Sure Electronics New Tripath Board 4*100W class-D Amplifier Board

[hubsand]

They sound amazingly good, when you get one that works! Item Audio swapped out my first board . . . they're really PSU-sensitive (as you might expect). You'll probably experience a transformation when you can slot in a better power supply

[pheonix358]

Hi, please see the attached two pics.

Taking note of the heatsink photo you can clearly see that three components have suffered heat failure. These are the TP2050 chips for channels 1 and 4 and the regulater chip. The question is what went first and why? Note also that more than sufficient heat transfer paste has been applied.

The second photo is out of focus but look at the left most chip. It is the TP2050 chip for channel 1. Notice that around 1/3 of the chips heat conductive surface was not touching the heatsink.

Now, my (still working at it) hypothesis is this.

At manufacture the daughter board is fixed to the main board and the header pins connecting the two are soldered. It is now solidly fixed in place.

Next a good amount of heat transfer paste is applied to all four chips and the heatsink is bolted on. In some boards the heatsink is bolted with sufficient force to cause the boards to flex. This is caused by the header pins being soldered in a minute fraction such that they are just a little tiny bit too high. The main board on mine was flexed enough to bow the main board my several mm. This force is enough to cause channel 1 TP2050 to come away from the heatsink by a tiny amount. This is enough to cause the chip to loose 1/3 of it's heat transfer area.

To make the problem worse, the TC2000 chips are located under channels 1 and 4 TP2050 chips adding to the heat problem. The heat builds and at some stage a catostrophic failure occurs taking the other chips with it. Both TP2050 chips for channels 1 and 4 are under increased heat load due to the location of the TP2000 chips which, being sandwiched between the two boards, has no chance whatsoever to dissipate any heat at all. The heat just builds until failure.

A solution would be three fold.

Redesign the daughter board such that the two TC2000 chips have some air circulation so they can dissipate heat.

Fix the heatsink to the daughter board before soldering the header pins to the main board.

Provide two more central fixing point for the heatsink to avoid the flex problems.

Any comments are welcome, please.

Terry

[audio1st]

I don't have one of these boards but looking at the diagram I see 100k resistors to ground after the input caps. This will give dc through the amp..
And no input caps via input blocks..

[pheonix358]

Originally Posted by audio1st

I don't have one of these boards but looking at the diagram I see 100k resistors to ground after the input caps. This will give dc through the amp..
And no input caps via input blocks..

Sorry, no they don't, they just set the input impedance.

Terry

[pheonix358]

Your other observation is rather critical to know. I wonder if this is a choice or an oversight.

Terry

[pheonix358]

Or it's so you can add any input cap you choose!

Terry

[audio1st]

Both problems have now been addressed.

From Sure:
Thank you so much for your mail.

We have introduced your design in the development of this amplifier board. We have removed R6,R7,R8 and R9 and replaced R12,R13,R16 and R17 with monolithic capacitors so that there will be more space.

We are grateful for your support.

[pheonix358]

Originally Posted by audio1st

Both problems have now been addressed.

From Sure:
Thank you so much for your mail.

We have introduced your design in the development of this amplifier board. We have removed R6,R7,R8 and R9 and replaced R12,R13,R16 and R17 with monolithic capacitors so that there will be more space.

We are grateful for your support.

Hi, Audio1st, could you please explain how this resister can put DC through the amp. IN all amps I've seen (Class A or AB) a resister at this point sets the input impedance. How does it work in a class D.

Should those with current boards make these mods?

Thanks, Terry

[audio1st]

Hi Terry,

If you think of it as an amp with a +/- 2.5Vdc supply. GND is connected to the -2.5Vdc supply. The actual 0V point is the bias cap pin. That is why you must use an input cap (dc blocker), after any input connection to GND.

The 100k resistor is not the main problem, Sure may have been able to dial out the dc from this with the trimmers?

The main problem would be shorting the terminal block inputs to GND with say a pot. If you use the rca inputs, then there should not be a problem..

Barry.

[obsoletemac]

So, should R6,R7,R8 and R9 be removed even when using the RCA-connectors? (They seems to be located under the heat sink)

//Christian