Amplify the hell out of it.

At that sort of gain level you have to start being really careful about power supply rejection, particularly at higher frequencies - I once built a circuit that could be cranked up to a gain of 10,000 at up to a couple of MHz, and figured out that I needed to to put RLC filters on each power pin, so that the current drawn by the output stage wouldn't feed back ripple onto the power rails that could leak through into the output of the input stage.

My calculated supply filters worked fine, and the final circuit was stable for every gain up to 10,000, though the outpty was a bit noisy at high gains - I did have to make one modification to the printed circuit, but that was to deal with the unspecified input capacitance of the Texas Instruments TLC2201, which was high enough to call for a compensating capacitor in parallel with the feedback resistor.

You've also got to be a bit careful with the layout - a little bit of capacitative or inductive coupling between input and output can be fatal, but an inch or so of physical separation is plenty if the relevant tracks run over more or less solid ground plane.

---------- Bill Sloman, Nijmegen

Reply to
bill.sloman
Loading thread data ...

See "instrumentation amplifier". You can roll your own from op-amps or buy a ready made one. Many off the shelf ones can be adjusted to gains in the many k range. Also note what others have said about bandwidth, etc.

Chris

Reply to
kmaryan

Some of the biggies:

Unity gain stable opamps have an open loop gain that rolls of proportional to frequency above a few hertz. So a closed loop gain of

1000 has 1/1000th as high a maximum frequency as a closed loop gain of 1 amplifier has.

Guarding, shielding, star grounding and other means to keep the output, power supply and other signal sources out of the input signals become a lot more important.

Very high gain implies very small input signals. Control of noise from all components becomes important.

--
John Popelish
Reply to
John Popelish

The most noticeable "issue", as you call it, will be the correspondingly lower cut-off frequency.

Reply to
dB

Instrumentation amps are fairly specialized with speed/noise/whatever capabilities being very important tradeoffs. If you want a good example of a high gain application of an instrumentation amplifier, google for EKG, ECG, electrocardiogram. There are plenty of examples online, most of which are simple enough that they can be executed by those just starting out with electronics. Typically, these see differential input waveforms in the range of 0.5-5mV, and use some combination of instrumentation amps and conventional amps to get that into the 1V range for a ADC. Interestingly, these usually don't use a single high gain stage, but rather a series of modest gain and filtering stages that usually work out to a bandpass frequency response from 0.1Hz-50Hz and a gain of 1000 or so. Some even use a unity gain first stage instrumetnation amp just to get the differential signal changed to single ended, then do the rest with successive normal op-amp stages.

The real questions you have to ask yourself to design this thing are "what is the signal that I'm looking at?" The main considerations being bandwidth, signal strength, is it differential or single ended, is there a significant common mode voltage? There may even be some aspect of what you are trying to do that requires special consideration, i.e. clean ECG measurements usually require some kind of feedback to deal with common mode problems. Then you ask yourself "what do I want to do with this signal?", what are your noise requirements? is it going into a ADC? if so, how many bits of precision? etc. Try asking some of the answers to these questions, then perhaps the solution may become clear.

Chris

Reply to
kmaryan

Greg,

In addition to opamp gain, noise will be a mayor factor. Especially noise on the ground lead. Look at

formatting link
for a discussion.

Dieter

Reply to
Dieter Knollman

I know how to build the basic inverting and non-inverting op-amp amplifiers, and they seem to work well enough for reasonable gains like in the 10's. But what if I want to amplify a small signal by a factor of 1000 or so? It seems like there must be a lot of issues swept aside in the intro books that will rear up-- non-linearities, zeroing, I don't know what else.

Any advice?

--
"Tell me, Dr. Einstein, at what time does Boston arrive at this train?"
Reply to
Gregory L. Hansen

I read in sci.electronics.design that Gregory L. Hansen wrote (in ) about 'Amplify the hell out of it.', on Thu, 17 Feb 2005:

It isn't normally a good idea to try to get 1000 times gain out of one op-amp. A cascade of two with gains of 31.46, or any reasonably comparable combination that multiplies to 1000, is a much better solution and solves many of the problems you mention (and some others) automatically.

--
Regards, John Woodgate, OOO - Own Opinions Only. 
The good news is that nothing is compulsory.
 Click to see the full signature
Reply to
John Woodgate

The more gain you need, the better you should know the signal, the impedance, the offset, the noise. For reasonably narrow bandwidth, a gain of few 10k is doable. Not necessarily with one amplifier alone.

Rene

--
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net
Reply to
Rene Tschaggelar

Yes. If it had a gain bandwidth product of 1 MHz, it would have had an open loop gain of 100 at 10 kHz, so its gain would be accurate to about 10% amplifying 10 kHz. At 100 Hz, it has an open loop gain of about 10,000, so it will provide about 10% accurate closed loop gain of 1000 at that lower frequency. Lots of applications need the open loop gain to be 100 times the closed loop gain, so that the feedback controls the gain to within 1%. So high gain implies a need for a relative faster opamp.

--
John Popelish
Reply to
John Popelish

$10 each from Digikey.

formatting link
?Ref=258766&Row=201592&Site=US

Unless the device is far enough from the first stage that there is quite a bit of common mode (in both lines) hum pick up, there is probably no good reason to use an instrumentation amplifier. A two stage opamp design (low noise front end, close to the source) with AC coupling or some other way to remove the DC offset will probably work just fine. The second stage can be a simple active filter with gain.

--
John Popelish
Reply to
John Popelish

I'm not sure I understand this. Do mean that, just to pull a number out of my butt, if the op-amp will satisfatorily amplify a 10 KHz signal with a gain of 10, it will satisfactorily amplify a 100 Hz signal with a gain of 1000?

--
"Very well, he replied, I allow you cow's dung in place of human 
excrement; bake your bread on that." -- Ezekiel 4:15
Reply to
Gregory L. Hansen

Oddly enough, that almost pre-answers another question I had.

My first impulse was to get some small signal, around a millivolt, amplify the hell out of it, and then apply filtering, like low-pass or whatever. Then I thought about the pre-filtered signal hitting the voltage rails and the post-filtered signal not being satisfactorily large, so I wondered whether it would be better to pass the signal through a unity gain pre-amp first, then filter it, then amplify the remainder. Or whether the pre-amp should have significant gain, maybe 10 or 100 or so.

And it would seem that's actually a good idea.

Besides cascading op-amps, are the conventional two-resistor amplifier designs still good?

--
"The main, if not the only, function of the word aether has been to 
furnish a nominative case to the verb 'to undulate'." 
 Click to see the full signature
Reply to
Gregory L. Hansen

That sounds like good advice. I remember the procedure in one group was an electrolytic capacitor on the power bus, and a small ceramic capacitor on each power pin because the ceramics responded much faster.

What I'm interested in would be well below a kHz, which I assumed would simplify just about everything, like inductive coupling.

--
"Never argue with a fool.  They will drag you down to their level and win 
by experience."
Reply to
Gregory L. Hansen

Oh, good. I knew there must be a way to add more op-amps to it. Thanks. And I've learned that very often, off-the-shelf is not only faster and easier, but better and cheaper than I could do on my own. Do you have a favorite instrumentation amplifier on a chip?

--
"Yes, I revere you much, honored ones, and wish to fart in response." -- 
Aristophanes, Clouds
Reply to
Gregory L. Hansen

I read in sci.electronics.design that Gregory L. Hansen wrote (in ) about 'Amplify the hell out of it.', on Thu, 17 Feb 2005:

Not exactly, in one way, and absolutely not, in another.

You need to understand that op-amps are always used with negative feedback (unless one is ill-advised enough to use one as a sluggish comparator).

The *open-loop* gain, with **no feedback** of a 'cooking-type' internally-compensated op-amp is (to simplify a bit) 3 million at 1 Hz, and falls to 300 thousand at 10 Hz, 30 thousand at 100 Hz, and so on, so reaches 1 at 3 MHz. It has to have its high-frequency gain killed like that, otherwise it would be an oscillator.

But you don't use it like that. If you apply negative feedback to reduce the **closed-loop** gain to 10, it will be approximately 10 up to 300 kHz. But if you apply feedback to set the closed-loop gain to 1000, that will only be maintained up to 300 Hz.

Of course, these days there are much faster op-amps, whose unity-gin frequency is much higher than 3 MHz, and there are op-amps for which you can control the roll-off of the open-loop gain with external components. But the same principle remains; very high gain from one op-amp raises difficulties that may well be eliminated by using two in cascade.

--
Regards, John Woodgate, OOO - Own Opinions Only. 
The good news is that nothing is compulsory.
 Click to see the full signature
Reply to
John Woodgate

Have you built something or are you talking about a simulation or mental exercise?

--
John Popelish
Reply to
John Popelish

An excellent approach. You may never know how much aggravation you just eliminated.

--
John Popelish
Reply to
John Popelish

I was looking at the AD620, I think it was. It looked like a nice all-in-one-package thing, but I was having a little trouble placing an order for less than a thousand units.

Signal: roughly 1 mV from a magnetic pickup, frequencies of interest in the hundreds of Hz to near zero. All things considered, I think it must be pretty simple as far as signals go, except for the high gain.

--
"Outside the camp you shall have a place set aside to be used as a 
latrine.  You shall keep a trowel in your equipment and with it, when you 
 Click to see the full signature
Reply to
Gregory L. Hansen

I read in sci.electronics.design that Gregory L. Hansen wrote (in ) about 'Amplify the hell out of it.', on Thu, 17 Feb 2005:

You are going to need to address 'flicker noise' in that frequency range. Than means a selected low-noise bipolar op-amp. What is the resistance of your pick-up? That will dictate the input circuit design for minimum noise.

--
Regards, John Woodgate, OOO - Own Opinions Only. 
The good news is that nothing is compulsory.
 Click to see the full signature
Reply to
John Woodgate

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.