Mark has given you fair warning.
Abandon this paralleled chipamp project and learn how to build a single chipamp correctly.
My guess was exactly the same as Mark's.
Personally I wouldn't even consider such a design without using DC servo's.
Although you can still have oscillation issues even if you are using them.
A perfect example of a works the first time can be found here,
http://www.diyaudio.com/forums/chip-amps/259591-bpa200-lm3886tf-detailed-diagram-2.html#post4019640
Pretty amazing considering all of our warning's and some doubt whether or not it was going to work!!
Layout is very critical with these things to get good performance and even more so sometimes just getting them to work at all.
Wires (long or even moderately long) from the chip to the board is not good!!
All connections to the chip must be kept as short as possible especially if you are trying to parallel them.
Everything is explained in the thread,
http://www.diyaudio.com/forums/chip-amps/252436-lm3886-pcb-vs-point-point-data.html#post3845470
Good luck and keep trying and you'll get it working.
And, Keep on DIY'ing !!!
FWIW
jer
Personally I wouldn't even consider such a design without using DC servo's.
Although you can still have oscillation issues even if you are using them.
A perfect example of a works the first time can be found here,
http://www.diyaudio.com/forums/chip-amps/259591-bpa200-lm3886tf-detailed-diagram-2.html#post4019640
Pretty amazing considering all of our warning's and some doubt whether or not it was going to work!!
Layout is very critical with these things to get good performance and even more so sometimes just getting them to work at all.
Wires (long or even moderately long) from the chip to the board is not good!!
All connections to the chip must be kept as short as possible especially if you are trying to parallel them.
Everything is explained in the thread,
http://www.diyaudio.com/forums/chip-amps/252436-lm3886-pcb-vs-point-point-data.html#post3845470
Good luck and keep trying and you'll get it working.
And, Keep on DIY'ing !!!
FWIW
jer
Last edited:
.
My idea that some of the chip's pins were shorted was obviously wrong, since the amps work individually. Put that one down to flailing about for lack of anything better to do.
http://www.ti.com/lit/gpn/lm3886
My thoughts turn vaguely to Cm, the timing capacitor for the mute resistor. You don't state that this is installed.
Also (still vaguely) to Cs. These are bypass capacitors specified for each power pin. The schematic doesn't call out values for these, but look on page 19 of the data sheet, "Supply Bypassing." The schematic shows only one bypass capacitor for each power pin, but page 19 suggests two. Are these present?
Also (still vaguely) to your power supply. Are bypass/decoupling capacitors present?
None of which might ordinarily be all that critical, you can get away with a lot. But when you hook two high-power amps together...who knows?
All of which is pretty much just more flailing. If it were me, the problem would be some wrong connection that I'd looked at 50 times but just didn't see. Not that I think other people do that, but I know I do. I also know that looking at it 50 more times wouldn't help.
So still if it were me, by this time I'd be thinking about cutting my losses. Starting fresh, the same error (if there is one) is not likely to occur a second time. Or anyway I like to believe that, on the grounds that being burned tends to make one more careful around fire and soldering irons.
.
My idea that some of the chip's pins were shorted was obviously wrong, since the amps work individually. Put that one down to flailing about for lack of anything better to do.
http://www.ti.com/lit/gpn/lm3886
My thoughts turn vaguely to Cm, the timing capacitor for the mute resistor. You don't state that this is installed.
Also (still vaguely) to Cs. These are bypass capacitors specified for each power pin. The schematic doesn't call out values for these, but look on page 19 of the data sheet, "Supply Bypassing." The schematic shows only one bypass capacitor for each power pin, but page 19 suggests two. Are these present?
Also (still vaguely) to your power supply. Are bypass/decoupling capacitors present?
None of which might ordinarily be all that critical, you can get away with a lot. But when you hook two high-power amps together...who knows?
All of which is pretty much just more flailing. If it were me, the problem would be some wrong connection that I'd looked at 50 times but just didn't see. Not that I think other people do that, but I know I do. I also know that looking at it 50 more times wouldn't help.
So still if it were me, by this time I'd be thinking about cutting my losses. Starting fresh, the same error (if there is one) is not likely to occur a second time. Or anyway I like to believe that, on the grounds that being burned tends to make one more careful around fire and soldering irons.
.
Thanks all for your words of wisdom and advice, and tremendous help!!!! I will change to the layout as suggested. One question I have though on that, do you guys prefer to etch your board or does a stripboard work as well?
I do not have experience etching using masking, and not sure if the results would turn out to be as good as seen on TV
Also, my heatsink has predrilled holes and the chips may not line up properly. Is there a workaround for that?
Thanks,
Mohit
I do not have experience etching using masking, and not sure if the results would turn out to be as good as seen on TV
Also, my heatsink has predrilled holes and the chips may not line up properly. Is there a workaround for that?
Thanks,
Mohit
Don´t want to get anybody angry but from an enginering point of view paralleling low internal impedance voltage sources is stupid.
Or at least poor Enginering.
There, I said it.
Problem is, of course, that one generator will try to source infinite current to keep its own definition of (slightly higher) voltage while the other will gladly try to sink sink all is sent its way to keep its own definition of (slightly lower) voltage.
The conceptually poor "solution" is to add series resistors to each (so internal impedance is not that low) and supermatch them.
And even so .....
BIG problem is that chip manufacturers want to sell many, so they push these marginal designs and noobs find them "easy" so they fall for them.
Don´t realize that by the moment they are building 4 chipamp (11 to 15 pins each + extra components plus *really* needed DC servos) parallel bridged amps , I have seen triple (6 chips) parallel bridged versions, they are as complex (or more) than same power discrete designs.
After all you can build a 300W RMS amp with just 5 x TO92 + 4 x TO247 transistors .
A grand total of 27 semiconductor legs plus a handful of resistors and capacitors
Or at least poor Enginering.
There, I said it.
Problem is, of course, that one generator will try to source infinite current to keep its own definition of (slightly higher) voltage while the other will gladly try to sink sink all is sent its way to keep its own definition of (slightly lower) voltage.
The conceptually poor "solution" is to add series resistors to each (so internal impedance is not that low) and supermatch them.
And even so .....
BIG problem is that chip manufacturers want to sell many, so they push these marginal designs and noobs find them "easy" so they fall for them.
Don´t realize that by the moment they are building 4 chipamp (11 to 15 pins each + extra components plus *really* needed DC servos) parallel bridged amps , I have seen triple (6 chips) parallel bridged versions, they are as complex (or more) than same power discrete designs.
After all you can build a 300W RMS amp with just 5 x TO92 + 4 x TO247 transistors .
A grand total of 27 semiconductor legs plus a handful of resistors and capacitors
.
I shall now rant.
Some say the circuits given in data sheets are lacking in one way or another, and need improvement. I suppose maybe so. Or maybe not.
These little black chips are ridiculously complicated on the inside. So much so that they can only be built by automatic machines working at a microscopic level. Obviously the cost to build a factory filled with such machines is huge, we're talking serious money.
Then how can we buy a chip for 10 bucks...or 2 bucks...or 50 cents? It's because the automatic machines turn chips out not by the thousands, but by the millions. The market is worldwide, the demand insatiable. So the vast number of chips sold pays for the factory with sheer quantity. The profit per unit is small, but there are millions of units.
But in order to serve their markets the factories have to do one critical thing, which is...wait for it...sell a chip that works. Hundreds of careers, and millions of dollars, depend on that little black chip doing what the salesman said it would do.
But nobody expects a chip to work just because you solder together some random parts. No, a chip is designed from the ground up to work in a specific circuit, and that circuit is...wait for it again...the circuit given in the data sheet. Which is why these circuits are not sketched on a napkin during somebody's coffee break. Teams of engineers use instruments costing tens of thousands of dollars to design, test, redesign, and test again. When you're selling millions of units there can't be recalls or warranty claims. Everything has to work, and it has to work every time.
So can anybody believe that the data sheet circuit won't work? Or that it's somehow mediocre? When the engineers who designed it knew full well that their mortgages, and shoes for the kids, depend on that circuit working?
Which brings us to the bottom line: if you build a circuit given in a data sheet it will work, and it will work as advertised. If it doesn't work, then you built something else.
So often we hear, "I built the data sheet circuit, except..." But there is no "except." If you used different parts, with different values, then you didn't build the data sheet circuit, you built something else. So everything done by those teams of engineers with their expensive test instruments has been flushed.
So the question becomes: can you successfully second-guess the engineers? Well, some say they can, and who am I to say that they cant? I'm thinking maybe they should send their suggestions to the factory, the engineers might be grateful.
.
I shall now rant.
Some say the circuits given in data sheets are lacking in one way or another, and need improvement. I suppose maybe so. Or maybe not.
These little black chips are ridiculously complicated on the inside. So much so that they can only be built by automatic machines working at a microscopic level. Obviously the cost to build a factory filled with such machines is huge, we're talking serious money.
Then how can we buy a chip for 10 bucks...or 2 bucks...or 50 cents? It's because the automatic machines turn chips out not by the thousands, but by the millions. The market is worldwide, the demand insatiable. So the vast number of chips sold pays for the factory with sheer quantity. The profit per unit is small, but there are millions of units.
But in order to serve their markets the factories have to do one critical thing, which is...wait for it...sell a chip that works. Hundreds of careers, and millions of dollars, depend on that little black chip doing what the salesman said it would do.
But nobody expects a chip to work just because you solder together some random parts. No, a chip is designed from the ground up to work in a specific circuit, and that circuit is...wait for it again...the circuit given in the data sheet. Which is why these circuits are not sketched on a napkin during somebody's coffee break. Teams of engineers use instruments costing tens of thousands of dollars to design, test, redesign, and test again. When you're selling millions of units there can't be recalls or warranty claims. Everything has to work, and it has to work every time.
So can anybody believe that the data sheet circuit won't work? Or that it's somehow mediocre? When the engineers who designed it knew full well that their mortgages, and shoes for the kids, depend on that circuit working?
Which brings us to the bottom line: if you build a circuit given in a data sheet it will work, and it will work as advertised. If it doesn't work, then you built something else.
So often we hear, "I built the data sheet circuit, except..." But there is no "except." If you used different parts, with different values, then you didn't build the data sheet circuit, you built something else. So everything done by those teams of engineers with their expensive test instruments has been flushed.
So the question becomes: can you successfully second-guess the engineers? Well, some say they can, and who am I to say that they cant? I'm thinking maybe they should send their suggestions to the factory, the engineers might be grateful.
.
Thanks again, all! @bentsnake, what you are saying makes total sense!
Some more questions for you experts. In my power supply, I do not have the earth connected anywhere, how do I connect it? If I connect it to the power ground, I get a very loud hum. Removing it makes the amp completely silent.
Second, when I unplug the source, let's say an input coming from a computer, it causes huge noise, is there a way to kill it? I do not have a preamp or volume control on my amp.
Third, I am getting a good audio output with both channels driven at low/med volume, but when I raise it beyond a point, considerable amount of static starts to show up. Changing to balance to either complete left or complete right makes it go away, but centering brings it back. BTW, I removed the paralleled chip and am just using one chip per channel. Layout is still the crappy one, but I intend to change it one day. Ground is star ground (load negative is power ground, is that correct?)
Last question, so many of you, in fact all of you, seem like pros...where do you get all this knowledge?
Thanks much,
Mohit
Some more questions for you experts. In my power supply, I do not have the earth connected anywhere, how do I connect it? If I connect it to the power ground, I get a very loud hum. Removing it makes the amp completely silent.
Second, when I unplug the source, let's say an input coming from a computer, it causes huge noise, is there a way to kill it? I do not have a preamp or volume control on my amp.
Third, I am getting a good audio output with both channels driven at low/med volume, but when I raise it beyond a point, considerable amount of static starts to show up. Changing to balance to either complete left or complete right makes it go away, but centering brings it back. BTW, I removed the paralleled chip and am just using one chip per channel. Layout is still the crappy one, but I intend to change it one day. Ground is star ground (load negative is power ground, is that correct?)
Last question, so many of you, in fact all of you, seem like pros...where do you get all this knowledge?
Thanks much,
Mohit
.
<< In my power supply, I do not have the earth connected anywhere, how do I connect it? >>
To cover this point only, I just happen to have some illustrative circuits.
First, just to cover all bases, is how grounds are usually drawn. Second is a "ground bus" that's never used with ICs (maybe tube circuits still use it, I don't know). Third is the correct star ground that's used with ICs.
There's also an extension of the star ground that involves two star points, or a "binary star," with the two points then joined into one. However, even though this binary arrangement has been around at least since the mid-90s, currently there are disagreements and what I consider some pretty bizarre ideas about how to implement it, so I'm ignoring the whole thing until ongoing research (mine) reveals what's what.
.
<< In my power supply, I do not have the earth connected anywhere, how do I connect it? >>
To cover this point only, I just happen to have some illustrative circuits.
First, just to cover all bases, is how grounds are usually drawn. Second is a "ground bus" that's never used with ICs (maybe tube circuits still use it, I don't know). Third is the correct star ground that's used with ICs.
There's also an extension of the star ground that involves two star points, or a "binary star," with the two points then joined into one. However, even though this binary arrangement has been around at least since the mid-90s, currently there are disagreements and what I consider some pretty bizarre ideas about how to implement it, so I'm ignoring the whole thing until ongoing research (mine) reveals what's what.
.
Attachments
I meant speaker negative.
Also, that static type noise at loud volume still exists. I rechecked the pins, individual chips work correct, but produce this when in stereo mode. How can I determine which channel is the culprit? Redid the ground also because I think that has a role to play in this.
Thx
<< good audio output with both channels driven at low/med volume, but when I raise it beyond a point, considerable amount of static starts to show up. Changing to balance to either complete left or complete right makes it go away, but centering brings it back >>
Hard to say without knowing how your balance control is tied in. Is the noise there when you rotate a control, and only when you rotate a control? Then probably it's DC (bias current, offset current, random errors...) getting to the pot.
Hard to say without knowing how your balance control is tied in. Is the noise there when you rotate a control, and only when you rotate a control? Then probably it's DC (bias current, offset current, random errors...) getting to the pot.
Last edited:
<< I do not have the earth connected anywhere, how do I connect it? If I connect it to the power ground, I get a very loud hum. >>
Classic ground loop. Try plugging everything (preamp, mixer, player, amp, whatever) into the same wall receptacle, possibly using a power strip or similar if there are too many plugs for one receptacle.
Classic ground loop. Try plugging everything (preamp, mixer, player, amp, whatever) into the same wall receptacle, possibly using a power strip or similar if there are too many plugs for one receptacle.
<< when I unplug the source, let's say an input coming from a computer, it causes huge noise, is there a way to kill it? >>
Probably none except turning things off before you plug or unplug.
You might try using RCA connectors, which don't short when they're plugged or unplugged (most others do), but the noise is probably different voltages and currents meeting in the middle, which causes a moment of chaos.
I suppose I'm forced by law to mention diodes on the input. Don't even think about it.
Probably none except turning things off before you plug or unplug.
You might try using RCA connectors, which don't short when they're plugged or unplugged (most others do), but the noise is probably different voltages and currents meeting in the middle, which causes a moment of chaos.
I suppose I'm forced by law to mention diodes on the input. Don't even think about it.
.
<< (load negative is power ground, is that correct?) >>
<< I guess it depends on what you mean by load negative >>
<< I meant speaker negative >>
Sorry to be picky again, but there is no speaker or load negative. There's only power supply positive (Vcc), power supply negative (Vee), and ground. Ground is created by the power supply, and we consider that it has no polarity, it's just ground. (which is not so, but we agree that it is, so close enough)
The ground created by the power supply is always ground, and the only ground (another fun li'l fiction), so the speaker connects to an output and THE ground.
.
<< (load negative is power ground, is that correct?) >>
<< I guess it depends on what you mean by load negative >>
<< I meant speaker negative >>
Sorry to be picky again, but there is no speaker or load negative. There's only power supply positive (Vcc), power supply negative (Vee), and ground. Ground is created by the power supply, and we consider that it has no polarity, it's just ground. (which is not so, but we agree that it is, so close enough)
The ground created by the power supply is always ground, and the only ground (another fun li'l fiction), so the speaker connects to an output and THE ground.
.
.
<< I do not have any balance on the amp, the balance is coming from my computer's sound card. When centered, there is noise, and noise disappears when either is driven in isolation. >>
Just guessing, I greatly suspect nothing about your amps, much about your input circuit(s). The two inputs should not be connected in any way, except at the single ground point. Certainly don't have them connected by being grounded to the chassis, again except at the single ground point.
However, if you can get everything plugged into one wall receptacle the problem might disappear.
.
<< I do not have any balance on the amp, the balance is coming from my computer's sound card. When centered, there is noise, and noise disappears when either is driven in isolation. >>
Just guessing, I greatly suspect nothing about your amps, much about your input circuit(s). The two inputs should not be connected in any way, except at the single ground point. Certainly don't have them connected by being grounded to the chassis, again except at the single ground point.
However, if you can get everything plugged into one wall receptacle the problem might disappear.
.
Last edited:
Hi Bentsnake,
You may be correct, but I cannot isolate the problem. Moving to the same receptacle may not solve it because I have tried plugging in another laptop running of a battery and also my phone - still the same problem.
In the meantime, I put in my original chip for the right channel, and still the same problem. My circuit is the one from the TI datasheet, it is identical IMO, and both channels are parallel circuits, but I am using only one half of each.
My power supply is not snubberized, so stays on for 3-4 secs after being switched off. The moment I turn off the switch, the noise is gone (music continues to play). Also, there is a constant noise (very faint) when the power supply is on, it is also gone the moment power is turned off (music continues)
Now what part of the power supply should I explore troubleshooting? I tried a different wall outlet, still the same problem.
Thanks much,
Mohit
You may be correct, but I cannot isolate the problem. Moving to the same receptacle may not solve it because I have tried plugging in another laptop running of a battery and also my phone - still the same problem.
In the meantime, I put in my original chip for the right channel, and still the same problem. My circuit is the one from the TI datasheet, it is identical IMO, and both channels are parallel circuits, but I am using only one half of each.
My power supply is not snubberized, so stays on for 3-4 secs after being switched off. The moment I turn off the switch, the noise is gone (music continues to play). Also, there is a constant noise (very faint) when the power supply is on, it is also gone the moment power is turned off (music continues
)Now what part of the power supply should I explore troubleshooting? I tried a different wall outlet, still the same problem.
Thanks much,
Mohit
I am so frustrated and disappointed, the problem still exists. I rewired the ground again, also thought it may have been a noisy power switch, bypassed it - still the same thing.
Redid the pins on both the chips (left and right channel). Redid the circuit of one of the channel (other is untouched and was working well earlier)
I don't know what else to do. As I said, powering the system off keeps it running for 2-3 secs (I have a total of around 60KuF capacitance for power supply) and the sound is very clean during that time.
Please let me know if anybody has an answer.
Thx
Redid the pins on both the chips (left and right channel). Redid the circuit of one of the channel (other is untouched and was working well earlier)
I don't know what else to do. As I said, powering the system off keeps it running for 2-3 secs (I have a total of around 60KuF capacitance for power supply) and the sound is very clean during that time.
Please let me know if anybody has an answer.
Thx
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.