TL062/64 replacement

[MT]

What can replace a TL064/62 who have the same parameters and "low price" but with better specified output swing?
TL064 have terrible out swing at min +/-10V for 15volt supply. Its the combination of low price, low power consumption,  input current bias and output drive. Its very hard to find anything really, CMOS often fails due to poor capacitive drive. (TLC272 etc)

[Kleinstein]

It depends on the application. If the low price is important one might still use the TL06x. For low power requirements one tends to use lower voltages today - lots of CMOS OPs to chose from for a 5 V supply.

For higher performance the OPA170 might be an option, but the price is likely not that low.

[David Hess]

The TLC272 is one of the inexpensive replacements for the TL062 assuming that its lower maximum supply voltage is acceptable.  But low power, rail-to-rail output, and good capacitive drive capability are largely mutually exclusive so I need to know more about the application to make a determination.  Maybe there is some minor change that could be made to accommodate a replacement.

If you were using the TL061, then a pair of bipolar transistors could be added driven by the supply pins to make a rail-to-rail output although I am not sure about its capacitive load driving capability.

[MT]

Primary use is as low gain, low Bw signal buffers here and there, mux buffers, etc , i dont have any space for additional components, The whole design runs from +/-7.5V from that i need an absolute min of +/-5V out swing. But i would certainly not say no to full rail to rail.

Just saw that TL064IDT and TS274CDT at pack of 500 units have almost identical price yet both past 6months been upped by 80% by the usual oligopoly suspects, have to look for other sources.These two are hard to beat price wise.

TLC274 and TLC2L74 and TLC2274,LMC660 is out, way ludicrously expensive.

[Zero999]

If all that's required is 20mV or so to either, but not both supply rails, a current source/sink, or often just a resistor, between the output and the positive or negative rail will do. I've done this before and it worked quite well.

Don't expect it to work, using a simulator program, as the most SPICE models don't replicate the output stage correctly. The work around is to make your own model, using the schematic shown on the data sheet. See the following thread, which shows how it's done with the TL072.
https://www.eevblog.com/forum/projects/better-ltspice-tl072-model/msg999066/#msg999066

[David Hess]

If all that's required is 20mV or so to either, but not both supply rails, a current source/sink, or often just a resistor, between the output and the positive or negative rail will do. I've done this before and it worked quite well.

If the operational amplifiers are configured as voltage followers as the TL061/2/4 usually are in this sort of application, then this will not work because the input common mode range will be exceeded; a rail-to-rail input or extended input range operational amplifier will be required.

Primary use is as low gain, low Bw signal buffers here and there, mux buffers, etc , i dont have any space for additional components, The whole design runs from +/-7.5V from that i need an absolute min of +/-5V out swing. But i would certainly not say no to full rail to rail.

I think you are going to need to change the design to reduce or isolate the load capacitance in order to find a suitable but inexpensive part.  Is a low value series resistance on the output acceptable?  Or two resistors and a feedback capacitor if the output resistance must be kept low at DC?

[MT]

I think you are going to need to change the design to reduce or isolate the load capacitance in order to find a suitable but inexpensive part.  Is a low value series resistance on the output acceptable?  Or two resistors and a feedback capacitor if the output resistance must be kept low at DC?

Output load capacitance compensation could be added if 0603 yet PCB redesign will be required.

As you may notice the datasheet numbers do not correlate with figure1 as it points towards amp operational to at
least +/-2,5V , i know that many opamps can be driven lower then usually recommended with decreased performance.
Sometimes datasheets graciously even tell. Anyhow figure1 seams to indicate that i will be OK. (added green lines). Well, have to build some prototypes using various batches of 064/62 and do some measurements i presume.

Don't expect it to work, using a simulator program, as the most SPICE models don't replicate the output stage correctly. The work around is to make your own model, using the schematic shown on the data sheet. See the following thread, which shows how it's done with the TL072.
https://www.eevblog.com/forum/projects/better-ltspice-tl072-model/msg999066/#msg999066

I'm familiar with the common visits of incomplete spice models , once i had to do my own CA3080.
Did a quick spice buffer test just to see 5k,100n, TL061C (TI model) input swing is OK to a least +/-6.15V.

[Zero999]

If all that's required is 20mV or so to either, but not both supply rails, a current source/sink, or often just a resistor, between the output and the positive or negative rail will do. I've done this before and it worked quite well.

If the operational amplifiers are configured as voltage followers as the TL061/2/4 usually are in this sort of application, then this will not work because the input common mode range will be exceeded; a rail-to-rail input or extended input range operational amplifier will be required.

Of course, the input common mode range applies but the original poster didn't mention that.

A buffer is simple. Build an amplifier with a gain of two and bias it to the positive supply and connect a potential divider to the input, to divide the input by two.

[David Hess]

If all that's required is 20mV or so to either, but not both supply rails, a current source/sink, or often just a resistor, between the output and the positive or negative rail will do. I've done this before and it worked quite well.

If the operational amplifiers are configured as voltage followers as the TL061/2/4 usually are in this sort of application, then this will not work because the input common mode range will be exceeded; a rail-to-rail input or extended input range operational amplifier will be required.

Of course, the input common mode range applies but the original poster didn't mention that.

I had to intuit it from context.  The TL series of JFET amplifiers are commonly used to buffer sample and holds for the exact reason MT identified; they are the least expensive low input bias current operational amplifier available although there might be some competitive CMOS ones now.  The CMOS alternatives would be ideal except for price and that their common source output stages drive capacitive loads poorly.  Something in the requirements has to give and fixing the capacitive load problem is probably the simplest and least expensive.

A buffer is simple. Build an amplifier with a gain of two and bias it to the positive supply and connect a potential divider to the input, to divide the input by two.

I admit to doing that once in a power supply design to avoid the common mode input problem but besides adding offset and gain error terms, the input divider defeats low input bias current which is an absolute requirement.