Amplifier Spikes

[mjeanrichard]

@T3sl4co1l: I tried to find a filter configuration that fits my problem but this was the best I could come up with. My Signal is about 2-5mV p-p. and between 10kHz and 80kHz. Most of the calculators would not give me the required gain (at least 50dB). Or only with a lot more stages. But I am a bit out of my depth here... So if you could point me into the right direction that would be very much appreciated!

[T3sl4co1l]

Expecting gain out of an active filter is tricky, because you need GBW sufficient to include the filter's behavior, plus the gain, and this very quickly limits your performance.  A 4th order Butterworth filter might need a Q factor of 3ish in one stage, which means GBW roughly that many times the cutoff frequency.  Which means, for 12MHz op-amps, you are limited to a gain of 40 in that stage.  (Which is still not unreasonable, but you should allow some headroom to keep distortion down, and the bandwidth accurate.)

If you could tell us more about your signal source and its nature, maybe a better suited system design could be considered.  Is the signal always exactly ~2mV?  Does the midband gain really have to be exactly 1000?  Would a cheaper gain stage using discrete transistors be overall better?  Would a variable gain stage (the equivalent of radio AGC) be desirable?

It's best to have a chunk of low-noise gain out in front, to raise your signal level above the noise level of the remaining signal chain.  Then you can apply whatever filtering and amplification is necessary, without having to worry about noise (which will be a modest concern for ~2mV signal levels, in this bandwidth).

Point being, you don't need a 10-100kHz bandpass on each stage.  The bandwidth limiting only matters to the stages where excess bandwidth is a problem: the first stage probably doesn't mind, because the signal is so small.  But if there are strong interfering signals around, they need to be filtered away, or even the first stage can be driven into clipping.  (To account for RFI, a modest CLC filter might be used, with a cutoff frequency of ~MHz.  It won't affect the desired signal bandwidth, but it will do a damn good job preventing interference.)  The bandwidth probably should be kept within the desired range before the final gain/output stage, because it's close to clipping, just from the signal alone.

Oh, and note that an oscilloscope with 10x probes will have basically no reading for ~2mV signals, while your circuit will be slamming into the rails at the output.  Just because you can't see it on your instrumentation, doesn't mean the circuit isn't seeing it!  That's really what's meant by "scope the rails": measure it to a degree satisfactory to your circuit's sensitivity.

Tim

[Habropoda]

Since you can use this with a battery and you can detect the spikes with your DMM, an easy check would be to move it to another part of the house or even outside.

I use the same circuit for bat detector experiments, and with the high gain and the fact that it operates well into radio frequencies means that I pick up all kind of stray signals around the house; TVs, computers, all those little power supplies and phone chargers, clocks.  The breadboard with all those wires makes it much worse.  I never realize how noisy my environment is in those frequencies.

[mjeanrichard]

@Habropoda: The "same" as in exactly the same or similar? I use it for the same purpose, and would like to sample the signal with a teensy 3.0 to analyze the frequencies of the calls... What kind of mic are you using?

[Habropoda]

Not exactly the same but very similar.  I’m using two stages since the op amps I use have enough bandwidth – yours might as well.  For microphones I use Panasonic WM-61a clones and the Knowles SPU0410HR5H.  Here’s a project I’m working on for the Hackaday Square Inch contest.

https://hackaday.io/project/8353-worlds-smallest-bat-detector

That is the circuit I use.  For this project it is way overamped and clips like crazy but that is just fine to drive the LEDs.  For your purpose you would need to lower the gain.

See the Resources and Further sections for other uses for this circuit.  As noted there I am very excited about using the preamp/RPi/Cirrus combo for bat call analysis.  For the contest deadline I will be adding some more video, photos and content over the next couple of days.

Keep in mind I am not an expert at this stuff and I look forward to more comments from T3sl4co1l and others on ways to improve it.  Using two stages for gain and two stages for active filters might work out well.

edit:  added schematic here for ease of discussion.

[mjeanrichard]

@T3sl4co1l:

The input signal comes from a mic. The 2mV is just what I can measure with the expected sound volume. Basically I just want to amplify the signal enough to be able to sample it with an ADC and to create a envelope detector. The precise gain is not really important nor is the exact passband. It is just important that the signal is attenuated above 100kHz to prevent aliasing. The highpass is there to prevent loud low frequency noise to "hide" the interesting signal.

The circuit has to operate for hours without user interaction so I don't think variable gain will help.

If I understand correctly, I should amplify the signal before the filter? Also I would like to keep the part count (and price) down as best as possible...

[mjeanrichard]

Since you can use this with a battery and you can detect the spikes with your DMM, an easy check would be to move it to another part of the house or even outside.

I use the same circuit for bat detector experiments, and with the high gain and the fact that it operates well into radio frequencies means that I pick up all kind of stray signals around the house; TVs, computers, all those little power supplies and phone chargers, clocks.  The breadboard with all those wires makes it much worse.  I never realize how noisy my environment is in those frequencies.

I finally got around to do just that. My DMM does not show any frequency when I try it outside. Unfortunately I cannot take my scope outside to check if the noise is just to small for the DMM to pick up or if it is really not there.

[T3sl4co1l]

ADC then detector?  You need over 200kSps!  Man, why not detect in analog?!  Kids these days!

Manually variable gain might be useless, but I meant automatic gain control (AGC).  Ideally, you'd start with the one high frequency signal, and process it into two signals: the detected audio output (which consists of the envelope of the 10-100kHz signal), and an AGC signal (which varies slowly, perhaps ~100Hz bandwidth) which ideally has an exponential relationship between AGC voltage and input signal.  In effect, you have the "sound", and its "volume", separated!

Indeed, you can apply enough gain that the detector is always reading a signal, even if it's ambient noise, or amplifier hiss.  Use a comparator on the AGC to determine what's loud enough to count as a signal (otherwise known as a squelch function).

You'd get enough gain from about 4 transistors, but a variable bias, high gain differential amplifier (using perhaps 7 transistors) would be more convenient (or, for example, using an LM13700, which provides variable gain control with a single pin).  The whole thing should be possible in under 1mA of supply current, though to make it easy, you might use more.

And, if you ever want an introduction to radio circuitry, it would be the perfect learning environment, because it does almost everything a radio does (except frequency conversion; which could still be done, but it's probably not worthwhile here).

Tim

[mjeanrichard]

Sorry I was not clear enough:
There are two signals going into two analog ports of a microprocessor. One is the envelope used to detect the signal, the other is the signal, which is sampled at ~250kSps and further processed with an FFT (but only if a signal is present...).

Yes, what you describe sounds exactly what I need. The volume would be used to check if a signal is there, the AGC signal would be used for sampling and then FFT.

My problem is, that I have no idea how to implement that. I just about understand how an opamp works...

[chris_leyson]

Is your Vcc/2 rail bypassed ? Try 100n or 1uF across R6. I see you wired up the unused amplifier as a unity gain buffer and connected it's +ve input to Vcc/2, that's always good practice. Had a quick look at the data sheet, the input common mode range for the TL974 is Vss+1.15V to Vcc-1.15V, or 1.15V to 2.15V if using 3.3V supply.

Some older generation dual and quad op-amps share a common bias generator for all of the amplifiers and in some cases if you take any of the amplifiers out of their common mode range this upsets the bias for all of the other amplifiers. I don't think this is true for the TL974 but I've had amplifiers do some very strange things when using a spare op-amp as a comparator AND not staying within the recommended common mode range.

Figure. 18 in the TL97x data sheet (Rev H), output voltage verses input voltage, started to ring alarm bells. I think this could be where the problem is. If you believe Fig. 18 then the TL974 isn't designed to run at low supply voltages. The transfer function of the unity gain buffer in fig. 17 changes sign at around 1V input !!! Tried running the TL972 spice model in LTspice and it looks OK, i.e no sign change at around 1V input, but spice models are not always accurate and there is no guarantee that they will include all non linear effects.

If all else fails try getting one op-amp to work and keep your wiring reasonably short.

Good luck.

[fivefish]

He's using a 3.3V total. with virtual ground, he's really just using +1.65 and -1.65V to power his opamp.
I still suspect the problem is somewhere related here but he said he upped the voltage to 5V and he's still seeing the spikes. 

[mjeanrichard]

Actually the spikes go away as soon as I go outside. I currently blame the Wifi Router that is in the same room as my workbench... :-)

[Audioguru]

A solderless breadboard uses many rows of contacts and many messy jumper wires that are antennas that pickup all kinds of interference, especially when you have a high gain wide bandwidth circuit.
Use a compact pcb instead.

[T3sl4co1l]

RF interference is more common with GSM phones nearby than wifi, but it is a bipolar amp, so that could be a contributor.  Ideally you'd have some amount of RLC filtering (say, 1nF to GND plane, 1kohm, or 1uH or ferrite bead, something around there) in effect around all lengthy traces/leads going to the amp.  But on a breadboard, you can't really do that...

Tim

[garre]

I have heard a number of good ideas and suggestions, but most likely your problem is from your electrical power.  100Hz is the first harmonic of your power supply in Europe.  The noise is likely being injected from some appliance or lighting in you home.  With a typical breadboard you wiring will likely allow either RF or magnetic coupling to inject the noise from an electrical appliance not your WiFi router. By reducing your breadboard lead lengths and putting the amplifier in  a shielded box with a good ground should eliminate the noise.

[DimitriP]

The strange thing is, that the magnitude of the spikes changes when i move my hand over the circuit.

I have heard a number of good ideas and suggestions, but most likely your problem is from your electrical power.  100Hz is the first harmonic of your power supply in Europe.  The noise is likely being injected from some appliance or lighting in you home.  With a typical breadboard you wiring will likely allow either RF or magnetic coupling to inject the noise from an electrical appliance not your WiFi router. By reducing your breadboard lead lengths and putting the amplifier in  a shielded box with a good ground should eliminate the noise.

Yeah...well... we are only on page three...we have ways to go...  1 + 1 is soo full of possibilities....

[mjeanrichard]

Yeah, it was not the wifi (tried switching it of...). Still don't know what it was, but as soon as I leave my office the signal disappears.
As long it is not some oscillation within my circuit i am happy.

Thanks for all the helpful tipps!!