• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

OPA1656: High-Performance CMOS Audio Op Amp

I have an amplifier that is using the opa1678 and am considering swapping it out for the 1656. Is this advisable? I believe that it is being used for the XLR outputs of this particular amplifier though I don't have the schematics so I can't be sure.
There is no other op-amp in sight. And it's a Class-D amplifier.

Any help would be greatly appreciated.
 
I stupidly killed the right part of one of my few 49720HA in my preamplifier yesterday because of a fault in my new DIY phono preamp. That was a day full of sadness.
I replaced it with an OPA1642 and am sorry to say that there is a big difference sound-wise. The old 49720HA sounds more direct and dynamic than the newer SMD opamp.

A forum member who used to work in development at National (and then got laid off after TI took over) tipped me off many years ago to try the metal can. I didn't think the packaging would make such a big difference.
Unfortunately, the TI John who started this thread seems to have gone into hiding as he has not posted here again nor responded to my PM as to why they discontinued the HA and if there is an equivalent replacement.

Reading the TI website, you find the most awesome marketing terms like Soundplus, Premium Sound, Burr-Brown audio. But then I wonder if they ever did any listening comparisons between the OPamps. Because then one would have a LME49720HA or LME49990 with security not adjusted. Controlling was at work here and not the audiophiles.

Many high-end audio manufacturers still use THT technology and avoid SMD at the sound-relevant points. Also, a LME49720HA that costs 5 times as much as the PDIP or SMD version should also be interesting in terms of cost calculation.

Now one is forced again to go on the search, and to find a replacement at least approximately equivalent, since the best parts are no longer produced. Concentration has never led to the best results. TI swallows everyone, and diversity falls by the wayside. Too bad.
 
Last edited:
  • Like
Reactions: 1 user
Stupidity did not kill the metal packages for ICs. Cost focused semiconductor customers killed them with their purchasing preferences.



Metal cans were still popular in the 1980s, but by the 1990, there was a movement to plastic packages for COST reasons ($$$). Customers wanted to pay less. Even the military moved from hermetic sealed metal packages to plastic package parts, because of cost in the late 1990s.

Even the original LM12 power op amp was in a 4 lead TO-3 metal package, with the die mounted on a heat spreader tab. In the 1990s, National moved the metal packaged power audio amps to plastic package for cost reduction to the customer. This was around the time of the LM3886 getting released.

Many more recent op amps (released after 1996) are less susceptible to RF EMI, even in plastic packages. You will see notes about EMI immunity in TI datasheets of IC Op Amps released in the 21st century.

The SMD packages became popular by semiconductor customers in the late 1990s, which really shrunk the PCB board space.
 

Attachments

  • c1.JPG
    c1.JPG
    45.1 KB · Views: 548
Last edited:
Member
Joined 2011
Paid Member
That's what "dielectric isolation" is: silicon on glass. The OPA627 is designed for applications requiring ultra low input bias currents. So it's fabbed in a crazy-expensive dielectrically isolated JFET + NPN + vertical PNP process: DiFET.


_
 

Attachments

  • difet.png
    difet.png
    237 KB · Views: 466
Should I put some OPA1656 in a metal can and see if anyone wants some :)
OR
you can shield the ckt and do much the same thing perhaps

Wonderful that TI can keep that DI process going after all these years.
I have never bothered to buy a OPA627/637 and listen to it in action, one day perhaps but I think I will not be able to tell the difference anyways, probably why I do not bother to try :)
 
The main reason for using metal cans was to get a hermetically sealed package & reduction of die surface stress. This reduces the chance of corrosion from water ingress & ionic contamination, such chlorine on the metal bond pads and bond wire interfaces.

Plastic molded IC packages have particles in the molding compound which can press on the die surface in random regions. This can lead to large Vos change with temperature. Sometimes a ceramic package with a fused lid is used to eliminate this stress without using a metal can.

Op Amps can conduct RF on the input node wire connections, which is then rectified by the input junctions (BJT or JFET) or input ESD protection diodes. Many early Op Amp designs (before ~ 1993) did not have RF EMI immunity as a design requirement goal.

Many newer Op Amp designs have this as a design consideration in the objective specification. You will find most new Op Amp design can tolerate RF EMI pretty well.

Dielectric Isolated IC process allows some interesting performance capability, such as low device leakage to the substrate at high temperature, higher speeds with less capactive coupling to substrate & noise isolation between device on the same substrate. It is an expensive $$$$ process. The process steps of the wafer bonding are more common these days for MEMS applications.

Here is one DI patent example from Harris in 1987, though it was used long before that, see RCA review from 1968 as one example.
 

Attachments

  • DI-Harris-Pat-1987.JPG
    DI-Harris-Pat-1987.JPG
    257.9 KB · Views: 467
  • RCA-isolation.JPG
    RCA-isolation.JPG
    135.4 KB · Views: 485
That's what "dielectric isolation" is: silicon on glass.

No, silicon directly on glass is impossible, silicon monocrystal won't grow on an amorphous substrate like glass. There was a silicon on sapphire technology, used by the early HP processors (in the 98xx computers the grandfathers of the PC) but it is long obsolete (could still be running somewhere for military stuff, though).

The dielectric isolated process was developed originally by BB for high frequency. It is roughly (bottom top) silicon (bulk wafer)->thermal silicon dioxide (which is not amorphous glass!)->epi silicon re-crystallized and gettered (difficult, since the delta between the silicon dioxide and silicon lattices is large, hence a lot of crystalline defects in the bulk), with islands defined by local oxidation of the epi layer, down to the bulk oxide. A very complex and demanding process, this is one of the the reasons why these devices are crazy expensive today.
 
If you buy the metal canned ones (not recommended for new designs by the way) you are in for a lot more than $40 each.

Yep. As I said before and got criticized for, there is really no reason to buy OPA627 today unless you are replacing a dead component or building an old circuit. OPA828 has lower noise, lower offset, lower drift, comparable input bias current, is faster, and cheaper.
 
Last edited:
True, although I have seen few projects using OPA828. I personally do not like the pinout of single OpAmps in SOIC-8 because the placement and routing of the feedback network and decoupling is not as straight forward as with duals, but it is an impressive part. I do not know if it is approved by the golden-ear audiophiles though, probably not expensive enough in comparison with OPA627...
 
No, silicon directly on glass is impossible, silicon monocrystal won't grow on an amorphous substrate like glass. There was a silicon on sapphire technology, used by the early HP processors (in the 98xx computers the grandfathers of the PC) but it is long obsolete (could still be running somewhere for military stuff, though).

There's a company here in San Diego (silicon on sapphire) for military/RF. My boss used to be pretty high up at Global Foundries, working on their SOI process. Although I'm sure it looks totally different to 36V analog process (that's probably still on 200mm).
 
Although I'm sure it looks totally different to 36V analog process (that's probably still on 200mm).

Can't imagine anything over 4-5" silicon on sapphire wafers, but I may be rusty since I worked in a fab (2" at the time, don't ask the cost :D)

There are many ways to build SOI substrates, one I worked with was a version of SIMOX, which is a silicon wafer (mechanical) substrate oxidized, then another wafer bonded on top of the oxide, then the top side was ground then polished to a thickness useful for devices, isolation between islands was still by local oxidation.

We also tried at the time oxygen implantation, worked fine, but only for low voltage applications, we never had a MV ion implanter with enough beam current to get the oxide layer deep enough.

Then we also tried deposited polysilicon re-cristallization and failed miserably.
 
My boss used to be pretty high up at Global Foundries, working on their SOI process. Although I'm sure it looks totally different to 36V analog process (that's probably still on 200mm).

Global Foundries has a few different SOI technologies, some with very thin Box (Buried oxide layer). They have been promoting the 22nm FD-SOI (Fully Depleted Silicon on Insulator) for quite awhile.

One of my engineering instructors once referred to Silicon On Sapphire as $O$.

Another company using SOI - pSemi - Intelligent Electronics for the Connected World
 

Attachments

  • GF-22nm-FDSOI-22fdx-26-web.pdf
    460.2 KB · Views: 211
  • GF-RF-SOI-INST-8SW-02.pdf
    1.1 MB · Views: 69