unfortunately for us audio enthusiast
assume a lot of components to disappear
As for a commercial design.
You would be looking at the usual landfill designs as I call them.
Likely a high frequency / inductor less Class D could replace the 3886
And would be powered by a switch mode supply.
Far as analog class AB simplicity there is also the LM2876
which I believe around 300 currently available
assume a lot of components to disappear
As for a commercial design.
You would be looking at the usual landfill designs as I call them.
Likely a high frequency / inductor less Class D could replace the 3886
And would be powered by a switch mode supply.
Far as analog class AB simplicity there is also the LM2876
which I believe around 300 currently available
Issue is that he wants an amplifier, and the chip amp needed is not available, and one year down the line, who knows?
So, the alternatives are:
Buy one second hand, and listen...easiest. And can modify as needed, may not necessarily be 3886 based.
Use another chip amp, and expect little difference. Harder.
The last is to build one using discretes, and for those also many designs, well proven and discussed on various forums, are available. Hardest.
Unless he wants to get stuck on the 3886, that is. Hardest, to the next level.
So, the alternatives are:
Buy one second hand, and listen...easiest. And can modify as needed, may not necessarily be 3886 based.
Use another chip amp, and expect little difference. Harder.
The last is to build one using discretes, and for those also many designs, well proven and discussed on various forums, are available. Hardest.
Unless he wants to get stuck on the 3886, that is. Hardest, to the next level.
That for sure
Yup.You have to jack around with calculations
It´s called "designing"
Not arguing, just showing my particular case
* not aiming for Audiophile Heaven but for Guitar amp duty (admittedly not what lauda needs) , I needed a "black box" which gets +/-35V , drives 50W RMS into an 8 ohm Guitar/Bass speaker with 1V or less input.
* 6 terminals in total which if needed can be wired to original LM38** -TDA7294 - TDA1514 (the original "problem child") PCB holes, go figure.
IN / Gnd / +V / -V / Spk Out / Spk return (which can return straight to PS Gnd if needed)
* no problem with < than 45mV offset so no need for input pair matching.
* No need to thermal track *everything* since I do not aim at DC amplification at all, only needed at biasing which is easily achieved by gluing diodes to heatsink , the "fancy" version adds a 6th bias transistor.
* footprint is about same as original chipamps, can be mounted on same heatsink, of course requires drilling 2 mounting holes.
* within its expected function, works like a charm
Again, just mentioned it as "Plan B" and a comparison, of course a Consultant must offer his customers a "commercially available" solution.
If they can stock a safe quantity and even better, secure a stable supply, specially straight from TI , or worst case, from a quality stable "second source", my blessings to them
@JMFahey: If it can take +/- 46V, please post the circuit, I have a blown STK 403 that needs replacing, and not available. Or PM it.
It is stereo, will have to make two with a common supply.
Also, another item is in short supply, Omron G6E-134P-US-24, a 20mm x 10 mm small PCB mounted relay, 0.4 A DC...
We managed to get some from RS here, element14 and Mouser are saying February 2022 expected.
It is not just the semiconductors, other things also are seeing supply issues.
It is stereo, will have to make two with a common supply.
Also, another item is in short supply, Omron G6E-134P-US-24, a 20mm x 10 mm small PCB mounted relay, 0.4 A DC...
We managed to get some from RS here, element14 and Mouser are saying February 2022 expected.
It is not just the semiconductors, other things also are seeing supply issues.
Last edited:
A lot of beginners are still at the chip amp phase. I sort of skipped the LM3886 revolution myself, since I was well beyond chip amps when they came out. I’ve got friends who are still in the chip amp phase, and one has been staying with the LM1875 family because of availability. I started with the TDA2002 and STK modules but that was in the late 70’s. Making your own STK modules is certainly possible, like Mr. Fahey has done, but you still need to know how to make them. Definitely a viable alternative to fakes on the market now when you NEED an STK module to get something old working. I’m sure there were users out there who used LM3886 kits to replace STK modules they could no longer get 10 or 20 years ago.
Let's not forget that chipamps are not just for "beginners". The LM3886 has a history of use in high-end audio dating back to its use in Jeff Rowland's Model 10. This tradition is being maintained and advanced today by a number of developers, including members in these forums.
It's also worth noting that in addition to cost effectiveness and convenience, the LM3886 has a GBW and other specs that depending on your budget may not be falling-off-a-log easy to get in a discrete design.
As always, know your options and pick the best one for the application.
It's also worth noting that in addition to cost effectiveness and convenience, the LM3886 has a GBW and other specs that depending on your budget may not be falling-off-a-log easy to get in a discrete design.
As always, know your options and pick the best one for the application.
If your project has big volumes, talk to a silicon foundry, and let them come up with supplies.
Maybe TI will allow license production.
But if your volumes are not large, you have a problem.
And like mentioned in posts above, the chip is not being used in large scale production any more, so you will be the only large consumer.
JRC was talking $200K for special production of regular product, someone posted that here, and it did not pan out.
Your order value and volume should be interesting for the seller to deal with you.
Maybe TI will allow license production.
But if your volumes are not large, you have a problem.
And like mentioned in posts above, the chip is not being used in large scale production any more, so you will be the only large consumer.
JRC was talking $200K for special production of regular product, someone posted that here, and it did not pan out.
Your order value and volume should be interesting for the seller to deal with you.
Building an LM3886-based amp is just as much DIY as building a discrete amp. The success or failure of a particular design has more to do with the skill and competency of the circuit designer than with the components.
One very nice feature of the chipamps is that they have no bias adjustment. That does make the DIYer's life easier. It also removes a lot of risk from a vendor/manufacturer's perspective.
Tom
One very nice feature of the chipamps is that they have no bias adjustment. That does make the DIYer's life easier. It also removes a lot of risk from a vendor/manufacturer's perspective.
Tom
Thank you for just not putting the currently offered class D solution.
Maybe some people out there can enjoy class D amplifier. It is the same for me as flickering LED automotive lights. They appear as a series of dots when I pan my head so they are easy to spot.
I recently encountered the most horiffic listening experience at an outdoor show. It seemed like there was sufficient hardware in place but the upper midrange and highs were the most scritchiest I have ever heard. I went home and looked up the speakers they used and I see a powered 3 way system with class D driving horn drivers and DSP crossover. I would love to blame poor setup but I just doubt it was that.
Maybe some people out there can enjoy class D amplifier. It is the same for me as flickering LED automotive lights. They appear as a series of dots when I pan my head so they are easy to spot.
I recently encountered the most horiffic listening experience at an outdoor show. It seemed like there was sufficient hardware in place but the upper midrange and highs were the most scritchiest I have ever heard. I went home and looked up the speakers they used and I see a powered 3 way system with class D driving horn drivers and DSP crossover. I would love to blame poor setup but I just doubt it was that.
Sorry, not that big
My goal was replacing unavailable 40-50W chipamps, I had to to keep them tiny since the idea was replacement, not new designs, and had to fit in original available space, mounted on existing heat sink, so two side by side TO220 packages possible, period.
Which is exactly what 3886/7294 cases are:2 x TO220 side by side, check it.
In fact "they" call the package TO220,which is nonsense
Body is 20mm wide by 17.5mm high, while TO220 is by 10.40 mm wide by 15.75 high
Best widely available power complement in that size was TIP121/126
TIP126 pdf, TIP126 description, TIP126 datasheets, TIP126 view ::: ALLDATASHEET :::
and it meets or exceeds 3886 specs:
LM3886 - TIP122/126
Peak current 7A - 8A
Supply +/-35V - +/-40V
Rth 1°C/W - 1.92°C/W (I use two in parallel so net ~1°C/W)
So, it works well in the intended application, but not necessarily beyond. (sorry nareshbd)
Not a FULL replacement: no Mute, no SPIKE protections, but those were two "bonuses" not strictly necessary either.
They can be added, like anything else, but then board will be larger.
To nareshboard: you can certainly design/adapt a normal power amp design, the simpler the better for space reasons, to run from +/-50V rails, pass necessary current, dissipate what´s needed, etc.
Beats having a good amp/receiver turned into a doorstop any day of the week.
I have seen very nicely made DIY Sanken replacements, with TO218/247 transistors mounted straight to heatsinks which is important for good dissipation, guess those were "just the power pack, no front end" version but design can be extended to include that.
I think the typical problem is that people become obsessed with minutely cloning STK ... in an electrical way that is .... problem is that leads to using SMT transistors mounted on a PCB which **explode** as mentioned above, my goals are simpler: signal in > speaker out.
OUCH!!!!! just found very complex and inusual STK403 ... 3 amps inside a package, wow!!!!!
In any case, it might be "doable" with some effort, but you must design it yourself.
BIG problem is heatsink footprint: original chip is 67mm wide, you can barely fit 6 x TO220 packages there, then you´ll have to fit the rest of each amp (diff pair and Vas) into 15mm x 25/30mm board space TOPS, forget short protection (which anyway is not a deal breaker).
Oh well.
To build something like that you stack two PC boards. One for the power transistors pairs (use Sanken darlingtons in TO-3P), bias multiplier, and emitter resistors, with the power packages mounted under the board. Run the mounting bolts thru the board. The second board containing the entire front end stacks on top with a pair of 0.1” single row headers. Nice little “sandwich” that takes up little more room than the original module. It comes out thicker than an original STK, but there is usually *room* in that dimension. All thru-hole parts which can actually dissipate something can be used.
I know that I have been having trouble getting digital chips and microcontrollers, but I never realized that an older design made on an older analog process like the LM3886 would become hard to get.
Jameco still has them listed on their web site. They do not show "out of stock" like they do on some other chips, but I don't know if they really have any.
Parts Express lists the chips for $15.98 each, but they are out of stock. They do have an amplifier board containing a pair of chips in stock for $32.98.....Too much for an $8 chip.
I looked in my parts box and I have a tube containing 18 National branded chips from at least 10 years ago, maybe longer. The date code reads 74AK. The tube originally held 20 chips, so I used two for something, don't remember what for sure but I did make some MIDI controlled guitar amps full of National SPI controlled EQ and other chips in the early 2000's.
I will never use them, too many other projects, not enough lifetime left. I don't want to send 1 or 2 here and there but I will send them all to anyone making a reasonable offer.
Jameco still has them listed on their web site. They do not show "out of stock" like they do on some other chips, but I don't know if they really have any.
Parts Express lists the chips for $15.98 each, but they are out of stock. They do have an amplifier board containing a pair of chips in stock for $32.98.....Too much for an $8 chip.
I looked in my parts box and I have a tube containing 18 National branded chips from at least 10 years ago, maybe longer. The date code reads 74AK. The tube originally held 20 chips, so I used two for something, don't remember what for sure but I did make some MIDI controlled guitar amps full of National SPI controlled EQ and other chips in the early 2000's.
I will never use them, too many other projects, not enough lifetime left. I don't want to send 1 or 2 here and there but I will send them all to anyone making a reasonable offer.
Out of curiosity, I tried to order some and was immediately hit with "We have no stock available on this item, this item is discontinued." In addition, Mouser's LM3886TF inventory (not RoHS) is now down to 23. The value of your tube of 18 is appreciating.
According to info posted by tomchr in another thread my chips are from 1997, which was during a time when I was building solid state guitar amps. They predate ROHS, and therefore are not ROHS. Doesn't matter to me since I solder with real lead, but might to some.
It is stated in some of his info that all TI chips are laser marked, and painted (silkscreened) chips are fake. That may be true of LM3886 chips made by TI, but National did screen the numbers on chips. Mine probably came from the National Semiconductor sales rep since I was a cell phone designer at the time, and National was good at "saturation bombing" us with samples. Since nearly every engineer in the plant had a National Semi LDO designers kit on their desk, guess whose regulators wound up in all the Motorola phones?
It is stated in some of his info that all TI chips are laser marked, and painted (silkscreened) chips are fake. That may be true of LM3886 chips made by TI, but National did screen the numbers on chips. Mine probably came from the National Semiconductor sales rep since I was a cell phone designer at the time, and National was good at "saturation bombing" us with samples. Since nearly every engineer in the plant had a National Semi LDO designers kit on their desk, guess whose regulators wound up in all the Motorola phones?
Actually I thought of mounting two 7293/94 or 49810 on the heat sink, it is big enough, and running a ribbon cable to the main PCB.
I bought it in scrap for 120 Rupees, US $1.75....The transformer alone is worth much more.
Also, my amps are working, and I have spares, so I am a bit lazy about it.
The main issue was the high voltage, power output is too high from the 403-90 or 100 (different countries had different STKs), that is not a deal breaker, but the high supply voltage precluded most mono chip amps.
So this adapter made by J M Fahey could have been used in place of the mono chip amps, and I could use one for each channel, with common power supply, for the stereo output.
I think 40 watts is enough for most use.
I will open it up and see, maybe I can put ready Mosfet kits on the heat sink.
It is a Sony 3 CD/cassette/receiver, the display lights up. Small cooling fan too.
I bought it in scrap for 120 Rupees, US $1.75....The transformer alone is worth much more.
Also, my amps are working, and I have spares, so I am a bit lazy about it.
The main issue was the high voltage, power output is too high from the 403-90 or 100 (different countries had different STKs), that is not a deal breaker, but the high supply voltage precluded most mono chip amps.
So this adapter made by J M Fahey could have been used in place of the mono chip amps, and I could use one for each channel, with common power supply, for the stereo output.
I think 40 watts is enough for most use.
I will open it up and see, maybe I can put ready Mosfet kits on the heat sink.
It is a Sony 3 CD/cassette/receiver, the display lights up. Small cooling fan too.
Last edited:
As for the 3886, it does look like it is finished as far as TI is concerned, maybe a 'Limited Edition' will be released in the future.
Don't depend on it.
Just see if any reputed copies are available for personal / repair use.
And look for substitutes if building a new chip amp, I think the 49810 series might still be available, and the 7293/94.
I did not check in the past year or so...
Don't depend on it.
Just see if any reputed copies are available for personal / repair use.
And look for substitutes if building a new chip amp, I think the 49810 series might still be available, and the 7293/94.
I did not check in the past year or so...
Thanks for the offer, but I think the availability Dxvideo pointed out is all I need short-term.
Longer term however I may pitch a design using the part to a company I provide consulting to. So, I'm more concerned about the stability of supply than immediate need. (I'll be reaching out to TI directly as things evolve.) The only suitable part I've found so far for the application is the LM3886.
LCSC was one of the vendors I checked, and I wasn't able to find availability there.
NTE7197 is a direct replacement for the LM3886