some T-amp modification suggestions
Nov 22, 2004 at 6:56 AM Thread Starter Post #1 of 4

sluggo

New Head-Fier
Joined
Oct 7, 2004
Posts
35
Likes
0
Some of you are interested in modifying the Sonic Impact "T-amp" for better performance. I don't have an amp to play with, but I have a lot of experience with the Tripath parts used in them. Some suggestions and what-not:

1. Do not modify anything around the output pins of the amplifier chip. This is a switching amplifier that operates at around 700 kHz and the placement of the components in the output circuit is critical. This is the single most important piece of advice I can offer. Get this wrong and you can blow up your amp and any speakers connected to it.

2. You CAN experiment with the main output inductor (as this is the place where most OEM's will try to save some money), but you should NOT try different values. Stay with the recommended value (10uH), but maybe experiment with a core with a larger saturation current. (Based on the photos, the cores used in T-Amp look to be pretty good.)

3. The 180uF supply cap should probably be beefed up a bit. I would also bypass it close to the chip with a small value high-quality film cap. Those of you going with mondo linear power supplies can ignore the beefing up, but adding a small film cap is still a good idea. A regulated supply is not required, but you don't want something saggy.

4. Replace the 2.2uF input electrolytics with a similar value film cap with at least a 12V rating. Electrolytics (even good ones) are usually not the best sounding cap in a signal path. Avoid using a smaller value part, as that will roll off the bass.

5. The part has built-in thermal protection and you will probably be well into audible clipping before the thermal protection kicks in. None the less, if I were a serious modder I would work out a way to attach a big chunk of metal to the solder blobs on this board. The part is designed to operate with die temperatures approaching 150 C, but a little thermal mass never hurts, even if only to allow the part to run comfortably when it's hot in the listening room. Old microprocessor heatsinks (passive) are great for this.

6. There's a general-purpose FAULT pin that's handy for problem situations. I don't know if it's used in the T-amp design, but if not it's a nice addition. Lights up for undervoltage on the power supply or charge pump, thermal warning, and overcurrent (shorted output). Pin 19, it can drive an LED directly through a 5k resistor.

7. Be careful about what you use for a power supply. The part can be damaged permanently by a supply that wanders much above 17 V. If your supply spikes when turned on, use a relay with a time delay to protect the amp chip or use a different supply.

8. Remember that the outputs of the amps are run in bridged mode, so no connecting the output "grounds" together. These are not grounded outputs, and tying them together will damage the amp.

9. The amp is pretty efficient and will run a long time on a standard 12-13.2V gel cell if you need to go mobile.

Good luck, and have fun playing around with this amp.
 
Nov 22, 2004 at 2:29 PM Post #2 of 4
Great tips and insite into this chip.

Maybe we should move this into the DIY forum?
confused.gif
 
Feb 25, 2005 at 3:28 AM Post #4 of 4
Since you have experience with these chips, I was hoping you could critique an idea I had. I want to try this amp with headphones that have the standard 3 contact jack. I was thinking of using 2 6V SLA batteries in series for the power supply and connecting the headphones as follows. Ground on the headphones would go to +6V on the power supply. The right and left channels on the headphones would be connected only to the positive half of the right and left channels of the amp. The negative half of each channel would be left floating or connected to ground (0v) through a power resistor. My assumption here is that the outputs of each channel switch polarity (+-12V) simultaneously so that each headphone channel would get a normal signal at half the amplitude. Do you think this would work? I am hesitant to test it because I worry that having different loads on the positive and negative portion of a channel might be damaging to the device or might confuse some kind of feedback system on the chip.
 

Users who are viewing this thread

Back
Top