Single Supply TL072 Op Amp -- Strange biasing

[jim_griff]

Hi all.

I'm currently learning how to 'properly' use Op Amps, but have found a strange way to get this TL072 working from a single supply by biasing the non-inverting inputs to around 783mV when run from a 9V supply. When run from a 13.8V supply, the bias voltage is slightly higher (about 850mV, I believe), which requires changing out the bias resistors.

I know it's very doubtful, but I'm wondering if anyone has seen Op Amps used in this way before? I understand that temperature will be a serious issue with this design, as the bias point is extremely sensitive to voltage changes. Just 5mV either way and it goes into saturation/cutoff.

(See attached image for schematic and photo.)

[Rerouter]

take a good look at your datasheet, the common mode range doesnt start until 3V from negative, on a TL072, your gain would not likely even be uniform at the 700mV

[poptones]

I'm surprised it's not latching up. You can't leave opamp inputs floating like that and expect the circuit to work reliably.

[poorchava]

I know that some people avoid bipolar power supply by biasinjg opamp inputs at VCC/2 and coupling signal to input capacitively. But that only works for AC signals. I remember that in some project where i needed am opamp to output like 26V, i just powered the TL074 (identical, but quad) from assymetric +30V/-5V supply and it worked as expected.

[kg4arn]

"ops amps for everyone" book (early editions are freely downloadable) will show you how to use op amps with a single supply.  Really there is no difference between dual and single supply operation but you tend to forget that the op amps negative rail is still its negative rail (even though you call it ground).  So what was said above about the common mode input range is very true.  And when you use a "single supply" you tend to reference the input to the negative rail (which you now call ground) when the op amp likes the input referenced to the midpoint bewtween the rails.  It's very easy to give the op amp an input that you are calling 1 V above ground, but in reality is 1V above the negative rail and the op amp is not designed to handle an input that close to the rail.

It's all very sematically tricky but the electronics is exactly the same whether you use a dual supply with a built in gorund reference midway between the rails, or when you use a single supply.

Op amps that are marketed for "single supply operation" are just optimized so that their Inputs and outputs can get close to the postive and negative rails and usually the common mode ranges are are optimized for this as well.  They are also optimized to run on narrower rail to rail voltage supply ranges.  You have to carefully read and understand the data sheet specs for the particular op amp you use. 

[T4P]

The TL072 would be okay within it's Vicr
But apparently if i remember clearly it can only swing within 2V of either rail (might be wrong)
If biasing it within 700mV the opamp starts acting weird

[poptones]

I doubt you'd want to use a tl072 for a phono preamp - the noise voltage is 18nv/hz. it'll cost you like 10db in s/n ratio over something like a 5532.

[T4P]

TL072 is great when you need a cheap JFET input opamp, for it's high input Z
Other than that the NE5532 is unreplaceable, quiet and cheap plus high output current capability
Only thing is NE5532's input Z is quite low

[T4P]

Just be careful where you use the NE5532, it's input Z is low as in 30k-300k nominal input Z vs 300k-1M on the cheap and no good for audio NE4558
But yeah .... why bother 

[T4P]

Not much difference i believe but i do have issues with some Signetics NE5532N
The only working NE5532 i have is a TI one NE5532P
Apparently my 7 NE5532N oscillates madly no matter which amp i put into
( TDA2050, LM3886, LM1875, Discrete 100W amplifier )

Okay, so i just went to widlarize 1 of them whos pins got broken off
And something doesn't seem right to me ...

[ciccio]

Are you sure it is working?
An op-amp with the inverting input unconnected and no feedback from output to inverting output input cannot work!!
I repeat: it cannot work. There is nothing to keep its dc bias point, nothing to fix its gain, nothing.
Period. Look for a suitable schematic.  Best regards

[poptones]

Well if it sounds good to you then have fun... but... you really might want to do some more research. I'm willing to wager that thing has no more than 60db s/n and you're loading the cartridge like crazy with that passive eq network. That is supposed to go AFTER the first gain stage. You should have a 5532 with 20db gain, then your eq network, then another 20db gain stage.

[Mechatrommer]

and opamp with no feedback to its inverting input cannot work. at least if you want high gain output, put high ratio resistor divider on the output and feedback to inv.inp. the chance it worked the first time is just pure (beginner's) luck, its just a matter of time before its stochastically drifting mad, just as you have/will experience.

[poptones]

You should find anything written by Walt Jung or Richard Marsh and digest it. National makes available on their website a free download of a $60.00 book by Jung, here's a link.

http://www.analog.com/library/analogDialogue/archives/39-05/Op_Amp_Applications.zip

There's example designs in there including active and passive riaa eq phono amps. When I was in my teens, National's op amp applications handbook was my bible; when I got a bit older (well, I was still a teen) I bought a copy of Jung's "Audio IC Op Amp Applications" which I wore to tatters. Using what I learned from National's application handbook I designed and built a phono preamp that was considered excellent even in some very high end salons around Ann Arbor. There are newer devices, but the basics in these books haven't changed a bit.

Here's another article, written by Marsh and Jung back when I was a teen. I was an Audio subscriber back then and I don't know how many times I read through these issues. This was all pretty new and somewhat controversial stuff when it first was published; now, 35 years later (yeesh), it's withstood the test of time and you can still find the articles reprinted online. Man, the internet is great.

http://waltjung.org/PDFs/Picking_Capacitors_1.pdf
http://waltjung.org/PDFs/Picking_Capacitors_2.pdf

[poptones]

Well, there ya go! You might want to change those 10K input caps to something closer to 50K.

I'm guessing you measured the noise output with a shorted input? Referenced to 5mV, the max S/N ratio you're going to get with a 600 Ohm source is going to be in the high 70's.

[poptones]

No, I mistyped. I dunno why I said caps as I was obviously talking about resistors.

Different cartridges respond to different loads. Some phono preamps have switched resistors and caps across the input to allow customization, but I really think that's straining at gnats.

[poptones]

If you're using 5534s for the input stage and you're not using a moving coil pickup there's really no need to invest in anything lower noise. The Johnson noise of a 600 ohm resistor (ie the coils of your MM cartridge) is going to be about 440nV and the noise contribution of your 5534 is going to be about 500nV - so, you see, even with a "perfect" amplifier on the input you could only improve the S/N another 3db or so. Now, if you're using a MC cartridge, which puts out about a tenth that of the MM design, you'll want to go to something like paralleled low noise FETs. Otherwise, enjoy!

[GK]

Hi

1M/100k is waaaaaayyy to large for the 1st gain stage feedback resistors. This will entirely defeat the low noise of the NE5532. Use 10k/1k instead.
 

[poptones]

That's what the bias and comp pins are for.

If you'd use  a 5532 you wouldn't have this problem.

[poptones]

Replace the feedback network with 10K|1K resistors and put a 100uF cap between the bottom of the 1K resistor and gnd.

[Mechatrommer]

is there reason why you need 100Kohm noninverting input impedance and cannot be reduced to 10K or 1Kohm? how much source impedance from I1? is that 100Kohm pull down before the opamp really necessary? is it part of the filtering circuit before the opamp? (after the I1)

[poptones]

Standard load impedance for a moving magnet phono cartridge is 47K.

[GK]

Hi

1M/100k is waaaaaayyy to large for the 1st gain stage feedback resistors. This will entirely defeat the low noise of the NE5532. Use 10k/1k instead.

I can't use anything smaller, because then I would need to pull up the non-inverting input back to ground with a much lower value resistor. It charges to -0.5V and beyond, which alters the bias and causes tonnes of distortion.

100k/10k means I have to use a 10k from non-inverting input to ground, else it causes the input to go into saturation. 10k,1k would mean I'd have to pull it to ground with a 1k resistor. This defeats the low impedance input. The NE5534 seems a bit crap, as it does not compensate the bias internally, as it claims...

EDIT 2: Also, if I use the schematic example in the datasheet, it tends to fart at loud volume when any audio signal level is put into it. I've tried capacitor values all the way from 3.3pF to 1000uF capacitor (which is overkill), and it just farts at different frequencies. Either all of my NE5534's are dead, or there's some sort of electronics magic I've not been told about haha. Well, I give up for now. Will do some more reading and see what books say.

You can terminate the non-inverting input to ground with a 47k resistor and delete the pot thing. The usual trick is to put a large value electrolytic capacitor in series with the Ri (100k in your current circuit) feedback resistor to give the gain stage a DC gain of unity, so that it does not amplify its input offset voltage.
Also, the 1nF comp cap you have added will massively de-compensate the NE5532 (which is already operating at a noise gain of 11) and cause heaps of unnecessary high frequency distortion.  It is also generally a bad idea to stick capacitors to ground directly at op-amp outputs. It degrades their phase margin and can cause oscillation and/or ruin transient response.
Although it doesn't matter right now as the dominant noise source is your 1M//100k feedback resistors, that 560 ohms in series with the input doesn't do anything but add noise. Also, the 10uF coupling capacitor may be large enough to give an edequately low, low-frequency roll off, but it is an unnecessary contributor to low frequency noise. At 20 Hz 10uF has a reactance of ~800 ohms.

[poptones]

You really don't want .5db at 20K in an audio circuit. Similarly, the 10uF cap is adding quite a lot of lf phase shift. But direct coupling the 5534 input to the cartridge is going to apply that input bias and offset across your cartridge, not a good thing.

I didn't even notice the 100pf caps on the outputs. The image is so big I hadn't really scrolled around on it much. A 5534 is internally compensated for gains above 3, you shouldn't really need any caps at all in that regard - and you definitely want to avoid those cheap ceramic caps in an audio circuit.

[SeanB]

I will add just one thing, in that there must be supply decoupling on both the positive and negative rails right by the opamp, using both a 10-100uF electrolytic and a 0.1-0.22uF ceramic. A lot of NE5534's can oscillate at around 1MHz with poor decoupling, and can make quite a good AM transmitter while still mostly working with only minor distortion on the output.