Corrections made to previous Measurements & Calculations on April Fool's Day

Well April Fool's Day turned out to be a constructive one for me after all. This morning after experimenting with Elsie and plotting the response of all the BPFs using firstly their nominal component values, then secondly using their L values adjusted to give a better response for each Band, I realised that the method I had used for measuring and calculating loss, which had been troubling me, was wrong. I should not in fact have terminated the input to the BPF with 50 ohms and then used the voltage Vin developed across it as the divisor when calculating the insertion loss from Vout. Instead I should have based the divisor on the voltage Vin developed across a 50 ohm load with only it connected to the output of the DDS with the BPF completely removed and disconnected. At first sight it might seem that this would produce an even worse result but in fact the reverse turns out to be true.
Since building my DDS and temporarily increasing drive to the 2N3866 for more output. I have been aware that although its output when measured on open circuit varies from a maximum of about 5 volts pk to pk at 2 MHz to a minimum of about 1.5 volts pk to pk at 20 MHz it does not behave similarly when loaded by 50 ohms, but instead it delivers a reasonably steady output of 1 volt pk to pk which is flat from about to 1 MHz to about 20 MHz. This would seem to imply that the internal Thevenin impedance of the DDS varies from about 200 ohms to 25 ohms over this frequency range. As a consequence the revised values used for Vin were taken from actual measurements made at each frequency of interest across  a 50 ohm load before connecting the DDS to the BPF and measuring Vout across the 50 ohm BPF output terminating load. The loss calculated using these two values thus being the true insertion loss of the filter when introduced into the system. Measured results now compare much more favourably to what I had expected and that predicted by the software model even though I have so far not made any further adjustments to the coils in the BPF since their alignment on Wednesday. I do intend however doing an alignment check again before I fit the coils for 160 Meters to ascertain if I can improve the performance of the individual BPFs still further. The results obtained at present being as follows:

80 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
3,300            1000              110          19.2
3,400            1000              340           9.4
3,500            1000              520           5.7
3,600            1000              520           5.7
3,700            1000              550           5.2
3,800            1000              420           7.5
3,900            1000              180          14.9
4,000            1000               90           20.9

40 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
6,900            1080              310          10.8
7,000            1080              510           5.8
7,100            1080              570           4.9
7,200            1080              540           5.4
7,300            1080              360           8.6
7,400            1080              190          14.4

30 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
9,900           1140                110         19.2
10,000         1140                250         12.4
10,100         1140                270         11.7
10,200         1140                150         17.1

20 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
13,900         1220               400          9.6
14,000         1220               500          7.7
14,100         1220               630          5.7
14,200         1220               700          4.9
14,300         1220               620          5.8
14,400         1220               580          6.4
14.500         1220               370          8.4

17 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
17,800         1100               190          15.4
17,900         1100               280          12.0
18,000         1100               340          10.2          
18,100         1100               300          11.4
18,200         1100               230          13.6
18,300         1100               190          15.4            

15 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
20,900          920                190          13.6
21,000          920                260          11.1
21,100          920                340           8.6
21,200          920                340           8.6
21,300          900                300           9.6
21,400          900                250          11.0
21.500          900                220          12.4

12 Meter Band
KHz       Vin mV pp   Vout mV pp  Loss dB
24,800          760                380          6.0
24,900          760                370          6.2         
25,000          740                370          6.0
25,100          740                380          5.8

10 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
28,000           640                215          9.4        
28,500           640                250          8.2
29,000           640                350          5.4
29,500           640                350          5.4
29,700           640                265          7.7
30,000           640                175         11.4

I have posted some Elsie files using nominal and tuned inductor values pertaining to the above at  http://groups.yahoo.com/group/BITX20/files/g3oth/  for comparison purposes. To view them you will need to download, save and then import them into the Elsie program when launched.          

Analysis of my BPF using Elsie

I havent done anything practical with my BPF since my last Blog, instead I have been scratching my head for reasons why with the exception of 12 and 10 Meters my measurements result in insertion loss values so much higher than I had expected. Either I havent tuned the BPFs correctly or the coils and / or capacitors I have used have a much lower Q than the norm, or the switching diodes have a much higher ohmic on value @ 10 mA than specified. I am not aware of whether there is a "right" way of going about tuning the BPF, the method I used was to use a succession of tweaks on all coils which because of interaction was repeated several times until I achieved a condition that provided a simultaneous minimum insertion loss at both band edges. The Qs of coils and capacitors I have used are quoted as being typically 85 and 2000, and the on resistance of diodes better than 1 ohm. I have not measured any of these independently, but I did try varying the DC bias current through the diodes from 5 to 20 mA and could barely perceive any change on the filter performance, so I doubt if they are responsible.
Although I was previously aware of Jim Tonne's Electrical Filter Software and had downloaded his free version of Elsie http://www.tonnesoftware.com/elsie.html some time ago I had never used it before today, so this morning I decided to explore it further to see if I could analyse my BPF using his software model. I discovered that by using the Manual Parts Entry feature in his program I was able to do this quite easily using the Schematic, Entry, Analysis, Plot and Tune features provided. As an example I have saved a file of a simulation of the 20 Meter Band Pass Filter which I have called 20MeterTuned.LCT at http://groups.yahoo.com/group/BITX20/files/g3oth/ which can be opened and viewed after launching the Elsie program. I am most impressed by this program, and am now going to model the other BPFs using it. The predicted results for the 20 Meter Band are better than I measured, even after substituting Qs as low as 50 and 500 for coils and capacitors respectively, so I suspect there must be something wrong with my previous tuning technique after all.
.

More BPF Measurements

I did some more random measurements on my BPF today as I was not happy with the earlier results. Although previously I had tuned all the coils such that a definite band pass was obtained for each of the bands, the insertion loss in most cases seemed to be higher than I had expected, As the tuning of the coils had remained untouched I decided to repeat some of the measurements on the higher bands again but with the BPF reversed. To my surprise with the noteable exception of the 12 and 10 Meter Bands the results in brackets shown as (*) for comparative purposes were quite different. Two thoughts came to mind, had I made an error whilst reading the range setting on one of the channels on my oscilloscope whilst taking some of the previous measurements (10 mV/cm rather than 20 mV/cm which would make an immediate difference of 6 dB) or is my tuning technique somehow to blame. So I need to take a deep breath and think about this some more and perhaps repeat all the measurements again before I start altering any core settings.

80 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
3,500            720                230           9.9
3,600            590                300           5.8
3,700            550                320           4.7
3,800            660                230           9.1

40 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
7,000            800                250         10.2
7,100            710                320           6.9
7,200            680                320           6.6
7,300            760                240           9.9

30 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
10,000          800                190         12.4
10,100          630                210           9.6
10,200          570                130         12.8

20 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
13,900        (1100)           (270)          (12)
14,000     860 (1060)   510 (340)    4.6 (9.9)
14,100     790 (980)     600 (430)    2.4 (7.1)
14,200     800 (940)     620 (510)    2.2 (5.4)
14,300     860 (1000)   550 (460)    3.9 (6.7)
14,400     900 (1060)   450 (350)    6.0 (9.6)
14,500        (1100)           (260)        (12.4)

17 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
17,900        (1040)           (170)        (15.9)
18,000     580 (960)     380 (220)   3.6 (12.8)
18,100     640 (960)     360 (210)   5.0 (13.2)
18,200     660 (1000)   300 (160)   6.9 (15.9)

15 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
21,000     480 (880)    190 (130)   8.0 (16.5)
21,100     430 (880)    230 (130)   5.5 (16.5)
21,200     440 (760)    250 (190)   4.9 (12)
21,300     480 (760)    240 (170)   6.0 (13.2)
21,400     500 (820)    200 (140)   8.0 (15.4)
21.500     480 (840)    160 (110)   9.6 (17.7)

12 Meter Band
KHz       Vin mV pp   Vout mV pp  Loss dB
24,800     280 (260)    195 (190)    3.1 (2.7)
24,900     290 (240)    200 (190)    3.2 (2)
25,000     275 (260)    205 (190)    2.5 (2.7)
25,100     350                190          5.4

10 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
28,000           325                150          6.7
28,500           215                190          1.1
28,700          (250)              (210)       (1.5)
29,000           270                215          1.9
29,500           365                180          6.2
29,700           365                150          7.7

Picture of the Test Setup used for my earlier BPF Measurements

I have added the picture above to show the test setup used for obtaining the measurements detailed in my previous blog. The picture was taken whilst testing the response of the BPF at 7.1 MHz. A crocodile clip supplying 5 volts to the 40 Meter BPF PCB Band Select Pin can be seen temporarily connected to activate the appropriate switching diodes for this purpose. Two terminating resistors (I cheated and used 47 ohm rather than ideally 50 ohm !) can be seen temporarily soldered to the input and output ports of the BPF on the underside of the PCB. The oscilloscope probes set at x10 are shown connected to and monitoring the voltage across these resistors on the two channels of the oscilloscope which are both set to display at 10mV / cm. The larger trace on the oscilloscope therefore represents an actual input signal voltage of 710 mV peak to peak and the smaller trace an output signal voltage of 320 mV peak to peak going to and coming from the BPF when connected to the DDS which is shown set to deliver a CW output of  7.100000 MHz. The insertion loss of the filter when measured under these conditions thus appears to be approx 6.9 dB. Also note the 180 degree phase reversal which has occured to the signal during its passage through the BPF.

G6LBQ 9 Band BPF Project

This project is based on a design by Andy G6LBQ and employs a prototype PCB supplied by Sunil VU3SUA and Inductors supplied by Spectrum Communications in the UK and is primarily intended for use in the G6LBQ Multi Band BITX but could be used in conjunction with any ham band receiver or transceiver where a diode switched BPF is required.

I received the prototype PCB kindly donated by Sunil from India during thr first week of March 2012. As I had earlier already purchased the coils in a joint buy with Andy from Spectrum Communications in anticipation of the PCB being produced, I was quickly able to confirm that all coils and components I had on hand fitted perfectly to the PCB.
The special BA243 PIN diodes and NPO ceramic capacitors required to complete the build were then procured from local UK distributors and Ebay auctions using the values that had been computed by Andy and checked earlier by myself. These were delivered and fitted without difficulty to the board the following week. The PCB is laid out to accomodate 0.2 inch (5mm) spaced capacitors throughout and I used Ceramic NPO Disc capacitors for the lower filter values and Ceramic Multilayer NPO capacitors for the higher filter values and Ceramic Multilayer Y5V capacitors for coupling and de-coupling purposes.
Rather than use the 470 ohm and 390 ohm resistor value combination to forward bias the PIN diodes as recommended in Andy's BOM based on operation from 12 volts I chose instead to use 100 ohms and 330 ohms so that I could still operate and switch them at 10 mA from 5 volts provided by the output band switch control PIC utilised in my recently completed VU3CNS DDS.

Apart from fitting the coils used for 160 meters, my BPF PCB assembly is now complete and I have carried out some provisional tests this weekend using my VU3CNS DDS as a signal generator and Tektronix Model 465 Oscilloscope to align and evaluate it's performance as a stand alone unit.

To do this I terminated both input and output ports of the BPF with 50 ohm resistors and across each resistor connected x10 probes which in turn are connected to each of the 2 channel inputs of my scope. I then connected the output of my DDS set to 3.5 MHz to the input port of the BPF and applied a 5 volt DC signal to the 80 meter Band Selector Port and observed the resultant traces on the oscilloscope whilst adjusting each of the corresponding band coil cores for maximum output from the BPF. I repeated the procedure several times using signals up to 3.8 MHz from the DDS until a compromise was reached whereby the reponse of the BPF was as uniform as possible and attenuation least between the edges of the band.
The remaining bands were then aligned similarly.
Whilst performing this alignment I noticed interaction between all three of the band coils, adjustment of the centre coil seemed to have most effect on the output amplitude. However near resonance the input voltage amplitude of the BPF was also found to be notably more diminished and this was also found to be most effected by adjustment of the core nearest to the input, presumably both effects being as a result of loading on the DDS output due to the input impedance of the BPF becoming lower at resonance than that provided solely by the 50 ohms fixed resistor when the BPF is not near resonance.
I have taken this into account when computing the attenuation loss in the BPF for all the Bands aligned in this manner and the results are as follows.

80 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
3,500            720                230           9.9
3,600            590                300           5.8
3,700            550                320           4.7
3,800            660                230           9.1

40 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
7,000            800                250         10.2
7,100            710                320           6.9
7,200            680                320           6.6
7,300            760                240           9.9

30 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
10,000          800                190         12.4
10,100          630                210           9.6
10,200          570                130         12.8

20 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
14,000          860                510           4.6
14,100          790                600           2.4
14,200          800                620           2.2
14,300          860                550           3.9
14,400          900                450           6.0

17 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
18,000          580                380           3.6
18,100          640                360           5.0
18,200          660                300           6.9

15 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
21,000          480                190           8.0
21,100          430                230           5.5
21,200          440                250           4.9
21,300          480                240           6.0
21,400          500                200           8.0
21.500          480                160           9.6

12 Meter Band
KHz       Vin mV pp   Vout mV pp  Loss dB
24,800          280                195          3.1
24,900          290                200          3.2
25,000          275                205          2.5
25,100          350                190          5.4

10 Meter Band
 KHz       Vin mV pp   Vout mV pp  Loss dB
28,000           325                150          6.7
28,500           215                190          1.1
29,000           270                215          1.9
29,500           365                180          6.2
29,700           365                150          7.7

The losses for some of the above bands seem to be higher than I would have expected whilst in other cases they seem to be lower, so I am open to any suggestions as to why this might or appear to be so and any comments regarding the validity of my measurement and alignment techniques would be appreciated.

The reason why I have not fitted the 160 Meter coils as yet is because according to theory ideally the windings on the coupling coils should each have an extra 3 turns as they are lacking on the standard coils chosen from those currently available from Spectrum Communications. Since these particular coils do not have screening cans fitted it would be comparitively easy to add the extra turns before the coils are fitted if later should I decide to do so.

In the meantime I would like to try in the next day or so to use the BPF in conjunction with an antenna and my G6LBQ transceiver PCB assembly and VU3CNS DDS to access it's performance in practice.

Finally a big thank you to Andy and Sunil without whose help this project would not have been possible.



         

   

DDS project

I have created this blog to record and illustrate various stages in the construction of my latest DDS project which is based on an Analogue Devices AD9851 controlled by PIC16F628A microcontrollers with firmware created by VU3CNS and PCBs supplied by Sunil VU3SUA.

The following picture shows test signals from the part assembled PCBs being observed and measured on my kitchen table in late January 2011.

The PICs were programmed using my Picket3 clone and the keypad used in the tests was taken from a scrapped telephone and the LCD module was bought new on Ebay from a supplier in Hong Kong.

Once it was established that this lashup functioned ok it was time to think about a suitable enclosure and how best to package up all the parts.

It was my intention to eventually use this DDS as a VFO for a multiband BITX project based on either the 2E0ZHN smd or a G6LBQ version both of which I have already built and tested earlier using a N3ZI DDS but which lacked an output to automatically switch BPFs or LPFs.

As I previously had some experience in using the Aluminium cases supplied by Maplin I decided that their box type AB13 order code LF14Q measuring 6 x 4 x 2 inches would be suitable, so on that basis before I did any serious metal bashing I made up a mockup of what I had in mind to get an idea of part placements using PCB spacers and Blutack to assist in locating and affixing all of them.

The photos above show the mockups. Note in order to squeeze the PCBs into the standard Maplin case, about 1/8  inch of the flanges needed to be removed from the top of the end sides of the lower half of the case, but otherwise it is a perfect candidate for the job.

The mockup D Sub Connector housing shown attached to the rear of the box is how and through which all the output and input control lines for the BPFs and LPFs will be routed. I realised early on that although this DDS provides positive 5 volt signals for this purpose they are limited by the PICs to 25 mA maximum and therefore in order to drive relays for LPFs etc some level changing would need to be employed eventually. This I reckoned could be best done externally and most neatly utilising a ULN2803 Darlington Driver Array as an add on module built into a D sub connector housing to suit individual interface circumstances as they arose.

The next job was to mark out and hand drill and file all the holes and apertures required for mounting the parts. All this work was done using hand tools on the kitchen table during late February and early March 2011.

The aluminium box was washed and cleaned ready for a trial fit of all chassis and panel parts and sub assemblies. Miniature LEDS used to display the Band selected were mounted on a piece of perf board and connected to the D Sub 25 way connector which in turn connects to the main DDS PCBs using Molex connectors.

The main PCBs were then stacked using metal and nylon spacers ready for connecting to the subassemblies and final installation inside the case.

The photos below show the DDS finally assembled, powered up and working.

DDS tuned to 28.000000 MHz

Scope display of output of DDS tuned to 28.000000 MHz

Scope display of output of DDS tuned to 7.000000 MHz

DDS tuned to 7.000000 MHz showing the LED for the 40 Meter Band illuminated



DDS tuned to 14.000000 MHz showing the LED for the 20 Meter Band illuminated

My next job is to use this DDS as a signal generator to align and test a prototype G6LBQ Multiband BITX BPF PCB which I recently received this month from Sunil and which I have since populated and assembled and am now ready to test.