LM1875 Chipamp Full Project

I built this audio amplifier using a dual-mono LM1875 chip configuration. The LM1875 can output 20 watts at less than 0.05% total harmonic distortion into an 8-ohm load. It is in a 5-pin package and usage is very simple - two for the differential power supply, one input, one output, and inverting input. All of these chips are simply high-powered op-amps.

Characteristics of my LM1875 amp:

• EMI filter in AC input salvaged from old PC power supply.
• 100 VA E-I transformer. 28 volt center-tapped secondary with primary on 240 volt tap, 52 volt when on 120 volt tap. Input from the grid is 120 volts, 60 Hz. Right now I use the 28 volt output, but it gives me the option to use greater voltage if I upgrade to bigger chips. After rectification this equates to about +/- 18 volts DC.
• Chipamp.com PC boards and parts.
• Discrete Diode Bridge Rectifier, with 0.01 uF bypass capacitors across each diode.
• 2 x 6800 uF capacitors for ripple smoothing.
• 2 x LM1875 amplifiers. Gain = 22
• 22 kOhm input impedance
• 100k volume control pot
• RCA input, 5-way binding post output.
• The chips are cooled using heat sinks from two old Intel Celeron PCs with natural convection.

Case Construction

The sides of the case are made from 0.6 cm thick, 15.25 cm wide oak planks that were purchased at a local home improvement store. They were already planed and smoothed and of the right width which made it very convenient. The base is a 1.5 cm thick piece of plywood to which all of the side panels are fastened to. The top consists of some oak stock purchased in the same section of the store. I cut the ends at 45° and butted them together to make a frame for the glass. The small triangular scraps that were left over were used as reinforcement of the side panels by connecting them to each other.

A thin pane of glass covers the top of the amplifier and allows the inside to be viewed.

Vents for cooling air intake are drilled near the bottom of the case and the exhaust vents are openings cut in the side panels and covered with black steel mesh.

An aluminum mounting panel on the rear is installed for the input/output jacks. The IEC power connector and fuse are mounted directly on the wood.

Power Supply Unit

Electrical energy from the grid comes into the amplifier case via an IEC connector fitted with an EMI (electromagnetic interference) filter, then passes through a fuse and switch to the transformer. A 0.22 uF capacitor across the hot switch terminals is in place to eliminate the "popping" sound (switch arcing) that is otherwise created when the amp is switched on.

The switch is a great big red toggle switch from an ancient PC, that makes a nice smack when flipped on. It adds a whole dimesion of "heavy-duty" to the amplifier case...

The transformer is an E-I style core salvaged from an old HP plotter (large printer). The transformer is probably the most friendly unit I have ever worked with. It was very easy to disassemble for rewinding and contains taps on the primary side which were designed to allow it to work with every voltage in the world. The 120 volts from the grid is stepped down to 28 volts, which is center-tapped to provide 14-0-14 to the rectifier board.

The current flows to the rectifier board where it passes through a discrete diode bridge rectifier (part of the ChipAmp.com parts package) and is converted to +/- 18 volts DC which is filtered by two 6800 uF electrolytic capacitors (these I added myself - one for each rail).

Amplifier Units

There are two amplifier boards, one for each channel.

LM3875 CHIPAMP SAMPLE SCHEMATIC

Above is a sample schematic of a chipamp using LM3875. Nearly the same circuitry can be applied to LM1875, only the pinouts of the chip are different. I did not draw this diagram, but it represents probably the simplest chipamp design, and I believe it is a great place to start. As of 5/19/2008, my LM1875 is based upon this circuitry. I plan to modify/play with more complex designs now that I understand the concept.

Tinkering with the Chipamp

Based on the above schematic, some basic modifications can easily be done:

• The GAIN (degree of amplification) of the chips is dependent upon the ratio R_NFB/R₃. If we decrease the impedance (it doesn't HAVE to be a resistor!) of R₃, our gain will increase. In the case of the schematic, it would be 22/0.68 = 32.6. So a given voltage input will be multiplied by 32.6 on the output.
• The resulting peak-to-peak voltage coming out of the chips cannot be higher than what is available on the power supply rails or the amplifier will clip (distort). Line-level sources are usually around 0.5-0.7 volts MAX, so take that into consideration when figuring out gain. Don't make it too high, but not too low either as you will not be taking advantage of the chip's true power. Some chips also will not be stable at very low gains (such as less than 10).
• The power that will be produced should also be calculated, by dividing the square of the resulting voltage by the speaker impedance (generally 8 or 4 ohms). The power levels should not be higher than what the chips are rated or what your power supply can pump out. It is best to be limited by the former because the chips are not 100% efficient (they are really about 60-70%) and a wimpy power supply will cause the voltage to sag and produce horrible sound near the upper boundary of its power limits.
• The capacitors C_S are bypass capacitors which exist to reduce high-frequency cruft on the power supply and do not directly affect the audio.
• VOLUME CONTROL is usually accomplished by placing a potentiometer inline with the incoming signal. This decreases the voltage reaching the chips. The gain remains the same, just more or less voltage is reaching the input pin. Pots from 10-500k have been used. I've found that 10k works well, anything less than that is probably not a good idea (will essentially short high-impedance sources)
• More esoteric designs will use a pot in place of R_NFB or R₃. These usually are not seen in chipamps, as there is a philosophy to keep the wiring associated with R_NFB (the negative feedback loop) as short as possible to reduce noise, increase stability, etc.
• To use an input capacitor, it is usually placed between R₁ and the chip's input pin. Polarized capacitors (a.k.a. regular electrolytics) are not very good for this, and electrolytics in general are usually avoided. I've read that polyester and polypropylene are good, and I've seen values from 0.1 uF all the way to 10 uF (the 10 uF's are almost always going to be electrolytic...or else it will be huge)

Description of Operation

For those who can understand words better than a schematic...

The amplifier board uses a basic non-inverting configuration (meaning that the output is the same as the input, the waveforms are not "flipped upside-down". Some chipamp designs do invert the signal. To your ear it will sound the same, however).

The audio signal comes in, and enters the input pin (+) of the chip. I decided to omit a capacitor input - yielding theoretically better sound but with the risk of amplifying direct current voltages on the input.

The inverting input (-) of the LM1875 is coupled to the output via a 33 kohm resitor, and is coupled to ground via a 1.5 kohm resistor. This forms the feedback loop, and basically gives the amplifier a gain factor of 22 (33/1.5), meaning that a 1 volt input would equate to 22 volts output.

Some builders have varied the resistance in the feedback loop with a potentiometer. This allows volume to be controlled slightly differently from the traditional method of placing the pot in the signal path. Technically, this violates the rules of the "Gainclone" because it extends the feedback loop and in gainclones it is to be kept as physically short as possible.

The only other components on the amplifier boards are two bypass capacitors on each power rail very close to the chips to eliminate any high-frequency noise. They are 0.1 uF tantalum and 100 uF electrolytic.

The output travels out of the chip and through the case via 5-way binding post/banana sockets.

The LM1875's are cooled using aluminum heatsinks from old desktop computers, using natural convection (no fans). At full output, they would each be dissipating somewhere in the neighborhood of 14 watts. Don't underestimate cooling - on LM1875 it is less of an issue but with the higher powered chips, be sure they are adequately cooled if you are running them at higher powers. Cooler chips will last longer and probably sound better (as high temperatures usually mean a high level of random molecular movement, but people love the sound of vacuum tubes, so maybe I'm wrong!) Semiconductors, however, are best cooled.

Resources

• LM1875 Data Sheet From National Semiconductor.
• Chipamp.com - PC Boards and Parts Kits