My passive low pass filters and their application.

[blackdog]

Hi,

I want to show you how I made my -12dB passive low pass filters and how I apply them.
I also hope that GEOelectronics will be a bit happier because these measuring instruments do not need 3458A DMM. 

I often use these low pass filters together with the two measuring amplifiers I have shown here this week and a opamp testbox.
And let's start with the opamp testbox that contains no more than a few resistors that set the gain to 1000x.
Of course, also the necessary decoupling capacitors and a output series resistor to drive the cables.

This is the opamp testbox with here connected to one of the low pass filters.



Specifications
The output load is intended for a scoop, DMM, AC RMS meter, so the load will need to be 1Meg or more.
Some of the filters have a relative low input impedance, for my application no problemo.
Almost always these filters are driven by a impedance of 50 Ohms or less.
As always, you should know your measuring instruments specifications.

This is the schematic of the first filter that i build, It is switchable between 100Hz and 1KHz, another frequency is of course also possible.
But a factor of 10 is desirable to keep the filter simple.
If I still want to have other turning points, I will build a new filter for this.
So that I can apply multiple filters at the same time to different points in my measurement setup.



I used the Fallstaf website to configure the filters:
The resistors used are 1% and I selected the capacitors with a capacitor meter.
http://www.falstad.com/circuit/e-filt-lopass.html


Ready to build the first filter, the schematic on the background is wrong!


Everithing build in the box and now its time for testing...



Let's start with the 1KHz filter, the generator is set to 100Hz and I use the AC dB function of the TEK DMM and also the relative key.



Now the generator is on 1KHZ as visible on the DMM, the level is now as expected at -3dB



At 10Khz -31.42 dB is reached.



Now the 100Hz filter, lets start at 10Hz.



As espected -3dB at 100Hz.



At 1KHz the espected more than -31dB



I drew both filter characteristics over each other here, I could show no deviations so beautifully they fall over each other.



10Hz Low Pass
Now lets go to a special one, and watch out, this has a low input impedance, this is not so important for my applications, but dont forget! (Ri=250 Ohm)


This one is build in a bigger Radial Box, i do not buy these boxes new, far too expensive, look for it at Hamfests...
The small white capacitor on the left side is used to set the red 10uF capacitors to 40uF.



This is the second node of the filter and its trimming capacitors



These are some boxes from a Hamfest, the big one is a Radiall 280 874 and the small one is a Radiall 280 873



What to do with these filters
I have already shown you my test box for testing a single opamps.
The following data are from a LT1250.
LT described this as a: Very Low Noise Zero-Drift Bridge Amplifier.

Let's look at the output signal of my 1000x test box.
I forgot to adjust the probe setting in the scoop, so mV has to be uV in the scoop picture.
The switching frequency is around 5.64KHZ and the signal has a large output of almost 60uVtt at 50uSec timebase.



And now at 10mSec timebase, you wil lose some signal if you dont adjust the sampling on your scope.
It is difficult to see the noise in the lower frequencies.



Now the 1KHz filter is included at the input of the scoop and now there is no chopper frequency visible anymore.
Timebase still 10mSec and the level is about 15uVtt.



Lets switch to 100Hz and it is now 1uV/Div, a totaly different signal from the first picture.



I hope you will find this topic interesting without the need for "high end" references or measuring instruments.

For those who also want to build these filters, if you want something on other frequencies please keep in mind
that there is a ratio of a factor of 10 between the two -6dB filters for the best result...

Kind regards,
Bram

[GEOelectronics]

GEOelectronics is very happy. Thank you Bram.

Geo

[Andreas]

Hello Bram,

Does your PicoScope (which one is it?) have a function generator?

mine has and it makes the "paperwork" for frequency response much easier (with some free 3rd party software).

Here a example of the AN83 noise amplifier.

with best regards

Andreas

[blackdog]

Hi Andreas, 

My picoscope is one of the blue ones, 12 bit range and has no generator on board and i'am happy with this instrument.

Furthermore, I find it a bit sad that some have a "generator" that only goes up to 20KHz... they must be kidding!
Also a lot of scoops that are only 8 bit, it is now 2018 and it should be at least 10 bit, also for the cheaper models.

I can't find a model in their whole range of scoops that I can use well for more applications with a reasonable price.
Also a generator on board with a good attenuator at the generator output, that is low level and low noise.

My Audio Precision Measuring set has GPIB, but i'am not so good in programming...

Kind regards,
Bram

[Andreas]

and i'am happy with this instrument.

I know ... (so probably a 4224)

that only goes up to 20KHz... they must be kidding!

depends on application. but of course you are right.

For the price.
For my first scope (2 channel analog 20 MHz) I had nearly to work 1,5 months.
For the current (4 Channel 200 MHz up to 16 Bit) scope less than 0,5 months.
So actually it is cheaper.

with best regards

Andreas

[e61_phil]

Nice builds, blackdog

Just one word for measuring frequency response with a scope and a generator: I do such measurements with FFT. You already know which frequency is the frequency of interested and the FFT will act as a very narrow filter. Furthermore, it increases the dynamic range of the scope due to the FFT gain.

Edit: Added a small example to show what is possible with a 8-bit Scope. It is a 10MHz AM signal with 1kHz modulation. I sampled 14MPts with my Rigol DS2000 scope. The dynamic range is around 80dB.

[GEOelectronics]

Look at those beautiful upper and lower sidebands. Nice.

Geo

[blackdog]

Hi e61_phil, 

Looks good for a 8 bit scoop.

If you want to measure the frequency band from 0.1Hz to 100Hz correctly, then go and do some shopping,
hang up the laundry and drink some coffee and if you have had luck, your first settings have been the right one.   

Of course I'm exaggerating a bit now, but you would like to measure from 0.01HZ with FFT, Pfff...

When I am back in my LAB, I will use one of my passive filters to measure the LT2057 again on the Pico Scoop and/or the Hameg 3044 .
This as you indicated to increase the dynamic range of the FFT with the filter in place.
The unfiltered switching frequency is the most dominant signal at the output of this IC.

Andreas
This is my PICOScoop: 3224

Some remarks
I still want some more pasive filters, but this is not likely to be done until later this year.

If you use the value that is in my 100Hz / 1Khz filter, then it is easy to make filters at other frequencies just scale the components,
dubble the capacitor value and the drops to 50Hz and 500Hz ect.
But pay attention to the input impedance of the filter, can your D.U.T. handle this load?

My starting point was that the filters should be able to be driven by a 50 Ohm impedance without too much deviation.
That's why my 100Hz/1KHz filter has an R1 value of 1K and the total output impedance below the cutoff frequency is about 10K.
This 10K is, in turn, a decent value to a drive a Scope, DMM or AC milivolt meter which usually have a 1 or 10Meg input resistance.

Handy tools if wise used 

Kind regards,
Bram

[e61_phil]

Hi Bram,

Of course I'm exaggerating a bit now, but you would like to measure from 0.01HZ with FFT, Pfff...

0.01Hz isn't possible that's true. But your graph starts at 20Hz and that is possible. We did some measurements some time ago starting from 0.6Hz with this technique (https://www.researchgate.net/publication/270342715_In-circuit-measurement_of_parasitic_elements_in_high_gain_high_bandwidth_low_noise_transimpedance_amplifiers I think everone knows how to get the full text..). For lower frequencies it is usefull to activate the boxcar averaging of the scope.

But I don't want to drive the discussion off-topic.

I hope we will see more from your stuff

Best
Philipp

[GEOelectronics]

"The unfiltered switching frequency is the most dominant signal at the output of this IC."

I hope I understand correctly.
Is it really there or is it being picked up parasitically? I can see the switching frequency clearly with the scope probe anywhere near the transformer.
Thanks



 By the way, I have a picture of a menu board at a restaurant in Thailand. It lists Dog and Blackdog.
When asked they said black dog was a delicacy. Oh and it came with peanut sauce and tastes like chicken. Also have a picture of the menu board at the restaurant in Amsterdam that only has marijuana dishes.

Geo

[blackdog]

Hi GEO,

There is no parasitics in play here, metal box and coax cable conections.
Do not forget i use a 1000x gain setting in the box.

My blackdog name comes from a picture my sister made for me when i was aboud 15 years old and started piracy on the medium wave,
around 1300KHz, yes i know, i'am a bad boy. 

Que! Blackdog a delicacy...
I will call the army to protect myself...

I know the Buldog in Amsterdam, and then I mean that they exist, I have absolutely no interest in drug.
I love to train my brain and not to demolish it.
Think of Johnny Five, Input, More Input! 

Kind regards,
Bram

[GEOelectronics]

Good. I set some records ion 1820 kHz.
Geo