| thoronx |
I've just assembled my LM3886t amplifier, based on the
schematic published in Elektor. When I switched the psu on
my first amp gives -vcc on the speaker out terminal and the
heatsink was hot, the second one finished his audio career
with pffffff sound, and a bit of smoke :-), -vcc was on the
output too. I checked everithing on the schematic, soldering,
components, psu and found no errors.
So I decided to disassemble the ic, I've simply cut out the ic
from the pcb, and what was found inside was a bit confusing for me
1. According to the lm3886 documentation from National pin 2,6 and 11
are NC , and I was just wodering why the guys from National put so thick copper wire on these pins ? (pin 6 is not visible on the cracked ic)
2. the die is at very "strange" angle, is it normal ?!
or is it just another fake , LM3886T, or in worst case another type
of ic in to220-11 case with (re)printed label ?? |
|
|
| Netlist |
The TA11B package is not isolated. If your heatsink touches ground you'd be running into problems.
Just a guess.
/Hugo |
|
|
| thoronx |
| Thanks, but the heatsink was far from anything |
|
|
| yangsmm |
I have another experience.
I am running audio shop on internet. When I ordered PCM1704K to TI's branch. and I get them from the branch. When I tested pcm1704, vcc and gnd were shorted. I asked the reason to branch.
Then I got the answer that there were not stock so they bought in the Cxxxxx and gave them to me.
So I returned all pcm1704.
woooooo |
|
|
| peranders |
| quote: | Originally posted by thoronx
I've just assembled my LM3886t amplifier, based on the
schematic published in Elektor. When I switched the psu on
my first amp gives -vcc on the speaker out terminal and the
heatsink was hot, the second one finished his audio career
with pffffff sound, and a bit of smoke :-), -vcc was on the
output too. I checked everithing on the schematic, soldering,
components, psu and found no errors.
So I decided to disassemble the ic, I've simply cut out the ic
from the pcb, and what was found inside was a bit confusing for me
1. According to the lm3886 documentation from National pin 2,6 and 11
are NC , and I was just wodering why the guys from National put so thick copper wire on these pins ? (pin 6 is not visible on the cracked ic)
2. the die is at very "strange" angle, is it normal ?!
or is it just another fake , LM3886T, or in worst case another type
of ic in to220-11 case with (re)printed label ?? |
Did you use a debugged pcb or was it some home-brewed solution? My guess is that you actually forgot something in your design
Did you ever check the PS before you connected the LM3886. Did you have proper voltages? Where all pins conected where they sould? Did you check your power ground and signal ground?
NC pins are thick but this comes from the manufacturing. You'll understand this if you know how they package the chip. |
|
|
| timpert |
The copper wires in such packages are usually made from a stamping out of sheet metal. In order to reduce cost, a universal stamping is used. Therefore, even if a pin is NC, the wiring may still be present in the package, because other IC types in the same package might need it.
I have no firm explanation for the strange die orientation, but it might just be to make some die connections accessible.
Did you verify your amp with a known good LM3886? Could you post a picture of your build? |
|
|
| peranders |
| In National documents that I have seen the chip is not twisted 45 deg. |
|
|
| SpittinLLama |
Peranders is right, the LM3876 and LM3886 die are not rotated in the package. Also, there needs to be some clarification on NC pins. The lead frame will have pins running up to the DAP (die attach petal) but not wires from the die to the lead frame. If you actually see wires then it is not from National. Lead frames are some what generic. The 11 lead TO-220 package used for the LM3886 is exactly the same one used for the LM4700, LM2876, LM3876, and other parts at National. But the bond out will be specific for the part so a NC for these parts will absolutely not have wires. Some other ICs will but then should clarify in the datasheet if the NC is truly open or used in testing and should be left floating, isolated, or GND on the PCB. Although not clearly stated in the LM3886 datasheet about the NC pins I know that they are floating and you can do as you please with them without affecting the die.
-SL |
|
|
| ionomolo |
| I don't see the LM3886 to be enough expensive/have enough similar counterparts to be worth faking it, may that be a manufacturing problem (The machine stucking the die did a wrong job)? |
|
|
| timpert |
Hi,
On the pictures posted, I can't see any bond wires, so there is no telling if they are actually connected or not. The lead fream doesn't look suspicious to me, but the die rotation might.
Faking is a real problem, not limited to the highest priced IC's per se. If re-labeling cheap devices as somewhat more expensive ones earns a few cents, then it's worthwhile for a faker to do it.
A little more searching yielded this thread:
Picture of a fake LM3886
Which adds more than a little bit of credibility to the "fake"hypothesis. Don't rule out circuit error though! |
|
|
| timpert |
| Do you have any intact samples of this device? If you measure a diode to substrate (V-) or to V+ (easily checked with a multimeter) on an NC pin, then the fake hypothesis gets stronger again... |
|
|
| poynton |
I can't get this to work.
Anyone else have problems ?
Andy |
|
|
| SpittinLLama |
They may also be remarked. The LM4700 is rotated like that. The LM3886 & LM3876 should have double bond wires to the ouptut, Vcc and Vee pins. If single wire it is absolutely not a LM3886 or LM3876. The LM4700 also has a Standby pin so see if there are wires there. If you can get a scope you might be able to read the name on the die. This will tell you what it is but this can also be faked.
It would be quite a big mistake to assemble parts wrong. And these parts are tested after assembly unlike some super small ICs where test is limited to wafer form only. Possible National made an error but more likely there is something 'funny' with these parts.
-SL |
|
|
| Netlist |
Andy, that server must be down.
/Hugo |
|
|
| robertj88 |
Last week I cracked a couple of LM3886T's and LM3886TF's, witch absolute can't be fake. Since then I had a new key-ring:clown:
It was exactly the same as yours, same die on an same "strange" angle. So I really think the chips are the "original" National ones.
Success with youre amp!
Robert |
|
|
| SpittinLLama |
hmmm, I will double verify my info and get back to this thread.
-SL |
|
|
| SpittinLLama |
Well maybe it changed some time but Robertj88 is right, the LM3886 is rotated just a little counter clockwise, just like your picture. The die ID should be at the bottom towards the right to look at if you have a microscope. So it is looking more like you do not have fakes, might want to spend time double checking circuit.
Sorry for the confusion earlier about rotation.
-SL |
|
|
| thoronx |
1. PSU (Suspected!)
The unregulated PSU was built based on carlosfm's psu design with small exception, I have used two rectifier bridges instead of one, the transformer is about 400VA, with center tapped secondary (3 wires), after I have assembled the psu, I mesured the voltages on the output without the amp connected,and without any load, it was nearly +-30V, far in the safe operation area of the LM3886.
|
|
|
| thoronx |
2. AMP
I choosed to build my amp based on the Elector 1998/7-8 publication, the PCB was made by me using toner-transfer method |
|
|
| thoronx |
| another view of the PSU |
|
|
| thoronx |
| the pcb of the amp... via is connected from the bottom of the pcb because it was a bit close to the heatsink so I decided to connect from down, the wire was in insulated tube (removed after for pcb debugging) |
|
|
| thoronx |
| the (via) wire is visible on this picture ... it is about 5mm from the copper wires of the pcb, far enough to make any short circuit :-) ... and it was in insulating tube. |
|
|
| timpert |
Hi,
I'd start with rewiring your power supply, because if you really built it the way you describe, it can't possibly work. Either disconnect the transformer's secondaries from each other (independent windings) and connect each winding to its own bridge, or connect them together to form a center tapped winding (like you have now) and use only one bridge.
As it is now, with the upper leg of the transformer being positive and the lower leg being negative, all seems fine, but you're charging the buffer caps in series. Any discrepancy between the current draw from the positive supply and the negative supply will make the supply voltage asymmetrical. It is not unimaginable that a situation arises in which your LM3886 gets such weird voltages on its supply that it decides to vent its smoke of life.
It gets worse when the upper leg is negative and the lower leg is positive. In this case, the diodes parallel to C1 and to C8 are in series and with forward bias, and almost present a dead short to the transformer's secondary! You perhaps fried your rectifier as well! Replace it, along with the 3886's. And don't apply power until the supply is corrected.
Good luck! |
|
|
| thoronx |
| thx timpert, I will try once again this weird psu, trying with load eg. adding 2 x 10 ohm/10W on the output that will drive some current (3A) to mesure voltage drops, and current on the resistor. |
|
|
| timpert |
Before connecting anything critical, try your rebuilt supply with no load for a while (half an hour or so) while monitoring it closely, looking for signs of distress. Nothing should get warm or hum excessively and the voltages should remain correct. Then, load only one output (V+ or V-) with the rated current (3A should be fine, mind the power dissipation in the resistors though, it will be in the order of 80-100 Watts), leave the other voltage unloaded and check voltages again. The load shouldn't affect the voltage on the unloaded branch by more than one or two volts. Things may get mildly warm now (except for the load resistors, which will get really hot), but still not too hot to touch. Repeat this test for the other voltage. Then load both voltages (take care that you stay within your transformer rating) and check if nothing overheats/blows up and voltages stay where they should. Only when your PSU survives this test, connect the amp.
Good luck! |
|
|
| AndrewT |
Hi,
I agree that a dual rectifier cannot be used with a centre tapped transformer.
Either use a dual secondary transformer with dual rectifiers
or
use a centre tapped transformer with a single rectifier.
And build up a light bulb tester. And use it every time you power up a modified or new project.
|
|
|
| timpert |
| If at all possible, I complete and test the entire power supply of a project before starting work on the rest of the electronics. This includes all regulators and stuff on a PCB. Resistors first and ICs last is a good way to build a kit with no clue of what's going on. Having the power supply ready enables me to test my makings step-by-step. I start with the low power stuff and then I work my way up to higher power. As soon as things connect to the raw power supply, I use a variac and current meter to increase the voltage from zero and see if everything follows nicely and nothing strange happens. This allows me more control than a lightbulb, but if you don't have a variac (or are too impatient to frequently check your work), then the lightbulb is indeed mandatory. Working step-by-step seems slow at first, but it saves so much grief that I think in the end I'm quicker than when I just build the whole thing and only debug when everything is in place. |
|
|
|
