DIY Chip Amp From Old Plate Amp Parts?

[pacificblue]

Originally Posted by crazifunguy

Could it be as simple as setting up the single supply in the same configuration as the split supply bridged?

Basically yes, but to make your life easier, don't use the bridge schematic from the datasheet. Build two identical single supply channels and then use Simplest Ever Bridging Adapter for Amplifiers. IF you leaf through datasheets, you will find that the TDA amps usually apply this method for bridging, too.

[danielwritesbac]

Here are some examples of single rail bridge amplifiers, and although these don't match the voltage requirements, I'm listing them so that you can look at more examples of bridging and single rail.
http://electroschematics.com/wp-cont...bridge-bcl.jpg
http://electroschematics.com/wp-cont...005-bridge.gif
http://www.fidelityforce.com/daniel/tda2003bridge.jpg

[danielwritesbac]

LM1875 (high gain) and LM675 (low gain) are similar 5 pin chips, but able to operate on your 42 volts. By looking at the gain setting on the examples, you could figure out which to use. Yes, those can be bridged, but the thermal interface is little, so they like 8 ohm speakers.

The TDA7294, which shows a split rail bridged application in its datasheet, contains a simple current limiter that operates quietly (no screech) and a nicely large size thermal interface (so you can use a wider variety of speakers). http://www.datasheetcatalog.org/data...qfjx9qjc3y.pdf On the 42v single supply, that device should be quite durable.

Edit:
These two amplifiers are my personal favorites due to a very friendly tone. The LM1875 is easier, even if you spend the extra 5 minutes to do a 2-chip parallel. The TDA7294 is extremely durable if you first make it run cool, which is pretty easy with a bit of power circuit fine tuning and a good heatsink. These two can be made to sound the same, but the little one is easier and faster to apply.
However, I still think you're looking for TDA7292.

[danielwritesbac]

@PacificBlue
Is there an easy way to simple bridge a single rail parallel amplifier similar to this method: Simplest Ever Bridging Adapter for Amplifiers

[pacificblue]

The very same. You need to add the components to each parallel channel. If you want to bridge two amps with 3 parallel LMs each, you need 3 times the added resistor and 3 times the 100 Ohm resistor to form 3 bridged pairs.

For the following component names I am refering to the LM1875 datasheet schematic on page 2.

For a single-supply parallel amp, you have two options.
1) You can use C6 on each amp and connect the speaker after C6. In that case you don't need the load sharing resistors, because there is no DC offset after C6. For the same reason you don't need to match the gain setting resistors. Since each channel sees effectively a lower load, you can reduce the output capacitance accordingly. C6 = C6 / no. of channels. E.g. 3 parallel channels use 680 µF per channel instead of 2200 µF.
2) Use the load sharing resistors before C6 and use a single C6 between the resistors and the speaker. You still have to match the gain setting resistors.

For a bridged amp you have another two options.
a) Use C6 only on one side of the bridge.
b) Use C6 on both sides, and double its value, because the amp sees a higher load.

That gives a total of 4 combinations, for which you need

1) a) 3 capacitors of 680 µF on one side of the bridge, 3 load sharing resistors and gain matching on the opposite side of the bridge.
1) b) 6 capacitors of 1500 µF, no load sharing, no gain resistor matching.
2) a) 1 capacitor of 2200 µF, 6 load sharing resistors and gain matching on all channels.
2) b) 2 capacitors of 4700 µF, 6 load sharing resistors and gain matching on all channels.

[pacificblue]

Oh, and the 100 Ohm resistor connects to virtual ground, not to real ground for split supply.

[danielwritesbac]

Thank you!
I am so fascinated.

Because multiple caps with similar signals could possibly make epic cancellations, I would shy away from options 1a and 1b. I would choose option 2a, which is 1 output capacitor. Well, except that my one capacitor would look like 47nF//10uF//3300uF//10,000uF as in pretty much a solid wire performance for AC. Whatever is good enough parts for a power supply, is probably good enough for an output cap, and vice verse. Probably I'll think up something less expensive momentarily. Edit: 22nF// Nichicon ES 2uF//22uF//220uF// Nichicon PW 2,200uF//2,200uF and that is a lot less expensive. Those parts often see active duty in high performance economy speaker crossovers and should perform brilliantly as an output cap. yay! EDIT2: Did it again. Nichicon ES 0.5uF and then any good 470uF//2,200uF//2,200uF. So, about $3.

Although there seems to be available bridging options with no output cap, I don't mind using them--it is cheap speaker protection.

For the feedback resistors, it seems that I would need at least a 100 pack of resistors of a peculiar value that only available in 1% values (because otherwise you receive 5% resistors with a 1% paint job), and even then, measure with the digital ohmmeter.

Even though I was expecting caveats, you made it sound so easy to do!
One interesting aspect is that the master amp needs to be a really good sounding amp before bridging. Probably will have to check that with Rod Elliot's Headphone Adapter, prior to bridging.
At about that same connection point. . .
I'm thinking that it would be a good thing to use the unified output of the 3 load sharing resistors of the master amp as the point to connect the 3 "bridging resistors" since the slave amp is a parallel amplifier and its chips run cooler if each receives an identical copy of the signal. Does that seem like a good plan?

[danielwritesbac]

Originally Posted by pacificblue

Oh, and the 100 Ohm resistor connects to virtual ground, not to real ground for split supply.

Can I use 1k stopper to the non-inverting inputs of all of the chips to help stop turn on thump? And, in this case, do I need to add the 100 ohm to the 1k?

[pacificblue]

Originally Posted by danielwritesbac

Although there seems to be available bridging options with no output cap,

To achieve that, you have to modify Elliott's schematic a little. Swap the places of resistor and capacitor in the NFB-to-ground leg, so that the resistor goes to ground and the cap to the inverting input. Then connect the 'added resistor' from Elliott's schematic between cap and resistor instead of to the inverting input. Replace the 100 Ohm resistor with one of the same size as the resistor between virtual ground and the master amp's non-inverting input. That should bring the offset between the amp outputs to a level that makes the output capacitor expendable. Adjust the resistor a bit for bias current tolerances, if the output voltage differences per bridged pair are too great.

Originally Posted by danielwritesbac

For the feedback resistors, it seems that I would need at least a 100 pack of resistors of a peculiar value that only available in 1% values (because otherwise you receive 5% resistors with a 1% paint job),

You should really complain, if your supplier does not deliver what you order.

Originally Posted by danielwritesbac

I'm thinking that it would be a good thing to use the unified output of the 3 load sharing resistors of the master amp as the point to connect the 3 "bridging resistors" since the slave amp is a parallel amplifier and its chips run cooler if each receives an identical copy of the signal. Does that seem like a good plan?

Remember that current through a resistor leads to voltage drop. The voltage behind the bridging resistors is lower than before them. The slave amp would produce an accordingly lower output than the master, so you get a voltage offset.

Originally Posted by danielwritesbac

Can I use 1k stopper to the non-inverting inputs of all of the chips to help stop turn on thump? And, in this case, do I need to add the 100 ohm to the 1k?

Check the LM4780 datasheet, page 7. There are three resistors called Rinp and Rbi for that. Start with Rinp, and if that alone is not sufficient, add the two Rbi resistors. You will have to adjust their values according to the power supply, but it should be possible to remove clicks and pops through them.

[danielwritesbac]

Originally Posted by danielwritesbac

I'm thinking that it would be a good thing to use the unified output of the 3 load sharing resistors of the master amp as the point to connect the 3 "bridging resistors" since the slave amp is a parallel amplifier and its chips run cooler if each receives an identical copy of the signal. Does that seem like a good plan?

Originally Posted by pacificblue

Remember that current through a resistor leads to voltage drop. The voltage behind the bridging resistors is lower than before them. The slave amp would produce an accordingly lower output than the master, so you get a voltage offset.

Offset voltage that varies by output current? oops!!
Of course my next thought is that one of the master amplifier's chips (perhaps the center?) drives all three chips in the slave amplifier, identically, which is required for thermal management.

And the thought after that was. . .
Take 12 copies of the feedback resistor, double them up for 6 resistors of exactly half value, put pairs of that in series with each other for a total of three center tapped bridging resistors, and fuse the centerpoints all together thus forcing identical voltage input and current input to each chip in the slave amp, insuring manageable thermal output in the slave amp.

Am I getting closer?