LM358 too slow!

[ian field]

Something I want to build with an OP-amp needs the "ground sensing" inputs
like the 358 but needs power bandwidth up to about 50kHz, and preferably
better O/P drive capability - any suggestions?

TIA.

 

[Walter Harley]

Something I want to build with an OP-amp needs the "ground sensing" inputs
like the 358 but needs power bandwidth up to about 50kHz, and preferably
better O/P drive capability - any suggestions?

"Ground sensing"? I think what you mean is that the input common mode range
extends to the negative supply rail (that is, to "ground" in a single supply
scenario), and the output also can reach the negative rail. That's what the
manufacturers typically mean when they call something a "single supply"
opamp.

There are plenty of single-supply opamps out there that can do this, as well
as plenty of "rail to rail" opamps that can reach both the negative and the
positive supply. Just use the parametric search functions for the web site
of whatever manufacturer you're fond of.

If you go for a rail-to-rail opamp as opposed to a single-supply opamp, be
aware that they can sometimes be less stable than "traditional" opamps. If
sensing to the negative rail is all you need, you're probably better off
with single-supply.

 

[ian field]

Thanks for the reply

"Ground sensing"? I think what you mean is that the input common mode
range extends to the negative supply rail (that is, to "ground" in a
single supply scenario), and the output also can reach the negative rail.
That's what the manufacturers typically mean when they call something a
"single supply" opamp.

"ground sensing" is quoted directly from one of the '358 datasheets I've
studied.

There are plenty of single-supply opamps out there that can do this, as
well as plenty of "rail to rail" opamps that can reach both the negative
and the positive supply. Just use the parametric search functions for the
web site of whatever manufacturer you're fond of.

Rail to rail might be handy as I want to drive a squarewave into the gate of
a small MOSFET and it needs to run off a 6V SLA - looking at the '358 O/P
stage I doubt it will perform as well as I'd like, who cares what
manufacturer as long as I get a part that does what I want.

If you go for a rail-to-rail opamp as opposed to a single-supply opamp, be
aware that they can sometimes be less stable than "traditional" opamps.
If sensing to the negative rail is all you need, you're probably better
off with single-supply.

It's to drive a squarewave and in this application a little ringing isn't
that serious a problem - if it rings less than the discrete design I'm using
I'll be happy!

A fast comparator with rail to rail O/P might do, but so far I've not found
anything suitable in the catalogues I have to hand.

 

[Lostgallifreyan]

It's to drive a squarewave and in this application a little ringing
isn't that serious a problem - if it rings less than the discrete
design I'm using I'll be happy!

A fast comparator with rail to rail O/P might do, but so far I've not
found anything suitable in the catalogues I have to hand.

I've used a CA3140 with positive feedback to make sharp rise/fall times.
Fast and accurate enough for feeding into a PLL XOR comparator input in the
audio range, anyway. I chose the CA3140 because it will swing fully to zero
to allow single supply, use low current input, and be fast.

 

[ian field]

I've used a CA3140 with positive feedback to make sharp rise/fall times.
Fast and accurate enough for feeding into a PLL XOR comparator input in
the
audio range, anyway. I chose the CA3140 because it will swing fully to
zero
to allow single supply, use low current input, and be fast.

Can it deliver rail to rail O/P to the gate capacitance of a small MOSFET
with single supply down to 5.4V at frequencies upto about 50kHz ?

 

[Lostgallifreyan]

Can it deliver rail to rail O/P to the gate capacitance of a small
MOSFET with single supply down to 5.4V at frequencies upto about 50kHz
?

No idea, it's cheaper to try one if you have one around, than to agonise
over it, probably. It has 10 mA output drive, a 9V/µs slew rate, and is
rated for over 3.5 MHz use, so even at reduced supply voltage it ought to
get by at ultrasonic frequencies. The positive feedback will accelerate the
slew to overcome momentary lows while charging a capacitance.

It can't do rail to rail, but that wasn't what you asked, at first. It can
get within 0.6V though. Whether this is enough I don't know, it's probably
best to try unless there is a specific reason not to.

 

[ian field]

No idea, it's cheaper to try one if you have one around, than to agonise
over it, probably. It has 10 mA output drive, a 9V/µs slew rate, and is
rated for over 3.5 MHz use, so even at reduced supply voltage it ought to
get by at ultrasonic frequencies. The positive feedback will accelerate
the
slew to overcome momentary lows while charging a capacitance.

It can't do rail to rail, but that wasn't what you asked, at first. It can
get within 0.6V though. Whether this is enough I don't know, it's probably
best to try unless there is a specific reason not to.

Maybe I'll have a look in my chip tin see if I have one to try, otherwise
I'm not even going to bother checking if Maplin carry them until they send
me another promotion leaflet with money off vouchers. The discrete design
isn't so bad but I ran out of E-line ZTX transistors and TO92 equivalents
take more space!

 

[Paul]

Maybe I'll have a look in my chip tin see if I have one to try, otherwise
I'm not even going to bother checking if Maplin carry them until they send
me another promotion leaflet with money off vouchers. The discrete design
isn't so bad but I ran out of E-line ZTX transistors and TO92 equivalents
take more space!

Don't buy anything from Maplins. My broadband started to go slow and my
ISP said I needed a new filter. Maplins said I needed new cable. I
bought new cable but the broadband speeded up again - must have been
BT. So I took the cable back. But because it had been machine packed
there is no way I would return it "in the same condition". The cable
WAS "in the same condition". But the idiots said the packing had to be
the same -- impossible!

If they want to get things back "correctly" they should use packing it
is possible to pack.

Every other company I've taken anything back to, in living memory has
accepted it back with full refund.

Except Maplins.

Don't even think of shopping with them.

 

[ian field]

Maybe I'll have a look in my chip tin see if I have one to try, otherwise
I'm not even going to bother checking if Maplin carry them until they send
me another promotion leaflet with money off vouchers. The discrete design
isn't so bad but I ran out of E-line ZTX transistors and TO92 equivalents
take more space!

Don't buy anything from Maplins. My broadband started to go slow and my
ISP said I needed a new filter. Maplins said I needed new cable. I
bought new cable but the broadband speeded up again - must have been
BT. So I took the cable back. But because it had been machine packed
there is no way I would return it "in the same condition". The cable
WAS "in the same condition". But the idiots said the packing had to be
the same -- impossible!

If they want to get things back "correctly" they should use packing it
is possible to pack.

Every other company I've taken anything back to, in living memory has
accepted it back with full refund.

Except Maplins.

Don't even think of shopping with them.

Why didn't you take the matter up with trading standards?

 

[ian field]

I've used a CA3140 with positive feedback to make sharp rise/fall times.
Fast and accurate enough for feeding into a PLL XOR comparator input in
the
audio range, anyway. I chose the CA3140 because it will swing fully to
zero
to allow single supply, use low current input, and be fast.

Having had time to seek out the data sheets, the '3130 looks a much better
choice. Maplin don't list it, I suppose now I want some they're obsolete!

 

[Lostgallifreyan]

Having had time to seek out the data sheets, the '3130 looks a much
better choice. Maplin don't list it, I suppose now I want some they're
obsolete!

I hope not! I should look at them more, myself. These two IC's in both
single or dual form, are very standard devices, it's worth placing an order
for a few. Although they won't be ideal for you in this case, check out the
LF411/LF412. As a drop-in equivalent to the 417, I find them amazing. Very
good general purpose, and in some cases better than more expensive
dedicated audio types. If I haven't got any, I make it a priority to get
some.

 

[Lostgallifreyan]

Meant replacement for 741..

 

[ian field]

I hope not! I should look at them more, myself. These two IC's in both
single or dual form, are very standard devices, it's worth placing an
order
for a few. Although they won't be ideal for you in this case, check out
the
LF411/LF412. As a drop-in equivalent to the 417, I find them amazing. Very
good general purpose, and in some cases better than more expensive
dedicated audio types. If I haven't got any, I make it a priority to get
some.

Thanks - always open to suggestions, I'll have a look for the sheets for
those as well.

 

[Walter Harley]

Rail to rail might be handy as I want to drive a squarewave into the gate
of a small MOSFET and it needs to run off a 6V SLA [...]

A fast comparator with rail to rail O/P might do, but so far I've not
found anything suitable in the catalogues I have to hand.

How about a plain ol' LM393 dual comparator (or the quad LM339)? Cheap as
dirt, supply voltage is 2V to 36V, supply current is low, it should be fast
enough for your 50kHz requirement without being so insanely fast that you
can't keep it from oscillating, input goes to *below* ground if you want,
and output is an open collector, so it'll saturate down to 0.3V or so.

If you're running an H-bridge rather than a single MOSFET it might not be
ideal, because you won't have equal charge and discharge currents so your on
and off times will be different and you'll get shoot-through. But for a
single MOSFET that shouldn't be an issue.