1. Home
  2. Electronics Design
  3. Favourite op amp for driving ADCs?
  4. Page 2

Favourite op amp for driving ADCs?

I'm planning on making a fair number of them in due course, like several thousand over the lifetime of the product. The AD8605 is almost good enough, for $1.50 in quantity 25. The OPA365 is definitely good enough, at about the same price, but has a 5.5V VDD abs max, which is a bit tight with a +5/-0.3 V supply range.

In general, I like doing good measurements, and there's an elegance in doing really good measurements with absolutely minimal apparatus that appeals to me. (*) This box has about 300 parts, though, so minimalist it ain't. Once you get out of being able to rely on the near-ideal performance of the parts, as in the original noise canceller papers, and have to start tweaking a few terms of each of several series expansions, life gets more complicated fairly fast.

Cheers

Phil Hobbs

(*) As the old jokey office poster says, "We the unwilling, led by the incompetent, have been doing so much, with so little, for so long, that we are now qualified to anything with nothing."

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs
Loading thread data ...

o
e

for you?

isn't the ad8615 almost the same but with twice the drive and half the sett ling time?

-Lasse

Reply to
Lasse Langwadt Christensen

Thanks. I'd be sort of worried about its accuracy though--A_VOL is only

40 dB, and it has a really nasty underdamped characteristic at unity gain.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

Why would you be worried about exceeding the "absolute max" if you are designing within the spec? The absolute max is not something that blows the chip away if exceeded... well, maybe it is, lol. But really, what is your concern, that the PSU will exceed the rated Vdd max or that being close to it can have consequences? I think the typical rating of a chip is +-10% on Vdd meaning it is ok for the PSU to be anywhere in that range. Is the Vdd for the specified parameters over a range smaller than +-10%? I just can't imagine the chip having a problem with a 5.6 V transient.

--

Rick
Reply to
rickman

When you're trying to start an instruments company, it's not very smart to be courting field failures by pushing abs max ratings like that. If I have to, I'll put in an extra diode.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

5.6V is above abs. max, so there is no guarantee that it won't violently explode 5V + 0.3 is close to abs max 5.5V and you have to take the accuracy of the 5V and the 0.3V diode into account

-Lasse

Reply to
Lasse Langwadt Christensen

"Pushing"? Don't you think the manufacturer has provided adequate tolerance in the part so that you can use the full range specified in the data sheet? Or are you worried that your tolerances will put the Vdd outside of the spec?

I tend to design conservatively, but I am happy trusting the manufacturers to spec their parts accurately.

--

Rick
Reply to
rickman

Absolute maximum ratings don't have an implied tolerance. They're, well, _absolutes_. And if they did, wouldn't you want to design for all the parts, including those whose maximum ratings were in the -10% end of the specs?

If you look at the failure statistics, you'd see a failure rate that starts rising _below_ the abs. max, and rising very rapidly in the vicinity of the abs. max. The exact shape and intercept of the curve have become more abrupt - less forgiving - as semi vendors push their processes to go ever faster using many more devices of ever thinner layers that once broken do not heal.

Reply to
Frank Miles

If all your 5V regulators are within 4% over time and temperature, and you're perfectly sure there won't be any overshoots due to line transients or stuff like that, you could argue that way.

Doesn't appeal to me.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

I think you are over stating the case. Data sheets need to be read with an eye to what they are trying to tell you. I can guarantee that this device will not fail at Vdd = 5.6 volts. There would have to be more margin built into the design for you to operate it at 5.5 volts which

*is* within the stated parameters.

All of the device characteristics are specified...

"ELECTRICAL CHARACTERISTICS: VS = +2.2V to +5.5V "

First page of the data sheet "2.2V TO 5.5V OPERATION"

Someone was being very conservative when they specified this part for

5.5 volt max. I would like to hear from TI that the part can fail early if run at Vdd-Vss of anything under 5.5 volts.

If they are producing units that will fail at Vdd = 5.6 volts I can assure you that many will fail in normal operation.

--

Rick
Reply to
rickman

Yes, but you can be overly conservative too. I've never seen a regulator I would consider using that wasn't within 4%. Otherwise it wouldn't make it into my designs. As to line transients, they will always be a hazard. You need to design for that or the circuit can fail even if you use the part at exactly 5 volts. If you are going to operate at 5.3 volts then you need to take that into account with the accuracy of the PDS.

As I said in another post you are treating 5.5 volts as if it were the actual failure limit which it is not. The manufacturer is adding margin to the failure limit so that you don't need to. The actual failure point is likely above 6 volts or more. It would be instructive to test one and find out.

--

Rick
Reply to
rickman

The manufacturer adds margin to allow for variations in their *own* processes. That means they think you'll eventually get one that is right on the specified limit. Professionals would not rely on that not happening to them.

Reply to
Clifford Heath

This is getting tiresome. They won't be pushing the actual spec at all. So running the part at 5.3 volts *which is in spec* should be safe.

--

Rick
Reply to
rickman

"In spec" - does that help with things like low failure rates and long lifetimes? I seems to remember a military reliability estimate that made great use of how close to abs max a part was used in calculating reliability.

I rather like the homespun "my car engine can go to 6000rpm and be "in spec" but that doesn't mean it a good idea to drive everywhere in 2nd gear"

Reply to
David Eather

Do you really think the 5.5 volt abs max is the real abs max? No, it is not even close. That is the point. Running my engine next to redline all the time would have eventually destroyed the thing long before the

208k miles it has on it currently. But running a 5 volt part at 5.3 volts had better not have an impact on it's lifespan or the maker is crap because it would have failed not much longer at the 5 volts it was intended to run.

The mistake everyone is making is thinking that the 5.5 volt figure is the value to use in such a comparison. The destructive breakdown is nowhere near the 5.5 volt spec in the data sheet even for that 1 in

10,000 part. The value to use in the reliability estimate is the *actual* breakdown voltage, not a figure plugged into a data sheet for convenience. Do you really think their chip design is such that 5.5 volts is just enough inside the actual limit? No, that is just a number they can give out that no one will complain about.

Running a 5.5 volt abs max part at 5.3 volts had better be acceptable and not impact the life of the part. Otherwise there are a lot of designs without adequate control of the Vdd which are going to fail and

*this* chip is going to look like the culprit. 
--

Rick
Reply to
rickman

You'd have to test all of them, because things like gate breakdown have a production spread. And while you talk about conservative design, you seem to think that the hairy edge of failure is a desirable neighbourhood, whereas I don't.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

Is this a real spec sheet (if so, which one?), or something that you imagined? Sample of 1 - I grabbed a spec sheet from my pile (OPA379). On the first page it says "Wide supply voltage range: +1.8 to +5.5V"; and on the absolute max ratings it gives 7V.

I probably am being conservative. But then, I like things that I design to be reliable (my customers like that too). Some devices may well tolerate voltages and currents at or slightly above abs max ratings. But fewer (I believe) as financial and competitive pressures force manufacturers into specsmanship games. I don't believe that current absolute limit numbers are conservative - at least when it comes to low-voltage semiconductor products. Worse, I'm not sure how we can resolve this question based on real data rather than conjecture.

Reply to
Frank Miles

Now you are being silly. We simply disagree on what constitutes the "hairy edge of failure". It is only rational that you can't have it both ways. If 5.3 volts is too close to abs max for a long device lifetime then 5 volts is not a lot better. That's my point. I don't believe that because someone put 5.5 volts on a data sheet that it means

5.3 volts is a lot worse than 5 volts. 
--

Rick
Reply to
rickman

This was from the spec sheet of the part we are discussing, the OPA365 which the OP says running at 5.3 volts is too close to it's abs max of

5.5 volts.

You are distorting what I am saying. I have *never* said a design should exceed the abs max. We are talking about using a part at 5.3 volts when the abs max is 5.5 volts. I'm saying that this is within the spec and should *not* shorten the lifetime of the part.

--

Rick
Reply to
rickman

Shouldn't, but does. "In spec" doesn't guarantee an infinite life. Everyone with experience learns to design conservatively - you're the one being silly.

Reply to
David Eather

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.