Topics

Problem with BPF

GM7
 

Hi All,

I've just finished building a 30m BPF.  The response curve looks good but it appears to have a very large insertion loss. As you would expect my receiver is deaf when it is in place.

I've tried going through the manual and troubleshooting, following the diagram etc. but I've obviously missed something.  I've also tried re-calibrating the NanoVNA with the test fixture in place (but not the BPF under test).

Does anyone have any suggestions?

Thanks

ajparent1/KB1GMX
 

I built two and both behave way better than that.  I did see attenuation (S21)
of 1.25 DB for both using my Rigol DSA815T.  My NanoVNA agrees.

Note tuning can be a bit fusy.

With the nanoVNA make sure you cal the THROUGH with all
cables and jig used!

Allison
-------------------------------
Please reply on list so we can share.
No private email, it goes to a bit bucket due address harvesting

Dave
 

The curve looks under coupled.  Over coupled would have a double hump that no amount of tuning will merge.  Check the coupling capacitors to see if they are in fact the values suggested.

Perhaps a little bit of gimmick coupling will help.

Dave


On Jun 29, 2020, at 18:02, ajparent1/KB1GMX <kb1gmx@...> wrote:

I built two and both behave way better than that.  I did see attenuation (S21)
of 1.25 DB for both using my Rigol DSA815T.  My NanoVNA agrees.

Note tuning can be a bit fusy.

With the nanoVNA make sure you cal the THROUGH with all
cables and jig used!

Allison
-------------------------------
Please reply on list so we can share.
No private email, it goes to a bit bucket due address harvesting

GM7
 

Hi,

Thanks for the replies.  I've spent some more time this evening but I still can't get a good response.

Dave, I tried to add more coupling capacitance as you suggested.  The curve I got has the classic double hump which suggests it is now over-coupled but it still has a very large insertion loss.

I've been testing this using a carefully calibrated fixture.  The continuity seems good as per the manual and it looks like I've wound the inductors as per the manual.

Does anyone have any suggestions of what to look for or how to test it further?  Could it be an impedance mis-match causing the insertion loss?



Thanks

Dave
 

That double hump looks not bad at all, now try tuning the little trimmers to peak into a single tight band pass.  One of my filters looked like that too until it was trimmed up.  This is where using a NanoVNA in transmission mode really shines.

Dave


On Jul 4, 2020, at 15:46, GM7 <geeemmseven@...> wrote:

Hi,

Thanks for the replies.  I've spent some more time this evening but I still can't get a good response.

Dave, I tried to add more coupling capacitance as you suggested.  The curve I got has the classic double hump which suggests it is now over-coupled but it still has a very large insertion loss.

I've been testing this using a carefully calibrated fixture.  The continuity seems good as per the manual and it looks like I've wound the inductors as per the manual.

Does anyone have any suggestions of what to look for or how to test it further?  Could it be an impedance mis-match causing the insertion loss?

<QRP Labs BPF 30m Response II.png>


Thanks

GM7
 

I can tune the response into a curve which looks good.  The problem is that it has a loss of 40dB.  I was expecting the loss to be less than 3dB.

geoff M0ORE
 

Can you please describe, with a diagram if possible, your test set up. Are you using a transmission mode analyser?  Forget about carefully calibrated test fixtures, just connect the input of analyser to output of same and do a sweep. If that is not zero loss you are doing something wrong.

On 04/07/2020 20:46, GM7 wrote:
Hi,

Thanks for the replies.  I've spent some more time this evening but I still can't get a good response.

Dave, I tried to add more coupling capacitance as you suggested.  The curve I got has the classic double hump which suggests it is now over-coupled but it still has a very large insertion loss.

I've been testing this using a carefully calibrated fixture.  The continuity seems good as per the manual and it looks like I've wound the inductors as per the manual.

Does anyone have any suggestions of what to look for or how to test it further?  Could it be an impedance mis-match causing the insertion loss?



Thanks

GM7
 

Hi Geoff,

I'm using an NanoVNA.  I've got CH1 connected to the IN of the LPF and CH2 of the NanoVNA connected to the OUT of the LPF.

When I replace the filter with a straight-thru connection I get a horizontal plot with 0dB loss.  In this case the output and input of the VNA are connected together.

The attached photos show the setup with the BPF in place and with the straight-thru connection.  I've also included the test plot.  That was a useful suggestion, it's good to rule things out.

I've also just tested a 17m Low Pass Filter and got a classic LPF response curve so it looks like the test setup is OK and the problem is with the BPF (or more likely a problem in my construction). 

Thanks

@CurtisM
 

Don't rule out obvious that maybe there is a missing solder connection - use your DVM to check each connection - not touching the particular part if possible.  I especially wonder if an inductor isn't connected.  thanks for sharing the plots with your nano VNA. 

73 curt wb8yyy

Dave
 

I noticed in the photo you posted that one trimmer is fully meshed and one fully unmeshed.  Ideally (does that ever happen?) both trimmers would be near their half way point when both LC circuits are co-resonant.  Perhaps the coil where the trimmer is fully meshed could use another turn, and the coil where the trimmer is fully unmeshed could lose a turn.

When I was tuning up a 15m BPF I found I had to have the trimmers at their minimum in order to get close to 21MHz. If I recall correctly I ended up with two fewer windings on both coils, the trimmers were about 50%, and I was right on the FSK portion of the band.

Dave


On Jul 4, 2020, at 20:11, wb8yyy via groups.io <wb8yyy@...> wrote:

Don't rule out obvious that maybe there is a missing solder connection - use your DVM to check each connection - not touching the particular part if possible.  I especially wonder if an inductor isn't connected.  thanks for sharing the plots with your nano VNA. 

73 curt wb8yyy

GM7
 

Hi Dave,

I don't think it's the trimmer capacitor settings.  They might need a bit of tweaking for a perfect result but adjusting them, or adjusting the inductors by squeezing the turns, doesn't seem to affect the 40dB loss.

The trimmers also aren't in the position they were in when I did the initial setup and got the first plot.  At that time they were mid-range.  I've re-attached this plot.  To me it looks like:

  1. The curve is a good shape, I think the fact that adding more coupling capacitance gives an over-coupled curve confirms this.
  2. The center frequency is correct which suggests the inductor and capacitor values are correct, or close to correct.  I measured both the inductors and capacitors during the build and they were all good.
  3. The loss is close to 40dB.  A straight-thru connection gives 0dB which suggests it's not a measurement error.
I'm confused at how #1 and #2 can co-exist with #3?

GM7
 

Hi Curtis,

Thanks for the suggestion.  I've tried this and tried following the schematic and PCB layout in the manual.  It all seems fine but I found it a tricky circuit to test because most of it is at DC ground due to the secondary windings of the two transformers .

I think it might be a problem with the impedance match on the input or output but I'm not sure how to test that idea.

geoff M0ORE
 

As the response is close to the expected frequency, it would tend to rule out component values or dry joints on inductors. Any problems here would give a response away from the wanted frequency.

To check for impedance mis-match, put your analyser into reflection mode on the wanted frequency range. Connect the filter input to the port of VNA and the output to a good 50 ohm load. This setup should show a SWR of near perfect and a RL of 30 to 40dB. Dis-connecting the dummy load will give a high SWR (RL of a few dB). To check the setup, just connect the dummy load to the VNA test port. A good, well matched filter will reflect the load Z at the tuned frequency. You can reverse the filter and check that it looks the same both ways. As you have a good LPF on a different band, you could look at the response of that in reflection mode.

On 05/07/2020 10:44, GM7 wrote:
Hi Curtis,

Thanks for the suggestion.  I've tried this and tried following the schematic and PCB layout in the manual.  It all seems fine but I found it a tricky circuit to test because most of it is at DC ground due to the secondary windings of the two transformers .

I think it might be a problem with the impedance match on the input or output but I'm not sure how to test that idea.

GM7
 

Thanks Geoff,

I'll give that a try and report back.

HB9FIH
 

I also had issues with low pwr.
Now have a VNA and will soon check the BPF and fix them.
Own 2*20m and 2*40m.
I am enthusiastic operating with them, asap I moved to SV8 / TA3 and plan to partcipate by challenges again.
--
---
73 de Erich

HB9FIH

HS0ZLS

Ronald Taylor
 

Is it possible that on one end or the other you got the transformer primary wire in the secondary hole and vice versa? 

Ron

On Sun, Jul 5, 2020 at 08:52 HB9FIH <erich.rieder@...> wrote:
I also had issues with low pwr.
Now have a VNA and will soon check the BPF and fix them.
Own 2*20m and 2*40m.
I am enthusiastic operating with them, asap I moved to SV8 / TA3 and plan to partcipate by challenges again.
--
---
73 de Erich

HB9FIH

HS0ZLS

GM7
 

Geoff, Your last post was really helpful.  I'm going to split my reply into two to avoid confusion between the LPF which works as expected and the BPF which is the one I'm struggling with.

I used the setup you described.  My first test was a dummy load.  While it's not an ideal response it does look good.  Note I still had the test fixture in place for this test with a straight-thru connection which explains part of the non-linearity.  This is probably more realistic because it's how I test the filters.

I then tested the LPF in both directions and got good response and good symmetry.  The impedance looks pretty much spot on too.  By 'LPF In' I mean the VNA connected to the IN of the LPF and the dummy load connected to the OUT.

Thanks to everyone else for their suggestions.

.


geoff M0ORE
 

Looks like Ron beat me to it, I was about to suggest that one or both transformers were wired back to front.i.e. the primary and secondary windings were reversed. This could give a good frequency response but high loss.

On 05/07/2020 17:12, Ronald Taylor wrote:
Is it possible that on one end or the other you got the transformer primary wire in the secondary hole and vice versa? 

Ron

On Sun, Jul 5, 2020 at 08:52 HB9FIH <erich.rieder@...> wrote:
I also had issues with low pwr.
Now have a VNA and will soon check the BPF and fix them.
Own 2*20m and 2*40m.
I am enthusiastic operating with them, asap I moved to SV8 / TA3 and plan to partcipate by challenges again.
--
---
73 de Erich

HB9FIH

HS0ZLS

GM7
 

The BPF doesn't look so good.  

The BPF Out in particular is completely wrong with an impedance of 0.  

I'm not sure if the BPF In looks reasonable?

ajparent1/KB1GMX
 

Its just not tuned.

Double tuned band pass filter tend to be hard to get right as tuning one side does interact with the other.
Often I shows as a hump well off to one side is badly tuned and if close but not quite, high loss in
the desired band and maybe a badly shaped skirts.

Note the trimmers are relatively small value so if far off (inductor or padding cap) you will not have
enough range of variable cap to correctly tune.

Allison
-------------------------------
Please reply on list so we can share.
No private email, it goes to a bit bucket due address harvesting