W into 8 ohms that the '3886 is good for I'm getting 1.8 W of dissipation.
28 V swing into 4 Ω would be 4.7 A RMS (accounting for 1.6 V drop-opt across the LM3886). 4.7 A in 0.22 Ω dissipates 4.8 W.

A 5 W rated resistor at 25 C ambient would reach nearly 250 C under those conditions. That's why I normally derate the resistors by 3~4x. So you'll need a 15~20 W rated resist.

With 0.1 Ω, the dissipation is 2.2 W. Now a 5 W type would actually have a chance of surviving a full-power sine wave test. A 7 W would be better...

You can, of course, argue that due to the crest factor of music, the power dissipation is less for typical program material at max output power. In the old days, crest factors on the order of 10~20 dB were common. These days with the hyper-compressed music, don't expect much above 3~6 dB.

Quote:
Originally Posted by sgrossklass View Post
Everyone and their dog is using .22 ohm 5 W emitter resistors in regular 100-150 W amps. Guess why?
The emitter resistors only see half the current as each side of the output stage conducts during alternate cycles of the sine wave. In a 100~150 W amp, you also generally have multiple output devices in parallel, each with their own 0.22 Ω resistor. This reduces the power dissipation per resistor even more.

~Tom
__________________
Modulus-86: Composite amp achieving 0.00018 % THD. Damn Good 300B, Novar Spud, 21st Century Maida Reg., Filament Reg., etc.
Neurochrome : : Audio - http://www.neurochrome.com/audio - Engineering : : Done : : Right

Last edited by tomchr; 16th March 2014 at 03:52 AM.
  Reply With Quote
Old 16th March 2014, 04:21 AM   #7
sreten is offline sreten  United Kingdom
diyAudio Member
 
Join Date: Nov 2003
Location: Brighton UK
Hi,

FWIW I don't know how the polarity of those feedback capacitors
makes any sense in that schematic. 0.1R resistors should be fine.

rgds, sreten.
__________________
There is nothing so practical as a really good theory - Ludwig Boltzmann
When your only tool is a hammer, every problem looks like a nail - Abraham Maslow
  Reply With Quote
Old 18th March 2014, 05:46 AM   #8
lgreen is offline lgreen  United States
diyAudio Member
 
lgreen's Avatar
 
Join Date: May 2003
Location: San Diego, USA
.1 ohm output resistors seem to be working great for me.
Use bigger heatsinks and match things closely (including input and feedback resistors) as you say.

Its not that complicated, and works, give it a try.
(no servo required)
__________________
Retired from DIY (2010) but still lurking now and again. My DIY audio projects- PartTimeProjects.com.
  Reply With Quote
Old 18th March 2014, 06:02 AM   #9
jcx is offline jcx  United States
diyAudio Member
 
Join Date: Feb 2003
Location: ..
with chip amps you pretty much need to parallel when bridging to handle the increased Iout required by the nearly doubled Vout
  Reply With Quote
Old 18th March 2014, 06:05 AM   #10
lgreen is offline lgreen  United States
diyAudio Member
 
lgreen's Avatar
 
Join Date: May 2003
Location: San Diego, USA
True, but you don't need to bridge when you parallel. The original post was just about parallel which works great to get more Iout.
__________________
Retired from DIY (2010) but still lurking now and again. My DIY audio projects- PartTimeProjects.com.
  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
Paralleling caps in x/o Milos P. Multi-Way 4 20th February 2012 09:40 AM
paralleling amplifiers. neutron7 Chip Amps 19 12th June 2005 07:13 AM
When paralleling opa549 bigmike216 Chip Amps 5 16th May 2005 09:07 PM
resistors for paralleling ejn327 Chip Amps 8 7th February 2005 09:10 PM
the perils of digital eq? diaspora Digital Source 12 3rd May 2004 02:48 PM


New To Site? Need Help?

All times are GMT. The time now is 10:00 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