LF357 substitude

[VanitarNordic]

it seems the LF357 is obsolete and can not be found easily except fake ones. it is a good opamp but it is not under production it seems.

I'm looking for a new substitute to introduce similar or better characteristics. it will be used as a high gain and sensitive amplifier in the circuit.

[Zero999]

What's your budget?

Does it need to be in a dual in line package (through hole)?

There's the AD825 and many others, but they aren't cheap and don't come in DIP.
http://www.analog.com/media/en/technical-documentation/data-sheets/AD825.pdf
https://www.digikey.co.uk/product-detail/en/analog-devices-inc/AD825ARZ-REEL7/AD825ARZ-REEL7CT-ND/4162145

[VanitarNordic]

What's your budget?

Does it need to be in a dual in line package (through hole)?

There's the AD825 and many others, but they aren't cheap and don't come in DIP.
http://www.analog.com/media/en/technical-documentation/data-sheets/AD825.pdf
https://www.digikey.co.uk/product-detail/en/analog-devices-inc/AD825ARZ-REEL7/AD825ARZ-REEL7CT-ND/4162145

It's Okay if it works well. I could find AD826 in the local market. budget limit is 5.0USD per unit.

I thought to use TL072 in a two-stage amplifier, but I have to use an inverting amplifier for both stages and the output of the second one is saturated. Maybe the gain is too high or naturally, it is not a good method.

What's important here is that the input signal has very low (few uV) variations and the Opamp should be sensitive enough to detect and amplify them, otherwise, just using a high gain opamp is not enough because it misses small changes (it introduces something like a threshold).

I don't know what term is used to determine this sensitivity for opamps but LF357 was sensitive enough for this purpose.

[Kleinstein]

We had that discussion before. So please give a link to the other thread and make clear what properties are really needed. As far as I understood from the other thread, the LF357 was not such a good choice anyway.  Using 2 amplifier stages instead of 1 with very higher gain eases the requirements on GBW quite a bit.

[VanitarNordic]

We had that discussion before. So please give a link to the other thread and make clear what properties are really needed. As far as I understood from the other thread, the LF357 was not such a good choice anyway.  Using 2 amplifier stages instead of 1 with very higher gain eases the requirements on GBW quite a bit.

Yes, I intentionally made this thread because the previous thread was going off-topic and this one should just discuss this. I mentioned what I'm looking for in the previous post.

I thought to use TL072 in a two-stage amplifier, but I have to use an inverting amplifier for both stages and the output of the second one is saturated. Maybe the gain is too high or naturally, it is not a good method.

What's important here is that the input signal has very low (few uV) variations and the Opamp should be sensitive enough to detect and amplify them, otherwise, just using a high gain opamp is not enough because it misses small changes (it introduces something like a threshold).

I don't know what term is used to determine this sensitivity for opamps but LF357 was sensitive enough for this purpose.

As I mentioned the high gain is not sufficient, but sensitivity to the input also matter. The LF357 was tested and was doing the job sufficiently but now I have to look for a substitute.

[David Hess]

The LF357 is a decompensated LF356 with a minimum stable gain of 5.  The LF355 is a low power LF356.  So what is needed is a 36/44 volt 20MHz 50V/us JFET input operational amplifier.  If offset null is required, then it needs to use pins 1 and 5 and go to the positive supply.

Linear Technology made improved replacements for the LF355 and LF356 but that is no help.

The LT1122 is very close and should be a drop in replacement with improved performance.  The AD744 could perform even better if it is externally decompensated however it would require changes in the offset null circuitry if used.

[VanitarNordic]

The LF357 is a decompensated LF356 with a minimum stable gain of 5.  The LF355 is a low power LF356.  So what is needed is a 36/44 volt 20MHz 50V/us JFET input operational amplifier.  If offset null is required, then it needs to use pins 1 and 5 and go to the positive supply.

Linear Technology made improved replacements for the LF355 and LF356 but that is no help.

The LT1122 is very close and should be a drop in replacement with improved performance.  The AD744 could perform even better if it is externally decompensated however it would require changes in the offset null circuitry if used.

offset null is not used in the main circuit. The replaced opamp should act the same or even better in terms of sensitivity and gain. the opamp has used as an inverting amplifier with a 1M resistor. High gain is not enough because the opamp should detect and amplify very small changes, otherwise it will neglect many of them and act as a threshold (I don't know the technical term for this input sensitivity). the LF357 was doing this quite well but now it is not manufactured anymore, it seems.

[Zero999]

The LF357 is a decompensated LF356 with a minimum stable gain of 5.  The LF355 is a low power LF356.  So what is needed is a 36/44 volt 20MHz 50V/us JFET input operational amplifier.  If offset null is required, then it needs to use pins 1 and 5 and go to the positive supply.

Linear Technology made improved replacements for the LF355 and LF356 but that is no help.

The LT1122 is very close and should be a drop in replacement with improved performance.  The AD744 could perform even better if it is externally decompensated however it would require changes in the offset null circuitry if used.

offset null is not used in the main circuit. The replaced opamp should act the same or even better in terms of sensitivity and gain. the opamp has used as an inverting amplifier with a 1M resistor. High gain is not enough because the opamp should detect and amplify very small changes, otherwise it will neglect many of them and act as a threshold (I don't know the technical term for this input sensitivity). the LF357 was doing this quite well but now it is not manufactured anymore, it seems.

A high gain is exactly what's required for amplifying small changes and a linear amplifier should have no threshold.

[VanitarNordic]

A high gain is exactly what's required for amplifying small changes and a linear amplifier should have no threshold

I have tested this in practice. configured an opamp for a high gain, it does the job, but it looks like it does not amplify all inputs and it misses the ones which are below a certain threshold. We can say it is not sensitive enough for those ranges.

[Zero999]

A high gain is exactly what's required for amplifying small changes and a linear amplifier should have no threshold

I have tested this in practice. configured an opamp for a high gain, it does the job, but it looks like it does not amplify all inputs and it misses the ones which are below a certain threshold. We can say it is not sensitive enough for those ranges.

What does it miss? It seems like something else is going on.

Could it be the bandwidth is too low? Have you plotted the gain vs frequency?

Going from the other thread the signal of interest is between 100Hz and 1kHz. Connect the output to an oscilloscope to monitor the amplitude and the input to a signal generator. Set the signal generator to a sine wave with the appropriate amplitude to give a clean sine wave on the output of the amplifier with no clipping and sweep the input frequency from 10Hz to 10kHz (I advise going in steps of 1, 1.5, 2, 3, 5, 7 i.e. 10Hz, 15Hz, 20Hz, 30Hz, 50Hz, 70Hz, 100Hz, 150Hz, 200Hz etc. to make life easier) and plot the amplitude. That will tell you whether your amplifier has the required bandwidth for the application.

[VanitarNordic]

A high gain is exactly what's required for amplifying small changes and a linear amplifier should have no threshold

I have tested this in practice. configured an opamp for a high gain, it does the job, but it looks like it does not amplify all inputs and it misses the ones which are below a certain threshold. We can say it is not sensitive enough for those ranges.

What does it miss? It seems like something else is going on.

Could it be the bandwidth is too low? Have you plotted the gain vs frequency?

Going from the other thread the signal of interest is between 100Hz and 1kHz. Connect the output to an oscilloscope to monitor the amplitude and the input to a signal generator. Set the signal generator to a sine wave with the appropriate amplitude to give a clean sine wave on the output of the amplifier with no clipping and sweep the input frequency from 10Hz to 10kHz (I advise going in steps of 1, 1.5, 2, 3, 5, 7 i.e. 10Hz, 15Hz, 20Hz, 30Hz, 50Hz, 70Hz, 100Hz, 150Hz, 200Hz etc. to make life easier) and plot the amplitude. That will tell you whether your amplifier has the required bandwidth for the application.

You are right, but small variations are not visible in the oscilloscope screen. of course, it is on the volt/division, then how few mV or even 10th of mV changes could be visible. That's why I say there is a sensitivity parameter which a non-suitable opamp might miss. The LF357 was sensitive enough for this purpose and it was tested.

By your definition, a TL071/081 can work instead but it does not in practice. it works but not as sensitive as LF357 (on the same gain setting).

[mzzj]

The LF357 is a decompensated LF356 with a minimum stable gain of 5.  The LF355 is a low power LF356.  So what is needed is a 36/44 volt 20MHz 50V/us JFET input operational amplifier.  If offset null is required, then it needs to use pins 1 and 5 and go to the positive supply.

Linear Technology made improved replacements for the LF355 and LF356 but that is no help.

The LT1122 is very close and should be a drop in replacement with improved performance.  The AD744 could perform even better if it is externally decompensated however it would require changes in the offset null circuitry if used.

LF357 is interesting and maybe bit unique in some aspects. Compared to AD744: input capacitance 3pf vs 5.5pf , input voltage noise 12nv/hz vs 22nv/hz and 20Mhz vs 12MHz GBW.
Overload recovery could be another issue..

[VanitarNordic]

LF357 is interesting and maybe bit unique in some aspects. Compared to AD744: input capacitance 3pf vs 5.5pf , input voltage noise 12nv/hz vs 22nv/hz and 20Mhz vs 12MHz GBW.
Overload recovery could be another issue..

Yes, you are right that's why I don't understand why it does not get manufactured anymore. More than 5 chips which I bought are fake 

I stuck in finding the best substitute. Somebody introduced these part numbers:

OPA627, OPA637, AD744, AD845, AD877, OP42, LT1122, TLE2071, TLE2081

[Kleinstein]

The LF357 is a rather old design. It's fast because it's de-compensated, but not that fast to really need it.

For most applications, there are suitable and likely better replacements. However it depends on the application because something like input capacitance, supply voltage and common mode voltage can be an issue.

From the old thread the application is rather low frequency and very high gain. So the high BW is not absolutely needed as an 2 stage amplifier would be a real alternative.  For a small signal level low noise might be more important than high GBW - quite some choices here if they can be slower.  No need to repeat the old discussion - things have not changes since last year. 

[StillTrying]

Just to add to all the confusion.

Hero999: "Going from the other thread the signal of interest is between 100Hz and 1kHz."

Kleinstein: "From the old thread the application is rather low frequency and very high gain. So the high BW is not absolutely needed"

From the dreaded other thread, the pulse repetition rate is only 100<1000Hz, but to the follow the shape of the pulses about 100kHz of BW is needed, according to my simulations there.

I don't think there's enough info as usual!

[mzzj]

Just to add to all the confusion.

Hero999: "Going from the other thread the signal of interest is between 100Hz and 1kHz."

Kleinstein: "From the old thread the application is rather low frequency and very high gain. So the high BW is not absolutely needed"

From the dreaded other thread, the pulse repetition rate is only 100<1000Hz, but to the follow the shape of the pulses about 100kHz of BW is needed, according to my simulations there.

I don't think there's enough info as usual!

The circuit on "the other thread" looks awfully lot like transimpedance amplifier... all this is pretty much useless guesswork without more info. 

[Zero999]

A high gain is exactly what's required for amplifying small changes and a linear amplifier should have no threshold

I have tested this in practice. configured an opamp for a high gain, it does the job, but it looks like it does not amplify all inputs and it misses the ones which are below a certain threshold. We can say it is not sensitive enough for those ranges.

What does it miss? It seems like something else is going on.

Could it be the bandwidth is too low? Have you plotted the gain vs frequency?

Going from the other thread the signal of interest is between 100Hz and 1kHz. Connect the output to an oscilloscope to monitor the amplitude and the input to a signal generator. Set the signal generator to a sine wave with the appropriate amplitude to give a clean sine wave on the output of the amplifier with no clipping and sweep the input frequency from 10Hz to 10kHz (I advise going in steps of 1, 1.5, 2, 3, 5, 7 i.e. 10Hz, 15Hz, 20Hz, 30Hz, 50Hz, 70Hz, 100Hz, 150Hz, 200Hz etc. to make life easier) and plot the amplitude. That will tell you whether your amplifier has the required bandwidth for the application.

You are right, but small variations are not visible in the oscilloscope screen. of course, it is on the volt/division, then how few mV or even 10th of mV changes could be visible. That's why I say there is a sensitivity parameter which a non-suitable opamp might miss. The LF357 was sensitive enough for this purpose and it was tested.

By your definition, a TL071/081 can work instead but it does not in practice. it works but not as sensitive as LF357 (on the same gain setting).

I repeat, it's nothing to do with sensitivity.

The TL071/81 most likely didn't work, because they don't have as higher bandwidth as the LF357.

LF357 is interesting and maybe bit unique in some aspects. Compared to AD744: input capacitance 3pf vs 5.5pf , input voltage noise 12nv/hz vs 22nv/hz and 20Mhz vs 12MHz GBW.
Overload recovery could be another issue..

Yes, you are right that's why I don't understand why it does not get manufactured anymore. More than 5 chips which I bought are fake 

I stuck in finding the best substitute. Somebody introduced these part numbers:

OPA627, OPA637, AD744, AD845, AD877, OP42, LT1122, TLE2071, TLE2081

The LF357 most likely stopped being made, because not enough people bought it. I was going to suggest the TL070, but it looks like that's also out of production.

What went wrong with using two TL072 stages?

[VanitarNordic]

What went wrong with using two TL072 stages?

The two stage amplification using TL072 was suffering from instability in the second output, either the signal was a flat line or ringing (when I was decreasing the gain) because the reference voltage is +12V, I must use inverting for both stages. I test this on the cadence PSpice with real parts, but in practice the results are different. if I could not find a good substitute, I'll try it on a new PCB.

[Zero999]

What went wrong with using two TL072 stages?

The two stage amplification using TL072 was suffering from instability in the second output, either the signal was a flat line or ringing (when I was decreasing the gain) because the reference voltage is +12V, I must use inverting for both stages. I test this on the cadence PSpice with real parts, but in practice the results are different. if I could not find a good substitute, I'll try it on a new PCB.

It sounds like a layout problem to me. It should be possible to use two separate gain stages to get the required gain and bandwidth. How this circuit is physically constructed can make a lot of difference to its performance.

Do you have a picture of the layout you used to test this? If so please post it.

[David Hess]

LF357 is interesting and maybe bit unique in some aspects. Compared to AD744: input capacitance 3pf vs 5.5pf , input voltage noise 12nv/hz vs 22nv/hz and 20Mhz vs 12MHz GBW.

The AD744 has a decompensation pin so in gain applications, its bandwidth and slew rate can be increased beyond that of the LF357.  It is an awfully expensive part though; see below.

Overload recovery could be another issue..

The OP will have to give more details to evaluate this.

I thought this was a replacement application but if it is a new design, then an LF357 or equivalent is not the best solution.  For instance a lower noise but slower JFET input operational amplifier operating with low gain and driving a fast current feedback amplifier would have better performance than a single LF357 operating at high gain.

[VanitarNordic]

I thought this was a replacement application but if it is a new design, then an LF357 or equivalent is not the best solution.  For instance, a lower noise but slower JFET input operational amplifier operating with low gain and driving a fast current feedback amplifier would have better performance than a single LF357 operating at high gain.

it is a new design, therefore I appreciate if you explain it a bit more about your design idea and suggestion. You mean two stage amplification with for example TL072, or it is something else?

[David Hess]

I thought this was a replacement application but if it is a new design, then an LF357 or equivalent is not the best solution.  For instance, a lower noise but slower JFET input operational amplifier operating with low gain and driving a fast current feedback amplifier would have better performance than a single LF357 operating at high gain.

it is a new design, therefore I appreciate if you explain it a bit more about your design idea and suggestion. You mean two stage amplification with for example TL072, or it is something else?

If one of the requirements is high input impedance, then it does not usually make sense to use two of the same high input impedance operational amplifiers in a cascade except to save cost and increase integration.  For instance a high impedance FET input operational amplifier can drive a low impedance bipolar operational amplifier.

Now designers do it all the time with the TL072 and TL074 but only because they are very inexpensive assuming that they have enough performance.