Low Signal Amplification: OP-07 & LM324 Preamp

[m718]

Do opamps amplify lineary down to very low signals like a few microvolts, the OP-07 for eg.
I have an LM324 in moving coil phono preamp and the + input in which the AC coupled signal is going to is also connected to the + supply(6v) thru a 100K resistor. Is this to bias the opamp input because of the very low signal input?

 

[berkeman]

Do opamps amplify lineary down to very low signals like a few microvolts, the OP-07 for eg.
I have an LM324 in moving coil phono preamp and the + input in which the AC coupled signal is going to is also connected to the + supply(6v) thru a 100K resistor. Is this to bias the opamp input because of the very low signal input?

Opamp inputs need to have some sort of biasing, to supply their finite (but small) input currents.

You can learn more about the subtleties of opamps in this thread:

(see for example, circuit B in post #39...)

.

 

[0xDEADBEEF]

Do opamps amplify lineary down to very low signals like a few microvolts, the OP-07 for eg.
I have an LM324 in moving coil phono preamp and the + input in which the AC coupled signal is going to is also connected to the + supply(6v) thru a 100K resistor. Is this to bias the opamp input because of the very low signal input?

The resistor is a pull up resistor that makes sure that the input capacitor is at a defined reference voltage when there is no signal. Normally this is 0 volts, but in your set up 6V might make more sense. If there is no such resistor then the capacitor can trap a charge, and have some floating voltage that is possibly very high. The resistor has nothing to do with linearity.

 

[Bob S]

m718-
Here is a website that reviews all (well, most) of the equations used in op-amp circuit design,
http://www.opamp-electronics.com/tutorials/semi_theory_ch_008.htm
About 3/4 of the way through this chapter is a section on input bias current, input offset current, and compensation resistor.
Bob S

 

[vk6kro]

Op amps are generally pretty noisy too, so it is unlikely you could use one at the microvolt level. The small signal could get swamped by the noise of the opamp.

This is especially so for general purpose ones like LM324s and LM741s.

 

[m718]

I made about 4 different low noise transistor preamps from web schematics none of them work.
Any schematic that had a transistor in it never worked when I made it like this one for eg.

 

[Bob S]

Op amps are generally pretty noisy too, so it is unlikely you could use one at the microvolt level. The small signal could get swamped by the noise of the opamp.

This is especially so for general purpose ones like LM324s and LM741s.

You can get low noise op-amps that are in the range of several nanovolts per root Hz noise levels. So a 10-KHz BW might be several 100 nanovolts noise.
Bob S

 

[m718]

You can get low noise op-amps that are in the range of several nanovolts per root Hz noise levels. So a 10-KHz BW might be several 100 nanovolts noise.
Bob S

Is this close to a good transistor preamp?
and do transistor preamps have lower drift than an op amp like AD8571(.005microvolt/C)?

 

[Bob S]

Is this close to a good transistor preamp?
and do transistor preamps have lower drift than an op amp like AD8571(.005microvolt/C)?

Hi m718-
The lowest noise-figure amplifier I ever used was a GaAs microwave amplifier with a 2 GHz BW that we cooled to about 10 kelvin.
Bob S

 

[vk6kro]

That "low noise amplifier" doesn't look like it would do anything because it has no biassing on the bases.

The outputs should be out of phase but then they are joined together. Not sure what that is all about. Quite a strange arrangement.

Seems like it would be "low noise" OK but low everything.

 

[berkeman]

I had trouble reading the schematic as well. Not drawn well at all, at least not for traditional modern schematics. Some of my older RF textbooks draw schematics like that, though -- quite the pain to try to puzzle out.

 

[vk6kro]

Yes, apart from being drawn strangely, I don't think it has a hope of doing anything.

I have redrawn it in case we were missing out on a little gem here, but I don't think so.

See attachment, but... DO NOT BUILD THIS... it is just nonsense.

It would be so easy to get it working, but as it stands, it would do nothing.

It was probably just someone's doodling and it was never built. Always a risk on Internet.

 

[uart]

Yes, apart from being drawn strangely, I don't think it has a hope of doing anything. ... It was probably just someone's doodling and it was never built. Always a risk on Internet.

I wouldn't dismiss it quite so quickly. It's certainly unconventional but it might just work. A key point with this circuit is that the 9v battery supply is floating, so it's actually possible for the positive and negative ends of the battery to be simultaneously driven at the signal frequency. This is in fact how the circuit works. Both Q1 and Q2 operate as CE stages (both emitters are AC ground) and both invert the signal from base to collector, so both ends of the battery are in fact driven in phase (with each other), and these voltage swings are coupled directly to the output.

The bias in this circuit was the weirdest part to figure out. It's a bit like a currrent balance between the transistors and parallel resistors and it needs fairly well match transistors to work ideally. (Ideally so that the battery voltage floats to +4.5 volts at Q1's collector and -4.5 volts at Q2's collector). In general there will be some mismatch and the battery voltage will float asymmetrically, though it still might work well enough since it's only dealing with relatively small voltage swings.

 

[m718]

I got the schematic from this site: http://users.ece.gatech.edu/mleach/headamp/

its the third and the first on that page is the original. Its supposed to have the noise equivelent of a 2.7R resistor.
if anyone can get that to work let me know if you used the exact same schematic they show.

 

[vk6kro]

Great bit of analysis UART. Maybe it could be got working.

Just playing with the drawing a bit, I came up with something a bit more conventional.
Wouldn't be too difficult to add a few components and get M718's circuit working.

See attachment.

 

[Averagesupernova]

I would say the original circuit posted should work. The biasing isn't all that tricky. Consider the base/emitter junction of each transistor wired in series with each other. If the transistors are a matched pair it should work. This circuit should work very well at keeping any power supply noise from getting in the output. Granted, there shouldn't be any power supply noise from a battery. Also, I would say that it stands a better chance of working well over a wide supply voltage in the event of a drained battery. vk6kro, your redrawn schematic would be a bit easier to understand if you faced the base of the NPN towards the base of the PNP.

 

[vk6kro]

OK about the bias arrangement. I see what you mean.

See attached picture. Something like that?

 

[Averagesupernova]

Yes exactly like that. Uart hit it pretty close with the voltage on the collectors swinging in the same direction, but I prefer my description of the bias. Maybe I've misunderstood what Uart was saying. I would also tend to think that this amplifier is a very low distortion amp. Any non-linearity in the transistors would tend to cancel out as long as each transistor distorts the same as the other since they are swinging in the opposite direction (one is going closer to cuttoff while the other is going closer to saturation). I think it is a very clever circuit.

 

[vk6kro]

I hope it works. That would be amazing.

This is like seeing a dead dog jump up and start barking!

The author complained that it had been used without his permission in some commercial equipment, so that sounds promising.

Not to forget M718 though. Maybe he could take some voltage readings and give component values for his actual circuit and it could be got going?

 

[Averagesupernova]

I'm guessing it didn't work because mismatched transistors.

 

[vk6kro]

Could be.

I guess he'll comment if he knows if they were balanced or not.

 

[vk6kro]

So, do you feel R4 is the actual load for both transistors?

 

[Averagesupernova]

Yes. Actually, R2 and R3 are too. All in parallel.

 

[uart]

I ran a quick pspice test on that amplifier if anyone is interested. Sorry my version of pspice is so old it doesn't even have schematic capture, so I'll have to post just the net-list.

The first attached image is of the input and output voltages (10mV 1kHz input, source resistance 10k and load resistance 2k).

The second attached image is of the two collector voltages and it shows the bias asymmetry due to mismatched (or as I should really say, not perfectly complementary) transistors. BTW, the simulation is a transient analysis with full DC bias point caclutation.

Test weird preamp ciruit
vin 11 0 SIN(0 10m 1k 0 0 0)
rs  11 1 10k
q1  2 1 3 q2n3904
q2  5 1 4 q2n3906
re1 3 5 4k7
ce1 3 0 10u
re2 4 2 4k7
ce2 4 0 10u
vbatt 2 20 dc 9
rbatt 20 5 1
cc1 2 6 10u
cc2 6 5 10u
rl 6 0 2k 
.lib ../eval.lib
.ac lin 101      10       1.000k   ; *ipsp*  
.tran/op 40.000u  .04      .03      40.000u   ; *ipsp*
.end

 

[m718]

I will just use op amps the small better performance of transistor circuits is not worth it.

 

[vk6kro]

I will just use op amps the small better performance of transistor circuits is not worth it.

Yes, but you can come back to discrete components later just to get familiar with them.

UART: thanks for that simulation. Looks like a useful gain of about 7.
That has to be the strangest amplifier circuit I've seen. Amazing PSpice could even model it.

 

[Bob S]

Here is a SPICE model and simulation I did on the "weird amp". In the first thumbnail I show the circuit. I have optimized (I think) the component values. The collector current (Q1 and Q2 are in series) here is about 12 mA (with the R2 and R3 47K base bias resistors). The collector current is very sensitive to the base bias resistors, dropping to 2.3 mA for 100k. The input impedance is very low, and R1 series= 10 ohms seems to optimize the gain response. In the second thumbnail I show the gain response from 1 Hz to 100 KHz. Both the voltage gain (155:1) and phase shifts are pretty flat above 20 Hz.

With the 12 mA collector current, I suspect that the floating 9-V battery will last only about 8 hours. The 10 ohm resistor in series with the battery is for checking the battery drain (Q1 and Q2 collector) current. Adjust it to maybe 10 mA by changing R2 and R3.

 

[vk6kro]

BobS:
Nice simulation. Did you try the original arrangement with input to the bases?
Not sure what the conclusion is, though. It certainly would work but why would anyone want to build it?
Imagine trying to repair it and drawing out such a peculiar circuit.

Have you ever written a description of how you do your simulations in PSpice?

 

[uart]

Hi Bob, that's actually a different amplifier to the one that had us all scratching our heads. You've done the common base amplifier that's first on the page linked by 718. This is a fairly conventional complementary CB design. There are three different amplifier designs on that page and it was the third one (a complentary CE design) that M718 originally posted here.

 

[vk6kro]

Yes, it was this one:
https://www.physicsforums.com/attachment.php?attachmentid=20574&d=1253124428

 

[Bob S]

Here is the SPICE model (see thumbnails) of the common-emitter version of the weird amp. The low frequency response is very sensitive to the capacitor bypass on the two emitters. With 4.7k emitter resistors (corresponding to about 1 mA collector current per transistor), the bypass had to be about 1000 uF to make the frequency response flat with minimal phase shift.
For posting these thumbnails from SPICE, I do a Print Screen of each SPICE image, paste them into Power Point, and save them as jpg.
Bob S
[Edit] As before, this circuit requires a floating 9V battery.

 

[Bob S]

Here is a redo of the common-emitter amplifier circuit in the previous post, except that the 9-volt battery has been separated from the collector signals, and the power supply bypassed to ground, so now this circuit can be coupled to a ground-referenced dc supply. Performance is about the same.

 

[vk6kro]

Thanks Bob, great simulations.

The second one looks a little more friendly although I find it hard to get away from a ground-referenced way of thinking.
I'm just a little worried that this may be a simulator artefact.

Did you get an idea of the input impedance for each common emitter setup?

 

[Bob S]

Here are some numbers for the last circuit. I put a resistor in series with the input voltage source, and found the corresponding output amplitudes:

0 ohms, -23.0 dB (below 1 volt)
1k ohms, -25.3 dB
2k ohms, -26.3 dB
5k ohms, -29.9 dB
10 k ohms, -33.7 dB

So 5K-ohm series input impedance is about half amplitude (6 dB down).

In the dc operating point measurement, the positive and negative rails are +3.76 volts and -5.24 volts, meaning the input would have a dc offset if the circuit runs on ground-referenced power supplies, so an input offset balance pot might be needed.

[Edit] I put two center-grounded 4.5 volt power supplies in series to replace the 9-volt battery, and the offset voltage across a 5k series input resistor was 6.8 mV, so there is an input bias current of 1.4 microamps.
Bob S