Equipment to testing band pass filters?

[SoftwareSamurai]

Does anyone know of a way to accurately test a passive band pass filter that doesn't involve buying/renting expensive equipment such as the AT6011? Perhaps a DYI method involving reasonably-priced equipment?

I'm designing some passive band pass filters in the MHz range. I've got a Quakko SFG-2120 DDS and a Hantek DSO, and while I can manually scroll through the frequency range while watching the FFT on the o'scope, I'm finding it very difficult to accurately plot the filter's curve in order to verify the design. (Perhaps I'm not using them correctly?) I would love to get an AT6011, but alas I'm not flush with such disposable income at the moment. I've though about renting, but that seems like a bit of a hassle, especially since I may need to do testing several times a month in unpredictable intervals.

Any thoughts/recommendations anyone?

Thanks!

[Kiriakos-GR]

How many Watts they will pass from it ?

[Psi]

What kind of mhz range?

100-500mhz is going to be more tricky than 1-10mhz

[SoftwareSamurai]

The BPFs I'm designing are for small signal applications. +5V, 100mA, 1-20 MHz range.

[Psi]

It probably wouldn't be that hard to build something that can sweep from 1-20mhz and record the output amplitude.

Off the top of my head, something like a micro connected to a 1-20mhz freq generator IC.
Then use a precision rectifier to convert the BPF output to DC and measure the voltage with the micro ADC.
Could sweep the range quite slow and record lots of amplitude points.
Could even connect a graphical LCD to the micro and plot the graph.

I guess it depends how accurate you need the data.

[SoftwareSamurai]

Yeah, that could work. Any idea how I could capture the phase shift as well?

[Zad]

Doing amplitude (scalar) response is pretty easy, you just need a sweep frequency generator (also known as a wobbulator) and a simple detector. Elektor published a design for one a few years ago. http://www.elektor.com/magazines/2008/october/rf-sweep-frequency-generator-spectrum-analyser.684208.lynkx?tab=4

If you are doing vector analysis (to obtain the phase) then things get somewhat more difficult and more expensive. Even relatively low budget amateur radio designs come in at several hundred dollars. Basically you have two programmable synchronous frequency sources. You pump the primary signal through the DUT, and then mix the output with a copy of the primary signal, and also the second source which is programmed to have a 90 degree phase shift.  Pass the outputs through low pass filters and digitise the levels to give you In-phase and Quadrature components.

[vk6zgo]

Do it retro-style!

Insert a series of discrete frequencies into the BPF & look at the output on your 'scope in its normal mode.

Write the frequency & input & output levels down on paper,& after you have a collection of values,graph them(by hand,or using an
Excel app.)

With a multichannel Oscilloscope you can check phase in the same manner--compare the zero crossings at input & output to give you the delay,then calculate the phase,then graph as before.

But wait!,there's more!

Amplitude/frequency & phase/frequency graphs don't completely characterise a filter---to do it properly,you also need to determine input & output impedance-v-frequency.

You can do this retro-style too!---the oldtimers did! You only need to make up an impedance bridge & again test at discrete frequencies.

Obviously a sweeper would make it all easier,but it can be done!

VK6ZGO

[ejeffrey]

Yeah, that could work. Any idea how I could capture the phase shift as well?

That is what the 'sync' output on your function generator is for.  Connect it to the trigger and/or second channel of your oscilloscope.  Phase is now calculated from the time delay between the sync signal rising edge and the zero crossing of the output.  This is also often necessary for measurements deep in the stop band where the signal amplitude may be too small to accurately trigger from.  You want to be looking at the time domain, not the FFT.  The FFT is useful if you want to look at non-linearities, but it is not the best way to measure linear transfer functions with a FG/scope.

Better yet is to probe the actual input to the filter right at the input: function generators, even DDS units are rarely the most accurate devices in either amplitude or phase: I have seen the output phase shift relative to the sync signal by over a degree when changing amplitude ranges.  The only tricky part here is making sure that your probe doesn't influence the signal.  Try to clip a 10X probe right to the filter input terminal.

It is fairly easy to write a computer program to sweep your FG and capture waveform data from the scope to measure the entire transfer function.  For simple characterization like just measuring the gain, 3dB points or the maximum ripple, it is just as easy to do by hand.

Whether you use a cheap scope or a fancy VNA, audio frequency or RF, it is always a good idea to measure the transfer function of a segment of coax before

[joelby]

Slightly cheaper than the Atten unit and better for testing filters might be the VNWA3 from SDR-Kits: http://sdr-kits.net/ . It's a 1.3 GHz vector network analyser, which is great for testing filters, complex impedances and more. I think mine was around $650.

[Tony R]

Do it retro-style!

Insert a series of discrete frequencies into the BPF & look at the output on your 'scope in its normal mode.

Write the frequency & input & output levels down on paper,& after you have a collection of values,graph them(by hand,or using an
Excel app.)

...

VK6ZGO

That's what i was thinking, i have done it before too, in fact if the band with was not to wide or i did not need that high of resolution. just remember, go in a logarithmic scale, kind of pointless not to, and kind of a waist of time if you graph it logarithmically, which bode plots are.

[Flavour Flave]

How about this cheap and very accurate diy VNA?



From here http://www.makarov.ca/vna.htm

[sub]

If you're keen to use an automated setup, Analog make a gain/phase detector IC, the AD8302 (AU$32.47 at Farnell).  I've not used it, but it might be something to look at if you can't afford a real VNA.

http://www.analog.com/en/rfif-components/detectors/ad8302/products/product.html

[SoftwareSamurai]

If you're keen to use an automated setup, Analog make a gain/phase detector IC, the AD8302 (AU$32.47 at Farnell).  I've not used it, but it might be something to look at if you can't afford a real VNA.

http://www.analog.com/en/rfif-components/detectors/ad8302/products/product.html

This peaks my curiosity! If only I had time to whip up a complete board & circuit for it. (Hmmm. I wonder if I could breadboard it just to get the numbers I need...)

[Zad]

Be for you jump in at the deep end, do some Googling first. Due to the way the AD8302 calculates phase, there is inaccuracy around the 0/180deg points, and in a simple setup, it isn't possible to distinguish the sign of the phase difference. So (for example) +30deg is the same as -30deg (or 330deg if you prefer). The gain detector is just two log amps and an op-amp. Cool little chip though.

[saturation]

I agree, I'd do it this way only because the tools to do it are usually available in your bench, a good FG + scope, and does the job if you need to do it infrequently.

For phase you can quickly eyeball it with a Lissajou figure.

Do it retro-style! ........
Obviously a sweeper would make it all easier,but it can be done!

[Mechatrommer]

For phase you can quickly eyeball it with a Lissajou figure.

complex FFTs provide you with both magnitude and phase difference. drawback... you have to code it yourself, if you are not willing to pay for a fortune.

[ejeffrey]

complex FFTs provide you with both magnitude and phase difference. drawback... you have to code it yourself, if you are not willing to pay for a fortune.
[/quote]

This only works if the detector and the function generator are synchronous.  Otherwise the phase of the FFT is meaningless.  The FFT function on most oscilloscopes only shows the magnitude for this reason.  In any case, using an FFT for network analysis is not usually a big advantage: it actually has a worse SNR than a typical swept-receiver measurement.  This is exactly because of the huge advantage FFTs have in spectrum analysis: they are recording data at every frequency at all times.  In a network analyzer with a swept source (rather than a white noise or impulse source), that signal is noise most of the time.

[SoftwareSamurai]

Be for you jump in at the deep end, do some Googling first. Due to the way the AD8302 calculates phase, there is inaccuracy around the 0/180deg points, and in a simple setup, it isn't possible to distinguish the sign of the phase difference. So (for example) +30deg is the same as -30deg (or 330deg if you prefer). The gain detector is just two log amps and an op-amp. Cool little chip though.

Yeah, I saw those limitation in the datasheet. Got me thinking that if I were to put two of them together in parallel, one with a modified reference signal that's +90deg from the original reference signal, then one of them should always be producing an "accurate" phase reference, and by comparing the output of both, the phase could be distinguished between +deg and -deg.

Theoretically, that is.

As for the gain detector, perhaps adding an amp to feed this chip (with an LC filter) could help distinguish the signal from the noise better and increase the gain detector's accuracy?

(You can see why I said this peaks my curiosity! I can see a whole lot of fun R&D-ing a board with this chip! Tons of possibilities here!)

[Mechatrommer]

This only works if the detector and the function generator are synchronous.

our dso got 2 channel right? ch1 = input (ie before the DUT, after the signal), ch2 = DUT output. to compensate for ch's phase discrepancy, replace DUT -> short. so, FFT1 = ch1, FFT2 = ch2... FFT1 - FFT2 = difference. sure my rigol cannot do FFT subtraction.

Otherwise the phase of the FFT is meaningless  ... The FFT function on most oscilloscopes only shows the magnitude for this reason.

its up to you to put a meaning to it or not.

it actually has a worse SNR than a typical swept-receiver measurement

ignore the (other) harmonics. feed in sine as clean as you have. run FFT, get the freq with max magnitude only, thats our carrier! now look at the complex number's phases for that element (frequency) in FFT1 and FFT2, substract... we got phase difference.

In any case, using an FFT for network analysis is not usually a big advantage:

it has! you dont have to pay a fortune for it

This is exactly because of the huge advantage FFTs have in spectrum analysis: they are recording data at every frequency at all times.

so thats the advantage, but you dont have to have that huge. as i said, ignore the harmonics. only thing to concern with FFT is spectral leakage, worst if it reaches the carrier frequency. i see thats the major drawback that makes FFT not suitable for all application. and you have to put the algorithm in hardware as the second.

now i'm not familiar with VNA or SA hardware design, and i know its the way to go. and i dont say FFT is the best. just giving a quick reply on another cheaper method and more accurate than eye sight and using protactor on the LCD. my 2cents, more detailed info are welcomed.