Oh, $12-ish if I throw them in with my next Mouser order. The limiting factor is still time.
Tom
So my main amp is one of these XY boards - with the normal powersupply board. Transformer is Triad Magnetics VPT48-2080.
I really like the amp - to me it sounds great, it's by no means crap. The sound is day and night compared to the stuff from the 70s and 80s (well that I can afford at least) Considering the price to me it's by no means a poor choice - just not the best.
I'm pretty pretty sure the LM3886 is the one with white markings - so counterfeit.
I would be interested in seeing the difference between two XY boards with one authentic LM3886 one not. Also might be interesting to see what the counterfeit chip looks like with your design.
I wonder if the chip is actually authentic in the sense it was made in the same fab but off hours to be sold as counterfeit. I'm guessing TI will never tell us what they find out.
Anyways really appreciate someone doing a real test on this thing!
I really like the amp - to me it sounds great, it's by no means crap. The sound is day and night compared to the stuff from the 70s and 80s (well that I can afford at least) Considering the price to me it's by no means a poor choice - just not the best.
I'm pretty pretty sure the LM3886 is the one with white markings - so counterfeit.
I would be interested in seeing the difference between two XY boards with one authentic LM3886 one not. Also might be interesting to see what the counterfeit chip looks like with your design.
I wonder if the chip is actually authentic in the sense it was made in the same fab but off hours to be sold as counterfeit. I'm guessing TI will never tell us what they find out.
Anyways really appreciate someone doing a real test on this thing!
Did you follow the original XY pcb Thread? Or incorporate any of the modifications mentioned in that Thread?
Is your XY the new smaller pcb, or the older bigger pcb?
Hi Andrew, I looked it up, but I can''t find a thread with modifications suggested for the XY pcb. Could you please post a link?
The LM3886 is a very good amp and the XY-board to me seems a pretty good implication if you can live with an output resistor to squelch oscillation. The decoupling caps are on the small side, a small ceramic close to the power pins could be considered, and I don't like the routing of the speaker return current through the board. The latter is easily corrected btw. But even then, without modification, the differences even between the XY-board stuck with a fake chip, and Toms board with a genuine chip are minimal and at any rate inaudible. Stucking the XY-board with a genuine chip is bound to make the difference, if any, even smaller.
This Thread started with the older bigger XY PCB. It looks at the changes when the smaller newer XY PCB was released.
http://www.diyaudio.com/forums/chip-amps/207971-bought-xy-lm3886-kit.html#post2930771
This Thread deals extensively in getting an XY PCB to successful completion. I don't think it discusses modifications to the XY PCB.
http://www.diyaudio.com/forums/chip-amps/206658-3886-ebay-kit-build.html#post2904935
http://www.diyaudio.com/forums/chip-amps/207971-bought-xy-lm3886-kit.html#post2930771
This Thread deals extensively in getting an XY PCB to successful completion. I don't think it discusses modifications to the XY PCB.
http://www.diyaudio.com/forums/chip-amps/206658-3886-ebay-kit-build.html#post2904935
As stated earlier, I'm committed to testing the XY board with a real LM3886. Right now time is the limiting factor. Paying work comes before volunteering. Sorry. That's reality.
I doubt it was "made off hours". Semiconductor manufacturing requires many, many steps (at least 10s if not approaching 100). Some of these steps (like anneals) take the better part of a day. After a few months, you get wafers that are then tested and the good dice go off to packaging. After packaging, the devices are tested again (final test).
I suspect the fake LM3886 delivered with the XY kit is a factory reject. It failed final test, i.e. does not meet the data sheet specifications. It was scooped up and sold on the black market.
Never say never. I'll get the canary to sing. Whether I'm allowed to record and share the song is another story.
You're welcome.
Except the output resistor doesn't do a very good job at squelching the oscillations. See the transient response in the first few posts.
I don't understand the opposition to Thiele networks. They're easy to make.
Tom
After staying out of the fray, I would like to finally post my experiences with the XY LM3886 amp module.
I don't doubt Tom's measurements or his comparison. While one could argue that he was comparing applies to oranges, he was just trying to qualify the performance difference between a well-designed board with high-quality parts to a mass market el-cheapo Ebay kit (fake parts or not). In that sense his comparison was completely legit and it showed the differences in performance between these two levels of cost and build quality.
With that said, the real question that many people are probably thinking is "can I hear the difference" and "how does it sound' (the XY kit). I have used several of these in the past, purchased as assembled boards from Ebay (a couple of years ago). I can't say for certain whether they came with genuine LM3886 or not, although I recall checking the lettering on the IC and thinking that it might be fake at the time. I did not push the LM3886 to its limits in my application, and used it for non-compressed music playback at modest home levels. Frankly, the XY builds sound just fine. IMHO one would be very hard pressed to point out any sonic signature or distortion. For the level of power that it can deliver (35W@8R, 65W@4R) the LM3886 is my go-to chipamp. I like to build active speakers and this can easily be incorporated into the speaker with modest heat sinking if the average power level is not high. The XY board makes for a compact footprint that works relatively well. Just look at Tom's measured results - hey, that's really not 'bad' by any stretch of the imagination.
After I used a few of these kits and was concerned about fake LM3886 ICs I was able to find an Ebay seller who was offering the unpopulated XY PCB. I bought something like 20pcs, and then got the same number of LM3886 ICs from Mouser, the higher quantity getting some discounted pricing. Now I can build up an XY kit using parts with known quality, and make modifications as I see fit to control gain, low frequency cutoff, add HF filtering, etc.
For those who are scared by the prospect of fake parts but might be interested in a compact LM3886 build for more than 2 channels, you might consider this approach. It will sound great and you can sleep at night without suffering from bad dreams about fake parts. 😉
Last edited:
Tom,
You misread what I wrote. I agree. Thiele networks are the way to make an amp bulletproof. That's why I wrote "when you can live with, etc.".
However, in many predefined situations, such as active speakers, a Thiele network is not required.
You misread what I wrote. I agree. Thiele networks are the way to make an amp bulletproof. That's why I wrote "when you can live with, etc.".
However, in many predefined situations, such as active speakers, a Thiele network is not required.
Link to thread
http://www.diyaudio.com/forums/chip-amps/207971-bought-xy-lm3886-kit.html
CharlieLaub - that's what I am wondering about.
Bought unpopulated XY's myself. Got real LM's etc. Haven't started building yet because I get the feeling its not worth it.
Do you have equipment to measure results of "fake" and "real" populated XY?
Even found a defective Sanyo DCA 303 to put it into. With big VU meters! Pretty.
Very interesting!
http://www.diyaudio.com/forums/chip-amps/207971-bought-xy-lm3886-kit.html
CharlieLaub - that's what I am wondering about.
Bought unpopulated XY's myself. Got real LM's etc. Haven't started building yet because I get the feeling its not worth it.
Do you have equipment to measure results of "fake" and "real" populated XY?
Even found a defective Sanyo DCA 303 to put it into. With big VU meters! Pretty.
Very interesting!
Except the output resistor doesn't do a very good job at squelching the oscillations. See the transient response in the first few posts.
I don't understand the opposition to Thiele networks. They're easy to make.
Tom[/QUOTE]
Will the Thiele network be instead of the output resistor?
I don't understand the opposition to Thiele networks. They're easy to make.
Tom[/QUOTE]
Will the Thiele network be instead of the output resistor?
The Thiele Network is either of two styles:
a.) R+C across the output to Return at the output on the PCB plus L||R in the cable feeding the output terminals.
b.) L||R at the output plus a C across the output to Return.
Dr Cherry did a paper on this pointing out that the two styles of the Thiele Network are at the end limits of a continuum. Cherry suggested that the F-3dB of the C+R should be around 100kHz to 300kHz.
Remove the .txt to leave the Zobel.xls for an excel format file.
The file was written while I still had MS excel on an old win95 PC.
It opens with Libre Office but the diagrams are slightly corrupted, but still legible.
It did open correctly with Open Office on my last PC.
a.) R+C across the output to Return at the output on the PCB plus L||R in the cable feeding the output terminals.
b.) L||R at the output plus a C across the output to Return.
Dr Cherry did a paper on this pointing out that the two styles of the Thiele Network are at the end limits of a continuum. Cherry suggested that the F-3dB of the C+R should be around 100kHz to 300kHz.
Remove the .txt to leave the Zobel.xls for an excel format file.
The file was written while I still had MS excel on an old win95 PC.
It opens with Libre Office but the diagrams are slightly corrupted, but still legible.
It did open correctly with Open Office on my last PC.
Last edited:
Yep. To keep the LM3886 stable, you really should add the R+C from output to ground (as low impedance to the local decoupling as you can get). Then add L||R in series with the output.
To keep the LM3886 stable as its output voltage approaches the supply rails, you need Cc across the inputs of the LM3886. To compensate for the effects of Cc (it mucks up the transient response), you need to add Rf2 and Cf as well.
You can find all the component values in the LM3886 data sheet. Some of the values, in particular the resistor in the L||R network, require a bit of tweaking to achieve good performance.
Tom
No, I simply put it together with the schematic that came with it.
Not sure about the PCB, the first one is the first kit I built (and gave to a friend)
second messy wiring one is what I currently use (I couldn't find a picture of them not upside down)
Reality we all know! Totally understandable. Didn't mean to sound pushy - I must have missed the part you said you intended to test with a authentic chip.
yeah - by 'us' I mean everyone here. I'm sure you will find out confidentially if it's a friend.
Either a lot of chips get rejected or something else is up. If it was like you said I would think the number of these things on the market would be pretty small. That said I don't know much about fabs as you have probably guessed 🙂
Attachments
Ooh! You received a schematic. Wow. I didn't even receive that much. 🙄
I used to work for National Semiconductor and hung on after the TI acquisition. I worked there for a decade, so I have some insight into how they operate and how semiconductors are manufactured.
I don't know what the final test yield is on the LM3886 and if I did, I wouldn't be allowed to disclose it anyway. Considering that the final test tests if the bonding and packaging went correctly, the yield should be pretty high. Let's for argument's sake say it's 99 %. You make 1M units/year, so you have 10k units that fail final test. Employee X grabs the discard pile and sells them on the black market for 10 cents each. Gets a $1k bonus... That's good money in the places the chips are assembled.
I'm not saying this is how the fake LM3886 ended up in the XY kit I ordered. I have no knowledge of how they obtained the IC or whether the "LM3886" does indeed contain an LM3886 die inside. All I'm saying is that it could have happened this way.
Tom
Would you need it with your knowledge???
They are not fakes then, just working rejects
Been reading this thread with interest, personally if you want to sell your product, then a comparison with a lm3886 from a known source in the xy board is a must. TBH with my aging ears, could I tell the difference of such a low claimed THD.....probably not😀
Sorry got this vision of a factory worker, with a swag bag, spilling lm3886's all over the place as he leaves the factory gates!
Last edited:
As I wrote in Post #1, I didn't need the instructions. I still think it's inexcusable to deliver a kit with instructions in Chinese only. There are plenty of threads here that go along the lines of, "I ordered XYZ from an eBay vendor and couldn't read the instructions. Now I'm stuck and the vendor won't answer my email. Please help." Clearly there is a need for instructions. I'd take even a machine translation of the Chinese instructions over no instructions at all.
On the other hand, the lack of support by eBay vendors does open up an opportunity for me. I do respond to email and provide tons of support. The design document that I deliver with my LM3886DR board is 34 pages. Some people appreciate that. For those who don't, there's always eBay (and DIY Audio when you get stuck).
Define "working". Do you know which part of the final test they failed (assuming they are factory rejects)? If the IC wasn't sold through the TI distribution channel it is not a genuine TI part. Not genuine = fake in my book.
And as I've said many times on this thread already, I will do that when time allows.
Depends.
🙂 I think Clint Eastwood should direct a movie about it.
Tom
Last edited:
So, (re: pirated rejects) my question is: when do the LM3886 packages get laser-marked or silkscreened? In legit production? Pre or post final test?
The XY kit 3886 ICs I have, show manufacturing codes (lot/plant/date) which bear no resemblance to any standard format on the TI website.
It's like somebody said, lets make up a random alphanumeric code to print there, which makes no attempt to copy even the format of a legit lot#.
I find this single aspect of the fake IC concept rather bizarre.
-Peter
The XY kit 3886 ICs I have, show manufacturing codes (lot/plant/date) which bear no resemblance to any standard format on the TI website.
It's like somebody said, lets make up a random alphanumeric code to print there, which makes no attempt to copy even the format of a legit lot#.
I find this single aspect of the fake IC concept rather bizarre.
-Peter
Post final test would be my guess. Why spend money marking parts that you'll throw in the trash? Every fraction of a penny counts.
Tom