This project is another chipamp, which is an audio amplifier using an IC, in this case a LM4780 from National Semiconductor. The LM4780 is a dual 60 Watt amplifier chip. The particular implementation uses one LM4780 per channel, with each of the two amplifier circuits per channel in parallel with the other to provide 60 WPC into 8 ohms and hopefully about 100 WPC into 4 ohms.
Due to the 2 parallel amplifiers per channel, this type of ampifier is called a "parallel" chipamp. The purpose of paralleling is to increase the output current capability over a single chip. The parallel LM4780 should be able to drive low impedances with, in most cases, double the current of a single LM4780 channel. This configuration is known to run "hot," so I've put this into a case with fan cooling, does it look familiar?
Note that the LM4780 internally contains two LM3886 "dies," but the package is not twice as big as an LM3886. Thus the LM4780 is not quite as good at power delivery as two LM3886 chips used separately because the heat buildup in the small LM4780 cannot be dissipated as well as two separatly mounted LM3886 chipamps. For a look at an amplifier using an LM3886, check out the LM3886 web page.I built this because I obtained a neat little Sun SPARC workstation enclosure and just had to fill it up with an amplifier. I wanted to try a parallel LM4780 because the next step is a "bridge-parallel" design which would use two LM4780's per channel (4 amps per channel) to increase the power even more. The boards I used are the LM4780 boards from chipamp.com. If you look closely you see that the chassis dips down in the middle, giving very low clearance for the feet. Because of this I decided to countersink every screw on the bottom panel. You also see that this is miniscule in size compared to my Krell KSA 50 Clone. Fascinating, both amps are about 50 WPC, the Krell's output stage and power supply are massive, but the LM4780's power supply is not that weak either, and its much much smaller and more efficient. I'll have to compare these someday.
Here is what I used for Parts.
|Single 330 VA Avel-Lindburg, dual 25V secondaries from Parts Express.|
|10 x 15,000 uF/50VDC Panasonic from JEA Capacitors|
|Eight MUR1520 from Digikey|
|two National Semiconductor LM4780 from Digikey|
|SUN SPARC Station IPS workstation.|
I expected this amp to put out about 60WPC into 8 ohms, just like the LM3886 chipamp. Unlike the LM3886, this should really drive 4 ohm and lower impedances with much more current, since its essentially 2 LM3886's in one. I measured about 25 V peak (50 pk-pk) into 2.5 ohms. Now the 2.5 ohms comes from my DMM in 200 ohm setting, so the accuracy is suspect. But this is about 124 watts RMS if that reading is to be trusted. I achieved the 2.5 ohms by putting two 5 ohm 10% power resistors in parallel, so one would expect about 2.5 ohms to be the resulting resistance.
Clipping at 2 Khz, 2.5 ohm load, 10V/Div.
20 KHz Square Wave, 10V/Div, 2.5 Ohm Load
The Square wave response should show no wiggles, and it doesn't indicating that the amplifier is stable...I think. Anyhow, I looked at a bunch of waves and all looked well.
Here are the relevant schematics and simulations that I drafted up.
|LM4780 Schematic||LM4780 Power Supply Schematic||Fan Power Supply Simulation|
Here are some pictures of the amp "under construction."
(not the actual unit used).
|Never took any pics, so here is a link to what the stock unit had in it before I took out most of the stuff.|
|Here is the small power supply for the amplifier (to be expanded on later) with connectors. The extra capacitors and resistors are for the "snubber" configuration. The red capacitors are very expensive "Black Gate" N type (4.3 uF) of which it is said, in bypass configuration improve the sound. We shall see.|
|Putting together the boards and the heat sinks. You see a brownish thermal grease is used. This is "Arctic Silver 5" which is supposedly nonconductive. However, I found that it is in fact conductive, and caused me to blow a chip up. Anyway, this is why the heat sinks were later isolated. You also see a temperature sensistive switch on the bottom of the heatsink. Its normally open circuit but shorts at 45C to light a red LED, indicating things are heating up, and alerting the user to potentially increase the fan speed. The two heat sinks are bolted togheter in a few spots to hopefully make this a single thermal unit. The sinks are no longer isolated from the chips which makes the heat sinks carry -VDC but reduces thermal resistance between the chip and sinks.|
|Working on the fan power suppy. The SPARC contains a fan, and I kept the original fan and used it in this project. It is a 12VDC fan but can be slowed down by lower voltage. I used the original "cage" in the SPARc and put in a variable power supply (schematic above) which could vary the speed of the fan if the switch was thrown. Note the limited space, I had to use tapered machine screws to fit the cage flush with the top panel. A 5A fuse is aslo included in here.|
|Here is the first attempt to check the positioning of the assembled components in the case. The diode bridge is shown in the front. The RCA inputs are installed, they fit right in the openings that were already there, but not by much.|
|This did not work correctly at first. The inner part of this case has a coating that, it turns out, is conductive (probably for shielding). The case kept charging to negative VDC, so I just finally decided to go with it. I left the heat sink at -35VDC (from the non isolated chips) but isolated it from the case with nyon shoulder washers, as shown. This is not as safe as an insulated heatsink but is superior for cooling the chips.|
|I added a large capacitor bank to the power supply. This inclues 8 capacitors of 15,000 uF, which should store a lot of energy.|
|I also kept the original "light pipe" dealie and added an LED to illuminate the front when this was turned on. Had to glue the LED in the right spot, but it worked out OK.|
|Here are pics of the "lunchbox" case opened up with everything in there, back panel connectors have also been attached. the switch on the back panel next to the potentiometer shaft are for slowing the fan down.|