An Ebay LM3886 PCB - comments invited

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The main use of an amplifier that produces acutance is for more intelligible TV/Movie vocals. These days, TV/Movies is the main use for amplifiers and speakers. And, in a house with more than one person, it is unlikely for there to be music replay in the same room as the TV (because the TV is on). Therefore, there's a huge market for cheap sharp sounding* gainclone kits, in the living room, with the TV. In other words, a poor bit of kit, with a very good use for it. They're not very good with music except in the case of overcoming somewhat dull speakers.
*decoupler size can be changed to likewise change the tone (bigger for tamer tone). Typical application is outright piercing as if made for nearly deaf people. Initial satisfaction is very high that way, although the sound could make ears get tired eventually.

To decrease the typical shout, at post#2, I tried to give that TV amp some Xanax to calm it down (less shout). Slightly less flashy first impression, but not thrown out so quickly either.


P.S.
A music amp has to be a lot nicer than that.
However, the market for nice Made-for-Music amplifiers is much smaller. So is the typical room for it. It is generally not used in the living room, because TV is on. Therefore, today, we see smaller scale music replay systems that fit offices, bedrooms, kitchens, etc. . .
 
Last edited:
Non-Inverting LM3886 Amplifier PCB | Chipamp Electronics

Here is a much, much better layout for $6.00.

If you take the time to read Tom's website, he explains what aspects of layout are most critical for the 3886, and he clearly explains why. This modest board from chipamp.com addresses at least one of them. You can even do a better layout on this board than what is recommended yourself. :) Tom's board addresses all of them and is very well sorted.

Tom has done some excellent, original research that is very relevant to people that are interesting in getting the maximum performance out of this chip. Thanks, Tom. :up:
 
Yes, I've had a look at those in the past. Interesting to compare - the chipamp.com layout also only has 100uF power supply bypass caps. Also they've omitted the L/R components on the output, though I suppose they could be added off-board.

National Semiconductor/TI recommends a supply cap of 470 µF per rail minimum in the LM3886 data sheet. They also recommend a smaller electrolytic or tantalum cap and a ceramic cap right by the IC pins. There has been tremendous improvements in passive components since the LM3886 data sheet was written, hence, I performed a comprehensive study of the supply decoupling network. I arrived at 1000 µF electrolytic || 22 µF electrolytic || 1 uF X7R ceramic (or 4.7 µF X7R ceramic if you can find it) as the optimum. See my simulations here: LM3886 chip amp supply decoupling.

The Thiele network (L||R) on the output is needed if you want to drive capacitive loads (long speaker cables, for example). Its function is to isolate the load capacitance from the LM3886 for frequencies above a few hundred kHz. I suppose you can go without it, if you know the load capacitance will be low. But for a production circuit, I think it's a bit reckless to go without the Thiele network.

~Tom
 
Yes, I've had a look at those in the past. Interesting to compare - the chipamp.com layout also only has 100uF power supply bypass caps. Also they've omitted the L/R components on the output, though I suppose they could be added off-board.

Regards
John

There's space on the board for 1000 uF caps, isn't there? Chipamp sells a 1500 uF/50 volt capacitor intended for this purpose. I don't think 100 uF is big enough.

Besides, you can improve the recommended bypassing, right on their board. You can substitute Os-Con capacitors for the ceramics; and put a ceramic capacitor on the back side of the board, soldered directly to the power supply pins of the chip.
 
Last edited:
................... they've omitted the L/R components on the output, though I suppose they could be added off-board............
I believe the better location for the L||R part of the Thiele Network is off the PCB. I try to keep it away from the "amplifier" and away from other metal panels.

The R+C part of the Thiele Network should be on the amplifier and located so that the high frequency currents can flow around the decoupling loop.

The Network that Neville Thiele discussed way back was a combination of C shunting the signal back to the amp and an inductor feeding the load. He showed that the C shunt could be both before, or after the L inductor. I don't recall comment on the need for low impedance in the C shunt route.

Dr. Cherry did a couple of articles expanding on N.Thiele's network. The articles I read were about 1995 & 1997, maybe in ETI. ( I was a subscriber at that time).
 
Last edited:
I don't argue with anybody's ideas or assertions, but the gist of the comments I've read seems to be, "That's just the factory circuit." So I guess it's in the eye of the beholder whether the guys at the factory know what they're doing. But after all, they did build the chip.

...Use [eBay] boards and get average performance...If you want world class performance, I suggest looking at my Modulus-86.... Bentsnake, you may be interested in reading my resume. It's available on my website...I think you'll find it informative...

There seems to be a bit of animosity toward me hereabouts. At least from billshurv and cogitech, with Fast Eddie D perhaps chiming in. Well, if that's what happens when you state a different view, then it is.

But I think some of this unpleasantness might be caused not by real events, but by problems with reading comprehension. So I think it's fair--even necessary--for me to try to clarify.

...please be so good as to leave me out of it. I know or care nothing about you or your product, so please don't twist the meaning my words in your attempts at self-promotion.

Apparently it's still not clear that I have no interest in anybody's product for sale. But not being interested doesn't mean I wish the maker ill. To the contrary, I hope they sell a million of whatever it is.

What I care about is somebody trying to sell a product I never heard of by using my words, quoted out of context from my posts on this forum.

If I were what the law calls a "public figure" this could bring severe consequences. Even in the case of a private individual the law prohibits "misappropriation of name or likeness."

But legalities aside the ethical question remains, and I say now what I said before. Using contrived quotes to link my posts to some product I know nothing about is seriously over the line.

So please don't confuse the two issues. The product, be it gold or Uncle John's Universal Remedy, doesn't enter in. What does enter in is that my posts on this forum are my attempts to help somebody understand what they didn't understand before, and only that. So as before, good luck with your product, but please don't try to twist my words to fit into your pitch.
.
 
Last edited:
Member
Joined 2014
Paid Member
You are still missing the point that, to get the spec sheet performance you need the factory circuit with the factory PCB layout. Get the layout wrong and you don't get the performance promised from the circuit. That is all that people have been trying to say here.
 
You are still missing the point that, to get the spec sheet performance you need the factory circuit with the factory PCB layout. Get the layout wrong and you don't get the performance promised from the circuit. That is all that people have been trying to say here.

Exactly!

In addition, the chips are tested and characterized on well-regulated laboratory power supplies. If you use an unregulated supply, you'll get different performance than that published in the data sheet. The data sheet shows the performance of the IC. The lab setup has been optimized to showcase the performance of just the IC under the best possible conditions. It is up to the system (amplifier) designer to minimize the impact of the other components in the system (power supply, PCB, etc). After all, National Semiconductor or TI have no control over the customer's system. They can provide recommendations and guidelines, but if the customer wants to shoot himself in the foot, there really is no way for TI to prevent that.

~Tom
 
Last edited:
You are still missing the point that, to get the spec sheet performance you need the factory circuit with the factory PCB layout. Get the layout wrong and you don't get the performance promised from the circuit. That is all that people have been trying to say here.

I'm confused, Bill - who's "you" in this sentence - there's several parallel threads going on now!

John
 
Member
Joined 2014
Paid Member
That was aimed at bentsnake, But I think all of us are on the same page now (hopefully). Just to summarise

You need factory cct and factory layout to get factory numbers (and regulated PSU)

Tom has a load of investigations on his website on getting good numbers which are freely available for anyone to use in their DIYing. He also sells a composite amplifier PCB but that is secondary to this thread.

I got off on the wrong foot with Bentsnake but hope he wont hold it against me.

Sadly I can't realise advise much to the OP other than what others have said.
 
Exactly!
In addition, the chips are tested and characterized on well-regulated laboratory power supplies. If you use an unregulated supply, you'll get different performance than that published in the data sheet. The data sheet shows the performance of the IC. The lab setup has been optimized to showcase the performance of just the IC under the best possible conditions. It is up to the system (amplifier) designer to minimize the impact of the other components in the system. . .

Tom, this is good information, thanks.

I also think they must have measured the LM3886T for those datasheet output power ratings. I've got both varieties. One of them can do it fine.

I like ya plenty and if I've a complaint it is that I wish you had a somewhat wider variety of products. One flavor excellent ice cream is a good start, but you know there will be someone who wants a different flavor (or in this case either a larger or smaller scale).
 
Yeah, Change 22k to 27k (at both input and feedback). Change 220u to range 1500u~2200u. Change 22u to range 220u~680u (and bypass with 10n).
Like this: Phil's LM3886
. . . First off, add 22 kOhm in series with 47 pF across the 22 kOhm feedback resistor. This is to prevent excessive overshoot on the transient response caused by the noise gain compensation (220 pF cap across the amp inputs). Adding these components also reduce the peaking in the frequency response around 100 kHz where the noise gain compensation kicks in.
If I have translated correctly then if/when the feedback resistor is 27k then the RC parallel to it is a unit of 27k series to 39pF.
Will that do it?
 
I also think they must have measured the LM3886T for those datasheet output power ratings. I've got both varieties. One of them can do it fine.

Both the LM3886T and LM3886TF can deliver the data sheet headline specs of 38 W, 68 W (8 Ω, 4 Ω respectively) on a ±28 V supply.

Here's the data. The "Small" heat sink is a piece of 1.6 mm thick aluminum sheet, 13x13 cm in size. The "Medium" is a Pentium Pro heat sink (no fan). The "Large" is a black anodized heat sink specified at 0.4 ºK/W. You can read all about the experiments in the LM3886 Thermal Experiment (with data) thread. The graph below is from Post #36 of that thread.
452535d1417941156-lm3886-thermal-experiment-data-lm3886_thermals2.png


If you're having trouble getting the LM3886T/TF to deliver the power specified in the data sheet, it's either because of supply sag, or, more likely, because you forgot the thermal grease (I recommend Wakefield 120 or 126) between the LM3886TF and the heat sink.
If you mount the LM3886TF dry to the heat sink, it overheats around 25 W (8 Ω). That has nothing to do with the LM3886 and everything to do with physics.

FWIW, i spent 15 years in the semiconductor industry and i know for a fact that we used to do 100% testing of all our chips after assembly before they are shipped to costumers...

Same here.

~Tom
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.