Topics

VK4PP uBitx LPF BANK add-on Board

Nick VK4PP
 

Hi All.
Hows this look:


73 Nick VK4PP

Nick VK4PP
 

More:

ARNAB SEN VU2CMV
 

Dear OM,
               Will you make it available as a kit ? DIY or Assembled ?
73s

ARNAB SEN ( VU2CMV )
6/35 JM SENGUPTA ROAD.
B-ZONE.
DURGAPUR-713205.
MOB:9434833047


On Fri 31 Aug, 2018, 09:10 Nick VK4PP, <nickpullen@...> wrote:
More:

Nick VK4PP
 

Probably just the boards, You need to re-use the LPF from the existing board.
Option to add a Firth filter is you want.

Relays are Omron g6s-2 12v

73 Nick VK4PP

Timothy Fidler
 

Nick
1. Wired OR diodes used for Relays J1A B  great idea. and I can see you seem to have allowed for relay contacts being opposite sensed. Ie J1A B set is active when Energised and Energised is out of cct.
2 . you have three signals and because of the wired or you have effectively four states without using a decoder. So how is the R5 to be energised. ?? Is it there for case of special software mod plus I2C or ....?

3. You need to define the relay functions.  From a Sota point of view it makes sense that J1AB set is a 20/30m Filter. Ie NO relay energised when working 20m.
then J2 becomes 80 m. then J3 40m  and J4 becomes 15/17 and J5 becomes Ten m after some magic.
4. You have allowed  spike suppression diodes for all relays except first pair. Forgotten or  there is a reason. ? 
5. Kees Talen is also talking of a Five relay board design.  With Decoding to be done Off board somehow. Perhaps you need to check up his thoughts.
6. Given  5.  I would suggest that you allow sufficient pads for a  74HCT42 decoder and  a 78L05 to power it.  These pads could be on basis of wired chip mounted as SM (legs become feet at 2mm up from tips) so the cost of the holes is not there.  You will also need 4 off BS250 P Mosfets  (simplest solution and they are cheap) to activate the last four relays sets.  The points where the Wire goes off to the Wired OR should be  a through hole rondel.  All this could be squeezed in up T LHS of board as presented without much increase in board space.

The final wiring could be done with  Cat E wiring cores stripped out for those who wanted to go this way . I would suggest the following mapping  000 as input to decoder  means nothing.  Same for 001.  010 activates relay set 2 and also energises J1AB via wired OR  .  ...etc all the way to 101. If you use PNP transistors, you will be up for one extra current control resistor per bit.  BS250s are logic level Mosfets  RDS on approx 14R.  Futurlec.com have  them Rockby in Australia list them but have to order them in. 

The mappings and Relay usage have to be defined up front or  the software will be a schmozzle. 

Is the current voltage going to K1 coil 12V .. the schematic calls it Tx voltage. It appears likely to be 12VDC but without details of the guts of the Raduino I cannot be sure.


Gordon Gibby <ggibby@...>
 

Two suggestions:

1. Looks pretty tight.   You might leave a bit more room for the hobbyist to reassemble the filters

2.  I don’t see RF bypassing on the coil wiring—so signal can bypass the filters on those wires.  Consider caps maybe even series inductors - I used smd 100 micro Henry




On Aug 31, 2018, at 03:29, Timothy Fidler <engstr@...> wrote:

Nick
1. Wired OR diodes used for Relays J1A B  great idea. and I can see you seem to have allowed for relay contacts being opposite sensed. Ie J1A B set is active when Energised and Energised is out of cct.
2 . you have three signals and because of the wired or you have effectively four states without using a decoder. So how is the R5 to be energised. ?? Is it there for case of special software mod plus I2C or ....?

3. You need to define the relay functions.  From a Sota point of view it makes sense that J1AB set is a 20/30m Filter. Ie NO relay energised when working 20m.
then J2 becomes 80 m. then J3 40m  and J4 becomes 15/17 and J5 becomes Ten m after some magic.
4. You have allowed  spike suppression diodes for all relays except first pair. Forgotten or  there is a reason. ? 
5. Kees Talen is also talking of a Five relay board design.  With Decoding to be done Off board somehow. Perhaps you need to check up his thoughts.
6. Given  5.  I would suggest that you allow sufficient pads for a  74HCT42 decoder and  a 78L05 to power it.  These pads could be on basis of wired chip mounted as SM (legs become feet at 2mm up from tips) so the cost of the holes is not there.  You will also need 4 off BS250 P Mosfets  (simplest solution and they are cheap) to activate the last four relays sets.  The points where the Wire goes off to the Wired OR should be  a through hole rondel.  All this could be squeezed in up T LHS of board as presented without much increase in board space.

The final wiring could be done with  Cat E wiring cores stripped out for those who wanted to go this way . I would suggest the following mapping  000 as input to decoder  means nothing.  Same for 001.  010 activates relay set 2 and also energises J1AB via wired OR  .  ...etc all the way to 101. If you use PNP transistors, you will be up for one extra current control resistor per bit.  BS250s are logic level Mosfets  RDS on approx 14R.  Futurlec.com have  them Rockby in Australia list them but have to order them in. 

The mappings and Relay usage have to be defined up front or  the software will be a schmozzle. 

Is the current voltage going to K1 coil 12V .. the schematic calls it Tx voltage. It appears likely to be 12VDC but without details of the guts of the Raduino I cannot be sure.


Timothy Fidler
 

https://www.altronics.com.au/search/all/?sr=Din%2BSocket&pg=16

RE LPF Board 
above as example page.. I can't be  bothered with time of  doing a Cut and paste just to make up a picture of something you know about.. These Green headers must be available all over surely and the plugs to fit em.

If sending signals and power over from main board (using the decoding scheme) you might like to make sure you can
get 12V, zero rail and the three bit signals over on a green horizontal  5.08 pitch header  (this assumes 5V for the decoder generated on the PCB ).. then the user just makes up a cable that picks up all the points on the main board .. direct soldered, Cable  fixed in some way to main PCB and then just plugs into daughter board.   Saves people mixing voltages if they pull the boards apart and then reassemble them as such headers are polarised. .......

Warren Allgyer <allgyer@...>
 

Suggestions based on experience and good practice are all good. It is my experience however that some fairly simple measurements can and need to be made in order to confirm a layout.

First, simply connect up the input and output on a completely unpopulated board and measure the attenuation. If you find less than 80 dB of attenuation at any frequency below 100 MHz, you are probably not going to like the end result. I would then add the relays and check it again. At this point your attenuation needs to be at least 70 dB. Add all remaining components except the filters themselves and check it again. Finally, one at a time, add the filters and check after each one is added.

If, at any step in the process, your input to output crosstalk gets higher than -60 dB, you are not going to be very happy with the outcome. Note all of these measurements are made with the relays deactivated so there is no actual intended path between input and output.

When working on a similar project with Tony Parks for an all-band Softrocks RXTX our first two prototype filter boards measured less than 30 dB isolation between input and output. And we thought we were following good layout and RF principles.

BTW, on the Softrocks project we finally got there........ only to realize that the radio would never make compliance goals without a similarly switched interstage filter prior to the RF drive chain. At that point we had pushed the cost of the radio beyond the target and the project was scrapped.

My bottom line reaction is: layout as best you know how but test, test, test.

WA8TOD

Glenn
 

All good points Warren.   I did some checks of isolation using a typical DIP relay, type HK19F. DPDT

I made a small jig to hold the relays and measure isolation across contacts etc.
Typically it was -34dB across open contacts,
-32dB from one set of contacts to the other. (common terminal)
-41dB from open terminals on each side
-39dB from common of one pole to open on the other pole.
-48db from common to coil.
 At 30MHz.  Better as freq drops of course.
That's without effects of a PCB and various other traces.

glenn

iz oos
 

I was wondering why the Softrock TXRX had just 3 close bands (mine 10-12-15m). Now I know there were good reasons for that!


Il 31/ago/2018 12:17, "Warren Allgyer" <allgyer@...> ha scritto:

Suggestions based on experience and good practice are all good. It is my experience however that some fairly simple measurements can and need to be made in order to confirm a layout.

First, simply connect up the input and output on a completely unpopulated board and measure the attenuation. If you find less than 80 dB of attenuation at any frequency below 100 MHz, you are probably not going to like the end result. I would then add the relays and check it again. At this point your attenuation needs to be at least 70 dB. Add all remaining components except the filters themselves and check it again. Finally, one at a time, add the filters and check after each one is added.

If, at any step in the process, your input to output crosstalk gets higher than -60 dB, you are not going to be very happy with the outcome. Note all of these measurements are made with the relays deactivated so there is no actual intended path between input and output.

When working on a similar project with Tony Parks for an all-band Softrocks RXTX our first two prototype filter boards measured less than 30 dB isolation between input and output. And we thought we were following good layout and RF principles.

BTW, on the Softrocks project we finally got there........ only to realize that the radio would never make compliance goals without a similarly switched interstage filter prior to the RF drive chain. At that point we had pushed the cost of the radio beyond the target and the project was scrapped.

My bottom line reaction is: layout as best you know how but test, test, test.

WA8TOD


Warren Allgyer <allgyer@...>
 

Glenn

That 32 dB number is the killer for the uBitx relay scheme. It gets at least 3 dB worse with each relay added to the chain and all four relays are in the chain.

WA8TOD

Jerry Gaffke
 

Warren,

The relays are not that bad, primary issue on the uBitx is how the board is routed. 

Look at the uBitx schematic, find KT1 near the 30m LPF of L11,12,13.
Assume the relays are in the relaxed state as drawn, for operation at 30m.

The 32dB number is for isolation from KT1 pin 14 to 16.
So when operating at 30mhz, we have 32dB of loss going from pin 14 to pin 16, 
through the 20m filter at L14,15,16, and then another 32dB of loss
going from KT1 pin 1 to 3 at the other side of the filters.  A total of 64dB down
to sneak around the KT1 filter by that path.  The third filter is not a factor for this
since it is an additional 64dB down.  

The roughly 40dB loss in going around the 30m LPF from KT1 pin 12 to 5
is the primary factor that the relays themselves contribute to blow-by,
that's acceptable unless the PA is putting out harmonics that are as strong as the fundamental.
That 40dB path disappears if we use 3 additional relays at the far side of the LPF's.
 
Blow-by when operating on 30m is the primary consideration,
operating at lower frequencies will have proportionally less capacitive coupling through the relays.



Someone proposed the scheme below a few weeks ago using a total of 4 DPDT relays
which I have named RA, RB, RC, RD.
I forget exactly who brought this up, it came as a web pointer to schematics for a board in production. 
There's blow-by when operating on 30m of 32+32=64dB through RB, RD, and the 20m LPF.
There is also a path of path across RA, RC, and the 40m LPF of 32+32=64dB.
And a path across the RB and RD relay sections of 40+40=80dB through that same 40m LPF.
With all three blow-by paths of 64,64, and 80 dB in parallel, we have a net of about 57dB of blow-by.
Not bad for just 4 relays, and the board routes sort out nicely.

                    o---------[10m]---------o
               +----o RB                 RD o----+
               |    o---------[20m]---------o    |
               |                                 |
         o-----+                                 +--------o
---------o RA                                          RC o---------
         o-----+                                 +--------o
               |                                 |
               |                                 |

               |    o---------[40m]---------o    |
               +----o RB                 RD o----+
                    o---------[80m]---------o

Jerry, KE7ER
 


On Fri, Aug 31, 2018 at 06:01 AM, Warren Allgyer wrote:
That 32 dB number is the killer for the uBitx relay scheme. It gets at least 3 dB worse with each relay added to the chain and all four relays are in the chain.

Gordon Gibby <ggibby@...>
 

I decided to test the  little daughterboard system that I built to move the input filtering off theboard.


Everything connected:  80 metersignal (has to go thru three relays on the board and 3 relays off the board):  S9+30 dB in my apprxoimately 70db attenuation system.


Remove the wire from the soldered-to-the-trace input to just the 80 meter LPF, lay it a couple centimeters to the side....


Signal  dropped to barely hearable and S0.


30 db for  the portion above S9  (verified with attentuators)

18 db more toget to S7.5  (verified  with attentuators)


Unknown from S7.5 to S0 but likely at  least 6db/Sunit (sincethose above7 were 12dB perS unit measured) --42 db


Total isolation at 80 meters roughly 90 dB

​at the very least (down to the verified  S7.5 point) 48 dB + whatever it takes to go from 7.5 to0  

Icom718.


So the isolation at 80 meters seems good....probably declines with frequency 


gordon



From: BITX20@groups.io <BITX20@groups.io> on behalf of Glenn <glennp@...>
Sent: Friday, August 31, 2018 6:40 AM
To: BITX20@groups.io
Subject: Re: [BITX20] VK4PP uBitx LPF BANK add-on Board
 
All good points Warren.   I did some checks of isolation using a typical DIP relay, type HK19F. DPDT

I made a small jig to hold the relays and measure isolation across contacts etc.
Typically it was -34dB across open contacts,
-32dB from one set of contacts to the other. (common terminal)
-41dB from open terminals on each side
-39dB from common of one pole to open on the other pole.
-48db from common to coil.
 At 30MHz.  Better as freq drops of course.
That's without effects of a PCB and various other traces.

glenn

Jerry Gaffke
 

In many places, I said "30m" when I should have said 10m (or 30mhz) in that previous post.
Here's an edited version:



The relays are not that bad, primary issue on the uBitx is how the board is routed. 

Look at the uBitx schematic, find KT1 near the 10m (10 meters, or 30mhz) LPF of L11,12,13.
Assume the relays are in the relaxed state as drawn, for operation at 10m.

The 32dB number is for isolation from KT1 pin 14 to 16.
So when operating at 10m, we have 32dB of loss going from pin 14 to pin 16, 
through the 20m filter at L14,15,16, and then another 32dB of loss
going from KT1 pin 1 to 3 at the other side of the filters.  A total of 64dB down
to sneak around the KT1 filter by that path.  The third filter is not a factor for this
since it is an additional 64dB down.  

The roughly 40dB loss in going around the 10m LPF from KT1 pin 12 to 5
is the primary factor that the relays themselves contribute to blow-by,
that's acceptable unless the PA is putting out harmonics that are as strong as the fundamental.
That 40dB path disappears if we use 3 additional relays at the far side of the LPF's.
 
Blow-by when operating on 10m is the primary consideration,
operating at lower frequencies will have proportionally less capacitive coupling through the relays.



Someone proposed the scheme below a few weeks ago using a total of 4 DPDT relays 
which I have named RA, RB, RC, RD.
I forget exactly who brought this up, it came as a web pointer to schematics for a board in production. 
There's blow-by when operating on 10m of 32+32=64dB through RB, RD, and the 20m LPF.
There is also a path of path across RA, RC, and the 40m LPF of 32+32=64dB.
And a path across the RB and RD relay sections of 40+40=80dB through that same 40m LPF.
With all three blow-by paths of 64,64, and 80 dB in parallel, we have a net of about 57dB of blow-by.
Not bad for just 4 relays, and the board routes sort out nicely.

                    o---------[10m]---------o
               +----o RB                 RD o----+
               |    o---------[20m]---------o    |
               |                                 |
         o-----+                                 +--------o
---------o RA                                          RC o---------
         o-----+                                 +--------o
               |                                 |
               |                                 |

               |    o---------[40m]---------o    |
               +----o RB                 RD o----+
                    o---------[80m]---------o

Jerry, KE7ER
 

Jerry Gaffke
 

I interpret this measurement from Glenn:
    "-32dB from one set of contacts to the other. (common terminal)"
as being from KT1 pin 14 to pin 16 as seen in the uBitx schematic.

Also, interpret his -39 and -41 dB numbers to be from one relay section to the other,
which I averaged as -40dB.    For example, KT1 pin 12 to pin 5.

I now suspect those assumptions are incorrect, but the analysis could easily be adjusted to for any change to those two numbers.

Jerry


On Fri, Aug 31, 2018 at 03:40 AM, Glenn wrote:
All good points Warren.   I did some checks of isolation using a typical DIP relay, type HK19F. DPDT

I made a small jig to hold the relays and measure isolation across contacts etc.
Typically it was -34dB across open contacts,
-32dB from one set of contacts to the other. (common terminal)
-41dB from open terminals on each side
-39dB from common of one pole to open on the other pole.
-48db from common to coil.
 At 30MHz.  Better as freq drops of course.
That's without effects of a PCB and various other traces.

glenn

Gordon Gibby <ggibby@...>
 

The measurements that I made  for the daughterboard  bleed-through indicate that for the ENTIRE SYSTEM,  including the printed circuit board, connecting leads, relays, coils, coil wiring and all the RF wiring,  the bleed  through is almost NEGLIGIBLE. (lifted the bottom green wire in the accompanying photo and measured bleed through that had to go through ALL SIX RELAYS).    Far, far below the pass through the filter....  down to unmeasureable on the ICOM718..whereas properly filtered harmonics  were depressed into the S5 or S6 range from 30dB overS9  fundamental (as predicted by others' measurements indicating the FILTERS had acceptable performance)     Measurement performed at 3.5  MHz.  Presumably capacitive coupling would increase related to frequency, perhaps 10 times the amount would be expected at 30 MHz --- but that would only be 10dB more signal, perhaps reaching to S2   I don't want to risk my finals by doing  the experiment over and over.   The bleed through with a daughterboard --- even with unshielded wires --- is far far far below  the requirements to meet FCC specs.    (-43 dB being  somewhere around S7.5 in my tests).    Get the relays separated, get the things apart physically that need to NOT bleed into each other....and it isn't that difficult to fix the harmonics.  (The spurs are a different issue).   





From: BITX20@groups.io <BITX20@groups.io> on behalf of Jerry Gaffke via Groups.Io <jgaffke@...>
Sent: Friday, August 31, 2018 12:17 PM
To: BITX20@groups.io
Subject: Re: [BITX20] VK4PP uBitx LPF BANK add-on Board
 
I interpret this measurement from Glenn:
    "-32dB from one set of contacts to the other. (common terminal)"
as being from KT1 pin 14 to pin 16 as seen in the uBitx schematic.

Also, interpret his -39 and -41 dB numbers to be from one relay section to the other,
which I averaged as -40dB.    For example, KT1 pin 12 to pin 5.

I now suspect those assumptions are incorrect, but the analysis could easily be adjusted to for any change to those two numbers.

Jerry


On Fri, Aug 31, 2018 at 03:40 AM, Glenn wrote:
All good points Warren.   I did some checks of isolation using a typical DIP relay, type HK19F. DPDT

I made a small jig to hold the relays and measure isolation across contacts etc.
Typically it was -34dB across open contacts,
-32dB from one set of contacts to the other. (common terminal)
-41dB from open terminals on each side
-39dB from common of one pole to open on the other pole.
-48db from common to coil.
 At 30MHz.  Better as freq drops of course.
That's without effects of a PCB and various other traces.

glenn

Warren Allgyer <allgyer@...>
 

Gordon

"Bleed through" or "blow by" at the operating frequency is not the issue. It is bleed through at the 5th and 7th harmonic that allows those harmonics to reach levels that are non-compliant.

WA8TOD

Gordon Gibby <ggibby@...>
 

Hi, I get that.   I but Icannot measure it at all frequencies.  I don't have a spectrum analyzer, I have a RECEIVER and I've calibrated portions of its response in order to use it as a poor-mans spectrum analyzer.    Presumably the blow-by is capacitive for the most part, but not at all parts.   Lacking a spectrum analyzer, I measure the effective blow by at ONE frequency and then draw assumptions at others based on a model that it is primarily capacitve in nature (which has been the major point of discussion for weeks now......)


So if you know that capacitive reactance goes down by 10X when you increase the frequency by 10X, and in both cases you are feeding into a 50 ohm load.......the voltage you're going to get will go up on the order of 10X.    


​Thus you can make a prediction (based on the model) of what the bleed through might be at higher frequencies.   Or you can risk your rig by making other measurements if you don't have any other way to do it (which is me).   Does that make more sense?


Cheers,


gordo



From: BITX20@groups.io <BITX20@groups.io> on behalf of Warren Allgyer <allgyer@...>
Sent: Friday, August 31, 2018 1:50 PM
To: BITX20@groups.io
Subject: Re: [BITX20] VK4PP uBitx LPF BANK add-on Board
 

Gordon

"Bleed through" or "blow by" at the operating frequency is not the issue. It is bleed through at the 5th and 7th harmonic that allows those harmonics to reach levels that are non-compliant.

WA8TOD

Warren Allgyer <allgyer@...>
 

Gordon

Your reasoning makes perfect sense. But I think you can measure at the actual harmonic frequencies which is where the issue would be . Your IC-718 can tune the third, fifth and 7th harmonics of 80 meters. So you can know exactly (within the limits of the S meter calibration and linearity) what the residual level is at the harmonic frequencies themselves. You have checked your S-meter and you know what represents 45-50 dB below S9+30 (normally one S unit = 6 dB but in real life S meters rarely replicate that). So just tune the harmonics directly and measure them. Or am I missing something?

WA8TOD

Gordon Gibby <ggibby@...>
 

Warren, you are right on, but you might not have seen all the posts where I did exactly what you are saying.   Both BEFORE the daughterboard and AFTER the daughterboard so i had proof of its effectiveness.   I also did some calibrations of the S-meter using BNC attenuators (which is all I have).   I then did the suggested measurements of today, measuring bleed-through (unmeasureable, whereas I was clearly able to measure the harmonics and able to measure their reduction by moving the input side of the relay chain OFF the board.)


Here, I'll see if I can find the posts (I'm not too used to that):


(1)Very first "before" measurements:   https://groups.io/g/BITX20/message/57840

(2)Complete baseline data:   https://groups.io/g/BITX20/message/57974

(3) Calibration of the S-meter and proof that the measurement setup itself didn't have bleed thru:  https://groups.io/g/BITX20/message/58070

  [This distinctly showed that 1 S unit in the range of 7-9 is 12 db onthat receiver on that frequency)

(4)Photos of the measurement setup:  https://groups.io/g/BITX20/message/58070

(5) DAta AFTER daughterboard:   https://groups.io/g/BITX20/message/58070



What I CANNOT prove is that the receiver maintains identifical response at all frequencies (hence, is "flat")  however it is reasonable that the differnces aren't mroe than a few dB based on specifications for rigs of this price range etc.   


Poor man's spectrum analyzer.   What is problematic is the 24MHz harmonic range --- here I achieved only 6 dB reduction so there is something else going on there...maybe a series resonance in a filter or somethign I haven't thought of.   Every other range was dramatically reduced by the daughterboard system.   So for my purposes --- a low pass filter above 20 MHz and I have every digital band I need for what I need out of the radio and I don't need to use the bands where the "spurs" are a thorn.   



Hope that makes more sense.   

Cheers,


gordon




From: BITX20@groups.io <BITX20@groups.io> on behalf of Warren Allgyer <allgyer@...>
Sent: Friday, August 31, 2018 2:24 PM
To: BITX20@groups.io
Subject: Re: [BITX20] VK4PP uBitx LPF BANK add-on Board
 
Gordon

Your reasoning makes perfect sense. But I think you can measure at the actual harmonic frequencies which is where the issue would be . Your IC-718 can tune the third, fifth and 7th harmonics of 80 meters. So you can know exactly (within the limits of the S meter calibration and linearity) what the residual level is at the harmonic frequencies themselves. You have checked your S-meter and you know what represents 45-50 dB below S9+30 (normally one S unit = 6 dB but in real life S meters rarely replicate that). So just tune the harmonics directly and measure them. Or am I missing something?

WA8TOD