LM1875 Stereo Amplifier with integrated PSU

Hi all,
I always wanted to create something for a beginner which is simple to build and very convenient and yet cheap.
what better chip than LM1875, 'the beginner's Hi-Fi' or ' beginners gainclone'.

Criteria was just a single cheap Chinese boardhouse pcb, left and right channels on the same pcb + PSU + all though hole components.

so just pcb + heatsink +center tapped transformer + some hardware and a cheap case is all that would be necessary to build your own 'my first hi-fi amplifier'

or it can also be a learning exercise for young ones who are interested.

One thing to note here , THT + all in one (stereo +PSU) PLUS small PCB dont go hand in hand. so one cant expect loads and loads of capacitance in power supply section.

I havent tested it, but layout follows the schematic perfectly. comments welcome.

Cheers to the spirit of DIY.:)

regards
Prasi
 

Attachments

  • lm1875stereo-sch-R3.png
    lm1875stereo-sch-R3.png
    12.9 KB · Views: 761
  • lm1875stereo-R3.png
    lm1875stereo-R3.png
    77.4 KB · Views: 798
I would recommend to remove the fuses from DC rails. If one fuse goes off it will take the woofers of your speakers with it. If you really need fuses then put them on the AC side of power input. Also instead of "spider" ground you can make a big copper pour instead covering all of the PSU filter caps pads. This will significantly reduce ground impedance of the PSU.

Edit: Even better would be to make GND plane over the entire top half of the PCB.
 
Last edited:
there are so many conflicting views on "to star" or "to ground plane"...I am not getting involved in another. Some people say making a plane with THT components is creating multitude of loops. + the datasheet mentions it is better to use different return paths for different components.
I have built a bigger SS class AB with such "star" configuration, it works splendidly well. So I am not going to make a "ground plane".

regarding fuse, leave it for users, who can just jumper it, but some prefer fuses on PSU. more advanced users can implement a properly designed speaker protection.
 
there are so many conflicting views on "to star" or "to ground plane"

I use a combination of both.
Ground planes connected to ground around sensitive audio signals.
Star grounding for the likes of zobel networks and smoothing capacitors.
The biggest culprit of hum is the charging impulses into smoothing capacitors. It can modulate the ground line badly.

I recently did a chip amp pcb.
I learned a bit from the first one.
Keep tracks to feedback resistor as short as possible as the tracks act like inductors and phase shift the feedback to the point where the chip amp oscillates.
Keep decoupling as close to chip amp pins as possible.
 
Does C5 actually do anything?
Should C6 and C7 form a tighter circuit around the IC supply pins for best HF performance?

Don't be afraid of SMT, you just need two irons to remove passives.

I agree about the fuses. You could consider smaller fuses, perhaps even resetting PTC types? Or are these too slow?
The general grounding, I'm not sure a full plane will be easy to do, you still need to think about capacitor current causing issues in the signal paths.
But a star for each channel might be an option?
 
....

regarding fuse, leave it for users, who can just jumper it, but some prefer fuses on PSU. more advanced users can implement a properly designed speaker protection.
I don't dislike fuses on the PCB, it's whether they should be in the AC or DC nets is the question.
If you moved one to each AC net, a blown fuse would result in the amps being fed with half wave rectified supplies, reduced power, probably a bit of hum, but no big DC on the speaker?
Whereas one blown fuse as drawn means the chip amp will probably go to a rail?
 
C5 DC block cap, as per TI app note, its 22uF, but generally its kept at 100u to 220u, sometimes even larger.

C6/C7 any closer, may require SMD, which I am not willing, as the whole purpose as stated in post #1, this is meant as a first amp for a beginner, a very easy one at that. fuses are ordinary PCB mount fuses (5 x 20mm) and holders are amply available everywhere. may be because the PCB is just 55mm wide, so fuse holders appear very big?


TO Nigelwright,
Yep I have done exactly that, i/p grounds connected with a plane. rest on its own to the star.
 
If we use external PSU we usually twist the wires. "Twisting" traces is more difficult thus the suggestion to use GND plane over all power and output sections. There are no signal traces in the upper and middle parts of the PCB which makes it an easy job. Just reroute the feedback trace from the IC1 output to R6 on the left side to go straight between the IC pads to avoid capacitive coupling to the GND plane in the upper part (if you make one). Move feedback parts R6'//C8' to the lower left of the IC2.
 
C5 DC block cap, as per TI app note, its 22uF, but generally its kept at 100u to 220u, sometimes even larger.....
Yes it's there in the app note.
The limit spec for input offset voltage is 15mV, with your gain of 22, this would give 330mV of DC on the output. Which is quite high.
Well it sounds quite high to me, I don't know what most people would consider to be acceptable.
I was expecting the offset voltage to be a lot smaller, the typical is, but the limit is not!
So, it's desirable to reduce the gain to 1 at DC.
 
The capacitors on the PSU are unable to supply current at a high frequency. That is why we need to use local bypass capacitors close to the chip on the amp board.

ok, so here it goes.....with lots of copper, but yet not over THT pads of different signals to maintain continuity.

I plan to use lot of copper for rail traces too but later (pro-bono :D)

to Chris,
I didnt understand what you mean by 330mV offset. Many many have built it over years using the 100u and above C5 without reporting any DC offset problems.

a general question, how high can be the value of R4 w/o deteriorating performance as 2u2 cap for C1 is a bit expensive compared to 1u i/p coupling cap. we would get the same -3db point, for example using 47k for R4...
 

Attachments

  • LM1875 stereo+PSU-new rev.png
    LM1875 stereo+PSU-new rev.png
    75.4 KB · Views: 398
Last edited:
...
to Chris,
I didnt understand what you mean by 330mV offset. Many many have built it over years using the 100u and above C5 without reporting any DC offset problems.

a general question, how high can be the value of R4 w/o deteriorating performance as 2u2 cap for C1 is a bit expensive compared to 1u i/p coupling cap. we would get the same -3db point, for example using 47k for R4...
The 330mV offset is what you might get if the chip only just passed its 15mV input offset spec, and the DC gain was 22 the same as the audio gain. Having a cap in the non-inverting side makes the DC gain 1, reducing the worst case DC offset.
I suppose we could select our chips for low DC offset, while the recommended circuit is intended to get max yield on a production line.

How big a resistor? Bigger resistor means more noise and also creates a voltage with the input bias current and an offset with the input offset current.Work out the noise voltages and see if it's significant compared to CD player noise and all that?

OTOH, you could shop around for your caps. In industrial terms, small, low voltage electrolytics are not seen as expensive. I'm not an audio designer so I've no idea why only certain designer labels are any good for piano music or whatever. A Nichicon 3u3 would cost me about an old British Shilling. But that's a whole different argument.
 
I am not that interested in chip amps anyway... this is meant for any beginner who wants build an entry level amp. so main reasoning is parts should be easily available to anyone. 1u film cap is universal, epcos/kemet 1u/63V film cap is available in a local electronic store near me, so are 0.1u, 0.22u and 0.47u. nothing above 1u sadly.
Generally Film caps are preferred for C1 duty.
 
Finally...done it.

Can someone tell from experience,
if the HS154 x 80 (depth 70mm) heatsink will be enough to take care of stereo channels on a single heatsink..
with 4 ohm speakers (worst case ) and home music listening ( not full blast, okay may be occasionally;)).

sorry forgot the link..
Welcome to Akash Enterprises
 

Attachments

  • new-LM1875.png
    new-LM1875.png
    84.8 KB · Views: 299
Last edited:
I would remove C11-C14 since they only bring the diodes ringing frequency down but do not eliminate it. You need a true CRC snubber at the AC side to cope with such ringing. Also why do you insist on having fuses in DC rails? It was already mentioned that it does more harm than good in this position. It will almost certainly fry woofers of both channels if one fuse goes off while the other is conducting and this is very likely situation for a beginner for whom you intend this project. Why don't you move the fuses to the AC side before the diodes? This will keep speakers safe in case if one fuse goes off and still provide the same level of protection in case of catastrophic failure. The PCB area is the same, so there should not be a problem here.

Regards,
Oleg