I was thinking of building a multichannel gainclone amp using bridged LM3886 chips. Normally you have to run one chip in inverting input mode and the other in non-inverting mode using .1% resistors to get the gain equal in both halves of the bridge. OR, you can get balanced differencial drive some other way.

SO I was wondering if one could use the balanced preamp output of the 990 to directly drive a pair of amplifiers in bridged output mode. Some of us may have a few mono-blocks not being used that could be bridged if this trick would work.

kscharf,

I've thought about your question a couple days and can't think of a reason that wouldn't work. Don't know of anyone doing that, but seems like it should work. My understanding is the Outlaw power amps are built by ATI, and I've read ATI's description of how their "balanced" amps work. It seems to me they are essentially bridged amps in each channel. Each channel of the amp consists of a "positive" and "negative" amp driving the load. If that's the way the 7500/7700/7900 are built, what you're asking of the 990 to drive gainclones is essentially what the 990 is doing with the Outlaw balanced amps when used with the balanced connections. The gainclones will need sufficient input sensitivity to be driven to full power by the 990 differential voltage output, but that should not be difficult to achieve. I haven't looked at the 3886 data sheet in awhile, but I think they're spec'ed for full output with a half-volt input.

Bill

The input sensitivity of the LM3886 is set by two resistors that determine the gain of the stage. So if the signal isn't enough I could always just increase the gain a bit. Normally people run these at somewhere between 15-30db gain (strangely enough if you make the gain TOO LOW the amp won't be stable!), but it can be set up for a bit more too.

From NS's excel spreadsheet design aid it looks like the lm3886 needs 1.150 volts rms to drive it to full output at 1% THD when set up for 26db gain.

It's not uncommon for amplifiers to become unstable if the gain is too low. The closed loop gain usually also sets the amount of feedback, and too much feedback can cause instability. I'd consider output at 1% THD too high. The amplifier may be starting to clip at that level. Typical THD for the amp, as I recall, is a couple tenths, so 1% is starting to overdrive it. I expect the 990 is good for at least 2 volts, so a volt should not be a problem.

National has made some interesting integrated amp devices over the years, but I've never fiddled with any of them. After I considered the cost to build one to the equivalent of a ready-to-run model, it didn't seem cost effective to me. Power supplies, heat sinks, and a professional-looking chassis always seemed to cost so much. I guess I could have salvaged the parts, but didn't get into that. Of course, I missed out on the fun of doing a home-brew project.

Good luck with the gainclone.

BIll

I work for Niles Audio (NilesAudio.com). They mostly make whole house audio systems, IR interconnect devices, amplifiers and speakers. I recently obtained one of their Si1230 amplifiers for parts that was being thrown out. This unit has 12 LM3886 chips (it's a 6 channel stereo amp that runs 30w/ch at 8 ohms.) The individual amp boards and the PS (large toroid transfomer) looked ok, the control circuits were bad. I can build a nice 5ch amp from these parts, the LM3886's should run 100w into 8 ohms bridged.

I agree that specifying power output at 1% is BS, but it seems to be a common yardstick. Actually if the graphs in the LM3886 specs are correct the power does not increase much between the .1% IMD and 1% IMD points. The Niles Si1230 is actually spec'd for 30 watts out per channel into 8 ohms and 38w per channel at 4 ohms with ALL channels driven at 1khz and 0.1% IMD. (IE: worst case). This is way below what NS says the chips should do at with a +/- 30v supply.

As I recall, the 3886 is good for at least 60 watts. However, there are two reasons the Niles amp may be rated lower. One, the power transformer may not be capable of the current necessary, especially driving a reactive load with a low impedance. I've heard of 4 ohm speakers that can dip to 1.5 ohms at some frequencies. Two, the heat sink capacity of the chassis may not allow for the required dissipation at higher power. Your goal requires 50 watts from each device, which the 3886 will do with the right power supply and heat sink. With +/- 30 volts, you'll need about 4 amps for each channel, a bit more those speakers with the low impedance dips. If the transformer will handle that, the next question will be if the heat sinks are big enough. My rule of thumb is the heat sinks should not get hotter than can be touched without burning oneself (they can get hotter than that, but that's a safe limit that will not overly shorten life expectancy). If the heat in your setup does push that, you'll need to resort to bigger heat sinks or fan-forced air flow.

True, 1% THD is a common yardstick especially with IC amps. Inflates the power out a little, but really the amp is starting to be over-driven. Your speakers will thank you if you never push the amp that hard into them. It's advisable not to use the amp in the part of the output curve that curves sharply to near-vertical.

Sounds like a fun project. I'm interested in how it turns out.

Bill

Niles is usually very conservative with their ratings.

The heat sinks used in the Niles 1230 amp are MASSIVE. They do have 4 LM3886 chips per heat sink (there are Three heat sinks and 12 chips in this beast). The limiting factor is probably the power transformer. It's a toroid type with 6 42vct windings. There are 6 pcb's each having two LM3886 chips, a bridge rectifier and two 10,000uf x 35v caps for the power supply. The amp is built as 6 stereo channels, but each stereo channel can be bridged for use as a single channel with 80w into 8 ohms. I'm guessing that the VA rating of the transformer is between 500-800 from it's size and weight compared with other known transformers.

I have an old Sony receiver that I've taken apart. I'll probably build a three channel amp into it's chassis (each amp channel having two chips bridged). The heat sink that was used in this amp should be big enough with a little help from some small fans.