Author Topic: SAW filter design cascading SAW filters  (Read 1536 times)

0 Members and 1 Guest are viewing this topic.

Offline GGMMTopic starter

  • Regular Contributor
  • *
  • Posts: 222
  • Country: fr
SAW filter design cascading SAW filters
« on: May 08, 2024, 03:07:04 pm »
Hello,

Good morning,

I made a spectrum analyzer (SimpleSA) according to the design of M0Wid.
You can see all the discussions on the ISO HBTE forum.

https://groups.io/g/HBTE/topics?sidebar=true

Unfortunately, the designer is not answering right now, and I have at least one point to clarify.
The analyzer can also do signal generator and also tracking.
The observation is that the spectrum analyzer works, but not the generator (and surely tracking).
I have been looking for a possible program problem for some time, but I am not a programmer and nothing special seen.

Today I think I detected the problem, I explain.
When one is in generator mode, the 2 SI4432 pass in emission.
One will be fixed at about 433.92 MHZ and the other will be at 433.92MHZ + FREQUENCE chosen. The mixer will make F1+F2 and F1-F2, the filter will let pass anything less than 433.92 mhz approximately.
  EX F1 443.92mhz and F2 433.92 MHZ will output a 10MHZ signal after the filter.

In fact, the assembly only works nice  if I shunt the 2 filters. (one by one has no effect).
So I wonder if the filter design is good.

Filter are ECPOS B3555
https://www.alldatasheet.com/datasheet-pdf/pdf/188917/EPCOS/B3555.html
!!!see the adaptation to 50ohm and how it's made on the Simplesa schematic. I thing strange  (?)


I wonder if waterfall design is a problem because it is different from the one I find on a Tinysa that uses the same principle.
Attached files the SimpleSA 433.92 filter and the rf part of TinySa, you can see also the two saw filter (Not completly the same)


I wonder if the waterfall design poses a problem. I tried with a nanovna, but the result is bad (no form bell), well the test was done directly on the circuit.
if you can tell me the best way to do it.

Thanks for your answer

 
The following users thanked this post: sab

Offline GGMMTopic starter

  • Regular Contributor
  • *
  • Posts: 222
  • Country: fr
Re: SAW filter design cascading SAW filters
« Reply #1 on: May 08, 2024, 03:18:04 pm »
when i said:  EX F1 443.92mhz and F2 433.92 MHZ will output a 10MHZ signal after the filter.
it's the filter after the mixer. The 2 SAW filters only the 433.92 MHZ pass.
see picture
 

Offline GGMMTopic starter

  • Regular Contributor
  • *
  • Posts: 222
  • Country: fr
Re: SAW filter design cascading SAW filters
« Reply #2 on: May 09, 2024, 08:06:07 am »
hello,

No answer at this time.
I simplify the question:
May you confirm that a SAW filter work on both direction.
Because my system is bidirectionnal.

I think yes, only confirm

cdt
 

Offline Gerhard_dk4xp

  • Frequent Contributor
  • **
  • Posts: 341
  • Country: de
Re: SAW filter design cascading SAW filters
« Reply #3 on: May 09, 2024, 01:08:40 pm »
For me, they are pretty symmetrical,
 but B3710 is a 50 Ohm filter without any ado.

... but watch your power level and allowed DC.

second pic is selection with 2 gain blocks and a variable pin attenuator
and 2 filters.
« Last Edit: May 09, 2024, 01:31:55 pm by Gerhard_dk4xp »
 

Offline GGMMTopic starter

  • Regular Contributor
  • *
  • Posts: 222
  • Country: fr
Re: SAW filter design cascading SAW filters
« Reply #4 on: May 09, 2024, 02:34:44 pm »
Gerhard,

thanks for your answer

The filters are B3555 433.92 mhz.
Can you confirm that the electrical connection scheme is good compared to the datasheet.  (see first post)

I really have no spikes with a nanovna in test.

Possible that the filters are bad or the values of the components. (Components in strip without any indication except on the bag)
I’ll try again, I have some left.
thanks
 

Offline RFDx

  • Regular Contributor
  • *
  • Posts: 158
  • Country: de
Re: SAW filter design cascading SAW filters
« Reply #5 on: May 09, 2024, 07:50:24 pm »
I made a spectrum analyzer (SimpleSA) according to the design of M0Wid.
The analyzer can also do signal generator and also tracking.
The observation is that the spectrum analyzer works, but not the generator (and surely tracking).

Show the whole schematic.

In fact, the assembly only works nice  if I shunt the 2 filters. (one by one has no effect).
So I wonder if the filter design is good.

Filter are ECPOS B3555
https://www.alldatasheet.com/datasheet-pdf/pdf/188917/EPCOS/B3555.html
!!!see the adaptation to 50ohm and how it's made on the Simplesa schematic. I thing strange  (?)

The matching network for the filter is 3.3p shunt-C + 39nH series-L. If you account for additional 1...2pF parasitic C (depending on PCB layout) in parallel to the filter input and output, you get the aproximate matching network (6.8pF + 33nH) shown in the datasheet.

I wonder if waterfall design is a problem because it is different from the one I find on a Tinysa that uses the same principle.
Attached files the SimpleSA 433.92 filter and the rf part of TinySa, you can see also the two saw filter (Not completly the same)

If you cascade two filters directly they will affect each other. Some isolation between them, like a small 50 Ohm attenuator or an impedance inverter, would be preferable.
 

Offline Gerhard_dk4xp

  • Frequent Contributor
  • **
  • Posts: 341
  • Country: de
Re: SAW filter design cascading SAW filters
« Reply #6 on: May 09, 2024, 09:53:17 pm »
For the B3555 filter, matching components to 50 Ohms are 33nH + 6p8 from the data sheet.
There are too many 50-ohm comments in the schematic and on the wrong places.
It may be worth a test to connect the 2 filters directly without filt1-> 50 ohms and 50 Ohms->
filter2. OTOH the LC should add selection and the high impedance may radiate.


I had 2 400 MHz filters back to back in a clock multiplier 100-> 400 MHz. That kinda worked,
but the far-off suppression was less than I had hoped for. No real improvement to 1 filter.
maybe the 2 filters were too close, mechanically,   or better shielding required.

I also tried 900 MHz filters by Yageo and got completely irreproducible results.
Maybe I burnt them with the hot air.  I should retry with vapor phase soldering.
« Last Edit: May 09, 2024, 10:06:19 pm by Gerhard_dk4xp »
 

Offline GGMMTopic starter

  • Regular Contributor
  • *
  • Posts: 222
  • Country: fr
Re: SAW filter design cascading SAW filters
« Reply #7 on: May 10, 2024, 09:14:56 am »
Hello  Gerhard,

Thank you for your reply.
I tested one of the inductance and one of the capacitor of the bags of pieces received.
The "Chinese" tester is not able to read a very low inductance of 33 NH and rather gives a resistance of less than one ohm.
For the capacitor of 6.8PF, it is the same does not manage to detect it, which proves at least that its value must be low.

I can do a 33nh/ 6.8pf resonant circuit and see at nanovna.

Anyway, I agree with you, it’s so small to weld the filter , only hot air works. I don’t know if they can be damaged, it’s more "mechanical", but I received 20 for a few euros. So I can try to change them.
cdt
 

Offline GGMMTopic starter

  • Regular Contributor
  • *
  • Posts: 222
  • Country: fr
Re: SAW filter design cascading SAW filters
« Reply #8 on: May 16, 2024, 06:56:43 pm »

hello,

I came across this interesting conversation about SAW filters.
There may be things to see about the adaptation for the input, the output and the connection between the 2 SAW filters.
The theoretical scheme LC 33Nh/6.8Pf  for filter B3555 may need to be revised (?)
Interesting thing about the component layout
Same for calculations, but it’s not clear to have what it is done. We are talking about ADS space with S parameters files
I don’t know what that is.
some help on this subject.

see here:
https://www.edaboard.com/threads/inconsistent-impedance-matching-network-of-a-saw-filter.408321/

cdt
 

Offline GGMMTopic starter

  • Regular Contributor
  • *
  • Posts: 222
  • Country: fr
Re: SAW filter design cascading SAW filters
« Reply #9 on: May 16, 2024, 07:06:29 pm »
I think ADS is like an equivalent of LTspice.
Don't known is one is better than the other.
As I have never practiced these softs, I think it gets complicated.
In short you have maybe a ready model of SAW filter.
to try ?

cdt
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf