An Ebay LM3886 PCB - comments invited - 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 10th March 2015, 09:48 PM   #1
ic0n is offline ic0n  United Kingdom
diyAudio Member
 
Join Date: Jun 2012
Default An Ebay LM3886 PCB - comments invited

Hi

Possibly a hoary old topic, but I can't find anything on this specific PCB. If I'm wrong, and there's a 50 page thread out there, please point me to it.

In a spirit of enquiry, I picked this PCB up from Ebay. It came with a kit of components including two LM3886TFs. Mostly no-name caps, except the input caps say 'Wima' on them. The PCB seems to mostly match the circuit diagram - one obvious exception is that the LM3886 share power supply decoupling caps.

Click the image to open in full size.

Click the image to open in full size.

There's some no doubt very useful notes, but unfortunately I can't read them.

Click the image to open in full size.

The circuit seems to be fairly close to the usual data sheet one.

Click the image to open in full size.

Any thoughts on values, layout etc?

Cheers
John
Attached Images
File Type: jpg IMG_20150310_170219.jpg (197.3 KB, 944 views)
File Type: jpg IMG_20150310_170249.jpg (133.9 KB, 519 views)
File Type: jpg IMG_20150310_170021.jpg (119.0 KB, 520 views)
File Type: jpg IMG_20150310_170057.jpg (140.5 KB, 513 views)
  Reply With Quote
Old 11th March 2015, 05:23 PM   #2
diyAudio Member
 
danielwritesbac's Avatar
 
Join Date: Nov 2007
Yeah,
Change 22k to 27k (at both input and feedback).
Change 220u to range 1500u~2200u.
Change 22u to range 220u~680u (and bypass with 10n).
__________________
Tools, Models & Software for DIYClipNipper boostLM1875 TurboPowerful TDA7293 kitTDA7294 pt2pt ♦ My post has opinion.
  Reply With Quote
Old 11th March 2015, 05:39 PM   #3
diyAudio Member
 
Join Date: Sep 2004
Location: Minneapolis, Minnesota, USA
I hate being the bearer of bad news but that PCB layout is pretty poor IMHO. And some of the component values are less than optimal, especially the power supply bypassing. I highly recommend you read the excellent analysis that TomChr did a few months ago, it's very thorough and informative. The link is here - Taming the LM3886 Chip Amplifier

Mike
  Reply With Quote
Old 12th March 2015, 12:27 AM   #4
tomchr is offline tomchr  United States
diyAudio Member
 
tomchr's Avatar
 
Join Date: Feb 2009
Location: Greater Seattle Area
The schematic is basically the Typical Application Schematic from the LM3886 data sheet. There's nothing wrong with it, but you can get better performance by changing a few values.

First off, add 22 kOhm in series with 47 pF across the 22 kOhm feedback resistor. This is to prevent excessive overshoot on the transient response caused by the noise gain compensation (220 pF cap across the amp inputs). Adding these components also reduce the peaking in the frequency response around 100 kHz where the noise gain compensation kicks in.

As Daniel writes, I suggest increasing the 22 uF capacitor to 220-1000 uF for lower THD at 20 Hz.
Also increase the two 220 uF supply decoupling capacitors to at least 1000 uF (National Semiconductor/TI recommends 470 uF per LM3886 in the data sheet).
Further, I suggest decreasing the 10 ohm resistor in the Thiele Network to 1.5 ohm to better dampen the ringing under capacitive load. Oh, and the eBay seller probably didn't bother to tell you how to wind the inductor... 15 turns of 18 AWG (1 mm diameter) enameled magnet wire on a AA battery will form an inductor of about 2 uH. Perfect for a Thiele Network.

I do not agree with Daniel's recommendation to increase the feedback resistor to 27 kOhm. First off, you don't need the extra gain. The amp as-is will clip around 1.65 V RMS and most hifi sources these days provide 2 V RMS. The MiniDSP is the only source I know of that strays from this standard with a 0.9 V RMS output voltage. Secondly, you get worse performance at higher gain due to the resulting reduction in loop gain.

The layout is pretty craptastic, though. The "ground plane" on the top layer is way too fragmented to do anything. It also only connects to the ground connector, so it's utterly useless as a ground plane.
The supply routes are skinny traces, rather than pours, leading to higher supply inductance. The feedback ground and input ground are routed separately ... to the power connector. You actually want them terminating at the output connector. It's interesting with the apertures in the solder mask on the output traces. That's an old trick to get a bit lower resistance of the traces. However, it only works if the board is wave soldered. The idea is that the solder will stick to the exposed parts of the traces as the board travels through the solder bath and lower the trace resistance.
Basically, the designer missed just about every opportunity to use the available copper to improve performance. They ended up with a connect-the-dots layout, just as most other people. They did dress it up a bit, but, sadly, not in a way that will do anything for performance.

I guess the good news is that the performance you'll get with this board is about the same as with most other LM3886 boards out there. I have yet to find anyone who have taken the LM3886 layout to the level I use in the Modulus-86. Of course, in the Modulus-86, I take the performance to a whole new level by adding another feedback loop and about 140 dB of loop gain to drive the THD down into the abyss.

You'll probably get quite a bit out of my Taming the LM3886 page as linked to by Mike above. I also suggest taking a look at my measurements of the THD of various layouts in this thread: LM3886 PCB vs Point-to-Point (with data)

~Tom
__________________
Modulus-86: Composite amp achieving 0.00014 % 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; 12th March 2015 at 12:35 AM.
  Reply With Quote
Old 12th March 2015, 10:45 PM   #5
ic0n is offline ic0n  United Kingdom
diyAudio Member
 
Join Date: Jun 2012
Oh dear oh dear. Sounds like a bit of a lemon. Never mind. So there aren't any 'basic' (just the LM3886 + datasheet components) PCBs, that are readily available on Ebay, that stand out as well engineered?

John
  Reply With Quote
Old 13th March 2015, 12:44 AM   #6
diyAudio Member
 
Join Date: Jul 2014
Quote:
Originally Posted by ic0n View Post
Sounds like a bit of a lemon. Never mind. So there aren't any 'basic' (just the LM3886 + datasheet components) PCBs, that are readily available on Ebay, that stand out as well engineered?John
I'm not seeing that you got a poorly engineered lemon. There seems to be at least some agreement that the circuit you have is approximately what the factory recommends, so how bad is it going to be?

I don't dispute statements made by those who have worked their personal magic on the LM3886 and other chips, but I do think it's fair to mention the engineering staff at Texas International. As far as I know TI doesn't hire from the bottom 10% of the class, and degreed engineers do tend to hold master's degrees. These are the people who created both the chip and the circuit it works in, so how wrong are they going to be?

I think it's also fair to look at the broader spectrum. In this age of little black chips amplifiers tend to have dead-straight response curves. But now look at the response curve for any given speaker. Even if your speakers cost in the $3,000 range--apiece--is .001% going to do you any good? That's one-one-thousandth of one percent.

Again, I don't argue with anybody's ideas or assertions, but the gist of the comments I've read seems to be, "That's just the factory circuit." So I guess it's in the eye of the beholder whether the guys at the factory know what they're doing. But after all, they did build the chip.
.

Last edited by bentsnake; 13th March 2015 at 12:50 AM.
  Reply With Quote
Old 13th March 2015, 05:56 AM   #7
tomchr is offline tomchr  United States
diyAudio Member
 
tomchr's Avatar
 
Join Date: Feb 2009
Location: Greater Seattle Area
Quote:
Originally Posted by bentsnake View Post
I'm not seeing that you got a poorly engineered lemon. There seems to be at least some agreement that the circuit you have is approximately what the factory recommends, so how bad is it going to be?
The board is not terrible. It's about average for an eBay board. The designers do get plus points for including Cc (across the inputs of the amp) and the Thiele and Zobel networks. Sadly, they didn't follow the recommendations for decoupling and forgot to compensate the overall frequency response for the addition of Cc. And they didn't do anything special to the layout. They connected the dots... This is what most people do.

Use the boards and get average performance. Nothing wrong with that. If you want world class performance, I suggest looking at my Modulus-86.

Quote:
Originally Posted by bentsnake View Post
I don't dispute statements made by those who have worked their personal magic on the LM3886 and other chips, but I do think it's fair to mention the engineering staff at Texas International. As far as I know TI doesn't hire from the bottom 10% of the class, and degreed engineers do tend to hold master's degrees. These are the people who created both the chip and the circuit it works in, so how wrong are they going to be?
TI = Texas Instruments. The LM3886 was made by National Semiconductor. TI acquired National in 2011. National tended to hire good people as well.

Speaking of... Bentsnake, you may be interested in reading my resume. It's available on my website, www.neurochrome.com. It's been over a year since I last updated it, but I think you'll find it informative...

Quote:
Originally Posted by bentsnake View Post
I think it's also fair to look at the broader spectrum. In this age of little black chips amplifiers tend to have dead-straight response curves.
Sure. In an optimized layout using a laboratory power supply, the ICs have good performance. This is the performance you see in the data sheet. However, with a poor layout and an unregulated supply, the ICs don't perform as well.

Below is shown the impact on the performance of the LM3886 versus various layouts. I described the experiment in detail in Post #30 of the LM3886 P2P vs PCB (with data) thread. The differences in performance is caused by moving the ground connections around. There were no schematic changes - only layout changes.

Click the image to open in full size.

The image below shows the THD+N of the LM3886 in an optimized layout using a regulated lab supply (Agilent E3632A) and using a regular unregulated supply (toroid + bridge rectifier + 2x22000 F). The data was captured using an Audio Precision SYS-2712. Click on the image for a larger view.

LM3886_THD_vs_Freq_vs_Supply_28W_8R.png

~Tom
__________________
Modulus-86: Composite amp achieving 0.00014 % 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; 13th March 2015 at 06:01 AM.
  Reply With Quote
Old 13th March 2015, 10:29 AM   #8
ic0n is offline ic0n  United Kingdom
diyAudio Member
 
Join Date: Jun 2012
Since I have all the bits, I'll make this one up with a few changes to component values, and stick it in a box.

I'm a bit stuck for the caps. The board 'designer' has put the footprint (0.2" pitch) for the two power supply bypass caps just a fraction too close together for two 12.5mm diameter caps to sit flush to the board - they have to **** outwards. 10mm works fine. If I'm prepared to tolerate wonky caps, the maximum value I can find is 1000uF/50V, or if I must have straight caps, 470uF/50V. Grrr.

The 22uF cap has, inexplicably, a lead pitch of 3mm. My normal UK supplier (CPC) has no capacitors with that lead pitch. Farnell/Element 14 (the same group as CPC); none. RS have a few, but then I'm into MOQ problems. At least the voltage rating shouldn't be a problem - if I'm reading it right, that cap only ever sees the maximum input voltage swing? So a 10V cap should be fine?

Cheers
John
  Reply With Quote
Old 13th March 2015, 11:11 AM   #9
diyAudio Member
 
danielwritesbac's Avatar
 
Join Date: Nov 2007
By all means, buy the right value caps. And when it comes to the install, take aim, and push. The 1000u may be enough for decoupler.

P.S.
Yes a lower voltage cap can work at -in coupler.
__________________
Tools, Models & Software for DIYClipNipper boostLM1875 TurboPowerful TDA7293 kitTDA7294 pt2pt ♦ My post has opinion.
  Reply With Quote
Old 13th March 2015, 12:33 PM   #10
diyAudio Member
 
danielwritesbac's Avatar
 
Join Date: Nov 2007
Quote:
Originally Posted by tomchr View Post
. . . First off, add 22 kOhm in series with 47 pF across the 22 kOhm feedback resistor. This is to prevent excessive overshoot on the transient response caused by the noise gain compensation (220 pF cap across the amp inputs). . . . I do not agree with Daniel's recommendation to increase the feedback resistor to 27 kOhm. First off, you don't need the extra gain.
Depending on the drive (in)capability of the source, such as anything modern, like a computer device, some more power amp gain may be useful so that the source doesn't distort itself very badly well before the amp potential is reached.
That sort of thing is typical today.
So I think we ought to give the amp owner the choice.
But I do have a question:
How does the amp compensation parts (with that 47pf) need to be changed if when the feedback resistor is 27k?
__________________
Tools, Models & Software for DIYClipNipper boostLM1875 TurboPowerful TDA7293 kitTDA7294 pt2pt ♦ My post has opinion.
  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
1625 pp amp layout - comments invited aardvarkash10 Tubes / Valves 18 17th January 2011 06:54 PM
LM3886 pcb on ebay.in viki_v2 Class D 0 15th May 2010 09:01 AM
Comments/guidance invited: TMMMM sealed box tcpip Multi-Way 67 4th June 2004 10:43 PM
Comments invited: proposed remote volume control tcpip Parts 2 2nd February 2004 07:07 AM


New To Site? Need Help?

All times are GMT. The time now is 04:30 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2015 DragonByte Technologies Ltd.
Copyright 1999-2015 diyAudio

Content Relevant URLs by vBSEO 3.3.2