# linearly decreasing gain with frequency

Discussion in 'General Electronics Chat' started by fez, Dec 24, 2010.

1. ### fez Thread Starter Member

Dec 6, 2009
47
1
I have looked around at a couple of datasheets for some audio-based op amps. They seem to have a linearly decreasing gain with (logarithmic) frequency over the human hearing range. And it puzzles me a bit, because isn't (good) gain supposed to stay at a constant value for all frequencies of concern? I looked at several such op amps and they have similar curves. So am I missing/forgetting something here? Is it good to have linearly-decreasing gain over frequency? Why?

(e.g. I checked AD823, LM4562, OPA2134)

2. ### beenthere Retired Moderator

Apr 20, 2004
15,815
283
Which figure in the data sheet indicates this? The OPA134 has am 8 MHz gain-bandwidth product, and a slew rate to 20 volts/μS. That would mean a gain of up to 400 at the upper audio frequency of 20 KHz. The response would be linear from DC to 20 KHz.

3. ### #12 Expert

Nov 30, 2010
16,704
7,354
You're supposed to use negative feedback instead of running the chip at 120 db of gain. That solves the problem very nicely.

4. ### fez Thread Starter Member

Dec 6, 2009
47
1
Beenthere: Um, I was looking at the open loop gain plot, on the top left of page 5 of the 2134's datasheet.

#12: I seem to have forgotten my op amp terminology. Are you by any chance referring to the closed loop gain plot just to the right of the open loop gain plot? (Incase you are, note that the closed loop gain plot doesn't show frequencies underneath 1k. Whereas I need to see atleast 20Hz.)

5. ### Jony130 AAC Fanatic!

Feb 17, 2009
3,993
1,116
Everything is alright
Look at this example

R1 = 1K, R2 = 99K
Voltage gain is equal
Av = 1 + 99/1 = 100[V/V]

So close loop gain will look like this ( click on the picture to zoom )

And Fc= 3MHz/100 = 30KHz

fez likes this.
6. ### marshallf3 Well-Known Member

Jul 26, 2010
2,358
201
Properly biased, and with the right negative feedback, even inexpensive good op amps can be extremely linear in this range.

Take a look at some of the simple examples in here: http://focus.ti.com/lit/ds/symlink/tl082.pdf

And to beat all you can buy this chip at (ugh) Radio Shack.

Maxim produces quite a few that are even better and they'll sample for free.

[EDIT:] Darn, I hadn't even opened up my last package of samples from them and I haven't a clue why I ordered them.

I found qty 4 each of MAX2014ETA+, MAX4000EUA+, MAX4105ESA+ and MAX4305ESA+

Now I've got to look them up to try and figure out what they are and why I ordered them.

[Double Edit:] OK, now I know why but that project is a long ways out.

Last edited: Dec 24, 2010
fez likes this.
7. ### fez Thread Starter Member

Dec 6, 2009
47
1
Yes, that! WHY/HOW is it okay? I would expect the gain to be a straight horizontal line, ideally. This one starts to decrease linearly from around ~10Hz. Which means that high frequencies will be amplified less.

8. ### fez Thread Starter Member

Dec 6, 2009
47
1
I saw the pdf. Seems you're right. Will definitely look into this opamp and others by maxim soon.

By the way, do you have to pay any shipping charges for the samples? I do not reside in the US and am planning to order some samples soon anyway, thought I'd ask about the shipping.

9. ### marshallf3 Well-Known Member

Jul 26, 2010
2,358
201
Maxim doesn't charge a dime to ship samples to the US, unsure about other countries. Just make sure to set up a proper business account name with them. You don't need to open an account but you need to look like you're serious about what you're doing. They won't check up on your business name as they realize a lot of people are just starting out.

I love the older TL08x series of ICs, far too easy and forgiving to design with. The single channel ones end up with extra pins for offset voltage but there's no problem just using an TL082 and tying the unused side's inputs to ground if not using the other half.

You achieve the flat response by the negative feedback and in some cases adding a cap into that will flatten it even more to your liking. In some occasions you'll also need a cap back to ground but experimentation is what it's all about.

Certainly inexpensive enough, but you also need to learn about being nimble with the manufacturers, most of who will provide free samples for all but their more expensive devices. Items in common DIP packages are often limited so if you have to go SOIC get the largest pitch available, just makes it easier.

If you do go to surface mount hit up a good supplier such as Mouser or Jameco and get a syringe of paste solder with the flux included. If you apply it to your board properly (only takes a tiny dot) a good hot air gun (not a hair dryer) will make the connections as will just getting a good hot iron very near to the pins. Some people even put them in a toaster oven but I won't go that route unless a lot are involved in which case the SMD components go on first then the rest can be put on by hand.

10. ### Kermit2 AAC Fanatic!

Feb 5, 2010
3,852
968
Do you really think an audio amp should have such a high gain? A 0.1 volt input would result in a 400 volt output or some other equally outrageous number.?

Open loop gain is just for comparison BETWEEN different op amps.

In circuit they are set to practical amounts of gain. Like 2, or 10, in which range they will have a flat freq response way past the human hearing range.

11. ### fez Thread Starter Member

Dec 6, 2009
47
1
Marshallf3: I looked at the pdf again. Once again, on page 11, there is a linearly decreasing gain, starting from ~20Hz. Now, while I do require a linear gain, I also require it to be HORIZONTAL/flat. You say I can achieve that by negative feedback. How can I know that without buying and testing it? Because there is no gain-frequency plot in the datasheet of the negative feedback mode.

And what exactly do you mean by "cap"?

------

Kermit2: Yes, I don't need 400x gain, 10x should easily suffice. But I need the gain to be constant across the hearing range. How does one find out/know from a datasheet that the gain will be constant, when the opamp is in negative feedback? Because I can't recall any gain-frequency plots for negative feedback, the datasheets all show open loop gain.

While we're at it, how does the open loop gain plot serve us for comparison when its the negative feedback mode we actually intend to use?

Lastly, are you, perhaps, saying that the gain is reduced so much in negative feedback (as compared to the open loop gain) that it is ALMOST straight over the frequency range? That is, it is still linearly decreasing, but the slope is so small in magnitude that the line is nearly horizontal?

12. ### marshallf3 Well-Known Member

Jul 26, 2010
2,358
201
You're looking at open lloop gain again.

Cap = capacitor = something you might need to use in parallel with the negative feedback resistor or from the negative input pin to ground if you're seeking absolute precision.

Thousands of op amp circuits both here and on the internet, they're almost always used in even the most high end audio preamps so you know that they come out as flat in normal design. If JBL hadn't dropped all their schematics for their car amps I'd show you a simple circuit that's perfectly flat but they took them away and I only have copies on a PC that isn't on my network at the time.

Just look around.

13. ### #12 Expert

Nov 30, 2010
16,704
7,354
I don't think anyone said this in this thread: op-amps work all the way down to DC. The fact that some graphs do not show the gain at DC does not mean op-amps don't work there.

Apr 20, 2004
15,815
283
See post #2:

15. ### #12 Expert

Nov 30, 2010
16,704
7,354
OK. Just pointing out that the graphs don't show everything. Learning to read a datasheet is like an exercise in detective work.

16. ### Kermit2 AAC Fanatic!

Feb 5, 2010
3,852
968

If that is the only way you can believe something has flat response is to INSIST that it is not flat. then by all means believe it.

So to answer your question. Op amps with High open loop gain, will have ALMOST flat response from the low to high end of human hearing and beyond.

For every step up in gain, you lose more of the high end response(a gain of 10 might have flat response to 200KHz, and a gain of 1000 only have flat response to 10KHz) When you get up to open loop gain, the freq response starts to fall with any increase in freq from DC.

17. ### marshallf3 Well-Known Member

Jul 26, 2010
2,358
201
http://www.rason.org/Projects/opamps/opamps.htm
http://sound.westhost.com/project88.htm
http://www.beis.de/Elektronik/AudioMeasure/AudioPreamplifier.html
http://www.shine7.com/audio/bpre.htm

That'll keep you busy enough for a while but as you can see most of these are flat from 20 Hz to well over 100 KHz.
It doesn't require the fancy op amps they specify either, so long as you keep with a TL082 or better you'll be fine.
Just don't make the mistake of trying to use anything older, a 741 starts having trouble at even a few Khz.

18. ### Papabravo Expert

Feb 24, 2006
10,340
1,850
The other way to think about the difference between open loop and closed loop is that you are trading off gain for bandwidth. If you multiply a gain of 1e5 by a bandwidth of 10 Hz. you get a Gain Bandwidth product of 1e6. Now in a closed loop system you drop the gain to 10 and you get a bandwidth of 100 kHz. which is far past the upper end of the audio range. Now that gain of 10 will be flat out to some frequency before it starts rolling off and goes to 3 dB down at 100 kHz.

19. ### Ron H AAC Fanatic!

Apr 14, 2005
7,050
657
Fez, Google gain-bandwidth product. Study several of the hits. They will explain how you can get flat closed-loop gain from linearly decreasing open loop gain.

Apr 5, 2008
15,805
2,389
Hello,

You can also read the attached application note from TI about opamp specifications.

Bertus

File size:
273.2 KB
Views:
162