Goals:
3 times higher power capability to ensure 1h key down digital mode without failure and rising LDMOS temperature beyond 70C.
Enviroment temp. 40C
Easy to repair, low number of spare and simple parts (no exotic hard to get capacitors)
Tuneable for marine frequencies if I dont get it right initially - 2,4,6,8,12,16 MHz
If controller fails, emergency call must still be possible, even if sending out harmonics.
Conventional 12V - control and I²C LPF interface compatible with SDR software and working identical to Hermes-Lite SDR I²C interface
Size must match the PA Heatsink (197x125mm), cost below 80$ including pcb
The Filter design is inspired by DJ0ABR, Kurt did a very good low loss Chebychev design. This attributes to a lot of not-dissipated heat in the design, but his pcb is not keeping up with my needs.
See very detailed instruction for the filters here: DJ0ABR
The current design is just the original Ham bands covered. Most of them should be ok fit for the Marine freq too, lets see this - the board is nicely tuneable later too.
The board layout is inspired by the latest version of W6PQL. My toroids are smaller than DJ0ABR and bigger than W6PQL. It is well overdosed for my 500W target, hope they stay cool. The board is definetly smaller than each of the predecessor designs. Also the darlington array including the diodes make it very few parts to place (to allow for more capacitor massaker).
The schema is made with LCPCB's design tool and with a BOM they can assemble without conversion. Feel free to clone, but plz feedback if you enhance someting or find an error Link: easEDA project file
rev1 will not be ideal capcitance, so I dropped it. I flipped for the relais of the original designs. still match the size, and the cost is even down. still stay well south of $80 total parts cost with pcb.
I think I will order rev2 to check the performance. Done.
I did test very thoroughly the interace options, the signal filtering, function and did stress test like 20ms switching through for a hour. This is of course far away from normal use case. But I wrote automated arduino tests to send I2C filter switches. So what.
Finally, had a severe bug in the layout and patched it on my first rev2 boards. Completed and tuned the whole thing to see what else might loom in the newborn. So far only labelling, added some good ideas guys gave me. LEDs - yes true, very cool to have that for diagnosis. And an additional connector to add more external relais, like antenna switch. Finally I found I missed fixing holes, didnt see that because my case is a pcb slide in concept which does not rely on screws. But I am adding a sandwich for all the other PTT, SWR meters,etc and act a shield. These two will be screwed together.
Now I believe the desing working very well and definetly meets my expectations for efficiency. Rev3 is updated on easyeda, I will order this with my next batch, but still need to complete the pcb design. But I am confident that Rev3 is valid to build now.
Boards arrived after 3 day production and 3 day shipping. Speed unbelieveable. Quality incredible, I have never had anything close to this quality, and it is 2$ per pcb. That is beyond my imagination today. The boards come out seems to me absolutely perfect like designed with the easyeda software of JLCPCB. The board comes 100% tested for electic contacts, so can trust this. Now I am looking how the coils will fit. So far no issues obvious to me. The boards came faster than my SMD order. Soldering has to wait for them, as that comes first.
ok, got SMD parts and soldered it on. Great experience, the footprints match perfect parts self-right 99% perfect with my simple heatgun. I fitted the interface options in two stages and tested systematically the performance of option 1. digital input. perfect low pass, 100uS saturation of signal. 40uS delay to the darlingtons. switched a few thousand times by an arduino test program. 2. option I2C, added the MCP23008 and the 3.3V regulator and ran a I2C test program. idle current 3mA, relais 70mA engaged. Signal filtering just fine. Only change I found: 2.2k I2C pullup instead of 1k gives a better signal dynamic for I2C. Finally soldered the filter caps. lot of fun simple to pick, centering well. So far no issue found, all works as expected to this point.
Argh, the first capital routing bug found. one side o the relais chain was hooked up the wrong way. damn, on the last few meters I messed it up. Now I will fix it in a further revision. But keep going on this one to tune the filters. that might bring some further additions, maybe add more spare pads for caps.
Went further to the winding and stiching. The values of the coils differ minimally from the original filters. I use smaller toroids T106, that leads to a turn or so less than on the original T200 cores. I used 1.12mm / 17AWG wire for all coils, which works nicely. This wire size is just still ok to manually wind. The air coils are wound on an EDDING 361 it has a diameter of 12.2mm. This results in about 13mm inside and 15.3mm outside diameter. The smaller coils are wound on a 10mm drill, resulting inside 10.5mm and outside 13.2mm . I have not really many tools, but the making was a nice afternoon job.
Originally I got questions why all air coil holes have same orientation. On this picture it is good to see what was on my mind. Simple an straight curls are easier to reproduce than fancy entry/exit turns. The 90deg change is achieved my making each coil right handed and left handed. This way we automatically get the 90 deg orientation. I happy so far with the concept.
Most values were easy to achieve by compressing or widening the coils accordingly. I measured all coils when making with a RigExpert @ 50Ohm. I found very little tuning effort finally, so I expect the measurements are ok within reason. The 7MHz i.e. has identical turns but the coils are comressed and widened to the opposite extreme, that did the trick properly.
I put here the designed values (theoretical) and what I came out with my work.
After this step I did a complete scan for SWR in the desired frequencies, as a low loss filter was most important for me (save battery:)
I have no networka anlyzer and my oscilloscope is max. 25MHz. It is all pocket stuff, quuite ok for the job but I am sure there are much equipped people. So if anyone builds this or a similar LPF I really would love to learn the results taken by more advanced equipment.
However, I have simple judgement: total energy ratio. So the harmonics will be tested with the full power later and the spectrum view of the oscilloscpoe.
The power measurements will be done by very good AD8307 meters, they are similar to the rig expert, and they stay inside for limit monitoring and cut off logic.
Just a real check, and total SWR measured by HL2 to the dummy load is approx. 1.1 SWR all requencies. Now lets look more detailed. (you see the MCP23008 listens to HL2 / SDR Console as expected - all perfect)
All capacitors were obviously exact right. No tuning was necessary for me, obviously this shows it is easy to reproduce the original results also with the changed layout. A bit sad, the design was so cool for tuning:) and I left spare pads even... maybe later I will find reasons to fool around with cap values...
A quick check on harmonics show a good damping whats expected. However, much more interesting is the losses of the fiter in the pass-through area. Everything is adding up relais, pcb capacities, impedance. This is the most scary point - is the prooven design compatible to the fancy SMD capacitor approach? does my layout mess it up?
So, my method is maybe a little rough. I am lacking cool tools and truly I am not a seasoned Ham for many years. If someone finds time and has ideas, tips for my process I would appreachiate if you post it to the forum on the page.
The smith shows my dummy load is centered where it should be - 50Ohms on thw range 0.5 - 30MHz.
The 160m is well suited also for Marine frequencies useable range 1.8 - 2.5Mhz seems to be the optimum around 1.05 SWR
80 is less wide, the Ham Band is excellent at 50Ohm. The Marine Band is 1.10. optimalö range 3.5Mhz - 4.2MHz
I mpressive how much of a range can be tuned just by the coils. I can get everything between 1.05 and 1.5 SWR on this band. Remember the same number of turns on these toroids. so turns arrangement is king here. You must use an analyzer to tune this band.
really good, flat from 4.8Mhz to 7.8MHz. That is really good for Ham and Marine frequencies.
8,10,12,14 MHz are the popular Ham and Marine Bands. This filter is good on marine, but not especially super for 20m. This may require another approach. 1.18 is not necessarily what we want.
1.05 - 1.10SWR on 15m The marine frequency is good, Ham 15m I have not used before.
12m right in, 10m still fine. In this area my equipment starts to become sort of useless. I guess the judgement is roughly ok, but I would not bank on my measurements above 20-30MHz.
Here it ends. I checked up to 30Mhz with the 6m Band filter, well ok no harm. but I cannot tell about its targeted range in the 50MHz world
Checking now if it still fits the box. Not bad, though it is heavy weight little box already. 3 more missing now for the protection and switching, interfacing to the radio.