# LM358 with Gain 30,000+ and single supply

#### drkblog

Joined Oct 4, 2012
109
My local provider doesn't has the OPAx134 in his catalog. I think I will have to buy it in the US and get it shipped to Argentina. In the meantime I will try with the NE5532 I just bought today. Thank you!

#### drkblog

Joined Oct 4, 2012
109
Great! I check that distributor's online catalog (Electrocomponentes) and it wasn't listed. So I'm calling them right now... thanks!

Nope... the don't sell it here. They can bring me those within 45 days and only if I buy 120 USD minimum. I'm going to buy them in eBay. Will let you know...

UPDATE:

Good news: I built this scary circuit: with all that little antennas there, in the breadboard. And yes, the 220V to 9V transformer is there at two inches from the circuit. Using the NE5532P now.

Results are far away from those with the LM358. I created a first stage with 19.3 gain, followed by a Sallen-Key high-pass for 100Hz, followed by a Sallen-Key Low Pass at 2.6kHz and finally a stage with gain 1000. And now this is the noise I get at the endpoint:

I see at least two obvious frequencies that I can discard: 50Hz and 100Hz The first one is the frequency line (probably induced from the 50VA transformer. And the second one is twice frequency line, which comes from the sensor as the datasheet states. I'm using 5% components in the filters. I will use 1% for the prototype board.

I know this could look a little ugly, but all this can be improved so I'm very positive right now.

Last edited:

#### drkblog

Joined Oct 4, 2012
109
New question

I have this Sallen-Key high pass filter built for a cutoff frequency of 190Hz:

This circuit works both in simulation and the real-life. But I need to lower the 50Hz and 100Hz noise even more. If I start increasing the order of this 2-poles filter, will I get more attenuation of those frequencies?

#### Audioguru

Joined Dec 20, 2007
11,251
New question

I have this Sallen-Key Butterworth high pass filter built for a cutoff frequency of 190Hz:
This circuit works both in simulation and the real-life. But I need to lower the 50Hz and 100Hz noise even more. If I start increasing the order of this 2-poles filter, will I get more attenuation of those frequencies?
Yes of course.
The 2-poles filter has a slope of 12dB per octave so 100Hz is reduced to about -10dB and 50Hz is reduced to about -22dB.

If you add another RC and make it a Butterworth 3-poles Sallen and Key active filter then 100Hz will be reduced to about -16dB and 50Hz will be reduced to about -28dB.

Here is a website that shows how to calculate one:
http://jpbedinger.tripod.com/projects/an/ds2-3polehp.htm

#### Attachments

• 20.4 KB Views: 30

#### Ron H

Joined Apr 14, 2005
7,012
Post a datasheet, or a link to a datasheet, of your sensor.

#### crutschow

Joined Mar 14, 2008
23,310
...................................

Now be tricky; look at the open loop gain plots for op amps and realize that the gains starts falling off at 20dB/decade at very low frequencies.

......................
What makes you say that? The gain does start to roll off at a low frequency as the frequency is increased, due to the one-pole compensation, but the gain is maximum at DC.

#### crutschow

Joined Mar 14, 2008
23,310
For higher order active filters FilterPro from Texas Instrument, which calculates all the component filter values for any order and frequency of filter, is a free download.

#### ramancini8

Joined Jul 18, 2012
473
What makes you say that? The gain does start to roll off at a low frequency as the frequency is increased, due to the one-pole compensation, but the gain is maximum at DC.
I say it because it is true and because many engineers get into trouble when they don't understand this. The GBW of an op amp is where the open loop gain is zero. Every decade you go back in frequency yields 20dB gain. The LM358/324 data sheet does not have enough data to do any reasonable calculations but the TL071 data sheet does have enough information to do the calculations and thus to learn. Read ://www.ti.com/lit/an/slyt146/slyt146.pdf for the full analysis.

#### ramancini8

Joined Jul 18, 2012
473
What makes you say that? The gain does start to roll off at a low frequency as the frequency is increased, due to the one-pole compensation, but the gain is maximum at DC.
I say it because it is true. Furthermore, the GBW of the op amp used for a filter must high enough at the noise frequencies to act as a noise filter.

Here is a reference for you. www.ti.com/lit/an/slyt146/slyt146.pdf

#### bountyhunter

Joined Sep 7, 2009
2,498
I say it because it is true and because many engineers get into trouble when they don't understand this. The GBW of an op amp is where the open loop gain is zero. Every decade you go back in frequency yields 20dB gain. The LM358/324 data sheet does not have enough data to do any reasonable calculations but the TL071 data sheet does have enough information to do the calculations and thus to learn. Read ://www.ti.com/lit/an/slyt146/slyt146.pdf for the full analysis.
Well known to audio designers. I think the GBW on the 5532 is something like 10 MHz ballpark and DC gain of about 100 dB, which is WAY higher than a typical op amp. That would mean the low frequency roll off pole is at about 100 Hz. Gain drops to 80dB @ 1k Hz, 60dB @ 10kHz, 40dB @ 100kHz, 20dB @ 1MHz, 0dB (unity) @ 10 MHz.

An audio designer must know this because the value of open loop gain (at a given frequency) determines how much gain differential there is at the frequency which is how much the amplifier reduces distortion at that frequency.

#### crutschow

Joined Mar 14, 2008
23,310
.............................
But I need to lower the 50Hz and 100Hz noise even more. If I start increasing the order of this 2-poles filter, will I get more attenuation of those frequencies?
You might consider notch filters set to 50Hz and 100Hz which can be made to have a very high rejection at specific frequencies. Here's an ap note that discusses them.

#### drkblog

Joined Oct 4, 2012
109
You might consider notch filters set to 50Hz and 100Hz which can be made to have a very high rejection at specific frequencies. Here's an ap note that discusses them.
Great! Because I see 50Hz and what seems to be harmonics at 100, 150, 200, 250... all this has to come from the power-supply's ripple.
Another thing I have to improve is the AC coupling from the sensor to the first stage:

IF output
IF resistance: 50Ω
IF noise power [IF =500Hz, IF load = 50Ω]: -134 dBm/Hz
IF noise voltage [IF =500Hz, IF load = 1kΩ]: 45 nV/√Hz

I currently have a 1µF capacitor from IF output to opamp's positive input and a 100k resistor to ground.

Last edited:

#### drkblog

Joined Oct 4, 2012
109
New question:

I've calculated values for a couple of filters (namely the notch and the sallen-key-butterworth) and the papers say I have to use 1% resistors. But the only provider selling those I can find, sells 100pcs of each value (minimum). Which is a waste of money in the development stage of the project. I may end up buying 100pcs of a give value and not using even one single unit. And I have 10 different values already.

Is there a work around for this?
Is it a good idea to combine standard values to get closer (for building a prototype board)?

#### crutschow

Joined Mar 14, 2008
23,310
New question:

I've calculated values for a couple of filters (namely the notch and the sallen-key-butterworth) and the papers say I have to use 1% resistors. But the only provider selling those I can find, sells 100pcs of each value (minimum). Which is a waste of money in the development stage of the project. I may end up buying 100pcs of a give value and not using even one single unit. And I have 10 different values already.

Is there a work around for this?
Is it a good idea to combine standard values to get closer (for building a prototype board)?
Certainly you can use standard values in series or parallel to get the desired value for the breadboard. But you should use 1% values in the final circuit since they are more stable.

The capacitors should also be 5% or better tolerance.

#### drkblog

Joined Oct 4, 2012
109
Certainly you can use standard values in series or parallel to get the desired value for the breadboard. But you should use 1% values in the final circuit since they are more stable.
The capacitors should also be 5% or better tolerance.
I'm sorry, I've made an ambiguous question I guess. Let me correct this:

First: I'm going to use 1% for the final circuit. Out of question.

Second: When I say standard values I mean E12 series at 5%. So I will be building the prototype using E12 5% values in series or parallel to reach an E96 value. Then I will buy E96 values 1% tolerance for the final product.

I guess your answer applies to this situation too, but just to be sure...

#### Audioguru

Joined Dec 20, 2007
11,251
You can't use a notch filter because it will not attenuate the harmonics. The values for a highpass filter will work fine with 5% resistors and capacitors.

#### drkblog

Joined Oct 4, 2012
109
These are the harmonics for a 19,333 gain and the low pass filter at 2600Hz (high pass filters removed):

Rich (BB code):
Freq(Hz) Volt(V)
====== ========
50   1.127595
75   0.081402
100 0.481089
125 0.098358
150 0.056265
200 0.080991
250 0.020163
300 0.023131
350 0.092906
400 0.055624
So, there must be a 58µV 50Hz noise in the input signal which is causing the 1.12V at the end. Unless I'm missing something (noise that's getting into the circuit during amplification or filtering stages). Which makes me wonder if my power supply is good enough. I'm using a 7805 and 7905 for ±5V supply filtered with two 1000µF

7805 Datasheet states:

Output Noise Voltage
f = 10Hz to 100KHz, TA=+25 oC
42 μV/VO

Ripple Rejection
f = 120Hz
VO = 8V to 18V
62dB

#### Ron H

Joined Apr 14, 2005
7,012
These are the harmonics for a 19,333 gain and the low pass filter at 2600Hz (high pass filters removed):

Rich (BB code):
Freq(Hz) Volt(V)
====== ========
50   1.127595
75   0.081402
100 0.481089
125 0.098358
150 0.056265
200 0.080991
250 0.020163
300 0.023131
350 0.092906
400 0.055624
So, there must be a 58µV 50Hz noise in the input signal which is causing the 1.12V at the end. Unless I'm missing something (noise that's getting into the circuit during amplification or filtering stages). Which makes me wonder if my power supply is good enough. I'm using a 7805 and 7905 for ±5V supply filtered with two 1000µF

7805 Datasheet states:

Output Noise Voltage
f = 10Hz to 100KHz, TA=+25 oC
42 μV/VO

Ripple Rejection
f = 120Hz
VO = 8V to 18V
62dB
You also need to consider op amp power supply rejection.

#### Audioguru

Joined Dec 20, 2007
11,251
A 50Hz frequency cannot produce 75Hz nor 125Hz as a harmonic.
The harmonics of 50Hz are 100, 150, 200, 250, 300, 350, 400, 450, 500, 550Hz etc.

How did you measure the frequencies and their amplitudes?