Exploding TA2022 Tripath Chips!!!

[bongoman]

Hi Everyone,

I've built a couple of 41hz.com AMP10b kits, based on the Tripath TA2022 chip. Both of these kits worked for a bit, producing excellent sound into 8 Ohm speakers. I haven't once managed to make the amplifier clip or distort in any way, when it plays it sounds lovely.

...However, I only seem to be able to get about a day's use out of them before they spontaneously explode!

A few notes:

• It's running off +/-33v rails from the 300VA torroidal tranny.
• I've had explosions using a standalone Creative Zen MP3 player source and from my DIY TDA1543 DAC (running from a seperate transformer), driven by a Squeezebox over TOSLINK.
• Explosions occur when I've unplugged the input source, or I've just been playing music loudly.
• The amplifier has been set to both "general use" and "high sensitivity" gain, as per the build instructions.
• All my components are in correctly as per the silkscreen.
• All my soldering is fine.
• I have currently not grounded the board, any heatsinks etc. It's just connected to the transformer (I'm wondering if I should)
• The chip failure is always the same, a small piece of the front blows off on the left hand side, and 7-11 pins look burnt (I've read this is the output FETs)
• I've had a chip pop even though I've fused both the 5v digital rail and the +/- rails (before the caps unfortunately).
• I'm using standard 3.3uF DC blocking caps.
• I've trimmed the speaker output bias to <0.01V
• I've substituted the mute pulldown resistor with a 1k, as per recommendations on the 41hz forum.
• The amplifier does occasionally mute itself if it is turned up.

Please can an experienced anyone make some suggestions as to why things keep blowing up, it's starting the get expensive as I'm about to install my 6th chip!!!? Surely these amps should either clip or mute, rather than killing themselves when things get loud, I'm assuming I'm doing something wrong but I can't see what.

(I would post on the 41hz forum, but can't register, and 41hz don't seem to provide support anymore or answer their emails. )

Thanks!

[john65b]

I have also had a few blow up just as you say - It was not a 41Hz amp, but a 6 channel prototype board I got my hands on a number of years ago - its power only required 12VDC tho...

Never did find out what the deal was...got about 15 of the 6 channel boards and most chips worked fine...

[kristleifur]

I don't know, but I'd try lowering the voltage a bit. Just to see.

As you know, 33V are definitely SUPPOSED to be OK, according to the TA2022 datasheet, but maybe the chips aren't as tolerant of voltage as they're supposed to be, and then maybe your household voltage swings a bit, and boom?

I THINK you could try to drop the voltage a bit with diodes, just as a check. One diode drops around 1-1.5V. I don't know if diodes maximize audio quality, but it might be an interesting experiment. Put a couple of diodes on the wires carrying power to the board, wait a few weeks and see if it still explodes

btw, I'm going to post a link to this thread to the 41hz.com forum. Hope you don't mind. There are some quality folks there.
http://41hz.com/Forums/topic.asp?TOPIC_ID=1567

[bongoman]

I can't see that the voltage is a problem, the TA2022 is rated to accept rail voltages up to +/-40v and we've never had problems in the house with voltage spikes in the past.

I'm starting to think that I am probably using too much gain, but shouldn't the chip's overcurrent protect either mute or clip it? As it stands I can't see that I am getting 60W RMS / channel. When compared to my Quad 405 (80W RMS / channel), it will only go about half as loud ....before exploding that is .

There are also overshoot diodes on the outputs, I might need to just triple check with the schematics that they are in the correctly.

Thanks

[bongoman]

Thanks kristleifur, I need all the help I can get!

[kristleifur]

Hey, I just came to remember a discussion of exploding chips at the 41hz forums. Here's a guy with exploding TA2022s:

http://www.41hz.com/Forums/topic.asp?TOPIC_ID=169

Check the bottom post. It was grounding issues, as I think you were starting to suspect.

[bongoman]

I came across that post, but sort of dismissed it because what he did seemed excessive. The chip datasheet and the Tripath eval board doesn't have all of the protection that he adds.

I was more looking for a simple solution, but also not rely on having to change fuses every day.

I'm thinking that grounding everything and keeping the gain down might be the best solution. Just accepting that this thing isn't that beefy.

[kristleifur]

I agree that plopla's method was excessive, but I think/hope that your simple solution can be found in it. I take plopla's post as an indication that this chip / design is sensitive to grounding, to the point of chip casualties. What I mean is that it's probably worth going over the ground connections of the amp and looking for things that are connected but shouldn't be. Basically we agree, I think Btw, grounding might be through the enclosure, jacks or pot, too.

A thought: indeed, fuses don't seem like a solution, as it seems to be voltage offsets that are blowing up your chips, rather than current flow. At least I'd think so, given that it also blows up when you're NOT playing things. Only guessing, though.

edit: And another thought: I doubt that it's the gain, not that I'm certain. Too high gain should AFAIK only raise the signal up to the voltage/current capacity of the amp, after which it will clip or shut down.

Good luck!

[bigwill]

If anything is running from a switch mode power supply, apparently the ground can be contaminated with quite high voltage spikes, so that things can blow up when you connect / disconnect them. I honestly don't know any more than this, I read it in a Rod Elliot article

[bongoman]

Nothing that is connected is running from the switch mode supply. The amp is running from the dual secondary transformer with the onboard AMP10b regulation, and the DAC is running from a seperate home made PSU with a normal tranny.

Do you think that connecting the socket earth pin to the ground on the board will have any effect, I mean is it totally necessary since at the moment it's derived from the centre tap of the transformer secondary?

Also nothing gets warm when idle, and even played loud the heatsinked TA2022 only gets luke warm.