FET input op amps

I am often asked about op amp based gain make up circuits for my passive EQs and I always suggest the TL072 because it has a FET input so it will not load the EQ which prefers a high impedance load.  However. it is so long since I designed with a FET op amp that the TL072  is the only one I remember and I know there must be much better alternatives out there that could drive 600 ohm loads directly or via a transformer.

The OPA2604 has been mentioned as one possibility but Farnell no longer stock it and they suggest the AD823 for new designs.

I know nothing about current FET op amps for audio work so could somebody suggest a short list of candidates that are readily available and not too expensive?

Cheers

Ian

You will find a few fans of the OPA2134 here. It drives 600R no problem and, correctly implemented, you will need a pretty serious analyser to see any nasties from it. It is quite hungry power-wise, so be warned. It is fast-ish @ 20mV and can oscillate without power supply decoupling and ancillary stability parts. From memory, Douglas Self rates it above the 2604.

As a straight swap-out for an 072 the OP275 works pretty well. It's generally quieter and swings nearer its rail limits (well, that's not difficult compared to an 072).  Mr Self is not a fan of the OP275 (nor do I think Samuel is). I tend to think of it as a slightly nicer 072.

edit - although I use them in spades, I haven't tried the Op275 into 600R. I tend to go for the 2134 when heavy lifting is needed.

Should point out that in the late 70's, early 80's, the pukka Broadcast Desk makers, Calrec, Neve, SSL used TL072 when low current (compared to 5532) was needed rather than HiZ.

OPA2134 doesn't quite have this advantage but has other good points.  Probably the least fussy to decoupling of the uber OPAs so it often outperforms better spec'd stuf in real life.  Also VERY good RFI immunity.

Of the new FET OPAs which I haven't tried, OPA140 family looks best on paper with low current.  It doesn't have the EVIL latching behaviour in TL07x and OP275 

But make up gain on a passive EQ doesn't really demand uber performance so practically any OPA will do.  Look up what TI suggest for a (probably) SMD TL07x replacement.

TLE2072

OPA134 has been my goto FET opamp for a "modern" TL071 replacement.

I haven't had any problems driving 600ohm transformer either and never had any issues using them in a circuit.

They are a bit costly though. I think I can buy them at $3.50 each.

The OPA1642 is not a 'cheap' amplifier,  but it is very clean, and seems to behave well, even in a non-inverting application. If one has to use a FET input amplifier for audio in a non-inverting stage, this would be my first choice today. Any superior alternatives would be welcomed...

Most of the time, an inverter would work much better, and often, this foregoes the need for a FET input. However, there are times where a non-inverting stage is required, and the OPA1642 fits that bill. It's not 100% clear, but from what I have read, the OPA1642 uses dielectric isolation to avoid the typical problems of common mode signal induced, input pair drain-to-substrate capacitance modulation that can cause relatively large amounts of distortion. So, this chip seems to be the best performer right now, in my estimation.

Again, I'd be grateful to learn of any other chips that can provide this level of linearity, along with this level (or better) of freedom from excess distortion when operated as a non-inverting stage, but this is the best that I can find now.

I have used the OPA132 series to good effect, and while they are vastly nicer than the ancient TL07x series, they are not as clean as the OPA1642. The OPA134 series is essentially the same AC-wise as the 132, but with relaxed DC specs. In general, FET input opamps are tough to design so that they are still clean as a non-inverting stage.

OPA827 would be another to check out of the newer FET designs, but have not tried it yet, or the OPA1641 / 1642. Another would be ADA4627-1. All these are SMD only.

For DIP packages:
OPA134  / 2134
LT1057
TLE2071 / 2072

What is the actual source impedance?

While most FET opamps indeed have higher input resistance than a typical bipolar part, I suspect this is of no relevance here. The NE5532/5534 has a typical differential input resistance of 100 kOhm. This value is bootstrapped by the loop gain. Even if we apply a conservative 40 dB, we end up at 10 MOhm. In most audio applications, this is negligible loading.

Samuel

Samuel Groner said:

What is the actual source impedance?

While most FET opamps indeed have higher input resistance than a typical bipolar part, I suspect this is of no relevance here. The NE5532/5534 has a typical differential input resistance of 100 kOhm. This value is bootstrapped by the loop gain. Even if we apply a conservative 40 dB, we end up at 10 MOhm. In most audio applications, this is negligible loading.

Samuel

Click to expand...

As it is a passive EQ, the source impedance varies over quite a wide range. What is important is the load seen by the EQ which needs to be high enough not to distort the frequency response. A. load resistance of not less than 470K is preferable.

Cheers

Ian

Looks like the OPA134 and siblings are well favoured and at least the data sheet actually list 600 ohms as a load. The TLE2072 seems a good choice too as it is intended to replace the TL072 and has better noise and slew rate plus it wilt drive a 600 ohm load.

Many thanks for the recommendations.

Cheers

Ian

ruffrecords said:

As it is a passive EQ, the source impedance varies over quite a wide range. What is important is the load seen by the EQ which needs to be high enough not to distort the frequency response. A. load resistance of not less than 470K is preferable.

Click to expand...

Sam is right.  Even 553x with its huge input current is happy with 1M input resistor in this application ... as will loadsa NJM chips.

I take it you aren't insisting on DC coupling & other Golden Pinnae fetishes  :

What you DO need to be careful about is some OPAs will latch on start up with HiZ on their input and at other times too.

The most EVIL are Self's beloved LME497x0 / LM4562 family.
http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=641&p=7427#p7427

Thankfully, TI have discontinued them in their pogrom on NS stuff  ;D

Though they are touted by some for LN, they are actually useless for this cos popcorn noise
https://e2e.ti.com/support/amplifiers/audio_amplifiers/f/6/t/415907

My only concern with the NE5532 would be output offset. With 200nA  input bias current, a 470K bias resistor and a gain of 10, I calculate the output offset voltage would be nearly 1V.

By the way, I have tin ears.

Cheers

Ian

ricardo said:

ruffrecords said:

As it is a passive EQ, the source impedance varies over quite a wide range. What is important is the load seen by the EQ which needs to be high enough not to distort the frequency response. A. load resistance of not less than 470K is preferable.

Click to expand...

Sam is right.  Even 553x with its huge input current is happy with 1M input resistor in this application ... as will loadsa NJM chips.

I take it you aren't insisting on DC coupling & other Golden Pinnae fetishes  :

What you DO need to be careful about is some OPAs will latch on start up with HiZ on their input and at other times too.

The most EVIL are Self's beloved LME497x0 / LM4562 family.
http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=641&p=7427#p7427

Thankfully, TI have discontinued them in their pogrom on NS stuff  ;D

Though they are touted by some for LN, they are actually useless for this cos popcorn noise
https://e2e.ti.com/support/amplifiers/audio_amplifiers/f/6/t/415907

Click to expand...

Why would anyone knock the 4562's behaviour when fed with a Hi-Z source? Surely, this is akin to criticising a hacksaw because it can't drill a hole?

thermionic said:

ricardo said:

ruffrecords said:

As it is a passive EQ, the source impedance varies over quite a wide range. What is important is the load seen by the EQ which needs to be high enough not to distort the frequency response. A. load resistance of not less than 470K is preferable.

Click to expand...

Sam is right.  Even 553x with its huge input current is happy with 1M input resistor in this application ... as will loadsa NJM chips.

I take it you aren't insisting on DC coupling & other Golden Pinnae fetishes  :

What you DO need to be careful about is some OPAs will latch on start up with HiZ on their input and at other times too.

The most EVIL are Self's beloved LME497x0 / LM4562 family.
http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=641&p=7427#p7427

Thankfully, TI have discontinued them in their pogrom on NS stuff  ;D

Though they are touted by some for LN, they are actually useless for this cos popcorn noise
https://e2e.ti.com/support/amplifiers/audio_amplifiers/f/6/t/415907

Click to expand...

Why would anyone knock the 4562's behaviour when fed with a Hi-Z source? Surely, this is akin to criticising a hacksaw because it can't drill a hole?

Click to expand...

Ricardo is not a fan of input bias cancellation circuitry, I do not disagree in principle while I have little hands on experience with modern uber op amps.

On paper several look quite good, making picking one winner difficult.

JR

thermionic said:

Why would anyone knock the 4562's behaviour when fed with a Hi-Z source? Surely, this is akin to criticising a hacksaw because it can't drill a hole?

Click to expand...

First I'm very specific about '4562'.  It is the LM4562 that is EVIL.  NJM4562 has at least as good noise though not the LM's supa THD performance.  The NJM dun latch or have popcorn noise.

When LM4562 latches, usually with a design optimised for HiZ sources (though I first came across this in a LoZ design), your hacksaw is of use only as a very poor quality hammer .. if that even.

Secondly, there are 2 reasons for lusting after LM4562 like Self & others ... Low Noise & supa THD.  I rarely need THD at the theoretical potential of LM4562.  When I do, other less specified OPAs often give better THD in real life

eg http://www.groupdiy.com/index.php?topic=37307.80  from about #41 to the end

For LoZ LN where you'd expect LM4562 to shine, the popcorn noise makes it totally unacceptable IMHO.

Please look up the links I posted earlier as well as

http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=423&start=150

CONCLUSION : bleh for LN unless LF noise is unimportant.  Sorta OK for supa THD provided you take as much care as Kingston on decoupling, earthing bla bla

PS : sorry Ian for hijacking your thread for a rant  :

ruffrecords said:

My only concern with the NE5532 would be output offset. With 200nA  input bias current, a 470K bias resistor and a gain of 10, I calculate the output offset voltage would be nearly 1V.

Click to expand...

I was guessing that as this make up amp would live among a forest of glowing bottles, you'd use a simple single ended supply (perhaps derived from your DC heater) and capacitor coupling.  A volt offset would be inconsequential  ;D

By the way, I have tin ears.

Click to expand...

Let's start the Church of the Pinnae Stannum ...  far superior to dem deaf Pinnae Aurum crowd  8)

ricardo said:

ruffrecords said:

My only concern with the NE5532 would be output offset. With 200nA  input bias current, a 470K bias resistor and a gain of 10, I calculate the output offset voltage would be nearly 1V.

Click to expand...

I was guessing that as this make up amp would live among a forest of glowing bottles, you'd use a simple single ended supply (perhaps derived from your DC heater) and capacitor coupling.  A volt offset would be inconsequential  ;D

By the way, I have tin ears.

Click to expand...

Let's start the Church of the Pinnae Stannum ...  far superior to dem deaf Pinnae Aurum crowd  8)

Click to expand...

I knew studying Latin would be good for something but how do you say phool in latin?

JR

ricardo said:

I was guessing that as this make up amp would live among a forest of glowing bottles, you'd use a simple single ended supply (perhaps derived from your DC heater) and capacitor coupling.  A volt offset would be inconsequential  ;D

Click to expand...

As I think I mentioned in the first post, this is more for other people building my passive EQ who want to use a semiconductor gain make up stage instead of a tube one. I always use a tube!

By the way, I have tin ears.

Click to expand...

Let's start the Church of the Pinnae Stannum ...  far superior to dem deaf Pinnae Aurum crowd  8)
[/quote]

The ear is like any tool. it is not how good it is that counts, but what you do with it. Despite having tin ears, I still manage to pick out things in recorded tracks that others have missed, often for many years.

Cheers

Ian

As you are no doubt aware, the OPAx134 op amps are now only produced as a surface mount package. Recently however, I purchased some of the 4134 DIL versions from eBay. They were described as original new Burr Brown. However they did come from China .... I haven't tested them yet. I am upgrading the balanced inputs of my Perreaux power amps.

aomahana said:

As you are no doubt aware, the OPAx134 op amps are now only produced as a surface mount package. Recently however, I purchased some of the 4134 DIL versions from eBay. They were described as original new Burr Brown. However they did come from China .... I haven't tested them yet. I am upgrading the balanced inputs of my Perreaux power amps.

Click to expand...

I did not know that. Looks like DIP is on the way out. farnell still have quite a few. perhaps time to make a lifetime purchase:

http://uk.farnell.com/texas-instruments/opa2134pa/op-amp-dual-audio-fet-i-p-2134/dp/1097574

Cheers

Ian