Introduction: Tales From the Chip: LM1875 Audio Amplifier
I love me some chip amps - tiny packages of pure audio power. With just a few external components, a clean power supply and some hefty heatsinking you can get truly hi-fi quality sound that rivals complex, discrete transistor designs.
I went into a little more detail about the benefit of chip amps in my LM386 tribute - that might be a good place to start. Here, I'll dive right into what makes the LM1875 so great and how to build a simple circuit. Ride, Dobbin!
Step 1: Say Hello to the LM1875
The LM1875 ("eighteen-seventy-five") is a monster of a chip in a very unassuming package, and another much-loved chip in the DIY audio community. The official datasheet (PDF) claims the ability to drive 20W into 8Ω loads given +-25V, and up to 30W supplied with an extra +-5V of juice... and all at less than 1% THD. And rare as it may be, I can confirm the boasting in the datasheet is spot on - those figures can be reached quite comfortably in reality (given some healthy cooling).
Step 2: Pinout
The TO-220 package, with only 5 pins, is dead simple to wire up:
1 - Negative Input (-IN)
2 - Positive Input (+IN)
Standard op-amp inputs, with the positive input receiving the audio signal and the negative input tied to ground.
3 - Negative Supply (-Vee)
5 - Positive Supply (Vcc)
Here you feed the amplifier, ideally with a dual supply. It can also be driven by a single supply by tying pin 3 to ground, however performance may suffer.
4 - Output
Here's where you dine on some sweet, sweet amplified signal.
Step 3: Schematic and BOM
Here's a simple schematic for a single channel - for stereo you'll need two of these.
R1 and R2 are the gain resistors attached to the inverting input of the amplifier. The values of 22KΩ and 1KΩ work out to a gain of 23:
Gain = 1 + (R1 / R2) = 1 + (22 / 1) = 23
To change the gain, simply swap R1 out with another resistor in the kohm range and plug it into the formula.
CIC1 to CIC4 are the decoupling capacitors for the LM1875. The smaller capacitor (100nF) filters out high frequency noise on the power rail, while the larger cap (220uF) provides a source of power to smooth out dips in the power supply. In a production circuit, these caps should be placed as close to the power input pins of the chip as possible. For more information, check out this surprisingly easy-to-understand article by Analog Devices on proper decoupling techniques.
Likewise C1, C2, R2 and R3 are there to filter out noise, while R5 acts as a pull-down resistor, allowing a path to ground if no signal is connected (hum reduction).
R6 and C3 form a RC circuit, a filter that removes radio frequencies from feeding back into the circuit and prevents oscillations from the speaker from returning to the amplifier.
R6: 1Ω, 1W
C1: 10uF electrolytic (or preferably, polyester/polypropylene film)
C2: 47uF electrolytic
C3: 220nF X7R / film
CIC1, CIC3: 220uF electrolytic
CIC2, CIC4: 100nF X7R / film
You'll need a way to feed audio in - I harvested a 3.5mm jack from an old device and made a breakout which plugs straight into a breadboard, or you could chop the head off an old 3.5mm audio cable, stick some headers on the ends and connect it directly.
Also, you'll need the usual jumpers, wires, a speaker/dummy load and a power supply - a decent variable bench PSU that can provide +/- 30V will be useful.
Finally - a heatsink! Most class A/B chipamps require significant cooling, so get a bigger heatsink than you think you'll need and keep it around for prototyping purposes.
Step 4: Breadboard Build
So here's my breadboard...
This is not the most optimal layout - ideally, the components should be much closer together, and the decoupling caps in particular are too far from the IC pins. However, I spread it out to make it easier to understand in the photos, and to make my awkward heatsink fit. The results are fine for short periods of testing.
I put both of the power rail strips on one side of the breadboard, so I could keep space around the IC for the heatsink. This has the added benefit of making the dedicated positive, negative and ground rails easily accessible along the bottom of the board.
Step 5: Don't Forget the Heatsink!
To prepare a heatsink, first line it up on the board and mark where the hole should go to secure it to the IC. Then drill the hole, and sand the entire contact surface with very fine paper until the surface is smooth and glossy.
Next, apply a dot of thermal paste to the contact surface and position the insulating mica on top with some tweezers - try not to handle the mica with your fingers.
Lastly, use a top-hat (or "bush"), a nut and a bolt to secure the chip to the heatsink. It should be just tight enough that the IC can't be rotated around the bolt, and no tighter!
Lastly, double check that the tab of the chip is insulated from the heatsink by doing a continuity test with your multimeter - with one probe on the heatsink tab and the other on the heatsink itself. No beep = good job!
Step 6: Test It!
Check and double-check that all your connections are solid, and ensure you're sending + and - voltage into the correct rails. Set the power supply to around +-10V, stand back and switch on!
If no shocking eruption of smoke appears, you've probably succeeded. Play some music and listen to your test speaker. If your bench power supply has a built-in ammeter, you can see how much current your amplifier is drawing at any given moment - try turn up the volume to see the current draw increasing.
At low voltages, you'll likely run into clipping or other forms of distortion sooner rather than later, and at higher volumes your music will sound fairly awful. Slowly turn up the voltage - the LM1875 handles +-25V like a champ, so if you have a decent heatsink there shouldn't be anything to worry about.
I ran the output into a gigantic dummy load (a 300W, 8Ω resistor) and scoped the output. With a 1kHz sinewave at 810mV peak, the LM1875 offered unto me a respectable, clean 20.15V peak (14.32V RMS) on the output - just a little over our gain setting.
In terms of clean power, I make that...
Power RMS = Vrms^2 / R
= 14.32^2 / 8
... just shy of 26W! Not bad at all.
At this point, I wanted to see if I could get to that mythical LM1875 30W mark, but first I needed to swap out the heatsink with something a little more reassuring...
Step 7: The Copper Monster
Check out this bad boy. It's a heatsink salvaged from an old graphics card - a pure copper monster with a high speed 12V fan built in. Perfect. I've never attached anything to it that could even make it break into a sweat. It's ugly. It's noisy. And it's very, very cool.
And bingo! At +/-30.5V with an input signal of 910mV, we get a clean 30.39 watts of power on the output. Lovely. What a chip.
Step 8: LM1875 Retail Kit
One more thing - once you've made an LM1875 on a breadboard, check out these awesome kits (sold everywhere - eBay, Ali, etc.). They're quite expertly laid out - a small, tightly designed PCB with a chunky star-grounding scheme and well placed components.
The resistors that come in the kits are a little dodgy - I'd replace them with decent resistors from Mouser or Digikey - but the chip seems to be completely legit. In fact, I was able to get not only my 30W target power out of two of these boards, but the noise level was utterly exceptional:
Using the Analog Discovery's spectrum analyser, that same 1kHz sinewave gave me a truly incredible -62.31dBc total harmonic distortion, which is around 0.07%! That's real hifi territory, although I'd need to measure that across the spectrum to be sure. However, this goes to show that the datasheet claims can definitely be achieved if you know what you're doing!
That's it! Next time, I'll dive right into the crown prince of the AB chip amps - the LM3886 itself.