Chips with built-in RFI protection

[mawyatt]

Noticed that some precision op-amps are stating they have some level of RFI built-in suppression. How widespread is this with newer chip designs?

Best,

[magic]

Seems that no one knows or cares
I have noticed that TI is making a big deal of it and datasheets of their newer opamps for industrial market (and some audio too, like OPA1656) show plots of "EMIRR".

NE5534 had some distributed capacitance over input stage collector load resistors 45 years ago; not sure what other purpose that could serve given the high RC corner.

[2N3055]

You noticed right, I also see this being marketed lately a lot, even for bog standard parts like TL071 where TI has new "improved" series TL07XH that claim " High ESD (1.5 kV, HBM), integrated EMI and RF filters"...

I was actually thinking of ordering some to see what it actually means, because datasheet doesn't specify a single number to that regard, except ESD resilience.
No mention what EMI and RF filters do, what are the characteristics or how it was measured and how can be verified..
Is it RF filters in signal path? Is improved PSRR on power inputs? What ?

But it has a typo where unity-gain bandwidth became  "Utility-gain bandwidth" in some tables ...

In layman terms, if that means that  bass guitar amplifier will be more resilient to those pulses you get when you put your GSM phone nearby, just by changing out old TL071 to new TL071H that would be great.
But it won't. That is all about circuit and layout of the components, grounding and shielding.

So I really don't know what they mean by it.

[T3sl4co1l]

I don't have anything quantitative to contribute, I've just noticed it in a few parts.

It does indeed appear effective against input rectification, particularly at the high frequencies where the attenuation is strong.

Most do not plot attenuation/immunity unfortunately; perhaps the same structures are used for all parts (why not copy-paste them, right?) and we can assume typical response?

In layman terms, if that means that  bass guitar amplifier will be more resilient to those pulses you get when you put your GSM phone nearby, just by changing out old TL071 to new TL071H that would be great.
But it won't. That is all about circuit and layout of the components, grounding and shielding.

It will!  Immunity certainly depends on all those things, but it is the product of all factors together, and dropping one of them by a solid 20dB or greater is an unambiguous win.  Not bad for not having to rev the PCB.

You do need it in all low-level signal paths, and you do need to avoid unfiltered signals into discrete semiconductor junctions, which might not be so easily managed in guitar pedals/amps specifically.  But at least you aren't being held back by the overabundance of RF-susceptible ICs, and on-board filtering and grounding can be relaxed to a certain extent.

Tim

[magic]

They define it as a sort of immunity to amplitude demodulation of out-of-band RF.

https://www.ti.com/lit/pdf/sboa128

That being said, given how laconic the TL07xH spec is, I'm not sure if it really was tested for "EMIRR" or "EMIRR IN+" by their own nomenclature... Also, no specification is given on the old part for comparison purposes. Good question to ask their tech support

I wanted to tear one down but they aren't very widely available yet.

edit
Is it just me or is the new part really a flicker noise monster with twice the amplitude and corner at several kHz? Are they even serious?

TL07xx Low-Noise FET-Input Operational Amplifiers

Fuck. Now that's starting to look like they have rebranded one of their CMOS opamps. Note the lack of J in front of FET. This cannot be real

[2N3055]

I don't have anything quantitative to contribute, I've just noticed it in a few parts.

It does indeed appear effective against input rectification, particularly at the high frequencies where the attenuation is strong.

Most do not plot attenuation/immunity unfortunately; perhaps the same structures are used for all parts (why not copy-paste them, right?) and we can assume typical response?

In layman terms, if that means that  bass guitar amplifier will be more resilient to those pulses you get when you put your GSM phone nearby, just by changing out old TL071 to new TL071H that would be great.
But it won't. That is all about circuit and layout of the components, grounding and shielding.

It will!  Immunity certainly depends on all those things, but it is the product of all factors together, and dropping one of them by a solid 20dB or greater is an unambiguous win.  Not bad for not having to rev the PCB.

You do need it in all low-level signal paths, and you do need to avoid unfiltered signals into discrete semiconductor junctions, which might not be so easily managed in guitar pedals/amps specifically.  But at least you aren't being held back by the overabundance of RF-susceptible ICs, and on-board filtering and grounding can be relaxed to a certain extent.

Tim

Tim, thanks!!

As I said, I plan to order a handful and try. If if does work that way, that is actually great news.  Even few dB less  for what is basically free upgrade is welcome.
But as you rightly noted, it is hard to suppress outside world influence on things that are directly connecting to the outside world, and guitars are quite good antennas for all kinds of crap....

Best,
Sinisa

[2N3055]

They define it as a sort of immunity to amplitude demodulation of out-of-band RF.

https://www.ti.com/lit/pdf/sboa128

That being said, given how laconic the TL07xH spec is, I'm not sure if it really was tested for "EMIRR" or "EMIRR IN+" by their own nomenclature... Also, no specification is given on the old part for comparison purposes. Good question to ask their tech support

I wanted to tear one down but they aren't very widely available yet.

edit
Is it just me or is the new part really a flicker noise monster with twice the amplitude and corner at several kHz? Are they even serious?

TL07xx Low-Noise FET-Input Operational Amplifiers

Fuck. Now that's starting to look like they have rebranded one of their CMOS opamps. Note the lack of J in front of FET. This cannot be real

Heh, good catch.. this DOES call for a closer look..

[magic]

Those assholes really did it
They think we are morons

Look at these two consecutive revisions of the datasheet.
CTRL+F "JFET" in SLOS080O: nothing found.

[JPortici]

Wow!
BTW: Current revision of the datasheet is "S"

There was this rumor i have no idea where i've heard it, might have been back in school so at least 10 years ago.. that TI had actually manufactured a metric crapton of wafers for the TL07x, LM13700 many years ago, they were not in production anymore and someday stock would have ended.
Yes it sounds a rumor from a guy whose uncle was working for TI but maybe.. maybe..

[DiTBho]

JFET -> _FET, this must be clearly a typo, ain't it? ain't it?

(or should I start worrying?  )

[DiTBho]

Anyway, "FET" is the super-class of devices:

• junction field-effect transistor -> JFET is member of FET class
• metal–oxide–semiconductor field-effect transistor -> MOSFET is member of FET class
• metal–nitride–oxide–semiconductor transistor -> MNOS is is member of FET class
• insulated-gate bipolar transistor -> IGBT is member of FET class

so, "FET" means nothing in this case, which one? -> it must be a typo, cannot be anything else 

[magic]

In Texas Instruments datasheets there is exactly one thing that "FET input" can possibly mean. Take it from the horse's mouth:


They know how to ride the audiophool hype train when they have an actual JFET product to sell:

There was this rumor i have no idea where i've heard it, might have been back in school so at least 10 years ago.. that TI had actually manufactured a metric crapton of wafers for the TL07x, LM13700 many years ago, they were not in production anymore and someday stock would have ended.
Yes it sounds a rumor from a guy whose uncle was working for TI but maybe.. maybe..

Well, the old TL072 is still for sale for now.

[2N3055]

In Texas Instruments datasheets there is exactly one thing that "FET input" can possibly mean. Take it from the horse's mouth:


They know how to ride the audiophool hype train when they have an actual JFET product to sell:

There was this rumor i have no idea where i've heard it, might have been back in school so at least 10 years ago.. that TI had actually manufactured a metric crapton of wafers for the TL07x, LM13700 many years ago, they were not in production anymore and someday stock would have ended.
Yes it sounds a rumor from a guy whose uncle was working for TI but maybe.. maybe..

Well, the old TL072 is still for sale for now.

Problem is that new "improved" one has almost order of magnitude more noise in audio range and going down to DC it shoots towards sky..
It has some better specs, it is faster, lower input bias current, etc.
It is different enough that it deserves different name : TLM072 

It might not be drop in replacement at all, depending of your circuit..

I'll stick to NJM072 ...

[magic]

Sadly JRC has nuked most of their jellybean products a year or two ago. I doubt that 072 made it through.

I suppose TI undercut them on price offering parts that more accurately follow the original 1970's specs
I found TI's jellybeans, including OP07, to have more flicker noise than competition. And I'm talking legitimate second source bipolar chips, no "next-generation" rubbish.

[Kleinstein]

The TL072H looks really a bit different than the normal one:  nearly more like an TLE2071.
There are some improvements in some aspects, but not so much for audio.  By todays standard the TL072 is no longer a good choice for audio - more like a low cost OP  that has input common mode range close to the positive supply.

EMI filtering is now quite common with AZ OPs. The ADA4522 has some specs on it.  It is not just the expensive one from Analog, but also cheap ones from Microchip. Chances are that without sich filtering the copper stabilized OPs may be quite sensitve to RF and with increasingly more phones and WLAN this can become an issue.

[magic]

It's not even remotely comparable to any classic JFET part and it smells a lot like CMOS.

Output saturation curve - straight line, starting from 0V at 0mA. Not only is it R2R, but a bipolar part would typically have some minimum saturation voltage and less output impedance when saturated.
Input offset vs common mode: discontinuous at mid-supply, like in R2R inputs but at a different threshold.
PSRR vs frequency: that one is weird too, equal on both sides.

[2N3055]

Sadly JRC has nuked most of their jellybean products a year or two ago. I doubt that 072 made it through.

I suppose TI undercut them on price offering parts that more accurately follow the original 1970's specs
I found TI's jellybeans, including OP07, to have more flicker noise than competition. And I'm talking legitimate second source bipolar chips, no "next-generation" rubbish.

NJR shows them as active product ..

https://www.njr.com/electronic_device/products/NJM072C.html

They have few thousands of them on Mouser..

[2N3055]

The TL072H looks really a bit different than the normal one:  nearly more like an TLE2071.
There are some improvements in some aspects, but not so much for audio.  By todays standard the TL072 is no longer a good choice for audio - more like a low cost OP  that has input common mode range close to the positive supply.

EMI filtering is now quite common with AZ OPs. The ADA4522 has some specs on it.  It is not just the expensive one from Analog, but also cheap ones from Microchip. Chances are that without sich filtering the copper stabilized OPs may be quite sensitve to RF and with increasingly more phones and WLAN this can become an issue.

There are still many devices in music world that use them. They are still good enough for guitar amps, even for a first preamp stage... Some pro bass amps (like MarkBass) use them in many models.
Even if you wanted to do better, only first stage would need to be a quieter amp... For tone shaping circuits they are all you need. 

[magic]

NJR shows them as active product ..

https://www.njr.com/electronic_device/products/NJM072C.html

Interesting, I will have to look at their lineup.

Note that it's a C suffix part. There is something weird going on with certain parts appearing with C suffix in parallel with old ones or after old ones have been discontinued (I can't find the old NJM072 anymore). I first noticed it with NJM5532, where the new variant in SO8 probably involved a die shrink or they would have offered a SO8 long ago. The C variants are only sold in SMD packages while the old ones in all the packages they have always been.

https://www.njr.com/electronic_device/products/NJM5532.html
https://www.njr.com/electronic_device/products/NJM5532C.html
https://www.njr.com/electronic_device/products/NJM4558.html
https://www.njr.com/electronic_device/products/NJM4558C.html

[2N3055]

NJR shows them as active product ..

https://www.njr.com/electronic_device/products/NJM072C.html

Interesting, I will have to look at their lineup.

Note that it's a C suffix part. There is something weird going on with certain parts appearing with C suffix in parallel with old ones or after old ones have been discontinued (I can't find the old NJM072 anymore). I first noticed it with NJM5532, where the new variant in SO8 probably involved a die shrink or they would have offered a SO8 long ago. The C variants are only sold in SMD packages while the old ones in all the packages they have always been.

https://www.njr.com/electronic_device/products/NJM5532.html
https://www.njr.com/electronic_device/products/NJM5532C.html
https://www.njr.com/electronic_device/products/NJM4558.html
https://www.njr.com/electronic_device/products/NJM4558C.html

They are not hiding NJM072C is modern redesign, but specs are very similar to standard TL072 and it is JFET for sure. That one I'm sure would be drop in replacement for classic TL072. In fact they usually have lower noise and are very good.

[magic]

Presumably there is nothing wrong with their new parts. But it's weird
One consequence is that you can no longer get a DIP 072 from them.

BTW, TI's could be called TLC072 in accordance with their old naming scheme.

[Kleinstein]

There is an TLC072 already  (looks more like an improved version of the TLC272).

The naming of the TL072 H is really confusing.  It looks more like a lower cost version of the OPA2991 / TLV9352.
They really should have given it a different name.

A rail to rail design with the cross over between the stages in the middle of the range is odd. That can cause quite some nasty surprised.

[magic]

Yep, I too went to their parametric search and these parts are about the only ones rated for 40V supply that they have.

I'm not really convinced that it's RRI because the discontinuity is less severe than in OPA2991 and there is some sign of trouble starting at 2V above the negative rail. But then I don't know what the mid-supply discontinuity is. Some X-Files stuff

It could simply be that they replaced the P-ch input stage with N-ch which is the reason for all the input-related differences (noise, offset) if the process was optimized for P-ch. And they sell it as 072 because it addresses the "I want something cheap with input range to Vcc" problem.

Still,

[magic]

JRC also plays the RF immunity game

https://www.njr.com/electronic_device/products/NJM8087.html

[T3sl4co1l]

It's not even remotely comparable to any classic JFET part and it smells a lot like CMOS.

Output saturation curve - straight line, starting from 0V at 0mA. Not only is it R2R, but a bipolar part would typically have some minimum saturation voltage and less output impedance when saturated.
Input offset vs common mode: discontinuous at mid-supply, like in R2R inputs but at a different threshold.
PSRR vs frequency: that one is weird too, equal on both sides.

Suspect the Vos vs. Vcm is an artifact of the measurement.  Wouldn't make any sense for that to happen if it's an RRI structure, how would it know what GND is?  It's certainly NOT an RRI type, as it doesn't include GND, but does include VCC as any other classic JFET input stage.

The output range is suspicious, and at the very least must be common-emitter, but could in fact be BiCMOS.  It's also PTC, i.e. the apparent Rds(on) increases with temperature.

Slew rate appears to be the boosted kind, with a high-slew bypass at high |Vin| and smooth settling otherwise.  Rates are somewhat asymmetrical, again suggesting a monopolar input stage.

Ed: oh also, the input bias current, is very clearly not a junction.  JFET gate leakage increases exponentially with temperature, from <pA at RT to ~nA at max.  The -H shows pretty flat ~pA, ticking up only a little at high temp.  That, plus the input noise being literally worse than a fucking TL431, is pretty damning.

Tim