Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Chip Amps Amplifiers based on integrated circuits

Lower chipamp distortion through PCB routing
Lower chipamp distortion through PCB routing
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 June 2018, 06:47 PM   #1
alexcp is offline alexcp  United States
diyAudio Member
 
alexcp's Avatar
 
Join Date: Jun 2009
Lower chipamp distortion through PCB routing
Default Lower chipamp distortion through better PCB routing

I wanted to apply to chipamp design what I learned from Bruno Putzeys' article "The G Word, or How to Get Your Audio off the Ground", originally published in Linear Audio Vol.5. The key learnings for me there were:
  • Each signal has its own reference;
  • A single GND pour not a good reference for any signal;
  • The signal and its reference should travel together; and
  • The signal and its reference should look into balanced impedances.
With these in mind, I designed two PCBs for LM3886, one with through hole resistors, the other with SMTs - see the attached photos. Because I've always liked the simple and elegant ChipAmp.com's LM3886 PCBs (unfortunately, no longer available), I used ChipAmp.com's form factor (2.9x1.2in), the same general layout, very nearly the same schematic (attached to the posts #4 and #5 below) and the same through-hole parts. For comparison, I built one channel on NOS ChipAmp.com's board that I happened to have - in the photos, it is the red one.

Attached are the spectra of the NOS ChipAmp.com's LM3886 board and two of my boards giving 18V RMS @ 1kHz into 8ohm resistive load. To give you an idea of my measurement setup noise and distortion, I also include the loopback (a.k.a GenMon) performance of my test setup with no amplifier connected (E-MU 0204 sound card with Pete Millett's sound card interface - see the photo of my test bench). The spectra were taken by ARTA software.

Note how a different PCB layout affects the THD performance. Not only the overall THD level is affected, the proportion of higher-order harmonics (which are bad for subjective performance) also changes. Same general layout, same parts, chips form the same batch. Also, note that the distortion and noise of my boards are not that much higher than my measurement floor...

UPDATE: a couple of spectra with a lower noise floor are attached to the post #26 below.

UPDATE2: boards are now available at HiFiOcean.com.
Attached Images
File Type: jpg IMG_0701.jpg (945.8 KB, 797 views)
File Type: jpg IMG_0702.jpg (925.5 KB, 760 views)
File Type: png ChipAmp.com 1kHz 18Vrms @ 8ohm.png (59.4 KB, 797 views)
File Type: png MyGC TH 1kHz 18Vrms @ 8ohm.png (60.8 KB, 751 views)
File Type: png MyGC SMT 0.1% 1kHz 18Vrms @ 8ohm.png (61.1 KB, 738 views)
File Type: png EMU0204+IF 1kHz loopback.png (60.7 KB, 243 views)
File Type: jpg IMG_0697.jpg (806.6 KB, 283 views)

Last edited by alexcp; 24th March 2019 at 09:41 PM.
  Reply With Quote
Old 10th June 2018, 07:00 PM   #2
DPH is offline DPH  United States
diyAudio Member
 
Join Date: Dec 2008
Location: Portland, OR
Just to make sure you've seen this and the accompanying material from TomCHR?

An open source layout for the lm3886 ?

Edit to add in my shortness, excellent work by the way! I meant to say that this complements what has been written/designed/measured as well.
__________________
Happy DIYing, Daniel

Last edited by DPH; 10th June 2018 at 07:08 PM.
  Reply With Quote
Old 10th June 2018, 07:34 PM   #3
00940 is offline 00940  Belgium
diyAudio Member
 
00940's Avatar
 
Join Date: Sep 2003
Location: Liège
That's pretty damn good. It's great to see more good options for the lm3886. Congrats.

Looking at your graphs, you're probably held up by your soundcard setup. My thd measurements on a lm3886 at the same voltage give a thd of 0.0013% and a thd+n of 0.0024%. I attach a pic of my amp at the same output (25vp, 50vpp) to give you an idea of what could be hiding under the noise floor. But it's so down under...
Attached Images
File Type: png 25vp.png (50.4 KB, 270 views)
__________________
All that is excessive is insignificant. Talleyrand
  Reply With Quote
Old 11th June 2018, 01:07 AM   #4
alexcp is offline alexcp  United States
diyAudio Member
 
alexcp's Avatar
 
Join Date: Jun 2009
Lower chipamp distortion through PCB routing
Default Schematic and BOM - through hole version

As promised, attached are the schematic and the board images.

The BOM is as follows:
  • All resistors except RZ - Phoenix SFR16S or similar compact metal film, 1% or better. Vishay CMF50/RN50 would be a good choice.
  • RZ - KOA Speer MOSX1 or a similar compact 1W resistor.
  • C1, C2 - Panasonic EEU-FC1H152L. Alternatively, the board will accommodate snap-in capacitors up to 22mm in diameter or smaller 100uF 50V capacitors (in which latter case you can also install C3 and C4.
  • C3, C4, CZ - Kemet PHE426.
  • CI, CF - Panasonic ECE-A1VN220U or similar bi-polar.
  • CM - Panasonic EEU-FC1E101S or similar; I used Epcos because I had them.
  • U1 - LM3886.
Attached Images
File Type: png LM3886_TH_R1_topsilk.png (75.4 KB, 237 views)
File Type: png LM3886_SMT_R1_top.png (55.0 KB, 206 views)
File Type: png LM3886_TH_R1_bottom.png (63.6 KB, 190 views)
File Type: png LM3886_TH_R1_sch.png (69.6 KB, 255 views)
  Reply With Quote
Old 11th June 2018, 01:29 AM   #5
alexcp is offline alexcp  United States
diyAudio Member
 
alexcp's Avatar
 
Join Date: Jun 2009
Lower chipamp distortion through PCB routing
Default Schematic and BOM - SMT version

Here are the schematic and the board images for the SMT version.

The BOM is as follows:
  • All resistors except RZ - thin film 0805, 1% or better. I used Susumu RR 0.5% for the test. I did not mount R5 and R6 for the comparison with the ChipAmp.com board.
  • RZ - KOA Speer MOSX1 or a similar compact 1W resistor.
  • C1, C2 - Panasonic ECE-A1VN220U or similar bi-polar.
  • C3, C4, CC - C0G 0805. None of these were mounted for the test.
  • C5, C6 - Panasonic EEU-FC1H152L. Alternatively, the board will accommodate snap-in capacitors up to 22mm in diameter or smaller 100uF 50V capacitors (in which case you can also install C3 and C4).
  • C7, C8, CZ - Kemet PHE426.
  • CM - Panasonic EEU-FC1E101S or similar; I used Epcos because I had them.
  • U1 - LM3886.
Two additional comments:
  • R5, R6, C3, C4, and CC are optional. A discussion of them can be found on TomChr's web site; look for the Those Extra “Stability Components” section.
  • Unlike the through hole version, this board can be used as inverting, non-inverting, or balanced input amplifier. However, there is a catch: its inverting input must be either driven by a low-impedance source such as an opamp output, or grounded. Floating the inverting input will make this board an oscillator. The reason is that floating the left end of R3 reduces the closed loop gain of LM3886 to less than one, but the chip is designed to be stable when operated at a closed loop gain of 10 or greater. For the test, I connected the inverting input to GND.
Attached Images
File Type: png LM3886_SMT_R1_sch.png (83.2 KB, 265 views)
File Type: png LM3886_SMT_R1_top.png (55.0 KB, 143 views)
File Type: png LM3886_SMT_R1_topsilk.png (70.6 KB, 57 views)
File Type: png LM3886_SMT_R1_bottom.png (60.5 KB, 63 views)
File Type: png LM3886_SMT_R1_bottomsilk.png (28.3 KB, 68 views)
  Reply With Quote
Old 11th June 2018, 01:47 AM   #6
alexcp is offline alexcp  United States
diyAudio Member
 
alexcp's Avatar
 
Join Date: Jun 2009
Lower chipamp distortion through PCB routing
Thank you DPH and 00940 for quick and positive replies!
Quote:
Originally Posted by DPH View Post
Just to make sure you've seen this and the accompanying material from TomCHR?

An open source layout for the lm3886 ?
Thank you - I have seen the thread. My motivation was different, as I wanted to apply to my own design a particular technique that I learned. A chimp amp seemed to be a good application, and I think the results are worth sharing.

I have also seen TomChr's LM3886DR board. Tom has done an admirable job of making his board a solid product, fully documented and characterised.

Quote:
Originally Posted by 00940 View Post
Looking at your graphs, you're probably held up by your soundcard setup.
I agree - my test setup is noisy and distorting on its own. I may want to run another test with a better oscillator, as my sound card's output seems to be limiting the noise floor more than the input. I would also like to run a test of HF linearity, such as two sine at 19kHz and 20kHz, but for this my current measurement tech is not very revealing.
  Reply With Quote
Old 11th June 2018, 11:31 AM   #7
00940 is offline 00940  Belgium
diyAudio Member
 
00940's Avatar
 
Join Date: Sep 2003
Location: Liège
Having a look at the schematic and layout, it is a bit of a pity that you didn't improve also on the overall design of the amp. Tight PS layout, HF decoupling, output inductor and the like are things that probably won't impact much the 1khz measurements but that could manifest themselves at higher frequencies or in IMD measurements. I understand the desire to compare apple to apple with the chipamp pcb but still...

It would thus be interesting to at least run a pair of other tests even if it pushes the limits of your soundcard (a 10khz test using it at 96khz sample rate).

Further playing the devil's advocate, it might be interesting to actually validate the link between balanced impedance and your results. Unbalance things by using different resistors values for example. My gut feeling is that you're pushing things further than needed with stuff like R6-C4 (smd version).
__________________
All that is excessive is insignificant. Talleyrand
  Reply With Quote
Old 11th June 2018, 06:00 PM   #8
alexcp is offline alexcp  United States
diyAudio Member
 
alexcp's Avatar
 
Join Date: Jun 2009
Lower chipamp distortion through PCB routing
All good questions.

Quote:
Originally Posted by 00940 View Post
It would thus be interesting to at least run a pair of other tests even if it pushes the limits of your soundcard (a 10khz test using it at 96khz sample rate).
Attached is a 19kHz+20kHz test of the amp. As a baseline, the same test of the soundcard is also attached.

Quote:
Originally Posted by 00940 View Post
Tight PS layout, HF decoupling, output inductor and the like are things that probably won't impact much the 1khz measurements but that could manifest themselves at higher frequencies or in IMD measurements.
My PS layout is tight, provided that the positive and negative supplies are isolated. A bad layout would result in additional non-linearity, as power supply currents would be coupled into the feedback loop.

HF decoupling and an output inductor may help stability, not linearity.

Quote:
Originally Posted by 00940 View Post
My gut feeling is that you're pushing things further than needed with stuff like R6-C4 (smd version).
R6/C4 allow me to use the SMT version in either non-inverting, inverting or balanced configuration, and the incremental cost is close to nothing in both money and soldering effort. I also think they add to the aesthetics of the design, but that's subjective.
Attached Images
File Type: png MyGC 2-sine 13.5Vrms @ 8ohm.png (66.0 KB, 132 views)
File Type: png EMU0204+IF 2-sine loopback low level.png (65.6 KB, 124 views)

Last edited by alexcp; 11th June 2018 at 06:15 PM.
  Reply With Quote
Old 11th June 2018, 08:33 PM   #9
00940 is offline 00940  Belgium
diyAudio Member
 
00940's Avatar
 
Join Date: Sep 2003
Location: Liège
I didn't understand you were using isolated supplies. Your routing makes more sense knowing that. I can't say I'd do it like that for flexibility reasons but it makes sense.

Thanks for the answer.
__________________
All that is excessive is insignificant. Talleyrand
  Reply With Quote
Old 11th June 2018, 08:36 PM   #10
nigelwright7557 is offline nigelwright7557  United Kingdom
diyAudio Member
 
nigelwright7557's Avatar
 
Join Date: Apr 2008
Location: Carlisle, England
To add to grounding techniques it is also important with chip amps to keep the feedback path short or you can get oscillations.
1mm = 1nH of inductance.
__________________
PCBCAD51 download or CDROM 2019 version out now http://www.murtonpikesystems.co.uk
  Reply With Quote

Reply


Lower chipamp distortion through PCB routingHide 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

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Does lower Qes, Qms and Qts speaker will produce lower distortion ? dipankar862 Multi-Way 110 28th May 2017 09:40 PM
Lower load resistance for lower distortion (6L6 PP) mrneedle Tubes / Valves 2 26th March 2016 03:22 PM
Lower efficiency = more distortion? chris661 Subwoofers 27 16th March 2014 06:11 PM
does lower B+ cause scratchy distortion? PRNDL Tubes / Valves 0 20th June 2008 11:13 PM
Excel W12 lower distortion than W15?!? capslock Multi-Way 0 24th February 2004 11:34 AM


New To Site? Need Help?

All times are GMT. The time now is 07:21 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio
Wiki