Bucks Bunny probably means that you can not rely 100% on protection by internal semis.
Same thinking. Only external protection with simple logic can be truely relied on but it adds to the cost.
However TPA3116/8 turned out to be pretty reliable.
I have seen and heard TA2020 amps that exploded when a speaker wire came loose. These also had protection inside. Not one that could be trusted.
Same thinking. Only external protection with simple logic can be truely relied on but it adds to the cost.
However TPA3116/8 turned out to be pretty reliable.
I have seen and heard TA2020 amps that exploded when a speaker wire came loose. These also had protection inside. Not one that could be trusted.
Last edited:
Good to know. I have a Lepai (Parts Express) that has a TA2020 in it. It also has a relay for the output to the speakers. I'm not sure if this is just to mitigate turn on thumps or if it also protects against DC on the output. My speakers are hoping for both! But without analyzing that circuit (not going to bother for times sake), that's just wishful thinking.
I still wish Bucks Bunny would have confirmed if his board was real or China TPA3118. I wouldn't trust the Chinese after receiving fakes. It was the 24 pin fake that burned my speaker with a short.
TA2020 is very sensitive. If your loudspeakers have value maybe you should bother it the amplifier has DC protection. Bucks Bunny was very clear IMHO.
You should trust the Chinese that produce the original chips 🙂 The nationality does not seem to have to do with it.
You should trust the Chinese that produce the original chips 🙂 The nationality does not seem to have to do with it.
Last edited:
You're right. I see now in post #24 (genuine).
I was a little confused that he said he'd not seen a failure and then that he hadn't seen a failure "until now". I'm assuming that his choice of wording means he's still never seen a failure.
And to the confusion between he and I where he tuned me out... I had thought perhaps that there was some type of circuitry within the chip AFTER the MOSFETs which could open circuit, or perhaps more likely crowbar the output transistors. I guess that's a no. But certainly it is possible to implement, especially a crowbar where the output is shorted + to - when DC is detected.
I was a little confused that he said he'd not seen a failure and then that he hadn't seen a failure "until now". I'm assuming that his choice of wording means he's still never seen a failure.
And to the confusion between he and I where he tuned me out... I had thought perhaps that there was some type of circuitry within the chip AFTER the MOSFETs which could open circuit, or perhaps more likely crowbar the output transistors. I guess that's a no. But certainly it is possible to implement, especially a crowbar where the output is shorted + to - when DC is detected.
In my professional environment we had monitoring electronics that were internally equipped with own monitoring and active protection circuitry. I frowned when I had the explanation of this then novelty.
When lightning struck kilometers away the next day a level indicator had a defective display. The internal monitoring said nothing. In the next 6 months all modules died and needed to be replaced and none of them gave any useful alarms. As they died the internal monitoring and protection circuitry also died. Since then we use redundant systems with redundant power supplies and external monitoring and protection.
What you say may be technically possible, it is not there in reality. Stuff needs to be cheap.
When lightning struck kilometers away the next day a level indicator had a defective display. The internal monitoring said nothing. In the next 6 months all modules died and needed to be replaced and none of them gave any useful alarms. As they died the internal monitoring and protection circuitry also died. Since then we use redundant systems with redundant power supplies and external monitoring and protection.
What you say may be technically possible, it is not there in reality. Stuff needs to be cheap.
True... perhaps. I'd think triacs built in wouldn't cost much when the die is made for the chip. Maybe I'm wrong. Of course that's a 1960s solution to DC on a speaker output. I'd think by now, maybe there'd be something better that TI's engineers have come up with to solve this problem within the semiconductor realm. A crowbar would likely ruin the chip but I'd sure rather toss a 3118 vs one or more expensive speakers.
Not perhaps. There are no triacs inside. The chip is sturdy but the story is not to trust internal protection for 100%. I think the MOSFETS in TPA3118 are pretty well made.
This.
Again and I guess we're at an impasse (you and I also) and no offense intended to either you or bucks.
But, perhaps you are both missing the idea behind what I had originally said. "After" the MOSFET (inside the chip). As in either:
1. Open circuit to + or - (between output device and speaker)
or
2. Short circuit to + and - (feeding speaker)
If either of those were implemented within the chip, it is possible to stop shorts at the MOSFETs from reaching the speakers. Of course only TI engineers would know this for certain. Can we say without a shadow of doubt that it's not implemented? Maybe we can. Perhaps I am optimistic that when I see DC output protection on the data sheet that neither #1 nor #2 above are implied.
EDIT: and one more obvious means that I missed: #3 open Vcc or Gnd when there's a fault at an output device.
But, perhaps you are both missing the idea behind what I had originally said. "After" the MOSFET (inside the chip). As in either:
1. Open circuit to + or - (between output device and speaker)
or
2. Short circuit to + and - (feeding speaker)
If either of those were implemented within the chip, it is possible to stop shorts at the MOSFETs from reaching the speakers. Of course only TI engineers would know this for certain. Can we say without a shadow of doubt that it's not implemented? Maybe we can. Perhaps I am optimistic that when I see DC output protection on the data sheet that neither #1 nor #2 above are implied.
EDIT: and one more obvious means that I missed: #3 open Vcc or Gnd when there's a fault at an output device.
Last edited:
There is one point missed in this discussion: The mentioned amps are single voltage and bridged. It depends of the state of the second amp how a defective output MOSfet will behave. If both blow and give full rail voltage to the output, the resulting voltage at the speaker is zero. One MOSfet would have to be grounded and the other one at full voltage to harm a speaker.
If such an amp is used single ended, the output capacitor will prevent a DC fault at the speaker.
I don't have practical experience with defective TPA3116 or TPA 3255 chips, so this is only a theoretical thought.
If such an amp is used single ended, the output capacitor will prevent a DC fault at the speaker.
I don't have practical experience with defective TPA3116 or TPA 3255 chips, so this is only a theoretical thought.
I don't think the block diagram shows HOW the DC detect protection is implemented.
Ok looking at the data sheet:
Which sounds like only the gates are controlled. Nothing after the MOSFETs and nothing cutting off the supply to the MOSFETs. Sad but true.
It's a shame that one more MOSFET couldn't have been implemented to switch off the 12-24V supplying the output transistors when a major fault were detected. MOSFET power is pretty cheap these days. I'd doubt that one more transistor inside would have added even a few cents to the retail price of the chip.
Which sounds like only the gates are controlled. Nothing after the MOSFETs and nothing cutting off the supply to the MOSFETs. Sad but true.
It's a shame that one more MOSFET couldn't have been implemented to switch off the 12-24V supplying the output transistors when a major fault were detected. MOSFET power is pretty cheap these days. I'd doubt that one more transistor inside would have added even a few cents to the retail price of the chip.
Apologies to bucks.
I think part of what was happening was a misunderstanding of your post #27:
"A catastrophic failure I did not observe until now."
In my culture, that means that you have now observed a catastrophic failure. Putting it into context, I think I misunderstood you. Pretty sure you meant that you hadn't seen a catastrophic failure to date. Sorry for my confusion and also for not reading back on your posts. I was in a bit of a hurry with several projects going at the same time (fixing diesel lawn mower before the grass takes over, working on siding, along with a few electronics projects). Lazy reading on my part. My apologies to you.
I think part of what was happening was a misunderstanding of your post #27:
"A catastrophic failure I did not observe until now."
In my culture, that means that you have now observed a catastrophic failure. Putting it into context, I think I misunderstood you. Pretty sure you meant that you hadn't seen a catastrophic failure to date. Sorry for my confusion and also for not reading back on your posts. I was in a bit of a hurry with several projects going at the same time (fixing diesel lawn mower before the grass takes over, working on siding, along with a few electronics projects). Lazy reading on my part. My apologies to you.
And yes, it's not an entire nationality that's the issue. But as a general rule, on ebay, Amazon, and never again AliExpress, I see China as the origin and I am hesitant.
I've ordered a dozen or so of the TPA3118 boards on ebay over the past 5 years and only the first batch or two were real. 75% or more have been sold as 32 pin and I received the 24 pin. Also obvious by the dimensions. Of course TI will oversee that this doesn't happen from their own factories. But China sellers in general have no warm place in my heart. Again, not the Chinese... no problem there, but the honesty of so many sellers and authenticity of their products has been very poor in my experience. I've never seen anything so bad. Beside the TPA boards I've seen:
-chips with sanded off numbers and fake numbers silk screened on
-fuses that are WAY off their rated values
-bait and switch on microphone diaphragms
and so much more. So again, nothing against the Chinese as a whole, but the reputation of their sellers and whoever they're getting their products from is not good.
I could just not order anything from China, but everything is made there. I long for the days where TeleFunKen was Germany, Mullard was Brittain, RCA, GE etc was USA. CTS pots were made in USA, Sprague capacitors were USA, and so much more. That era will never be again.
I got it now that this may be misleading. Sorry I am not a native speaker.
Right. When I first found the chip, it was through Parts Express (Lepai 2020A+). I then discovered the $5 ebay boards which at first were excellent. The mono boards with through hole caps and 330 marked inductors (and real chips). As time went on, all were fake after a certain point and it became very frustrating. I wasn't equipped for surface mount soldering or I would have built a board. If there were a through hole everything else, I'd have no problem soldering ONLY the chip with an iron if nothing else is SMD. I will dig a little for PCBs with through hole. Maybe the link above. I will check it out. Thank you.
Bucks, your English is great. Of course little nuances like this can throw an American off who isn't familiar with German phrasing. FWIW, German makes more sense than English. At least in German, only one letter is used for the K sound.
I have German in my lineage but of course like most Americans only speak 1.001 language. The extra 0.001 since it doesn't really count. We're all required to learn French or Spanish, but almost nobody here really can speak either after one high school semester of the language.
I have German in my lineage but of course like most Americans only speak 1.001 language. The extra 0.001 since it doesn't really count. We're all required to learn French or Spanish, but almost nobody here really can speak either after one high school semester of the language.