BPF 80m


Martin DK3UW
 

assembled my 80m BPFand carried it to a friend to align it. The Frequency is about right but  insertion loss is about 5 db. The trimmers make hardly any adjustments.That is too much. any ideas, what can be wrong ?

73s
Martin
DK3UW



















Ronald Taylor
 

Martin, that particular filter uses 30 pF trimmers in parallel with 470 pF fixed capacitors. So there will be very little frequency adjustment from min to max settings of the trimmers as you saw. The main question would be, “Does it peak at a minimum or maximum setting of the trimmers?” If peaking at a minimum, then you need a little more capacitance in parallel. If it peaks at a maximum you need a little less fixed capacitance. On the other hand, you can modify the turns count on the coils for the same effect. If trimmers are peaing at maximum capacitance there is also too little inductance and you can add a few more turns to the larger coils. Likewise if it’s peaking at minimum settings then you can remove some turns. 

All that said, if it’s peaking mid-range on the trimmers then it is tuned properly, and that insertion loss does seem a little high to me. I typically see 3-4 dB insertion loss on these. 

On Fri, Feb 14, 2020 at 12:23 PM Martin <m.sellschopp@...> wrote:
assembled my 80m BPFand carried it to a friend to align it. The Frequency is about right but  insertion loss is about 5 db. The trimmers make hardly any adjustments.That is too much. any ideas, what can be wrong ?

73s
Martin
DK3UW



















Martin DK3UW
 

Hello Ronalsd,

peaking is not what it does. Its just the sides moving slightly. The frequency is spot on for the peak and the curb looks ok.  If I refer to the doc. I should expect 1.2 to 1.5 db loss.

So I wonder, what else could be wrong.

Martin


N3MNT
 

Check connectivity on T1. Use a meter to check connectivity of each set of turns by probing the solder connection and not the end of the coil wire.  A poor solder connection on T1 will manifest itself  with little change in the readings.


Ronald Taylor
 

Bob, I think he’s talking about the standalone bandpass filter, not the BPF in the QCX.
Martin, I don’t know how your friend tuned it, but if he has access to a spectrum analyzer with tracking generator, it’s very easy to tune both peaks (this is a double tuned filter) to overlay each other at the desired frequency. If they can’t get to the right frequency then you at least will know if they are low or high and what parts values need to be changed to get there. 
If it truly isn’t even trying to move up in level as you tune the trimmers, then possibly you have a bad solder joint or, the old infamous issue with not getting all the enamel off the wire ends on the coils. Check for continuity from the input to ground and the output to ground. These should show continuity through the primaries of the coils. Then check from the high sides of both of the trimmers to ground. This will show continuity through the secondaries of the two coils. If you’ve done all that and are sure everything is in order, one of the trimmer capacitors could be faulty, but this would be rare.

Ron


On Sat, Feb 15, 2020 at 9:34 AM N3MNT <bob@...> wrote:
Check connectivity on T1. Use a meter to check connectivity of each set of turns by probing the solder connection and not the end of the coil wire.  A poor solder connection on T1 will manifest itself  with little change in the readings.


Andrew Lenton
 

Hi Martin,

How are you measuring the 5dB loss, do you have a plot for this band pass filter, I do not have ones of threse filters, however, I will order one, and measure it and post the plot, here is one for 15M

73

Andrew



ajparent1/KB1GMX
 

80m band filter
My measurements of coupling capacitor vs insertion loss and bandwidth are shown in this table.
Capacitor  Bandwidth   Insertion loss
47pF          465KHZ      1.27dB
42pF          428kHz       1.12dB
32pF          352kHz       1.00dB
22pF          246Hz         2.27dB
12pF          197kHz       5.45dB

Chosen value: 47pF.

Initial centre frequency with 34 turns on the main windings was 3.2MHz.
Removal of two turns from each side resulted in 3.386MHz centre frequency.
Removal of one more turn from each side resulted in 3.547MHz centre frequency.
Final removal of one more turn from each side resulted in 3.649MHz centre frequency.
You can see that each turn removed shifts the centre frequency by something like 100kHz (very
approximately). In this case the number of turns was therefore reduced equally from both transformers,
from 34 turns to 30 turns. It would not necessarily always be the case that the reduction is equal on both
transformers.

If this was read you sorta get a picture.  Two things wrong number of turns (or some variation is wire
distribution) and the choice of the coupling cap.  This can result in a wider filter and if mis-tuned
(there are two tuned circuits) it will be broader but also lossier.  Tuning is not totally by the caps
as a higher value trimmer is hard to get and bulky.  Also as the coils are closer to tuned that
trimmer has greater effect.

Allison
-------------------------------
Please reply on list so we can share.
No direct email, it goes to bit bucket due address harvesting in groups.IO


Andrew Lenton
 

 Hi,

 

I have made the 80M BPF I have checked the hot ends of the coils and the direction of the windings all good , and the best I can get is below response:

 

Components:

 

C2 and C5    470 PF

C3/4            49PF

T1 and T2 3.8 µH

 

I was hoping for less than 6dB loss, but in the big scheme of things, when listening to 80M on my K3S; I normally have some 12 dB of attenuation in, the important issue is to minimise out of band signals.

73 Andrew



ajparent1/KB1GMX
 

I built up an 80M BPF using LPF kit and  and got with minimal effort 2.9db loss
based on 30 turns T37-2, 470P and 47 (not 4.7) pf.

I got 450khz bandwidth (to -6db) and insertion loss at 3.75 of 1.9db.  By the book (manual).
Also the fall off above  4.5mhz was steeper and greater.

the trimmers have minimal range of adjustment so spacing of turns and adding/removing
one is the tuning technique

IF the link winind is not the same both sides losses are higher and it give better result
if wound on the grounded end of the larger winding.  Also wound with the same direction.

Either a measurement error of the test setup does not terminate the filter in 50 ohms
both sides.  Test setup must have good ground plane and I used for the quick test a
coax from and to both ends to the respective filter to the VNA or SA/TG

The NanoVNA and my Rigol DSA815T were within less than .1db of each other.


Allison
-------------------------------
Please reply on list so we can share.
No direct email, it goes to bit bucket due address harvesting in groups.IO


DL2ARL
 

Hello,
I can confirm the insertion loss stated above for the 80m BPF, the one intended to work with the U3S Rx (not the one in the QCX). As you can see in the picture attached , my BPF had even slightly more insertion loss than that the 5dB of DK3UW above.

I can not confirm the behavior of the trimmers: mine worked as intended, did what they are supposed to do. But one has to know where to look after the effect of turning the trimmers: the S21 transmission curve (red in the picture attached) does hardly changes it's absolute position. To see what the trimmers do you have to watch the S11 reflection curve (blue in the picture). The effect of the individual trimmers is clearer to be seen watching the two distinct "dips" of the S11. This is the nature of the beast.

The blue curve showing the reflection factor S11 at the input of the BPF also tells us something else. The reflection factor in the middle of the curve is about 30% on the linear scale chosen (corresponding to a 10dB Return Loss or, if you prefer, to a VSWR of abt. 1,8).  With these figures in mind one can say that the mismatch is the cause of some of the insertion loss. This assumption is right, but the loss due to a VSWR so low is only about 0,4dB, not more.

Probably the rest of the uncommon high insertion loss comes from the red toroids used. I have experimented changing the fixed capacitors with no success and the trimmers will be the same (not checked).

The -3dB bandwidth of the BPF being almost a half of MHz (well suited for the purpose), increasing the coupling to decrease the insertion loss is not an option. There must be something in there eating up the Q Factor: the design is straight forward and can not be blamed.

But the issue is of no concern, at least not for me. As Hans states in the manual, the filters are a huge field for experiments. Myself, intending to use the Rx together with an U3S WITH a LPF multi band switching board, I think of replacing the BPF with a HPF. What do I mean by that? Let me explain:

Using a multi-band U3S with a monoband Rx would be a "no-brainer". Unless there will be somewhen a switching board for the BPF's too, I think of following: Replacing the BPF  with a 5 or 7 pole High Pass Filter (HPF) with a cut-off frequency of about 3,2..3,3MHz would get me rid of that nasty QRM coming from below, all those switching power supplies, computers and what's left from the poor good old AM Radio.

Together with the LPF that is in front of the U3S, the "combo" would give me some sort of Band Pass Filtering, at least for the lower bands (80..60..40..30m) near the cut-off frequency of the HPF. Yes, I know, there is lots of QRM in-between those HPF-LPF corners, but ... I think I can live with this and I prefer this solution to having to manually change the BPF each time I change the band of the U3S. Used for band observation, his Rx is not supposed to become my ultimate DXCC chaser. Any better ideas welcomed however.

Thank you for the bandwidth!
73, Razvan DL2ARL


Martin DK3UW
 

thanks for all the feedback.

Hajo and I had other things to work on, thats why we did not come back. I am on the road again next week but then I will come back to our findings hopefully.

73's
Martin
DK3UW


DL2ARL
 

...and because it was so much fun, I did almost the same for 40m. A bang of a band here, full qrm. But for this band, I lacked a BandPass from qrp-Labs and feeling no desperate need for one at the moment, I thought I give it a try my way, by cascading the original LowPassFilter sitting in the U3S Filtershotgun and a High Pass Filter I rolled myself with somewhat more nervous coefficients, not caring about the resultig ripple.

This is what I got: the picture should be as self explaining as the one in the former message. It is only weird that the I measured the HighPass as low as 1MHz whereas the original qrpLabs LowPass is only measured 5MHz and above.
This is my way of thinking qrp which has become a state of the mind. Why should I be interested in the behavior of a LowPass way down below?

What I got perfectly suits my needs. The QRM I have at my location comes mainly from below and getting rid of those ugly ones below 3MHz helps a lot.

"whach out for the chimneys"

Yours friendly, Razvan DL2ARL


DL2ARL
 

Hello group,
well... not wanting to stay theoretical, I implemented the thoughts  in the message above #43560 and cascaded the Low Pass Filters (LP) sitting on the filter switch board of the U3S with static, non-switchable HighPass FIlters sticking into the Rx PinHeaders where a BandPass is supposed to go to.

What I was curious to see is what the difference is between this combo and the standard, official BandPassFIlter (BP). From these I only built one for 80, but it is enough for the question here.

This is what I got:

The picture below should be self explaining. Important are the red (BP) and the green (LP&HP) curves. The other two curves are the "root" of the LP&HP combo, the individual frequency responses of the original LowPass from qrp-labs and my own HighPass based on the same idea, but with nervous coefficients (Cheby instead Butterworth). Cascading the filters yields the green curve.
Compared to the red BandPass filter.... well there is no free lunch. I could have pulled the skirts of the two filters nearer to each-other, but the combo depicted here perfectly fits my needs, for a while...

About the -5dB insertion loss of the original BandPass.... there is somewhere in this net an ongoing discussion. Some said that you might need "heavy" groundplanes below the filter when measuring it. I did the experiment, with and without "heavy" groundplanes and must say that I do not subscribe to this oppinion, without wanting to start a chest-beating discussion here (the coupling cap used is thea standard one prescribed by the manual.. I recall 47pF). 

This is what I got, your mileage might vary. I need this configuration of cascaded filters NOT as a big improvement (because it is obviously not a betterment), but because this is what I need at the moment. For dedicated setups on single bands I will probably build other and other U3S and corresponding Rx'es. It is so much fun!

Yours friendly, Razvan DL2ARL


geoff M0ORE
 

Just to be pedantic and I appreciate that English is not everyone's first language ( and this will result in a lot of comments to give me something to read now we are in a lockdown situation again ), a negative loss is a gain.

Hence "-5dB insertion loss" is a 5dB gain.

Have a good weekend.

On 11/09/2020 21:42, DL2ARL wrote:
Hello group,
well... not wanting to stay theoretical, I implemented the thoughts  in the message above #43560 and cascaded the Low Pass Filters (LP) sitting on the filter switch board of the U3S with static, non-switchable HighPass FIlters sticking into the Rx PinHeaders where a BandPass is supposed to go to.

What I was curious to see is what the difference is between this combo and the standard, official BandPassFIlter (BP). From these I only built one for 80, but it is enough for the question here.

This is what I got:

The picture below should be self explaining. Important are the red (BP) and the green (LP&HP) curves. The other two curves are the "root" of the LP&HP combo, the individual frequency responses of the original LowPass from qrp-labs and my own HighPass based on the same idea, but with nervous coefficients (Cheby instead Butterworth). Cascading the filters yields the green curve.
Compared to the red BandPass filter.... well there is no free lunch. I could have pulled the skirts of the two filters nearer to each-other, but the combo depicted here perfectly fits my needs, for a while...

About the -5dB insertion loss of the original BandPass.... there is somewhere in this net an ongoing discussion. Some said that you might need "heavy" groundplanes below the filter when measuring it. I did the experiment, with and without "heavy" groundplanes and must say that I do not subscribe to this oppinion, without wanting to start a chest-beating discussion here (the coupling cap used is thea standard one prescribed by the manual.. I recall 47pF). 

This is what I got, your mileage might vary. I need this configuration of cascaded filters NOT as a big improvement (because it is obviously not a betterment), but because this is what I need at the moment. For dedicated setups on single bands I will probably build other and other U3S and corresponding Rx'es. It is so much fun!

Yours friendly, Razvan DL2ARL


DL2ARL
 

 geoff M0ORE

I humbly excuse my english as not being breed at Oxford.
As for the insertion loss being negative, I know a word that fits  marvelously: in german they say to it Korinthenkacker.
Your fine remark has helped a lot to  clear the problem described above, thank you.
Enjoy your digestion.

Yours Razvan DL2ARL


geoff M0ORE
 

Hi Razvan,

I am not saying that insertion loss is negative, it is the negative sign that I was referring to.

Geoff

On 13/09/2020 00:14, DL2ARL wrote:
 geoff M0ORE

I humbly excuse my english as not being breed at Oxford.
As for the insertion loss being negative, I know a word that fits  marvelously: in german they say to it Korinthenkacker.
Your fine remark has helped a lot to  clear the problem described above, thank you.
Enjoy your digestion.

Yours Razvan DL2ARL


Alan G4ZFQ
 

it is the negative sign that I was referring to.
Geoff

Did you look up what Korinthenkacker means?

Razvan knows he used a double negative but what does it matter?
Everyone knew what he meant.

73 Alan G4ZFQ


KI7MWA
 

This discussion has been very informative, but has failed (as I see it) to respond to the original question, which is, what are the potential sources of insertion loss for the QRP-Labs standard bandpass filter kit.

What are the potential sources of loss, and how would one identify them?  

One potential source mentioned is the T37-2 toroid.  How would you (anyone) go about testing this?

Are there other potenital sources of loss?  (Other than poor construction technique.)  And how would one go about testing them.

I have built a number of these bandpass filters, and, with the exception of the 80 meter filters, the insertion losses have all been below 2dB.  I have been unable to achieve a loss of less than 5dB on either of the two 80 meter filters I have built.


Alan G4ZFQ
 

I have built a number of these bandpass filters, and, with the exception of the 80 meter filters, the insertion losses have all been below 2dB. I have been unable to achieve a loss of less than 5dB on either of the two 80 meter filters I have built.
I cannot answer your question but why worry?
On 80m it will not matter, antenna noise is high, an attenuator is probably more useful rather than lower loss.

73 Alan G4ZFQ


KI7MWA
 

Well...okay, I guess...  Still, it would be interesting to learn how Hans was able to achieve an insertion loss of 1.27 dB (as reported in the bpf construction manual) while I have been unable to get less than 5 dB loss.

But, thank you, none-the-less.