Why I have three soldering irons and use two for SMT.

Well.. first of all I took care of the obligation to the family:  Monopoly in which I got destroyed in short order.

My two oldest boys (left to right, Brandon, 9.. Matthew, back, 8) and my wife Lisa (who won).

Anyway I got back on the SDR-Cube Yahoo list after losing big and saw an interesting (and informative) debugging thread on the first working cube built by others (i.e. not George or Juha) ... KE4NYV .. and in his case he had a buffer tilted (a criticism of plate reflow soldering?  Nah, not from me but you do need to check for proper surface tension "suck in" of the devices.  KE4NYV is guilty of getting me playing with paste solder actually... :O) the ICD2 header put on the wrong side of the board (!!! YES I PASSED THE DISEASE ON TO OTHERS !!!) and well.. following the original instructions.

This is what came out of that.  Once the first problem was fixed, he set the contrast and fixed the last problem, his board set officially came up as #1.  N1RX was #2.  If I only didn't NUKE that right angle connector...sigh...

Anyway here was the second problem.  This actually makes me glad I wasn't #1...

ACTION:  Only install R4, R5 and R6 if you will be using the onboard Si570 clock generator (IC1).  Otherwise, do not populate the R4, R5 and R6 positions on the Controls board!

REASON:  When R4, R5 and R6 are in place, the dsPIC controller expects to see an onboard Si570, instead of none or an external one on a Softrock.  So if no Si570 is found on the Controls board, the status read operation always comes back as “busy” and the Cube will hang during the boot-up process.   


So if you were like me and followed the original instructions.. here are the unofficial rework instructions.


1. Remove the Control board from the case if you "stored" it there like I did by taking the encoder knob off and all of the standoff screws.


2. Take the LCD panel off the front of the standoffs.  Carefully tilt up the LCD display.  Remove the bottom right (circled in blue in the photo below) standoff completely and also remove the rightmost grey key cap.


3. Take your two soldering irons (which is why I have two good ESD safe irons for SMT work) and simply float off the components.  R4, R5 and R6-- circled in RED.


4.  Use flux and solder wick to completely clean off the solder from the pads.


5.  Use solvent to clean off the flux.

(Taken through my ring magnifier light.. not bad quality.. I'm impressed)

Looks as good as new, doesn't it?

6. As you are doing this.. place the removed components on a card to save them:

7. Reinstall the rightmost bottom standoff.  Carefully bend the flex cable back and reinstall the LCD.  Make sure the flex cable "bubble" inset isn't rubbing on anything.

8. Reinstall the grey key cap.

Done..  I reinstalled it into the Cube case since I want to protect the LCD.

I really want to start on the DSP card.. but tomorrow would be a smarter day for it.  Until I get the replacement connector I'm not firing it up anyhoo...


2 mins to 2011 in central time zone...signing off

Sneaking in the Build Environment

Wife and kids are now occupied.. I will sneak in this post...

One of my goals with this project (probably the number two goal after operating "mainly" home built and experimental equipment as my mainstay) is the educational aspect.  A long time ago, before my title was Senior Software Engineer, Software Engineer, Systems Engineer (at previous employer) I was just a "Electrical Engineer" at my first job (previous, previous employer).

At that job, I did some DSP work to control machinery, in machine language/assembly on a 320C50 fixed point DSP.  The company was a small one and could not afford thousands of dollars on tools and "C" compilers at the time (the situation now is much better as some C compilers are low cost or free) so I did NON-adaptive filters for mechanical machinery control.  I was fresh out of school and did okay despite not knowing anything but what I learned in my controls classes (which was strictly linear control.)  Luckily,  most of the DSP algorithms are fairly commonly available.  I enjoyed that but haven't done much since with it except for a very short stint coding VHDL in FPGA's.

Juha has kindly made his code GPL so there is ample opportunity for the members of this project to learn DSP as it applies to radio on small fixed-point microcontrollers.  There is a hell of a learning opportunity here.

You see, my commentary about porting to the PIC32MX (and maybe other controllers) is also about needing to understand the code well enough to be able to do this.  If I do start adding filters I'll document my thoughts on this later...I'm kicking around a manual notch filter for the SDR-Cube by using a 4-tap IIR filter..the thoughts on this are not complete yet as I need to analyze the filter coefficients more to see if I can reduce the number I'll need or to see if I can calculate them in code.

But you can do stuff like add audio filters in the code.  This is the real benefit of the promise of SDR.  I hate the term SDR.. did I say that yet on this 'blog?  You still need the radio hardware component to be a radio.  Until we get 60 MHz+ 32-bit+ A/D's (and D/A's) to get the dynamic range needed for "real radio" (which is like 140 dB) these radios will either be limited in Dynamic range (such as the Flex radios and the SDR-Cube) or flexibility as a hardware filter and AGC circuit is needed to compensate for the 90-110 dB of Dynamic range the current CODEC's have.  There are also issues with the direct conversion zero-IF approach (least of which is the reason for the two I/Q adjustments on the cube).

Anyway I just digressed.  Later I might write an essay on that subject.  If you, the reader is lucky, maybe I wont! :O)

The first step about understanding the SDR-Cube IMHO is to get the source code, the free MPLAB IDE (indirect link.. get most current version from bottom of the page!) and the "Lite" C30 Compiler (go to bottom of the page and download the MPLAB C Compiler for PIC24 and dsPIC v3.25 in LITE mode 
by clicking on the link of that same name) .

Small digression here: Even though dealing with Microchip on a professional level has been light years better than ATMEL (which I actually think has technically better products.. another post if you are unlucky) which is really bad in my experience to deal with (parts never in stock, won't sample unless you are going to use 5000 devices a year, etc.).. with Microchip I do have a major ethical problem with some of the C compilers (or at least the PIC32MX compiler, but I suspect the C30 as well) in that they are based on GCC but somehow Microchip is charging real money for them (*they SHOULD NOT!*).  Luckily, if we can live with optimizations disabled, they now have some free compiler options.  This wasn't the case until a couple of years ago.. and that helps us little guys greatly.

Anyway.. after downloading the nearly 100 Mb of goodness from Microchip,  FIRST install the IDE by using the windows installer.  NEXT install the C30 Compiler.  In it's own directory off of the root C: drive (I use C:\SDR-Cube) unzip the source code.  You will end up with a directory like C:\SDR-Cube\SDR-Cube-Source-v1.0 as the project directory.  Click on SDR-Cube.mcp to bring up the project.  To build, I typically suggest doing a "Clean" first and then doing a "build-all".  The compiler and linker are different versions than is on Juha's desktop.. so you will get a string of error messages that state that the compiler and linker are not found and a suggestion to use the compiler/linker at a different location.  Say yes to all of these!  You should then see a successful compile (one warning in one module, but that's a don't care).  When you are done you should see a screen like this:

Screen shot from MPLAB 8.63 with the "LITE" C30 'C' Compiler

Notice my screenshot shows "PICKit 3" on it instead of "ICD 2".  Yup I have a clone PICKIT 3 coming from China as it has a nice programming pod and cables with it for less money and isn't nearly as ugly looking as the real thing.  It's reported to take Microchip firmware and the MPLAB IDE doesn't know/care.  To change the programmer/debugger all you have to do is go to the programmer tab, change it and then save the project workspace.  You should, of course, save the workspace after you compile anyway so all of those links you fixed by clicking "yes, yes, yes, yes" above stay saved!

I decided to go ahead and spend the $40 on it because I do work with the PIC32MX processors as well, and my little PICPROG type serial and parallel programmers from 10 years prior are getting too long in the tooth to use.

The SDR-Cube has a minimally modified serial boot loader on it (Microchip AN1094).. what this means is that as long as you don't completely hose up the code (it's pretty hard.. I've never done it on my similarly boot loaded PICMX32 MX440 CUI32 (sparkfun) and MX460 UBW32 (sparkfun) boards..USB HID in the case of the PIC32MX's)  The USB audio example on the CUI32 is interesting.. I wonder if I couldn't make that work as an adapter from a "soundcard" in Windows to direct digital input on the cube (the "digital NUE-PSK port)... so many projects/toys.. so little time... I digressed again...

Anyway what that means is that you CAN play with the code (add your callsign to it.. a change filter coefficients from the standard or add more) without a programmer.  Just a serial port.

Pretty cool stuff actually...

One last parting note.. There are a few free coefficient/filter generator programs out there.  I will link the one I was looking for (that I have on my work computer) later as I can't find it.  But this one seems fairly promising as well even though it's not been updated since 2004.  MATLAB and it's free clones have routines to calculate filter coefficients also.  FIR and IIR filters are simply delay lines.. FIR's don't use feedback, IIR does.  A standard Crystal and LC filter literally is an analog version of an IIR filter.  This should get you started...

You don't like one or all of Juha's filters?  Do your own 101-tap FIR and replace his coefficients with your own.  Cool eh?  Want 8 filter selections instead of just four?  It can be done fairly easily.. there are a lot of little tweaks like this possible along with more major things like my idea for a tunable manual notch filter...

Control Board Assembly. No sweat this time...

Busy at work today.  I decided to "go for it" on the control card.

Three soldering irons and a hot tweezers?  Yup.  Now I still need a hot air rework station...

I think I might have to do family stuff tomorrow for New Year's Eve though I'm dying to build most of the DSP board (minus the right angle connector.. sigh..)  But I did get about 4 hours tonight uninterrupted to get the control board done-- which is difficult.

Okay.. well.  I went really slow and mostly followed the instructions step by step.  The exception to this was rearranging the mounting screws for the LCD screen as discussed on the SDR-Cube Yahoo group mailing list.  Using all eight fillister head screws on the LCD and panheads on the rest of the standoffs was the correct way to build these with the parts supplied-- it looks good that way.

Anyway.. here is a picture showing the TSSOP level buffer IC's.  It's obvious how they mount but there is no picture of it on the SDR-Cube site instructions as they are covered up on an assembled control board by the LCD panel...

Top of the board prior to mounting the LCD panel.

Note the orientation of the IC's and how they should look properly tacked on and "flow soldered" and properly cleaned.  I also mounted the contrast pot on the back side of the board.  George N2APB stated this is optional because contrast only has to be set once.  I could see no reason NOT to do it this way especially since the potentiometer faces one of the switches.. it would be harder to get to if on the front of the board and couldn't later be easily adjusted...even if later adjustment isn't necessary... why not mount it in the easier to get at location?  I considered also putting the I/Q balance potentiometer on the back.. but it's not that hard to get to and who knows if it will get in the way of something.  The physical design of the cube is such you cannot assume anything will fit right if you change something.

All of the unmounted parts are the DDS option and the Si570 option.  I think a DDS option might be in my future.. but retrofitting this board to add the additional parts shouldn't be too bad...

Here is the back of the board at the same state.  It looks pretty good all washed up like that.  I'm now using Denatured Alcohol since my beloved chlorinated hydrocarbons (Trichor) isn't available anymore.  I suppose it is just as well.  This is the first kit I've built that has had plastic parts that started to melt with the Trichor.  (The Philmore audio jacks on the I/O board seem to be sensitive to Trichor and so does the buzzer.)  The Denatured Alcohol works well except on the dual row through hole connectors since the alcohol seems to need a little scrubbing.  The Trichor didn't really need any scrubbing.

Anyway.. the next step was to mount the LCD.  With the commentary on the Yahoo SDR-Cube group it was easy to figure out how to mount.  I'm a little disappointed with the flex cable.  I'm used to cables made out of Kapton which are better quality.  This will be adequate for these boards and are a better idea/solution than using yet another connector on the boards.

Here is a detail of the flex cable being folded "inwards" as is in the build instructions.
Make sure the fold in the cable isn't rubbing against anything.  Mine barely clears everything.

Starting to look like a radio here, eh?  Another view of the board with LCD mounted.

Anyway.. the next step here would be to put on the standoffs.  The standoffs apparently are a tad bit too short so the instructions have us putting a 2-56 nut on and then the standoff.  This works and puts the LCD up flush against the cutout in the case.  One thing I think you should do at this point that was not in the instructions is to test mount the board in the case.  I'm actually storing my board that way as well.  Yes, I will have to take it apart and take off the screen protector prior to final assembly.. but it'll have to come apart anyway probably to test out the final "smoke test".

It looks like it fits.  It's really starting to look like a radio now.

The "proper screwdrivers" mentioned in the instructions are a small jeweler's common for the four potentiometers.  The shaft encoder knob goes on after the board is in the case in the final assembly (so it will need to come off when I do the final "smoke test").  The "proper screwdriver" in this case is a 1/16" Allen wrench (hex wrench) for the encoder knob.

The four potentiometer knobs need to be brought out as much as possible.  Unfortunately, they cannot come out as much as I like due to the short shafts on the knobs.  The knobs go though the case which is an odd mounting.  The other odd thing is there is a "indicator" on the plastic shaft of the potentiometers.  The marking needs to be 180 degrees out of phase from the markings on the knobs to properly work.  But it seems right and looks alright if you carefully space out the knobs so they are still firm on the shafts and line up properly.  It takes a little while to get them right.  Luckily they go though the front panel so this only needs to be done once.

The knob on the encoder has to be raised up enough so you can push in and still turn the knob while pushing in and not have it rub on the case.

OK.. you knew this was coming... <SOAPBOX>

The other thing I've noticed on these boards, by they way, is that the solder mask on them is nothing to brag about.  On my Control Card above.. in a "blank area" under the LCD there was a light spot on the masking that actually had (nearly) no mask on it.  One spot on the back of the control card has a extra thick spot of masking on it.  This is poor quality control at the board house.  These boards are obviously from China.  Usable and probably necessary to keep costs down for us, but special care should be taken.

On the I/O board, the small amount of rework I did to remove the stubbys of the connector I trashed in a few places stripped off the mask.  I mention this as on the flowing of the SMT IC's having good mask it vital.  I'd be careful around the IC's until they are completely and properly mounted.

One other complaint and I think this is a biggie.  When rigging up the box.. I was going to take a picture with the I/O card set in the box.  I dropped the I/O card (the damn thing is cursed) about 1 foot on to the table.  Just that amount of "shake" or "vibe" broke off one of the leads on the switching module.  Those leads have to be the most brittle I've ever seen on an electronic device.  But that's not the complaint.  The mass of that switching regulator module is too big to be held on only with solder leads.  I imagine on final assembly we will need to also use that double stick tape to tape it to the aluminum subchassis.  That's not good enough for shock though.  So it looks like I'm going to have to design and build a bracket to mount that to the PCB and also probably tie wrap or bracket it down somehow to the subchassis.  I've done this sort of work enough to tell you that, in my humble opinion, if you don't do this and you backpack or use it mobile long enough, you'll eventually have the switching module broken off in the bottom of the box.

I didn't mean to be critical of the design, though.   Likely (and I'm guessing here.. no insider knowledge) that the original 'Cubes used a 7805 and that drew too much current for battery (or ran too hot or both) so this module was a simple retrofit.  What should have been done is that a switching regulator circuit should have been designed on the board or at the very least the whole (retrofit regulator) device should have been tacked down (with the tabs on the top too) to the PCB.

I'm predicting that regulator to be a major failure point of the radio in the field.

Once I come up with a good mechanical solution I'll let everyone know here.  I re-soldered on the broken lead and re-"smoke tested" the board and we are still good to go.

</SOAPBOX>

CCI Amp

I now have about 5 homebrew or kit QRP rigs lying around in the 1-7W range and a 5W Flex-1500.  My antennas at 20m and above are pretty efficient.  Okay my antennas are bogus, pathetic and sad for someone who lives out in the rural landscape and owns 35.63 acres of land.  But they are much better that the typical suburbanite with less than 80' trees has.  (I'm from the Midwest.  where are there 90' tall trees to get your wires up 80' at anyway?  30'-40' up in trees is that best I've ever done in Illinois, Indiana, Wisconsin and Iowa...)

Having said that.. I've done some decent DX on 20m on my Cushcraft A3 (or slightly redesigned by me version).. but also missed a couple of rare ones while on the Flex-1500 on PSK.  On 40m and 80m.. I'm weak even to stateside stations at 5W.  My antennas there work but aren't terribly efficient.  At 100W, I've done Australia, Guam, New Zealand and Hawaii at 40m on my half square.  So.. you see... a little more power makes the difference in a lot of cases.  Missing Indonesia and China on 20m PSK when 10db more would have nailed them makes me think that operating with the least power necessary means that it's OK when more power is necessary.  There are times that life is too short for QRP.

I've decided that, since I was 16 years old, I wanted to build a solid state PA.  The Communication Concepts parts kits for the AN762 amplifier looks like the best way to go.  The EB63A is another version of the same amp but the "single diode" biasing isn't really a good way of doing that.. and all I'd give up on the AN762 is the on board COR... which I could do on a hand etched PCB or perf in about an hour anyway.

I just received updated pricing.  It's $108 plus $14 a band for filters (which I may build myself with relays and microprocessor control anyway) and then $10.50 to ship.  Since I'd want a filter bank to test it out.. I'd likely buy a 20m filter (maybe 40m too..) that's about $132.50 for one filter bank or $146.5 for two.. it's tempting.  I imaging that the 20m filter would work at 30m and the 40m would work at 60m.. so.. maybe that would be a good start to a bigger project.

My mother sent me $100 for Christmas and told me to buy something fun.. this may be it.  Thanks Mom.

I should state that the ultimate in a 100W QRP afterburner is the unit designed by Juha of the SDR-Cube fame.  He really designs stuff like I'd do it if I didn't have as much ADHD as I do and had the time.  If I had more money (see first post about one-income and a wife and four young kids) I'd order off the $600+ kit (with shipping, money conversion, etc.) and have the most beautifully designed and kitted amp I've seen...  but I don't and will have to go "ghetto" on this one...

I didn't really mention this before.. but I want to build a HF rig to do my primary operating with.  Will it be the SDR-Cube, (a 5W module to replace the TXPA), the CCI amp and the rebuilt SB-200 at 700W on 20m???.. well it could be....

I/O card parts pictures...assembly.. and "making up a few words"...

Here's the I/O board kit pictures and the infamous "right angle" 0.100" header that is, unfortunately, no longer with us.

R.I.P. Right Angle connector.

Why?  Well.. when I built the RXAMP board I put on the 4-pin straight connector on the wrong side of the board.  I then fixed it and noted that the silkscreen on the PCB's show, with the white outline, what sided of the board the connector's top should be mounted on.  That stuck in my brain. Like cyanoacrylate esters apparently.   The instructions for the I/O build, which I read no less than 5 times prior to the build... said "important:  Mount on the back of the board for it to fit in the cube.".  (Interestingly enough there is a issue with this... I'll come back to it.*)  Well.. somehow in my soldering and building zen trance I work myself into while constructing.. I did the autopilot thing and put the right angled connector on the wrong side of the stinkin' board again.. I caught myself with only about 10 pins soldered.. but it was too late.. even with my best solder sucker, to get that right angled sucker off of the board.

RTFM.. LEARN THE F'IN MANUAL.. LIVE THE F'IN MANUAL.  You'd think I'd learn that.

If I had.. the job would have been less than 30 minutes to build the board.  It was about 50 with the mistake and R.I.P the connector.. I stole the one off the DSP Board.  That'll make the interconnect test not doable until I get a replacement.*  I'll come back to it I said...

Anyway.. I recommend putting on all of the "flat" 0805 and 1206 parts (which on this board the parts are 1206) first.. it's easier to finagle the parts into place if building this way.  Also cleaning the board at this stage is a great idea to get off the flux and other nasties on a flat surface.  It gets harder if you try to do it all at the end.  The next two pictures show the board at the "flat parts" on stage... cleaned with the last of my beloved fluorinated hydrocarbons.  I'm starting the ultimate experiment on "what should replace" my "environment destroying" (or isn't it offsetting global warming?) wonder chemicals.  R.I.P Good board cleaner...

Ah, the sweet smell of Trichlor.  Anyway.. this is what I suggest.  As long as it's not headers, I'd trust me.  The method I used to mount the caps on this board, by the way is flux.. solder drop on both sides.. place the cap, use two grounded ESD irons to "surface tension" draw the part into it's perfectly centered space.  Notice how straight the parts are and how well centered they are while using this method.  It's highly recommended.

BUT NOT ON THE ELECROLYTICS!

If you attempt to do this on the Electolytics.. WELL you can but it will be very hard and take forever.  The way I recommend doing those parts is to flux both sides.. solder blob one.. carefully position the electrolytic.. and then flow it into the blob and adjust several times until you get it perfectly centered.  Tack the other side and when done and really pretty-- reflow the original side.  Clean the flux off of each electrolytic as you do each one.

The "inventing the new (swear, curse) word" line is actually out of George Heron's instructions.  He's right.  But the less creative (swear, curse) words came later in the build with the @#$#ing right angle connector.  It's worth noting that his comments about the NUE-PSK Mini DIN connector is true.  The footprint on the PCB is.. well a bit hosed.  The case ground holes are a tad bit too small and one of the side ground holes is too close to the inside of the connector.  This means some trimming of the ground leads is needed and one of the has to be formed to be pushed into the board.  Then by doing this the case will pry off of the connector on one side.  Simply use a common screwdriver bend it back on.  If done right it doesn't look that bad unless you get up to it really close.   Even then it's not too bad.   All well that ends well.

It passed it's "initial smoke test".. and I didn't seem to damage the board in the rework of the right angle connector.  The connector from the DSP board looks pretty good here, doesn't it?  ;O)

Here is the final product of my build.  Again -1 point for the rework.  *  I'll get to that later...

* OK.. I'll get to it now.  George is going to send me another connector.  (He's a swell dude.  What can I say?)  It did turn out that he did forget to send me the TXPA kit.. and I hope I'm not blowing a secret here.. but if I am it's coming out anyway.. the aluminum sub-assembly in the SDR-Cube housing I didn't get either.. apparently up to the point that I discovered that and informed George.. no one did.  (oh, boy!)  So.. he's throwing in the connector at the same time.  Which is better than waiting for a sample from Samtec...:O)

The Spinner Household "junque box" has a lot of stuff in it but not right angle 0.100" headers.  Since I could have used them on this project and the sr63ng both.. I guess it's time to start searching e-bay for a Chinese vendor to send me a lifetime supply of them just like the draw full of straight headers and chip resistor and cap kits and the other goodies I've lately stocked up in.

But.. probably.. the next part I break on a kit will be one I don't have... it's an inevitable part of kit building.

It's late.. I need my beauty sleep.. and more luck tomorrow.  The build of the next two boards will be trouble free!  (I hope)...

DSP Board Parts..

The Meat *and* Potatoes of the whole Enchilada if you like mixing metaphors as much as I don't.

Anyway.. Here are some pictures of this potatoy meaty goodness....  If I'm really lucky I'll complete this tomorrow or New Years Eve minus one stinking right angle connector.  More on that in the next post.

Control Board Pictures...

Before I started to build (yesterday) I decided to take some pictures of the boards.  The first set is the control boards.  There's quite a few parts here and I'm going to study the construction manual thoroughly.  I really need to follow every step exactly... when I get to the build of the I/O boards you'll see why "auto-pilot" isn't a good mode to be in when building these.  All I can say so far is that the unique 3D nature of how the boards fit together need somewhat unusual attention to the mechanical detail.

For an idea of why.. look at the vast array of control parts you get with this board:

A further close-up:

The Card 'O' SMT parts and the "business" side of the PCB:

I'm going to tackle this one tomorrow.  Making NO assumptions this time.

RXAmp built for 20m.. RTFM next time

Well... all my comments about that your boards should look like they came off a wave came back to bite me on this one.  There is a problem with the kit or current schematic on the SDR-Cube site.. will discuss that later..  The bigger issue for me was that you need to put the three pin connector on before mounting the molded transformer on the board.  Less obvious to me still is that you need to put the 2x2 connector on the top of the board.  Yes, I removed the transformer (with no damage) to put on the three pin using hot tweezers.  Put it in and then remounted the transformer OK.  Then I noticed that I had the 2x2 on the wrong side.  I tried to use the tweezers to remove that and I put a slight scorch in the board.  I discovered that this brand/style of header can easily be taken apart pin by pin and put back together so that is what I did to remount it.  Except for the scorch by L1 the board turned out perfect.  I guess when I build a 40m and 80m one I'll not make those mistakes again.  I guess what I said was to make it look as close as you can to a commercial build!  :O(

The issue with the kit or schematic is that there was no C3 in the SMT parts card and there was an R101.  Referring to the only information that is currently up, C3 is needed as if R101 is supposed to be a replacement it would screw up the biasing of the RF AMP transistor.  The Spinner Household "junque" box stocks 0.1uF (100nF) 0805 caps, so C3 went in and I still have the R101 for future possible rework.

I'm sure this will be cleared up with the assembly manual.  I'll update the post when I know the definitive answer.  I built it to Rev C of the schematic.

Pictures:
Board Top:

Board Bottom:

RXAMP plugged into sr63ng board:

Board with left over parts, R101 (?) and the R7's I didn't choose.  I did choose 18dB gain.. seemed about right.. can always rework it if I need more or less gain:

XLPF Build

OK.. I'm off now on a break so I sort of have to start this if I ever want to get done.  The technical manual and assembly manual still aren't done so this is sort of at my own risk, I know.  This is why I'm starting with the smaller assemblies first; I really want to get the hang of it before we get to the DSP board!

The XLPF was easy.  Only two minor issues:  I was missing a 2 pin header (but I have a drawer full of snap off headers... it's a stock "junque box" item at the Spinner household.)  The second issue was the mislabeling of C301 as C100 in the 30m parts sheet.  No biggie either... it was obvious.

This one was easy.. the RXAMP as it turned out gave me fits.  See the next post.

Lafayette.. it has arrived...

 I received most of my SDR-Cube order today.  I even received a couple of extra RX AMP boards which I do appreciate.  Everything I ordered except for the accessory cube (enclosure) and the TX PA board was neatly packaged.  George and crew did an EXCEPTIONAL job on packing up the kits.  We are looking at modern day Heathkit here...


The above pictures show how the box was packed.  The picture below is what I received.  The K'NEX roller coaster in the background was a present from Santa that my 8 year old engineer built.

Sinve I am off this week on a plant shutdown I will have to get busy.  My plan of attack is to do all the little boards first, and then work up to the I/O and DSP cards.  It'll take a little while to build this up right, but this is certainly a good start.

PIC32MX? SDR-Cube Next step?

One of the things that fascinates me about the SDR-Cube project is that it is based on the Microchip 16-bit (fixed point) DSPic33.  I've done some work with the more powerful but in some packages pin compatible PIC32MX 32-bit processor.  It's too bad, actually, that George and Juha went with the easier to solder 0.5 mm pitch DSPic part as there is an exact pin compatible part in the 0.4mm package that, if it would have been used instead on the DSP board on the SDR-Cube could have been a drop in.  I would have likely bought a second DSP board to try it and do a relatively simple code port.

I'm thinking of adding some features and doing some customizations to the code... so I'm starting to set up the development environment here at home.

While looking around to get ready for the PIC work on my home computer.. I found this.

It is cheap.. apparently $99 plus $25 for shipping.  It's sexy... it has a stereo CODEC similar to the TI one on the SDR-Cube.

Front View:

Rear View:

There are also two other versions with more peripherals, the version with the PIC32MX7xx series part is especially interesting as it has the Ethernet interface and would be capable of TCP/IP...maybe even remoting given the PIC32MX is fast enough.  I don't know the answer to that at this time.. that board is $149+$25.

The only issue I see with these for a port of Juha's code is that the board uses the MONO MIC input on the board...  so to make this work with I&Q the stereo line inputs on the CODEC chip would have to be tack soldered to a small off board circuit (RC filter I'd guess)....at the least.  I suppose worst case you could run the TI CODEC in the board via SPI as well.  It looks like most of the I/O's from the PIC come off the board so it would be easy to add off board I/O cards and peripherals.

It's a thought for after the initial project is working.  A learning opportunity maybe.  The graphics library port would be.. interesting.. at least...

On the message board of this manufacturer, they stated that the PIC32MX Graphic board is exactly 2x as fast as their (and they do have one) version of this board that runs on the DSPic33.  In addition, the 32-bit processor would likely perform all of the math in (fixed point) 32-bits.. meaning that the full 24-bits of the CODEC likely could be manipulated.  It's an interesting thought.

In the future, though.. I'm wondering if a board like a Beagle board running on a dedicated processor would not even be a better way of doing this, or an ARM or a TI DSP.  But this is all about the design criteria.  The SDR-Cube the way it is more likely to be a better unit for small QRP portable battery operation than a  board like this or especially an ARM or Beagle Board would be because the current draw of the unit is very low for what it is doing.  This would be somewhat lost with a full color TFT and a 2x faster clock rate and MIPS core in the PIC32MX.

I'm more interested in the more computing power options, but many of the QRPers and SDR-Cube is not.

What I really like about the SDR-Cube project is the Code that Juha has written.  I'm just guessing now about another path that this code can easily be ported to provide.

"Yoyodyne" sr63ng SMT Softrock 6.3 "clone"

Well.. after discussing the hotplate and past method of reflow soldering on the SDR-Cube Yahoo group list and maybe scaring some people (I think mainly George Heron N2APB as he's having sleepless nights as it is with the complexity of the kits.. and he doesn't need any toasted PC boards in the mix, I know) off because they may have thought that I meant that the reflow method was the only way to SMT solder.  Now, I think that once the process is refined more it will be my preferred method, but stick soldering SMT is still possible for all parts EXCEPT Ball Grid Arrays (BGA's) and QFN packages.  QFN's in my experience in industry are worse than ball grids.

But all of the parts on the SDR-Cube are fairly benign and totally stick solderable.  Because the sr63ng (build instructions here) had SMT parts on both sides, and I was feeling guilty about scaring George Heron and maybe his customers off... I decided to completely stick solder the unit.  No problemo, dude.  This board has TSSOP's on it, like the CODEC on the SDR cube.  Really easy to do with the flux, roll solder ball and (if needed ) solder wick off. Being nearly 100% SMT it was faster to build than the same circuit portion of the Ensemble.

SMT Stick soldering review here:  flux the part (including caps and resistors.. everything) liberally.  Put a small ball of solder on the opposite side of the IC than you are handed... I.E. if you are right handed go from left to right.  Roll the small ball across all pins (don't touch the pins hard when rolling the ball--or at all if possible)-- avoiding adding more solder for as long as possible.  At the end try to pull the remaining ball (bridge) off of the pins by drawing the solder from the IC package towards you.  If that doesn't work do the same motion with some solder wick.  Fluxing the solder wick isn't a bad idea either when doing this.  DO NOT draw the wick across the pins like you did the solder ball.. particularly with finer pitch parts like the TSSOPs.  You can easily bend the pin into the next one over and it's a BEAR to fix as you have to either hot air remove the part or contend with reforming the pin with a hot soldering iron.  Always, ALWAYS wick from the package to you.

Anyway.. you can see that the stick solder SMT job is good.. but some of the parts are a little crooked compared with the "surface tension self centering" of the griddle reflow.  But these are 100% okay and reliable.  It's suggested that the SDR-Cube be built this way by it's creators and it's low risk that way.

Here is the sr63ng topside  The side mounted cap, unfortunately is a small demerit for George in Canada.. of all things I was short one 16V leaded electrolytic.  I only stock 35V parts in my junque box:

Again, notice that cleanliness is next to godliness when it comes to building PCB's up.  I again insist that you must de-flux the boards to make sure that you don't have questionable solder joints and/or bridges.  It is even more important when stick soldering because you used a lot of flux when doing this.. DIDN'T YOU?!?   Notice how much nicer and professional the mostly SMT board looks compared to the Softrock Ensemble RXTX (which looks more like a Japanese transistor radio from the 70's by comparison.. all of those resistors standing on end always remind me of an old AM radio.. I can't help it.).

Anyway the back side of the board is also used... here is a picture of this.  Some day I need to wire up a convection toaster oven and a PID controller so I can go paste/reflow on two sided boards like this.

Looks pretty good.  Keep your building neat.. it's worth the nominal extra effort.

Because of the design of this board.. the Si570 needs external I2C control.. it's not really worth testing this beast until I have the SDR-Cube DSP board going and probably not until I have the RXAMP and PA built for it.  Hopefully all will go well.  It was actually easier to identify the chip parts on this kit (they were extremely well packed and labelled by George in Canada at Yoyodyne---- and at least for the resistors were marked with numbers( compared to those annoying no contrast 1% resistors that are in the Softrock kits.   (Sorry again to Tony Parks.. regardless of the annoying toroids and resistors he is a saint providing a valuable service!).  So chances are that this one will fire up right away.

Anyway.. until I receive the goods that this will be built into (the coveted SDR-Cube kit).. I'll sign off for now.

Softrock Ensemble RXTX

Okay.. this is a picture of the build of the Softrock RXTX.  I did the bottom side with solder paste (from e-bay -- unfortunately a little old and dried out.. but that can be fixed easily for home use) and a hotplate (kitchen griddle).  The technique worked very well.

I did the SOICs with a stencil made pushing the IC through aluminum foil.  It worked well.  Other IC's and the Si570 were done by putting "dots" on by hand.  I did other SMT (caps, resistors, etc.) with stick solder.

I used an IR thermometer from Harbor Freight to monitor temperature but that isn't needed.  What you want to do to emulate the reflow "curve" that is done by realizing why it is the way it is.  The solder paste has little "balls" or "speckles" of tin/lead solder (or lead free if you are in a zone where you are forced to do so..but the temperatures are much higher and you have to be more careful!).  Those are suspended in flux and solvent.  If the solvent dries out you can carefully reconstitute the paste for home use.

You heat the board until the flux flows.  Unplug the griddle for 30 seconds at this point and reposition the IC's if necessary.. flux should completely flow.  Plug it back in.   Solder will eventually melt.  Most parts, if positioned close to where they should be should "snap" into position on the board be the action of surface tension of the liquid solder.  If not.. a little nudging can be quickly done at this point.  This is why I like this technique for SMT.. if done correctly all the IC's will be precisely centered on the pads.  If you would like this centering but don't want to mess with solder paste, you can solder on parts with stick solder, clean it up.. apply a small amount of flux on the pins and reflow with the griddle the same way.  Parts should pop into place.  Do this initially on old boards and you will get the hang of it.  I did not ruin any boards while learning but I could have.

Notice how the IC's are nice and straight and the Si570, is well, perfect.  The capacitors and other 2 and three terminal devices are done with stick solder and two soldering irons.  If you do that the capacitors will surface tension reflow just like the IC's did on the hotplate.  If you don't stencil, I personally believe it's faster to place them that way.

The top side of the board I built all with stick solder.  Not much to say here but flux well, and clean up really well.  Please always clean your boards.  Flux can corrode in some cases and it always can hide cold solder joints.  I often will clean.. rework a half-dozen connections and clean again.  Your boards can and should look like they came off a professional wave solder machine.  Once you start building this neatly your first time success rate in testing will be nearly 100%.

Top side:

The board has been partially tested.  The Si570 circuit and microcontroller is working.  My only complaint and fear with these boards is the awkward mounting method and complex toroidal transformers.  I think I have them correct.. hopefully in the next couple of weeks I can verify and get this on the air.

Someday, I hope the Softrocks will go to Coilcraft and/or Mini-Circuit type RF transformers and chip coils.  For that matter once you get used to building SMT... it is really quite a bit easier and faster to build and the final product comes out better.  I am not criticizing Tony Parks (who is a saint) for designing and kitting these this way as most hams fear SMT.  Hopefully I can help convince them otherwise.  The day of the through hole part has passed and like it or not, they will dwindle in availability until they mostly go away.  But, actually once you learn how to build SMT, you will like it better...

I was originally going to build my SDR-Cube with this rig external.  I will eventually try it as I'm going to configure it for both internal and external.  But.. since I decided that building SMT is better and really the 6.3 series Softrocks are more compatible with the Cube design (and a little backpack rig would be fun to have).. I've built a sr63ng, which is the "MOBO" version done by George Boudreau.  George Heron agreed to sell me an RXAMP board and a TXPA board even though I passed on the base SR6.3 kit.. so I will have a "cube" built as intended before I start to hack it for other uses.

The sr63ng will be the next post... 

Welcome to Fred's

When searching for some pictures to post of the sr63ng and the Softrock Ensemble II RXTX that I've assembled recently, I figured I'd say "thank you for shopping at Fred's". This was outside of Huntsville, AL where we vacationed and visited in-laws earlier in 2010.

Hopefully I will be able to keep this entertaining and informative.  I'm also waiting to see how long it takes before people find these ramblings.  I'll only post it to other lists once I start seriously building the SDR-Cube.

I'm not responsible for damages caused by shopping carts, either.  Nor damage due to following my "hints", "instructions" or even silly opinions.  But this is what a blog is for, is it not?

25 years? No Way? (Preface)

Well, after (almost) 25 years** of being licensed, and a few years of doing mundane things like adopting four kids from (South, or the Republic of) Korea, moving out to the country, building a house from scratch, getting a MBA while working an average of over 50 hours a week and still getting a 3.95 GPA at it... oh and a gastric bypass and 150 lbs lost... have kind of slowed down my favorite hobby, Amateur Radio for about the last 8 years.  I managed to get a small tower up* (1/2 of what was originally planned and sad for someone who lives on 35.63 acres with no restrictions...) and got really active for about June and July.  I intend on at least getting simple awards like WAS and DXCC... that'll force me to operate some too..

I'm now shifting back to my real love, which is building radio equipment from scratch.  My current project is going to be an SDR-Cube by George Heron, N2APB and designed by Juha Niinikoski, OH2NLT.  I hope to contribute hardware (probably PA's) and software (DSP notch filter) to the project eventually.  I'd also like to port it to the 32 bit PIC's...the PIC32MX series.  That could take some effort and I'm not sure if I'll have spare time to get there.


I think it is actually SDR done right.  I could go in to the problems I have had with my FlexRadio Systems FLEX-1500.. but that's only been like teaching a pig to sing.  It only annoys the pig and it still doesn't sing.  I've never seen a group quite so defensive of a technical product as the users of their radios.  The company is a bit better, however, and does try really hard to provide service to the customers.  I had a good, quick turn around on my 3-month old Flex-1500 when the PA stopped functioning on 20 and 30m.  Now that the newest software works (2.0.16) software is acceptable (first version that was for me) and the rig was fixed and tweaked a bit, I had fun with it on QRP PSK-31 last weekend.  I will admit that the IMD reported was the best I ever received.. so the DSP does do it's magic...It's just that the almost five months it took me to get it working right (and I do this for a living, remember) really soured me on the "Flex" "PC" approach to amateur SDR.  If only a DSP had been part of the design.. sigh...


Unlike the Flex philosophy, the SDR-Cube uses a dedicated DSPic33.. a 16-bit device to do real time (and therefore low latency) operations.  Ham radio, particularly at HF and even above in what hams currently typically use do not use much bandwidth and therefore don't need much processing power.  The Flex and other PC based topologies (like the Softrocks, Genesis radios, etc.) rely typically on Windows for the platform to do DSP on.  Windows has about the worst deterministic qualities of any OS out there, so random delays can interrupt the DSP that is on going.  So literally your requirements for a dual-core PC (and an AMD one at that) to properly operate a Flex-1500, 3000, 5000 are really emulating the processing power of the PIC on the SDR-Cube that is dedicated to the task of demodulation.  It's rather a testament of the effort that the Flex team has put into PowerSDR that it works as well as it does.  However, it's not the correct way to do the SDR in my humble opinion.


I'm hoping that the Software that Juha has done in the SDR-Cube is the basis for much growth.  I think he's done it correctly for current ham bandwidth signals.   


We will see..


George Heron is always a blast to work with.  I wish I could be 1/2 as positive as he is all the time..


I've started this to 'blog site to document the build of this project and maybe update some of my activities and other projects.  Anything that comes out of that that should be saved for prosperity will end up on my website:  fredspinner.com.  It needs to be updated.. it's hopelessly out of date but I'll try to fix that in 2011.


It's Christmas Eve.. and the wife and kids are getting anxious.  Merry Christmas to all and Happy New Years.


73,


Fred M. Spinner, W0FMS


*the small tower.  It's a lot better than nothing.. which is what I had for years...

** I'm kicking around the idea of joining QCWA... at 41.. I'd have to be one of the youngest members...I am eligible 1/1/11  :O)