Searching \ for '[EE:] Cheap Quad opamp. wanted with high Input bia' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/ios.htm?key=input
Search entire site for: 'Cheap Quad opamp. wanted with high Input bia'.

Exact match. Not showing close matches.
PICList Thread
'[EE:] Cheap Quad opamp. wanted with high Input bia'
2009\07\09@172725 by Richard Prosser

picon face
I'm looking for a cheap quad opamp with the same pinout as the LM324.
The killer is that I want one with a high input bias current. I'm
using a LM324 (LM224 actually to get the high temperature capability)
at the moment as a sensor amplifier and use the voltage rise caused by
the bias current to detect if the sensor goes open circuit.
Trouble is that some of the existing opamps have sufficiently low bias
current to balance the leakage current and don't hit the sensor fail
threshold. If I use a high speed opamp like the MC33274 it works fine,
every time but a) costs more and
b) is so much higher in transition frequency that I'm worried the RF
immunity will suffer. -  I'd probably have to get the item retested.
I can add additional pullup resistors but that would mean a pcb change
and also (probably) retesting. So I'm looking for a cheap quad opamp
with a transition frequency in the 1 MHz  area with a minimum input
bias current somewhere around 100nA or higher

Richard P

2009\07\09@183637 by olin piclist

face picon face
Richard Prosser wrote:
> I'm looking for a cheap quad opamp with the same pinout as the LM324.
> The killer is that I want one with a high input bias current.

That's really bad design.  You're not going to find a opamp that specifies a
minimum bias current.  If you need minimum current, put a resistor there.

> I can add additional pullup resistors but that would mean a pcb change
> and also (probably) retesting.

That's the way it goes.  That's the price of a design error getting this far
thru the process.  Either rework the existing boards or make new ones.
Maybe you can add the resistor elsewhere, but kludging this with a opamp
that appears to work from that production lot at your temperature is just
going to make things worse.

> So I'm looking for a cheap quad opamp ... with a minimum input
> bias current

Not gonna happen.  Go fix it the right way.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2009\07\09@210705 by Jesse Lackey

flavicon
face
I don't see any need for you to flame the poster.  He did not ask for
your design review opinions.  What you consider "design error" others
may consider cleverly getting a feature for free.  The poster may not
have even designed the circuit, and is instead trying to get more life
out of something 10 years old.  You have no context to make your
assertion that he has to either rework or redesign: is this a first
proto or are there 5000 boards already produced?  What is the failure
rate of the last 10,000 sold in the last 10 years?  Maybe this trick has
worked great.  You don't know.

Now I agree that this is a risky design, as is anything that counts on
an unspecified parameter that (in my limited knowledge of analog IC
circuitry) probably varies considerably with temperature and production run.

There may well be no reasonable replacement and he will have to
redesign.  I consider it sensible to try get more life out of this (or
any existing, field-proven) design, if possible, knowing the risks of
using parts in unspecified ways, than just junking it on the principle
that anything not designed to all specs should never leave the shop.
That is a business decision that every engineer and company gets to make
for themselves.

I would like PIClist to remain reasonably friendly for everyone to ask
questions, particularly somewhat unusual things like the OP did, without
getting flamed for doing a design (or trying to fix someone else's
design) that you dismiss out of hand as (apparently) so poor that they
need to be commanded to go fix, as though they are your junior employee.

Tone matters.  Even in engineering.

J






Olin Lathrop wrote:
{Quote hidden}

2009\07\09@213705 by Richard Prosser

picon face
Jesse,
Thanks for your support.

I've got a pretty thick skin, and pretty much agree with Olin anyway.
But I thought I'd ask the list as
a) I might get a solution
b) it's an example of getting an additional feature "for free"
c) It's an example of what can bite you when you try 2. above.


FWIW, I did the product design, borrowing heavily on previous designs
by others - including this feature. I think we've used this method for
about 10 years now without anyone noticing a problem.  It's not of
grave concern as the problem only arises if the sensor is disconnected
(which only happens if the cable is damaged or the sensor unplugged,)
but it's of enough concern to want to get it put right.
There's probably about 20k of these in the wild so far (2 years
production) and we would expect another 4-5years life at increased
production levels. Unfortunately we've just done a pcb update and
there are no other major  problems to fix that we know about.

Richard P


2009/7/10 Jesse Lackey <spam_OUTjsl-mlTakeThisOuTspamcelestialaudio.com>:
{Quote hidden}

>

2009\07\09@214317 by Bob Blick

face
flavicon
face
Jesse Lackey wrote:
> I don't see any need for you to flame the poster.

Let's try to keep it on topic here. This is an EE thread, feel free to
discuss the merits of the circuit.

The admin address is .....piclist-ownerKILLspamspam@spam@mit.edu if you think there's an
administrative issue, bring it up with the admins.

Please don't push the thread completely offtopic.

Thanks,

Bob

2009\07\09@225040 by SME

picon face
a) I might get a solution
> b) it's an example of getting an additional feature "for free"
> c) It's an example of what can bite you when you try 2. above.


Interestingly, the typical bias current and maximum offset current are
about 3:1 for the LM324/224 so the original method would be expected
to work often - as it indeed did. Alas, as you note, there is no
minimum bias current specified.

1. LM3900/2900 may be suitably different enough to prove better, or
worse, in practice.
Amplifiers from different manufacturers may prove better or worse.
Even date coded parts from an earlier era may do the job if the
problem has only occurred recently. A die shrink or process change may
have 'improved' the product and contributed to the previously
successful kluge now sometimes not working.

2. Don't try this at home / YMWV:    To add to the conniptions of the
prior commentator you could consider adding a 'blob' of a 'suitably
high resistance when set' compound at an appropriate point in the
circuit - depending on the number of channels being used for this
purpose and which amplifiers in the package were used and polarities
etc it could be between inverting and output or non inverting and
ground or ... .  Something with 10's to 100's of megohm/square at an
application thickness of semi-random and final characteristics
dependant on temperature, humidity, phase of moon and proximity of
deceased fish would add to the excitement and naughtiness of such a
solution. And it might even be able to be made to work reliably enough
to be useful. Or not.

3. Add on resistors would be less exciting. Just maybe these could be
suitably sized devices added at locations where there were not
intended to be resistors present. Such a solution approaches the
naughtiness of the prior one. Some fish may be harmed in its
execution.



                R

(This PICList account being used due to exigencies of current systems'
status)(statii?).

2009\07\10@073123 by olin piclist

face picon face
Jesse Lackey wrote:
> I don't see any need for you to flame the poster.

I didn't.  I said nothing about Richard, only that this particular idea was
bad design.

> He did not ask for your design review opinions.

Actually by posting here he essentially did.  If you don't want someone to
comment on what you say, don't say it to 2000 people.

> What you consider "design error" others
> may consider cleverly getting a feature for free.

True, but I stand by my assesment.

> The poster may not
> have even designed the circuit, and is instead trying to get more life
> out of something 10 years old.  You have no context to make your
> assertion that he has to either rework or redesign: is this a first
> proto or are there 5000 boards already produced?  What is the failure
> rate of the last 10,000 sold in the last 10 years?  Maybe this trick has
> worked great.  You don't know.

Since I wasn't judging the poster, only the idea, it doesn't matter who
designed the original circuit.  As for this trick working, apparently not,
else he wouldn't be here asking.  And even if it appears to work in a few
test cases, there is no guarantee that it will continue to work for any of
the next 1000 units.

> I would like PIClist to remain reasonably friendly for everyone to ask
> questions, particularly somewhat unusual things like the OP did, without
> getting flamed for doing a design (or trying to fix someone else's
> design) that you dismiss out of hand as (apparently) so poor that they
> need to be commanded to go fix, as though they are your junior employee.
>
> Tone matters.  Even in engineering.

Oh get over it.  Bad ideas are going to get trounced on and picked apart on
their technical merits.  This is a good thing.  If you disagree that the
idea was bad you can defend it on technical merits and pick apart the
criticism.  Anyone that can't handle this kind of discourse or gets offended
by it should probably find a different profession.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

More... (looser matching)
- Last day of these posts
- In 2009 , 2010 only
- Today
- New search...