Let me tell you a story. It was a dark and stormy night and not a creature was stirring, not even a mouse. And yet, in a little lab in the middle of nowhere, somewhere out beyond the back of Burke, a little Aussie VK Ham was trying to get an order in with a PCB manufacturer before China closes down for Chinese New Year.
So in my infinite wisdom I needed to rush a 2nd board onto my order to make it more cost efficient on shipping. I only had one board ready with updates and fixes, and I have been meaning to do a PA board for quite sometime, so i fired up Diptrace and opened up the schematics I was going to use, and got to laying out 2 separate designs published by Ruben VK5FE which can be found here
So anyway, the two designs were slightly different in components and the two fets had different footprints and in my cut and paste zeal while doing the schematic capture, i forgot to change the components on the IRF510 board from RD16HHF to the IRF510 component and footprint.
Whats funny in this whole story is that after I posted the above picture to a friend of mine, I joked that the mechanical design was such that if i had made a bomb the PCB would shield me from the FET shrapnel and i should not lose an eye.
So lucky for me when it came time to smoke test this thing I use current limiting power supplies and had set the current limit to 2amps, because when the drain and source pins are swapped on the IRF510 you have made a heater as it will draw infinite current until the wires coming from the PSU overheat, catch fire and burn your house down.
So my board looks pretty but does not work. The other board designed for using RD16HHF1 fets should work a treat assuming I have not screwed anything up there 🙂 But before i go and build one of those, i will be checking it over 3 times as I do not want to waste expensive fets for nothing. When IRF510’s costs pennies and the RD16 costs pounds, you dont want to be exploding things to often LOL.
Anyway, the moral of the story is don’t rush. Check things many times and cont use cut and paste when putting 2 similar designs on the one board. Now i just have to wait a couple of weeks for China to re-open for business and send off the board files I have sitting here now just waiting.
So I am sitting here this morning drinking a massive cup of espresso coffee and thought i really should post up where I am at with the receiver project. Its getting closer now to final testing and making noise. In fact, the final board will arrive today from china and as long as it does what its meant to, i might only be days away from hearing beeps
So anyway, onto the IF amps and mixers. I find the whole amp design, input and output impedance matching and how one can effect the other to be slightly confusing. So I went with an IF amp design that eliminated that confusion for me, the TIA amps. Termination Insensitive Amplifier by Wes Hayward
The mixers will be ADE-1 Mini Circuits jobs as I have a bunch of them here doing nothing and figured why not. I do have a board here I am designing with a number of different mixer topologies on it that I will eventually have made and test out homebrew mixer designs more fully, but for now I will use the commercial product.
I built the 2nd IF amp first, so I can have a direct conversion receiver happening in short order once I have a working audio board.
Well it works as designed and does the business obviously. There is a missing trace on the board, so I will eventually have to respin them and I want to fix the layout somewhat also and change the transistor footprints to wide not narrow. I do not like those small pads to solder to. LOL
A couple of months back I bought a pair of cheap lab power supplies off ebay that turned out to be crap. 20mv quoted ripple was closer to 200mv. I managed to clean one up with some massive filtering, but it was impractical and still the ripple and switch noise was to high. So while searching for something else i came across these on ebay. 28v, 10A and a linear supply. They weight about 8Kg each. So did some googeling to see if there are any issue and there is one, over shoot when you cycle the power on and off and there is an easy fix to that by adding a cap. But hey, who cycles power supplies on and off anyway. So i grabbed one and then checked the ripple, about 10mv. Thats perfect, so grabbed another one, cause you just cannot have enough power supplies.
So it looks like I am becoming a HeeHaa fan boy. 2 power supplies and soldering and reword station. Oh the humanity LOL
I updated the board and so its a bit better laid out than previously and have shortened the rf path by moving the SMA connectors Gerber files for the board can be downloaded from here: http://robs-blog.net/Files/Dual_Xtal.zip
I am going to redesign this board. The layout is sub-optimal. But for now, I am going to assemble it and see if I can make 2 filters worthy of use.
The one thing that has always frustrated me with many SSB/CW homebrew rigs is that you get one filter and one filter only. I have seen some use varactor diodes to solve this problem, by varying a voltage on the diodes you can vary the capacitance and this the width of the filter. I thought about going down that route myself for a while, but in the end, settled on having 2 filters as I have lots of relays already.
The plan is for an SSB filter that is about 2200hz wide and a CW filter about 500hz wide. I am also using the good crystals for these also. 4.91520Mhz xtals sorted to within 10hz of each other.
Each crystal was loaded up into my you beaute xtal tester and the frequency measured to within an inch of its life.
My function gen also has a frequency counter that counts down to 1hz resolution. So I measured my crystals and sorted them into piles within 10hz of each other. If i cannot make acceptable filters with such tightly grouped crystals, then there is no hope in hell i can ever make a good filter ;
Out of this pile of crystals I have 6 groups of 8 crystals all within 10hz of each other and a few more groups all within 20hz of each other. Buying these crystals was certainly well worth it. The pile in the bottom right corner are the outliers. It seems even good quality crystals have some in each batch that should be rejected.
Got the parts all sorted, just gotta build it now. Will update this post once i have built and tested each filter.
Part 2: Well i kind of expected that to not take forever, but anyway here is the update.
The build is pretty straightforward and all went to plan
Other than the width being a little wide in both cases and a bit to much ripple, I will say that I am happy with the shape of both filters. I can narrow both up easy enough, just need to change the cap values and the ripple, well it might be a bit of work, I will need to play with transformer values. I used 4:1 impedance transform, but 3:1 might actually be closer to the mark. I will have to play there and see what i can come up with, but a better match will remove some of that ripple.
For now though, I can keep on building the rest of the project as these will be acceptable to get things working. I can make them perfect later. Next job is IF Amps and mixers.
Being invasion day here in Australia with not much doing, between watching the cricket, sleeping and doing not a lot more I managed to build and test the bandpass filter board.
Overall, well its MEH. The filters are nothing to rave home about, they will work for now, but I do think i am going to redesign the board using a different topology and get much better attenuation in the stop band.
I ran all the filters though the bode plotter and remembered to screen grab 3 of them which i will now present here for laughter and ridicule.
40m bandpass filter is a 2 humped camel. The width between the peaks is fine, but i will have to play with the coupling cap value to see if i can take some of that hollow out.
20m bandpass filter looks ok, its just to narrow.
The 15m band filter is also to narrow and the high side attenuation is kind of poor at best.
The 80m filter was also MEH but i forgot to screen grab it, It was double humped, but only -3db down in the middle and the high side attenuation was also poor. Anyway, these will allow me to get building and testing more parts of the radio and think about changing the topology and laying out a new board and having it made. I have 2 other board sitting there waiting to be sent to JLCpcb, the 3rd iteration of the audio board and the first of a final PA board. More on those to come.
So between the first iteration and the 2nd, I have changed from using Sprint Layout to using Diptrace. I have tried a few different programs from Kicad to the free version of Altium and I just could never get the hang of how to make components and footprints in either of them
Diptrace on the other hand I find to be much more intuative and straight forward to use and I can make components and footprints in it with easy, as will be seen later when i get the latest batched of board back from china and write a blog post. So many new parts all created by me in the one library so i can find all my parts easily.
Anyway, to the board at hand. Its still got a couple of minor mistakes on it, the cap footprints were in 2.54mm not 5mm which the caps I use mostly are, the 5v regulator and caps are in the way of the right angled SMA connector and I am not sure that the variable RF preamp is actually varying. Circuit wise it should work, but the spectrum analyzer was not showing any gain change.
Anyway, the design will not be changing now, its built, tested and it works. And i can stay like that :
Got to do some building on this the other day and knock up the LPF board. Ran it though a battery of tests and other than Elsie screwing up the values of the filters, everything worked at designed. I gotta suck the inductors out and add some turns. Not a fun job but its gotta be done. Gerbers for the board can be downloaded, from HERE: http://robs-blog.net/2018/12/16/4-band-lowpass-filter-board/
This picture will give a bit of an idea on the stackup and how these things well all fit together. Using the 10w ebay PA for now for testing until I find a transmitter design i want to use and lay a board out for it. But as an idea, it will be using a pair of RD16HHF1 conservatively biased for 10w or so.