Question about BPF #filter


danielu@upcnet.ro danielu@upcnet.ro
 

Dear colleagues ,
I have the following set up: VFO generator on 10Mhz, LPF kit with standard settings, with only 6m BPF in position 1 and 30m BPF in position 4. Only relay 4 is activated and checked to be . I attached a photo (photo 1) with the signal at the VFO output and with the FFT which clearly shows harmonic no 5, although harmonic no 3 is barely visible. If I invert the two LPF filters between them the signal is much cleaner. I'm attaching a single photo(photo 2) with this second case. In this case the FFT does not show any harmonics. Obviously it is not about the precision with the oscilloscope in question. Preset 4K7ohm has no effect. Why are things different in the two cases?


Andrew Lenton
 

Hi,

 

I have looked at you “plot2” is it possible to send the FFT plot with the LPF in the suggested order I.E the Highest frequency last? The comments I have are:

 

·         The FFT plot does not have much dynamic range

·         You are hitting the scope rather hard at 20dBm

·         What spec to Rigol claim for this instrument in FFT mode?

 

I will build the same set up and measure, give me a week or so, can you borrow a spectrum analyser?

 

73

 

Andrew G8UUG

 


Andrew Lenton
 

  •  

30M BPF response:



Adrian Scripcă
 

Hi Andrew,

That looks more of  LPF than a BPF response.

73! Adrian, YO6SSW

On Wed, Mar 4, 2020, 10:52 Andrew Lenton <a@...> wrote:
  •  

30M BPF response:



danielu@upcnet.ro danielu@upcnet.ro
 

Adrian,
Obviously we are talking about LPF, but it is my mistake that I wrote BPF. In fact it is not in the shop BPF filter on 6m.
Andrew,
Even if I lower the level the situation remains. The oscilloscope accepts 5Vpp (18dBm) in the 50Ohm setting and the measurement is done on the penultimate domain. The 20dBm indication is just the scale. I attach FFT with 30m LPF in position 1 and 6m LPF in position 4. It looks harmonic on 30Mhz and nothing in 50Mhz.
Maybe I had to choose the Blackman window for higher resolution on the vertical.
First of all I think this is a rectangular signal at the exit of SI5351 with one of the fronts incorrect . It remains to investigate with a very good oscilloscope in the next day


Hans Summers
 

Hi

In my opinion, the FFT function in digital oscilloscopes (DSO) is just a bad joke inflicted on us by the first DSO engineer as his contribution to humanity's artistic body in the study of dark humour. Subsequent DSO designers had to keep up with the Jones'es and implement the horrible thing otherwise their new creation would not look good enough. 

Doubtless the FFT is good for some purpose, like, er, um, he... (suggestions, anyone?)... well anyway, moving swiftly on... I think it probably causes 10x more confused people than those who actually use it for something useful. 

One of the most interesting aspects of digital technology such as digital signal processing, is understanding the limitations. When is what you see on screen a product of the digital approximations involved, and when is it real? Actually analogue has limitations too... of a different type but somehow they are less stark than in the digital world. 

Yes I know yours is a Rigol and Rigol is one of the best manufacturers; and it looks like a very nice 'scope at that; but, nonetheless the FFT is still a joke... 

Most oscilloscopes have an 8-bit ADC. On the basis of 6dB per bit, the maximum available dynamic range is 48dB (8 x 6dB = 48dB). Practically, the least significant bits are subject to some noise. And your input signal is unlikely to be at the right amplitude to make maximum use of the available range and so you can knock off another bit or two for that also. Then you have to add on to all this, all the artifacts inflicted by the DSP. In the end these FFT have no practical use for us in RF because the dynamic range of what they can show is nowhere near adequate for the kind of range we use in RF. A comparison between the results on a real spectrum analyzer and an oscilloscope FFT are quite revealing. Maybe I will do some measurements someday when I have time, just for making it clear. I mean... do you really think all those little spikes only 15dB down from the carrier, and the deep nulls on the harmonics, are real? 

My first experience with a DSO was a Hantek USB 'scope which I hated with an enormous passion, so much so that when it failed within a month or two of using it, I didn't even want to send it back under warranty. I retain some marginal interest in one day opening it up to look inside (as is my normal habit) but after more than 7 years have passed it is still sitting there on the shelf, very much unloved. That was my first experience of the FFT function and was educational enough by itself. After that I had a 100MHz old analogue 'scope. 

My next DSO was my 100MHz Owon XDS3102A which I purchased on leaving Tokyo Japan, simply because our 20-foot shipping container was going to take 10 weeks to sail around the world to Turkey and, the thought of being without an oscilloscope for 10 weeks was only marginally less horrific than hanging upside-down having my toenails pulled out one by one while being forced to read the Hantek DSO manual repeatedly. The Owon 'scope solved that by weighing only 2kg, which was small enough that, along with my soldering iron and a selected mini-junk-box and one of each kind of all QRP Labs kits, could negotiate (with the XYL) a safe passage in our suitcases and come with us on the plane. 

This Owon XDS3102A distinguishes itself by having a 12-bit ADC; which is unusual in its class (reasonably low cost desktop standalone DSOs); this was not an accident, I purposefully wanted the higher ADC resolution, rightly or unnecessarily, it felt important to me and worth the extra $150 for the increase in dynamic range. Before long, I hooked it up to my QRP Labs VFO/signal generator kit and played around a bit; the FFT function is one screen I had a look at ONE time, just for the entertainment value... just for several seconds, just long enough for me to be marginally less unimpressed than I had been with the Hantek FFT... the additional 4-bits of ADC resolution make some small difference but not a lot. That wasn't why I bought the extra 4-bits anyway :-)    I have never visited the FFT function again... 

Finally in the line-up I now have a 200MHz Siglent SDS 1202X-E DSO which was a gift from Siglent, as one of the sponsors of the homebrew heroes award 2019 https://homebrewheroes.org/ of which I am the proud inaugural recipient... and I don't like to say anything bad about it because one shouldn't look a gift horse in the mouth, should one. Anyway it has an 8-bit ADC and I did not look at its FFT joke, because I don't care about FFT jokes anyway, and my expectation in this regard is precisely 0.0000. The chain of causation since the engineer who first inflicted the joke on us and all the other following 'scopes had to follow it is so long now, we could hardly blame Siglent anyway could we. The rest of the 'scope is very nice and useful in the lab. 

I am lucky to own a rather old n'th hand Advantest R3361C Spectrum Analyzer which though not very modern, is accurate and performant, and I love it. But what I am saying here is. if I did not have that... then I would look to find some other way to look at harmonic levels, not the FFT of a DSO. These days there are a lot of very low cost pieces of test equipment available which maybe not professional grade but certainly come a long way to get close to it. 

Now I have that rant off my chest - back to the original question: 50MHz is VHF. Board layout becomes more and more important, the higher up in frequency you go. The relay-switched filter board was originally designed for low-mid HF. If you want to do well at VHF you use lots of shielding, short traces, coaxial connectors, etc. If you don't, then signal will leak past filters. What happens in the VFO/SigGen and Relay-switched LPF combination there is anyone's guess, but it really would not surprise me at all, if the results are different depending on which order you plug the LPFs in. I think this is normal and expected. Just choose the configuration which works best :-) 

73 Hans G0UPL

On Thu, Mar 5, 2020 at 1:55 AM danielu@... danielu@... <danielu@...> wrote:
Adrian,
Obviously we are talking about LPF, but it is my mistake that I wrote BPF. In fact it is not in the shop BPF filter on 6m.
Andrew,
Even if I lower the level the situation remains. The oscilloscope accepts 5Vpp (18dBm) in the 50Ohm setting and the measurement is done on the penultimate domain. The 20dBm indication is just the scale. I attach FFT with 30m LPF in position 1 and 6m LPF in position 4. It looks harmonic on 30Mhz and nothing in 50Mhz.
Maybe I had to choose the Blackman window for higher resolution on the vertical.
First of all I think this is a rectangular signal at the exit of SI5351 with one of the fronts incorrect . It remains to investigate with a very good oscilloscope in the next day


Dave New, N8SBE
 

This may or may not have anything to do with your LPF issues, but I must admit that I managed to build a couple of mine incorrectly, by placing the caps in the wrong holes on the PCBs.

I spent a lot of time head-scratching and swapping LPFs around, until I realized my mistake.

The LPF documentation talks about the way the capacitor holes accommodate both narrow- and wide-spaced capacitor leads, and somehow I foolishly thought that a narrow-spaced lead set was supposed to straddle the middle of the four holes, instead of going in to either hole 1-2 OR hole 3-4.  By placing them in hole 2-3, it effectively took the capacitor out of the circuit and detuned the filter horribly.  The LPF PCBs are 4-layer, and so it is not obvious how those four holes are connected.

To compound matters, when I pulled one of the 56 pf capacitors, I ended up destroying it, and so at the moment I've tacked a 56 pf silver mica in there until I can find a decent source of NPO capacitors with the right footprint.  I'd like to order one of those kits with a bunch of different values, but I'm suspicious that the ones I've found are not really NPO, but sweepings off of some factory floor.

This is on my six-band U3s.

Hope that helps.

73,

-- Dave, N8SBE


Hans Summers
 

Hi Dave 

Minor correction:

The LPF PCBs are 4-layer, and so it is not obvious how those four holes are connected.

None of the QRP Labs kits are 4-layer. All are 2-layer (top and bottom layers). 

So it is always possible, in any QRP Labs kit, to visually trace were all signals are connected. 

73 Hans G0UPL