The yet tiniest single-sided LM3886?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi folks,

after seeing a lot of the LM3886 amps that were assembled directly on chip, I decided to create a simple "cleaner" version on a single-sided PCB which schould make it easier for the beginner to assemble it. I would be grateful to hear your thoughts!

The PCB was designed with Designspark (tried it for the first time)

Values are as following:

VN: -25.5V negative supply voltage
VP: +25.5V positive supply voltage
Out: output to speaker
Gnd: self-explanatory...
In: input to the amp

Cz: 100nF, 100V, polyester
Cin: 4.7 uF, 50V, polyester
Rg: 22k, 1W, carbon
Rf: 22k, 0.25W, carbon
Ri: 680R, 0.25W, carbon
Rm: 10k, 0.35W, metalfilm
Rz: 2R7, 2W, carbon
Rbr: 10R, 0.25W, carbon

The values were taken from this Gaincard-clone idea:
Mick Feuerbacher Audio Projects

The PCB is 30.46 * 28.18 mm

The power supply with caps and diodes is on a seperate board.

An externally hosted image should be here but it was not working when we last tested it.


Best regards,
hurtz
 
And here is the PSU, I used five 35V 4700uF caps since I've got a lot of them lying around, diodes are MUR860, Transformer will be 18V, 80VA.
This should result in +-25.5V, (8 Ohm) 30W output; 15W dissipation.

I see many people are using small around 100nF caps in parallel, what are they for?

Again, any input is welcome :)

An externally hosted image should be here but it was not working when we last tested it.
 
Thanks :)

If there are no major objections/suggestions I will try this design, and if it performs well, I will publish the etchable version of the board, and by request of the PSU although I don't think many people have 5mm caps lying around :)

Since it should stay tiny, there is no room for a ground-star :-/

cheers!
 
Neat. I'd look at all those ground tracks though, a star is a much better solution.

+1 (here you probably want to turn the whole layout inside out, and keep the INs and OUTs on opposite sides)
dont forget 2x Cs! they really need to be close to the chip! the value and part quality is esp important for the power amp stability and the sound. Sorry but this blows up yer small PCB...deadbug style is looking good.
 
So as I understand it the decoupling capacitors are there to provide current to chip really fast right?
Never actually thought a cable could be too slow for this :-/ but, OK.

@infinia at what would I look when selecting them? I could make a guess and say a low internal resistance, but this is as far as my wisdom carries me. Could I also use my 35V 4700uF 85°C ? There is also something called ripple current: 2710mA 120Hz, which as I understand is äquivalent to the internal resistance?

@discrete I calculated for a maximum of 4.44A (thats as much as the transformer can handle) at 35um copper thickness and 200mm track length a width of 2.348mm so it should be fine, or am I missing something? (I actually used 2mm until now, so new design is mostly 5mm which should be more than enough)


So here is revision 1.1 of the PSU and the board. as Cs I used caps of the same size as on the PSU board.

PSU: 80 x 60mm
AMP: 50 x 40mm

PSU:
An externally hosted image should be here but it was not working when we last tested it.


Amp:
An externally hosted image should be here but it was not working when we last tested it.
 
@discrete I calculated for a maximum of 4.44A (thats as much as the transformer can handle) at 35um copper thickness and 200mm track length a width of 2.348mm so it should be fine, or am I missing something? (I actually used 2mm until now, so new design is mostly 5mm which should be more than enough)


...more like this (please excuse the freehand doodle).

Think of it as an RCRC filter, where the tracks are really small Rs.
A star for the ground would be another improvement.
 

Attachments

  • netzteilv11a.png
    netzteilv11a.png
    24.7 KB · Views: 1,367
Just another Moderator
Joined 2003
Paid Member
So as I understand it the decoupling capacitors are there to provide current to chip really fast right?
Never actually thought a cable could be too slow for this :-/ but, OK.

No not really, The purpose of the decoupling capacitors is to shunt any residual ac noise on the DC power supply rails off to earth. By putting them as close to the chip as possible it will maximise their effectiveness.

from page 19 of the datasheet.

It is advisable to keep the output compensation
components and the 0.1 μF supply decoupling capacitors as
close as possible to the LM3886 to reduce the effects of PCB
trace resistance and inductance.

Note that the 0.1uF caps deal with high frequencies, the larger (assuming you have them) decoupling caps should also be close to the chip.

Tony.
 
Last edited:
Chip amp power supply decoupling needs 2 caps per side.
A small value cap is for high freq and the other is needed to keep output modulation ie feedback from creating stability or oscillation problems. I think the NS app sheets give some values. I would use a 0.1 uF small ceramic or film (closest to IC pins) the other larger cap could be polar electro or tantalum from the high 10's to several 100's of uF, also close by on the same PCB. Some folks use larger values here, up to 2K uF , claiming bass sound is improved.
 
@discrete, (thanks for your reply) I see now what you mean, but why would I want my PSU to behave like this? A quick number juggling tells me that this RC-filter gives me a cutoff around 7Hz... (23500uF, 1 Ohm <- wild guess). Also I want to avoid a ground-star to avoid jumpers... may seem silly but I like it to be as simple as possible.

Thanks for the program link, this is quite awesome!

I also looked through the data-sheet again, the tip concerning the decoupling was very helpful. as per NS the small one should be about 0.1uF, and the large one above 10uF. In my design I chose 22uF for the large one.
I saw many designs which use around 1000uF for this, is there any significant change in sound going from the advised 10uF to 100 times the value?

So new updated values are:

Cs1: 22uF
Cs2: 0.1uF

Cs3: 22uF
Cs4: 0.1uF

The updated AMP:
An externally hosted image should be here but it was not working when we last tested it.



P.s. I use 10 4700uF caps for the PSU since I have about 35 of them lying around...

P.p.s
@AndrewT I feel with you... since this happend to me two times in a row I almost never forget to copy the whole text just before hitting the "post" button.
 
Last edited:
Ex-Moderator
Joined 2002
...When will the Forum properly investigate why the system logs out "logged in" Members?

Sorry Andrew, but I suspect the problem is at your end. If it was something the S/W was doing at this end I'd expect many more complaints than we see.


Also I want to avoid a ground-star to avoid jumpers... may seem silly but I like it to be as simple as possible.

So you're set on having it smaller even though it might significantly degrade performance? Seems a shame to me.
 
Sorry Andrew, but I suspect the problem is at your end. If it was something the S/W was doing at this end I'd expect many more complaints than we see.

It's not just Andrew, I've had it in the past. If you take longer than about 20 minutes creating your post then when you come to press submit it looses it all. Pressing back just gives you an empty reply. Very frustrating.

Now if I know I am going to take a long time to write a reply I copy the reply before hitting submit.
 
@discrete, (thanks for your reply) I see now what you mean, but why would I want my PSU to behave like this? A quick number juggling tells me that this RC-filter gives me a cutoff around 7Hz... (23500uF, 1 Ohm <- wild guess). Also I want to avoid a ground-star to avoid jumpers... may seem silly but I like it to be as simple as possible.

In the simulation below, do you want to feed your amplifier with V1, V2 or V3?

Using V1 is a good way to have mains hum on your speakers.
The PSRR of the amp chip may (or may not) bring this down to inaudible levels, but why not do it the correct way?

Using a star ground is left as an exercise for the reader :).
 

Attachments

  • rcrc.png
    rcrc.png
    41.5 KB · Views: 859
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.