Ah, another person with 53.5v supply, yes I am sure that isn't overvoltage from reading the datasheet and checking the stated voltages of the caps.
Yes, wondered why I never get a big switch on pop, just checked wit a test speaker and only the slightest movement at turn on, barely noticeable. Also noticed offset voltage I measured at the output terminals of the channel with the speaker attached had dropped to 0.1v so I removed the speaker and it went back to over 3v. I am measuring with multimeter probes over positive and negative output terminal set to DC volts.
Then measured terminal to ground, on one channel its 1/2 the VCC on both terminals (speaker connected and not connected), on the other its 1/2 on both with speaker connected but not connected one terminal is 1/2 VCC, the other close to 600v!!! What the?
Not sure what is going on now!
Will try dropping voltage later but I think there are other issues now.
Yes, wondered why I never get a big switch on pop, just checked wit a test speaker and only the slightest movement at turn on, barely noticeable. Also noticed offset voltage I measured at the output terminals of the channel with the speaker attached had dropped to 0.1v so I removed the speaker and it went back to over 3v. I am measuring with multimeter probes over positive and negative output terminal set to DC volts.
Then measured terminal to ground, on one channel its 1/2 the VCC on both terminals (speaker connected and not connected), on the other its 1/2 on both with speaker connected but not connected one terminal is 1/2 VCC, the other close to 600v!!! What the?
Not sure what is going on now!
Will try dropping voltage later but I think there are other issues now.
Last edited:
Hi ChrisMmm, sometimes, one should not trust that much a DVM...
Better trust a scope if you can grab one around...
Unfortunately no scope!
OK so dropped in some diodes bringing the voltage down to 50.3v. So channel 1 stays around 1v across terminals and each terminal to earth 1/2VCC with no speaker. Channel 2 0.1v across terminals and 1/2VCC terminals to earth with speaker BUT with no speaker around 500v across terminals, 1 terminal to earth is 1/2VCC other around 525v.
I don't seem to get same results each time which is driving me crazy.
But lower voltage simply lowers the readings a little, clearly THE problem remains.
Ready to bin the amp board (no great loss, I have no current use for the amp) but I would dearly like some sort of explanation as to what may be occurring or what the fault is
Oh and left the amp on for a bit to monitor temp, after an hour 40C (20C above ambient).
OK so dropped in some diodes bringing the voltage down to 50.3v. So channel 1 stays around 1v across terminals and each terminal to earth 1/2VCC with no speaker. Channel 2 0.1v across terminals and 1/2VCC terminals to earth with speaker BUT with no speaker around 500v across terminals, 1 terminal to earth is 1/2VCC other around 525v.
I don't seem to get same results each time which is driving me crazy.
But lower voltage simply lowers the readings a little, clearly THE problem remains.
Ready to bin the amp board (no great loss, I have no current use for the amp) but I would dearly like some sort of explanation as to what may be occurring or what the fault is
Oh and left the amp on for a bit to monitor temp, after an hour 40C (20C above ambient).
Last edited:
Hi ChrisMmm,
Don't bin your board yet. Your amp isn't likely to actually have 600V or so at its output otherwise the output LC filter caps would already have popped open with some other side effects...
It may only be caused by either some undamped resonance glitches on that channel (better check if the output RC network between faulty node and GND has not failed open), or your measurements with the DVM are not all made wrt the exact same GND potential (I read you are measuring wrt to Earth: forget it and stick your (-) DVM wire to the Minus of the incoming supply voltage and just don't change place while measuring all output nodes...
Anyway, these switching things are not meant -and do not like sometimes- remain unloaded, was it with as much as several tens of kOhms; so no real worry about this open-output measurements.
I'd personally be more concern about any DC voltage at the output (between speaker nodes) while some 4...100 ohms load is connected.
Don't bin your board yet. Your amp isn't likely to actually have 600V or so at its output otherwise the output LC filter caps would already have popped open with some other side effects...
It may only be caused by either some undamped resonance glitches on that channel (better check if the output RC network between faulty node and GND has not failed open), or your measurements with the DVM are not all made wrt the exact same GND potential (I read you are measuring wrt to Earth: forget it and stick your (-) DVM wire to the Minus of the incoming supply voltage and just don't change place while measuring all output nodes...
Anyway, these switching things are not meant -and do not like sometimes- remain unloaded, was it with as much as several tens of kOhms; so no real worry about this open-output measurements.
I'd personally be more concern about any DC voltage at the output (between speaker nodes) while some 4...100 ohms load is connected.
Your amplifier is fine. Your DMM can not read DC volts with 600Khz 50volts PP on the terminals with no load. The output filters assume there is a load.
Many years ago I was working on an antenna amplifier on top of a 20 meter tower in my back yard. I was getting a poor signal on the tv. At the power supply beside the tv I was getting the 18Volts for the antenna preamp. At the top of the tower with a VOM I was getting 150. WTF. A break in the cable was the issue. RF does crazy things to meters.
I have another TPA3255 for another project. It will use 38Vdc SMPS. I will measure it with and without load tomorrow.
Many years ago I was working on an antenna amplifier on top of a 20 meter tower in my back yard. I was getting a poor signal on the tv. At the power supply beside the tv I was getting the 18Volts for the antenna preamp. At the top of the tower with a VOM I was getting 150. WTF. A break in the cable was the issue. RF does crazy things to meters.
I have another TPA3255 for another project. It will use 38Vdc SMPS. I will measure it with and without load tomorrow.
Hey, thanks guys for the great feedback, a few red herrings here obviously.
So what is consistent is the voltage across terminals with a load (6ohm speaker), this is 0.12v and 0.15v. Is that cause for concern? The datasheet states max 60mv so I am more than twice that. If its an issue any way to resolve?
Also, going back to why I started digging under the bonnet, would this offset be enough to create the heat in the crossover that I observed?
So what is consistent is the voltage across terminals with a load (6ohm speaker), this is 0.12v and 0.15v. Is that cause for concern? The datasheet states max 60mv so I am more than twice that. If its an issue any way to resolve?
Also, going back to why I started digging under the bonnet, would this offset be enough to create the heat in the crossover that I observed?
Hi ChrisMmm,
Even if one would like it to be lower, this 0.15V offset voltage is too low to induce any warmth into such a low load that your speakers or xo are.
This may sometimes be caused by unwanted HF auto-oscillation being present at the amp output even with no input signal (or shorted input)
Too bad you have no scope available.
Even if one would like it to be lower, this 0.15V offset voltage is too low to induce any warmth into such a low load that your speakers or xo are.
This may sometimes be caused by unwanted HF auto-oscillation being present at the amp output even with no input signal (or shorted input)
Too bad you have no scope available.
Thanks again Kalamin, guess the amp is OK then, may yet see more service
So if this oscillation is present is it likely to be generated by the source even when not outputting a signal or internal to the TPA3255 board?
Again the heating does not happen with the amp that replaced it which is also class D. I note the design doc for my new amp states the input stage will "remove some of the very high, above the audio frequency range components, and allow only signals below 24KHz to pass". Would this cover HF oscillation?
BTW just checked the offset on one of my heavily modded TPA3116 amps, 5ma with or without load.
So if this oscillation is present is it likely to be generated by the source even when not outputting a signal or internal to the TPA3255 board?
Again the heating does not happen with the amp that replaced it which is also class D. I note the design doc for my new amp states the input stage will "remove some of the very high, above the audio frequency range components, and allow only signals below 24KHz to pass". Would this cover HF oscillation?
BTW just checked the offset on one of my heavily modded TPA3116 amps, 5ma with or without load.
Last edited:
I had a look at your speaker x-over.
As I told you in my first post, the tweeter resistor can not heat up from DC, only AC.
You obviously do not understand x-over basics, as there is a capacitor in line with this resistor, blocking any DC.
Let me explain something to you, even as you do not like to listen to wise recommendations, like maximum voltages for amplifiers:
An amp is not limited by it's parts voltage range, but by the combination of all parts together. Most parts have a linear working range where you can use them as calculated. If you leave this range, they may not instantly break, but behave strange and erratic. Like amps starting to oscillate. Yours oscillates, as I told you in my first post.
The Chinese developer of these amp was not able to get it running safely on higher voltages, that is the cause for the recommended voltage range.
In a perfect implementation the TPA3255 could run near the limits from the data sheet, but this is not perfect, it is made in China.
In China intellectual property is worth nothing, so a developer does not refine his products over years as we are used to it. Once it is in production, anyone copies it, stealing the profit from the developers work and he gets a kick in the ***. A clever developer may even implement faults, so he can sell his design once more to someone else, working more reliable or with higher power output.
3e-Audio is a very strange company. You can not buy most of their products, which makes no sense. No sell-no profit. Also, there is no real development any more. Probably something went wrong on their side. If you use their stuff as intended, expect it to work quite well, but do not leave the region it is made for, they have no build in tolerance for safety, like you are used from Western products.
As I told you in my first post, the tweeter resistor can not heat up from DC, only AC.
You obviously do not understand x-over basics, as there is a capacitor in line with this resistor, blocking any DC.
Let me explain something to you, even as you do not like to listen to wise recommendations, like maximum voltages for amplifiers:
An amp is not limited by it's parts voltage range, but by the combination of all parts together. Most parts have a linear working range where you can use them as calculated. If you leave this range, they may not instantly break, but behave strange and erratic. Like amps starting to oscillate. Yours oscillates, as I told you in my first post.
The Chinese developer of these amp was not able to get it running safely on higher voltages, that is the cause for the recommended voltage range.
In a perfect implementation the TPA3255 could run near the limits from the data sheet, but this is not perfect, it is made in China.
In China intellectual property is worth nothing, so a developer does not refine his products over years as we are used to it. Once it is in production, anyone copies it, stealing the profit from the developers work and he gets a kick in the ***. A clever developer may even implement faults, so he can sell his design once more to someone else, working more reliable or with higher power output.
3e-Audio is a very strange company. You can not buy most of their products, which makes no sense. No sell-no profit. Also, there is no real development any more. Probably something went wrong on their side. If you use their stuff as intended, expect it to work quite well, but do not leave the region it is made for, they have no build in tolerance for safety, like you are used from Western products.
Last edited:
Turbo, I appreciate what you are saying and will take that on board. I guess hearing somebody else was running 53.5v I was eager to believe that was not the issue.
I do understand that capacitors block DC, what I am not really up to speed with is AC (presumably above the audio range?) in the circuit and I gather what you are saying is that the AC presumably from oscillations are heating up the resistor.
I also take on board your comments about the of all the parts rather than my black and white view of individual component ratings, this makes sense to me, and more a thing with digital circuits (which I lack experience with) I would think. So I won't just pop the top back on and forget it, I will try a reduced voltage supply. What do you consider the maximum safe voltage to run? Also any way of determining if its oscillating or not without a scope?
I do understand that capacitors block DC, what I am not really up to speed with is AC (presumably above the audio range?) in the circuit and I gather what you are saying is that the AC presumably from oscillations are heating up the resistor.
I also take on board your comments about the of all the parts rather than my black and white view of individual component ratings, this makes sense to me, and more a thing with digital circuits (which I lack experience with) I would think. So I won't just pop the top back on and forget it, I will try a reduced voltage supply. What do you consider the maximum safe voltage to run? Also any way of determining if its oscillating or not without a scope?
Last edited:
The oscillation heats up the resistor in front of the capacitor, leading to the tweeter. you are lucky your tweeter did not burn. Imagine the heat on the large resistor on the tiny voice coil. In your case one of the impedance correcting parts after the capacitor (C4, R5) took most of the power. A voice coil is an inductor too, blocking AC. So most of the power heated up the resistors (R0 and R5). A simpler x-over would not have saved the tweeter. These parts are not for security, but linear sound reproduction. In your case they accidently served as tweeter protection.
With your problem, things are so clear, someone can even see them looking on the floor, through the middle of the earth into your amp. Like from Great Germany to New Zealand.
Your amp is oscillating. Probably the internally generated 12V service voltages go over the top.
An oscilloscope is a great tool for some people, but not necessary in all cases. If your amp is hit by a sledge hammer, you do not need an oscilloscope to prove it is broken. Compare over voltage to hammering on your amp...
You measured the offset voltage with an open output. This is very wrong in this case. I
could not expect you doing that. So in fact the offset is quite normal and does not affect the speaker in any aspect.
In your case I would try to get a good 36V SMPS and call it a day. If you like to listen crazy loud all the time, maybe even try an 48V SMPS. I would not give the amp much credit for reliability with that voltage and not try it. But that is me.
Even as this is still a field for discussion, experts lean to the opinion that SMPS are the superior breed for audio today.
Most simple customers think high weight is similar to value. In a few years there will be no more heavy transformers and amps will sound just as good or even better than today.
At least both power supplies are on par. I lean to a good SMPS for a reason.
An SMPS keeps voltage up until the current limiter is reached, a transformer theoretically drops in voltage the more you get near it's maximum power rating, with increasing ripple.
At idle two 350W 48 V supplies are the same, under maximum load the transformer may only deliver 40 Volt. Theoretically only 34V, but this is not realistic with audio. So the SMPS made for audio has more reserve for impulse power, even driven very hard.
In fact you win nothing in sound quality from 30% more amp output. Maximum power is limited by your speakers anyway.
If your speaker / amp combination does not sound as expected, first get rid of any MP3, streaming, wireless. Take a CD with a player from a well known brand. Kid's don't understand that almost any of these sources have a reduced quality no audio system can recover. There is no free CD quality audio.
I had a guy complaining about his new high end car sound system. To prove it he played dozends different tracks of music from a CD he burned by himself. 40 Hours of music on it. He insisted it was just as good quality as a regular CD, only smarter recorded. At his home audio system he could not hear any difference CD-MP3.
You can not even use reduced data recordings for a first listening impression.
Sound quality is all about fine details, while data reduction is all about getting rid of fine details not primarily necessary for listening.
Finally, if your amp/ speaker combo does not sound good, do not expect it to get better on higher amp supply voltage. Never. Maybe the other way round...
With your problem, things are so clear, someone can even see them looking on the floor, through the middle of the earth into your amp. Like from Great Germany to New Zealand.
Your amp is oscillating. Probably the internally generated 12V service voltages go over the top.
An oscilloscope is a great tool for some people, but not necessary in all cases. If your amp is hit by a sledge hammer, you do not need an oscilloscope to prove it is broken. Compare over voltage to hammering on your amp...
You measured the offset voltage with an open output. This is very wrong in this case. I
could not expect you doing that. So in fact the offset is quite normal and does not affect the speaker in any aspect.
In your case I would try to get a good 36V SMPS and call it a day. If you like to listen crazy loud all the time, maybe even try an 48V SMPS. I would not give the amp much credit for reliability with that voltage and not try it. But that is me.
Even as this is still a field for discussion, experts lean to the opinion that SMPS are the superior breed for audio today.
Most simple customers think high weight is similar to value. In a few years there will be no more heavy transformers and amps will sound just as good or even better than today.
At least both power supplies are on par. I lean to a good SMPS for a reason.
An SMPS keeps voltage up until the current limiter is reached, a transformer theoretically drops in voltage the more you get near it's maximum power rating, with increasing ripple.
At idle two 350W 48 V supplies are the same, under maximum load the transformer may only deliver 40 Volt. Theoretically only 34V, but this is not realistic with audio. So the SMPS made for audio has more reserve for impulse power, even driven very hard.
In fact you win nothing in sound quality from 30% more amp output. Maximum power is limited by your speakers anyway.
If your speaker / amp combination does not sound as expected, first get rid of any MP3, streaming, wireless. Take a CD with a player from a well known brand. Kid's don't understand that almost any of these sources have a reduced quality no audio system can recover. There is no free CD quality audio.
I had a guy complaining about his new high end car sound system. To prove it he played dozends different tracks of music from a CD he burned by himself. 40 Hours of music on it. He insisted it was just as good quality as a regular CD, only smarter recorded. At his home audio system he could not hear any difference CD-MP3.
You can not even use reduced data recordings for a first listening impression.
Sound quality is all about fine details, while data reduction is all about getting rid of fine details not primarily necessary for listening.
Finally, if your amp/ speaker combo does not sound good, do not expect it to get better on higher amp supply voltage. Never. Maybe the other way round...
Thanks for your reply Turbo. Guess I was lucky with the tweeters, good thing as I really like the speakers (not my main system but I end up listening to them a lot of the time).
I totally agree with a lot of what you say. SMPS for example, I agree a well made for audio SMPS is better than a trafo. I have 3 Connex Electronics amps with their SMPS in each case, all run flawlessly. I think they make some of the best SMPS out there and match their amps well, they are really good at stating intended voltage ranges for the amps so you have no doubts as to what supply to use. It was the Connex CxD2160 I just built that replaced the TPA3255, very pleased with it.
Have to say while I was building TPA3116 amps a few years back I supported trafo supplies but I am an SMPS convert now.
My listening is streaming either from local FLAC files or through Tidal Hifi which is FLAC/CD quality also. I do have some hires files but in all honestly cannot tell the difference, maybe my ears have seen (heard) better days!
Regards the TPA3255, the PS was put together using parts I already had to save money, I never intended to have such high voltage, in fact I was quite nervous about using it - rightly as it transpires.
I have just done some experiments and not too sure what conclusion to draw. First I loaded the TPA3255 output with a 10watt 15ohm resistor assuming any oscillation present would result in it heating up. It heated fairly quickly and I turned it off at 70C. Next as a control I did the same thing with one of my TPA3116 amps and that did not heat the resistor at all. Then I installed an lower voltage trafo to the TPA3255 getting a 33.5v supply. I repeated the test with the resistor and it heated up as well to 70c but more slowly.
Is it fair to conclude that the oscillations are there at 33.5v also?
I totally agree with a lot of what you say. SMPS for example, I agree a well made for audio SMPS is better than a trafo. I have 3 Connex Electronics amps with their SMPS in each case, all run flawlessly. I think they make some of the best SMPS out there and match their amps well, they are really good at stating intended voltage ranges for the amps so you have no doubts as to what supply to use. It was the Connex CxD2160 I just built that replaced the TPA3255, very pleased with it.
Have to say while I was building TPA3116 amps a few years back I supported trafo supplies but I am an SMPS convert now.
My listening is streaming either from local FLAC files or through Tidal Hifi which is FLAC/CD quality also. I do have some hires files but in all honestly cannot tell the difference, maybe my ears have seen (heard) better days!
Regards the TPA3255, the PS was put together using parts I already had to save money, I never intended to have such high voltage, in fact I was quite nervous about using it - rightly as it transpires.
I have just done some experiments and not too sure what conclusion to draw. First I loaded the TPA3255 output with a 10watt 15ohm resistor assuming any oscillation present would result in it heating up. It heated fairly quickly and I turned it off at 70C. Next as a control I did the same thing with one of my TPA3116 amps and that did not heat the resistor at all. Then I installed an lower voltage trafo to the TPA3255 getting a 33.5v supply. I repeated the test with the resistor and it heated up as well to 70c but more slowly.
Is it fair to conclude that the oscillations are there at 33.5v also?
Last edited:
The single advantage to using a linear power supply would be in most cases it's easier to replace the electrolytic capacitors when they dry out, also in this case Chris already had the parts needed for it vs. having to buy something new.
Chris, an effective voltage reduction and at the same time save a bit of money you can use a few extra diodes in series with the load. Make sure that they can take the intended current. Much cheaper than voltage regulation for smaller adjustments, each diode in series reduce the voltage by a fixed value so it's easy to get right.
Edit:
Seems you have other issues as well though.
Last edited:
Your SR71 is a very nice speaker. Seas components are found in the worlds most expensive loudspeakers. So it is no miracle you use them that often.
CD quality is a very clear border, higher resolution can not really prove to sound better, but lower resolution does sound worse. At least if you have well working ears.
If you have a large subwoofer and actively cross the SR71 over around 100 Hz, you will question the need for any larger system, as they can play ridiculously loud.
I fear that you partly blew the TPA3255.
If you look very close, maybe enlarge good pictures of it, you may find burned parts. If it is an IC, there is a good chance they scraped off the marking.
I don't think you have the means and experience to repair SMD, maybe just sell them at eBay with a honest description. Often you get good money for bad parts. Once you started to "repair them" they are worthless. If you do not know what you do, don't do it. Otherwise this is called "messing around..."
Our Chinese friends are so deep into making copies of everything, they do not supply schematics for the copied stuff they sell. That makes any repair a reengineering project.
I have developed my own philosophy: If it is not worth a clean schematic, it is not worth anything. Not my work, not my money.
Selling DIYS amp modules without schematic is no small mistake, it is ridiculous.
Also it collides with the "right to repair" you have.
Using diodes to reduce a voltage is not state of the art, just a last resort, a quick and dirty fix.
CD quality is a very clear border, higher resolution can not really prove to sound better, but lower resolution does sound worse. At least if you have well working ears.
If you have a large subwoofer and actively cross the SR71 over around 100 Hz, you will question the need for any larger system, as they can play ridiculously loud.
I fear that you partly blew the TPA3255.
If you look very close, maybe enlarge good pictures of it, you may find burned parts. If it is an IC, there is a good chance they scraped off the marking.
I don't think you have the means and experience to repair SMD, maybe just sell them at eBay with a honest description. Often you get good money for bad parts. Once you started to "repair them" they are worthless. If you do not know what you do, don't do it. Otherwise this is called "messing around..."
Our Chinese friends are so deep into making copies of everything, they do not supply schematics for the copied stuff they sell. That makes any repair a reengineering project.
I have developed my own philosophy: If it is not worth a clean schematic, it is not worth anything. Not my work, not my money.
Selling DIYS amp modules without schematic is no small mistake, it is ridiculous.
Also it collides with the "right to repair" you have.
Using diodes to reduce a voltage is not state of the art, just a last resort, a quick and dirty fix.
That was my thoughts when I got the results from a 33.5v supply. After examining with a loupe no sign of burnout.
Well I will be philosophical, I feel I have learnt a lot at the cost of a dead amp board thanks to some great advice from this thread. May scavenge some parts, those LC inductors look worth saving and the heatsink as well. And I already have a use for the case it frees up.
Cheers.
ChrisMmm I finally got around to measuring my Tpa3255 that's on the bench waiting to be finished. It's the cheapest one I found on eBay, i think I paid $38 for it. With a 38volt SMPS it measures .0054 volts DC across the speaker outputs. With and without a load. Both channels are within a 100 micro volts of each other. There is about 1.1 volts A.C. at around 460Khz at the outputs. I say around because my scope is analogue.
The living room Tpa3255 With The 53.5 volt linear ps after being on for 24 hours continuous measures .0024VDC and .0064VDc with load. Measurements taken with a Fluke 87.
I use the linear ps as it was what I had. The transformer was taken from an old dead 5.1 channel receiver. Case is barely warm, ambient is 21.5C.
The living room Tpa3255 With The 53.5 volt linear ps after being on for 24 hours continuous measures .0024VDC and .0064VDc with load. Measurements taken with a Fluke 87.
I use the linear ps as it was what I had. The transformer was taken from an old dead 5.1 channel receiver. Case is barely warm, ambient is 21.5C.