Protecting Buffer Op Amp

[amptramp]

Ah, I think I get it now! Thanks guys!

I've been continuing to read about ESD protection, and have a new question: would a dedicated transient voltage suppression diode be inappropriate for this application? The 1N6270A (http://www.onsemi.com/pub_link/Collateral/1N6267A-D.PDF)
seems like it could work; it clamps at 9.1V. I can't find any capacitance information on the datasheet, however. In any case, it certainly looks like an interesting option.

ESD protection diodes usually have high junction capacitances (>1000 pF) except for the type with antiparallel silicon diodes in series.  These latter types are used to protect data lines.  The one you listed has no capacitance spec implying that it would be large.

[N9]

Maybe. What is the capacitance of the diodes? The datasheet should have it. The opamp input itself has some stray capacitance, also on the datasheet. The capacitive losses only count with the 2.2M resistor, as the opamp input is probably much higher.

The frequency rolloff of the input is then F = 1/(2*pi*2.2M * (sum of the capacitors)). Maybe it matters, maybe not. Have to do the math.

The maximum capacitance of a single 1N4148 is 4pF.
The datasheet for the OPA703 doesn't explicitly list the input capacitance in its tables, though I do notice that the input impedance is shown with the units of "Ω || pF". Is this the input capacitance? It's 4pF.

Assuming 4pF input capacitance:

F = 1/(2*pi*2.2M*20pF)

F = 3617.16

Does that seem right?

[R.G.]

The maximum capacitance of a single 1N4148 is 4pF.
The datasheet for the OPA703 doesn't explicitly list the input capacitance in its tables, though I do notice that the input impedance is shown with the units of "Ω || pF". Is this the input capacitance? It's 4pF.
...
F = 3617.16
Does that seem right?

Yes. However, in doing the numbers to check, I realized that I messed you up. Sorry - it wasn't intentional.

That is the correct rolloff if the input signal comes through the 2.2M resistor. I'm always harping on people to consider the unseen impedances, those being the source and load impedances. The reason to have a 2.2M is to have that be 10X or more bigger than the source impedance driving the input, so it won't load down the input.

A guitar pickup is on the order of 1-4H of inductance, and might typically be something like 100K at the 7kHz frequency that's where many of them peak before starting to roll off from internal capcitance. That's one reason 1M is such a common input impedance for pedals; it's 10X the pickup impedance, so produces negligible loading.

We can expect your unspecified pickup to have an output source impedance of about 100K to 200K. That says that the actual frequency cutoff from the 20pF loading is better by 10X to 20X in round numbers. So the cutoff we'd expect would be in the range of 37kHz to 70kHz, roughly.

Sorry for the inadvertent misdirection.

[N9]

Yes. However, in doing the numbers to check, I realized that I messed you up. Sorry - it wasn't intentional.

That is the correct rolloff if the input signal comes through the 2.2M resistor. I'm always harping on people to consider the unseen impedances, those being the source and load impedances. The reason to have a 2.2M is to have that be 10X or more bigger than the source impedance driving the input, so it won't load down the input.

A guitar pickup is on the order of 1-4H of inductance, and might typically be something like 100K at the 7kHz frequency that's where many of them peak before starting to roll off from internal capcitance. That's one reason 1M is such a common input impedance for pedals; it's 10X the pickup impedance, so produces negligible loading.

We can expect your unspecified pickup to have an output source impedance of about 100K to 200K. That says that the actual frequency cutoff from the 20pF loading is better by 10X to 20X in round numbers. So the cutoff we'd expect would be in the range of 37kHz to 70kHz, roughly.

Sorry for the inadvertent misdirection.

Ah, ok. No worries! It looks like we are fine in terms of frequency rolloff.

I was reading just now that capacitive loads on op amp inputs can also cause instability.

The math goes over my head, but this article calls for compensation when dealing with even small capacitance at the input (about 3/4 of the way down the page):
http://www.analog.com/library/analogDialogue/archives/38-06/capacitive_loading.html

The OPA703 has a GBW of 1MHz, which is fairly low, so it may be safe from this. I'm not sure.

[PRR]

> putting R4 before C3 make any difference?

No.

A case can be made that the 34K-47K ought to be "after" R3, and right AT the chip pin. As-drawn you have a 1:0.98 attenuator. Some nit-picks would decry the 0.2dB loss of signal.

Capacitance? If you have a cable or a *pickup* in the system, you already have hundreds of pFd. I agree you should know the diode pFds, but I bet they are tiny compared to cables or audio coils.

And I think you are over-fretting this protection. 34K in series requires 340 Volts to pass 10mA, which the chip is rated for in fault. >340V on stage is a hazard to many things (and people), not just a 5-buck chip.

[N9]

And I think you are over-fretting this protection.

Apologies if I seem too hung up on this. I'm treating this little project as a learning exercise for reliable op amp circuits, so I've kinda been taking things to their logical extreme and then seeing what's reasonable.

34K in series requires 340 Volts to pass 10mA, which the chip is rated for in fault. >340V on stage is a hazard to many things (and people), not just a 5-buck chip.

My concern is more with regard to ESD than a continuous current, though tolerance for ugly input lines is a bonus.
You're definitely right in saying that 34K of series resistance is good protection; I was just curious as to what other solutions there can be.

[PRR]

> reliable op amp circuits, ... seeing what's reasonable.

My un-reliable experience is: jacks go bad, pots go bad, once a solder-joint went bad (I was younger then). Also a 22V chip fed 34V will work for a decade and then quit.

Touring in hot humid summers and also overheated nylon-carpet labs, I can't recall ever killing a opamp chip.

[samhay]

^Apologies if I seem too hung up on this. I'm treating this little project as a learning exercise for reliable op amp circuits, so I've kinda been taking things to their logical extreme and then seeing what's reasonable.

This has made for an interesting discussion, so I wouldn't have thought an apology is necessary.

Two thoughts on making bomb-proof circuits: Socket your ICs, and think about adding a true bypass option (a toggle switch perhaps in this case) so that if something blows up and/or you lose power, you can still pass signal.

[Gus]

^Apologies if I seem too hung up on this. I'm treating this little project as a learning exercise for reliable op amp circuits, so I've kinda been taking things to their logical extreme and then seeing what's reasonable.

This has made for an interesting discussion, so I wouldn't have thought an apology is necessary.

Two thoughts on making bomb-proof circuits: Socket your ICs, and think about adding a true bypass option (a toggle switch perhaps in this case) so that if something blows up and/or you lose power, you can still pass signal.

If you want reliability don't socket your ICs solder them in.  I can often change a soldered in IC (like in effects 8 pin)in minutes most of the time is removing the board and heating up the iron.
Buy good switches and jacks. 
IMO toggle switches don't belong on effect boxes unless protected they will break.
Think about what happens when a heavy person stomps on the effect box will it hold up?
Think about what happens if soda or beer or ? get spilled on the box can you make the box fluid resistant?
Will the controls or jacks vibrate lose on stage?

Search the web and look in books for terms like stability and followers, emitter follower stability, opamp buffer stability 
EFs can become unstable look at the wha circuit the resistor in the collector leg of the EF part is for stability

[samhay]

Gus - IMO, there is a good chance that a socketed transistor or other 3 pin device will come lose if you shake the box around enough, but an 8 pin DIP in a decent socket should stay there for a long time. It's pretty easy to get a screwdriver out before a gig and re-socket something, but how many people carry a soldering iron around with them?
If you are worried about stomping on something you shouldn't, a toggle switch placed on the side of the enclosure next to one of the jacks will only break in-use after the cable and/or jack. At this point it is a moot point anyway.

[Gus]

samhay

Thermacycling moves chips out of sockets(cold car to warm place etc) pins corrode in sockets, toggles break when stuffed in a box for transport

If you are selling something you need to think worse case, if it is for yourself then you know what not to do

I built, adjusted some things for pro player friends I had to use lockwashers and thread lock to keep the jacks and potentiometers from getting lose(the stage volume was very high)

People can seem to find a way to break something you might not have thought  about or thought  made a difference
Maybe in the studio a person might want to plug an effect in to a preamp with 48VDC phantom etc.  you will need to protect the output in this case

People etch enclosures, how much weaker does the enclosure get? will it collapse with a person stomping on the switch?

[greaser_au]

Gus - IMO, there is a good chance that a socketed transistor or other 3 pin device will come lose if you shake the box around enough, but an 8 pin DIP in a decent socket should stay there for a long time. It's pretty easy to get a screwdriver out before a gig and re-socket something, but how many people carry a soldering iron around with them?
If you are worried about stomping on something you shouldn't, a toggle switch placed on the side of the enclosure next to one of the jacks will only break in-use after the cable and/or jack. At this point it is a moot point anyway.

If your'e worried about a socketed device coming out, tie it down with lacing cord (with a wipe of nail polish/varnish/superglue on the knot).   That said, it is more than once  I have fixed equipment in the field by lifting & reseating all of the socketed devices... 
david

[N9]

Yeah, there are a lot of mechanical issues to consider in stompbox construction/design.

I use machined IC sockets; they retain chips quite well, and being able to easily remove the IC is a big bonus for DIY projects.
That said, I would probably secure devices into the sockets in some way if the effect needed to go on tour. I like the lacing cord idea!

The issue of 48V phantom power being connected to the inputs/outputs is something I hadn't fully considered.
Wouldn't you need large input/output resistors and beefy diodes to fully protect against a 48V supply? Perhaps a crowbar circuit would be better?

[samhay]

^The issue of 48V phantom power being connected to the inputs/outputs is something I hadn't fully considered.
Wouldn't you need large input/output resistors and beefy diodes to fully protect against a 48V supply?

If you have non-polarised input and output capacitors that are rated for >48V, then this shouldn't be fatal should it?

[PRR]

> The issue of 48V phantom

Which is why I like to plan for >100V. 10mA blow-current in 34K is 340V, OK. You could go 10K; for guitar, 20K-40k is also useful for cutting radio reception without adding more capacitance to the already cappy guitar load. As Sam says, you need 50V NP caps to be sure. (The cap alone is not enough: you can't get a steady 10mA but you will get a surge limited only by resistance.)

That said..... Assuming even a little (and sometimes no) protection around chips, the main gig-stoppers are mechanical. Stuff falls aparts, or doesn't stay-together good-enough to keep contact.

[merlinb]

Thought I'd add this here:
I notice this TI application note
http://www.ti.com/lit/ug/tidu034/tidu034.pdf?DCMP=hpa-pa-opamp-thehub&HQS=hpa-pa-opamp-thehub-20140318-tidesigns-en
uses an SMBJ15CA bidirectional transient voltage suppressor for input protection.
https://www.jameco.com/Jameco/Products/ProdDS/1542338.pdf

[PRR]

Interesting that a TI "industrial products applications engineer" didn't see the advantage in putting D1 after the 100 Ohm resistor.

[amptramp]

Interesting that a TI "industrial products applications engineer" didn't see the advantage in putting D1 after the 100 Ohm resistor.

The text of the app note says he was using thick film 1206 1/4 watt resistors that would stand a good chance of arc-over and may even be vaporized by a significant voltage hit.  Think of a zener set for 15 volts and someone hits it with 50 volts.  The 100 ohm resistor would see source - zener voltage and the current would be this over 100 for a 100 ohm resistor.  This would be 0.35 A at 35 volts or 12.25 watts.  There would be a smoking hole in the board if the transient lasted very long.  However much the hit on the zener would be reduced with series resistance, in a lot of cases, it is better to but the zener in harm's way.