The Filter That Never Was

One upon a time in a land far far away, LOL. Well, this is one of those stories and it reads like a horror story. I have spent all day on this, pulling my hair out and trouble shooting and wondering what the fuck is going on here. Is it me, was it my design, was elsie screwing me with stupid values. Well, hold onto your hat Batman, this is going to be a bumpy ride 🙂

It all starts here. Designing a filter in elsie. Bandpass filter, 7.150mhz center frequency, 400khz bandwidth, chebychev response with 0.001 ripple. All good, simple even. I do not like the inductor values it spits out, i rescale the network to use 500nH. And begin to wind up the coil forms.

Everything goes to plan, i wind up 9 500nH coil forms, LCR meter says I am a genius and I agree with it, not only am i super smart and know everything, I am also good looking and a legend. 😉 That was this morning. Now let me rewind things back 4 weeks ago when this plan started to evolve. This should not be the problem.

After having elsie tell me what to do, I laid out this schematic in Diptrace and proceeded to produce a pcb to use. Schematic looks fine to me, looks the same as what elsie says i should do. This should not be the problem.

This is a small screen shot of the PCB layout, if you close your eyes and squint you can see the series capacitance and the shut capacitance and the inductance are all in the right places electrically. This should not be the problem.

Elsie gave this plot as an idea of what the filter should look like. Looks fine.

Never to take the word of a machine seriously, I checked the design in LTspice, just for sanity sake. Looks fine.

Then i built the filter, just one mind you, there are 4 on the board, but i have learned the lesson many time not to jump ahead of myself, because as the Mythbusters would say, failure is always an option. Becides, things going to shit are the norm around here. Today was no exception, bloody abortion of a filter was on 14mhz. Its like 1/2 my inductance and capacitance just vanished into thin air and left me with a pile of shit. And i had no idea why.

Being the clever idiot that I am, i did say i am a legend right, I began by melting the plastic cases on the relays, I mean pulling all the caps off the board and measuring them. Frequency went up, so that means less capacitance and less inductance. I am the boss of making strays of everything on my boards, so i usually have more than i need. Caps turned out to be correct.

So i rechecked the value of the inductors.  Actually i wound the slugs all the way in and as expected the frequency shifted lower. LCR meter says that i had 600nH. More than i needed but I was still on 14mhz, not 7.

Then after much wailing and gnashing of teeth, I grabbed out an inductor of a known value and measured it. It was wrong. Not just a little bit wrong, but a lot bloody wrong. I calibrated the LCR meter and checked the known value again and BOOM right on value. Checked the other coils i wound and they were 160nH not 500. Well there is your problem. I screwed up royally. I did not calibrate my crap and thus made a huge pile of it and wasted half a day on what should have been a 3 hour tour, a 3 hour tour.

So there you have it. That has been my day, fun, entertaining and well, i learned a valuable lesson. Never trust your meter not to lie to you. And with that, i am off to have a shower and cook dinner. The filter board can wait now for another day. I am done 🙂

Oh one last picture, i put some 500nH crappy inductors in the circuit and yeah its in band. Now i need to rewind my nice coil forms and make some filters that do not suck as bad as this nano vna plot looks.


I Now WSPR On 3 Bands

This has been a fun little journey. What started out as look at me i can WSPR on one band, is not looking like, look at me I can WSPR on 5 bands. So what we have here is the 3 band WSPR TX that does 20, 15 and 10m and with an extra PA board and a little bit of code i can add 80 and 40m into the mix also, all this from just 2 antenna.

Here is the schematic for the PA board, less the power supplies. It requires 7V for the fets and 3.3V for the bias. The 10K pots and 3.3v bias supply allow for really fine changes in bias voltage. If memory serves me correct, 600mV was enough to run the fets at about 400mA draw each. The fets themselves are for VHF to GHZ range, so running them at HF is not ideal but its all about what you can find. These are 50 cent parts and what is used in many dual band HT’s as their 5w PA. The fact that I am getting 5w out at HF is rather nice.

Here is the PA board sitting in the box I made for the never ending story transceiver LOL. Note the 5A buck converter being used for the 7V power supply.

Here is a more broad view of how things are looking. 2w out using the old LPF board that was more loss than pass. Built new filters and things are now much better.

Fresh built LPF board without the losses of the old board means I am not wasting 1/2 the power in the filters. I was always going to be happy with a couple of watts, but the fact its making 5w, is nice. Note the heatsink stuck to the back of the PA board. A tone of vias are uses to wick the heat away from the fets and into the heatsink. Its been running for 24 hours straight and temps have remained steady around 50deg C.

Close enough to call that 5w. Its a little less on 20m, more like 4. But that is ok.

And finally, no spots on 15 or 10m, that is to be expected, but we have a number of spots on 20m. So i am calling this a success. Learned a bunch thought this project and the outcome was better than I expected.


Using NanoVNA To Measure Inductor Q

How good are those inductors you just wound? What about those inductors you just bought from your favorite online store? Don’t know? Well knowing the Q of the part is going to help you know whether they are going to be useful in that bandpass filter or oscillator and it could not be simpler if you have a nanoVNA.

So what I have here is a T68-6 toriod with 19 turns of enamel wire on it. 19 turns gives a theoretical inductance of 1.7uH, the actual measured inductance with my cheap LCR meter is 1.747uh yeah that is surprisingly close and not something that I typically find that happens. Usually 1 or 2 turns either way is required to get to the value I want.

For this measurement you want the inductor to be in series. So you can see from the picture above my test jig the inductor goes from the center conductor to the shield of the coax.

So the most important part is how to make the measurement. First is to re-calibrate the nanoVNA for the frequency range of interest for the part. For me, Captain HF, 3 to 30Mhz is where I keep a saved calibration for. Next, the measurement is an S11 1 port measurement and the plot you require is R+jX.

So from here the math is rather simple, Q = X / Rseries for the frequency of interest. So for the above at 14Mhz, for simplicity sake lets call it Q = 200 / 1 || Q = 200. In reality the Rs is less than 1 and the X is less than 200, but thankfully NanoVnaSaver does the actual calculations for you and displays them in this case the actual Q was 150. Which is pretty damn good for a hand wound toriod inductor. Now the good thing is, you know how to measure Q and you can start checking all sorts of parts you might have sitting around.

Take these cheap ebay variable inductors as an example. I built a filter with them for 40m and it was crap. Q at 7Mhz was about 20. But at 14Mhz though to 150Mhz the Q was about 50 to 75 on average. So they would be useful to use at higher frequencies and certainly not useful at lower ones. Unless you have a datasheet that tells you what the Q is for a given frequency range is, you will never know unless you measure them.

And lets face it, measurement is king. Even a non perfect measurement with a non perfect instument like the nanVNA is vastly superior to having no measurement at all. Oh and one last nugget, now that you know the Q of your actual parts, you can then use that value in Eslie when designing filters. Being able to measure things can really improve everything you do.


OpAmpia: The Op-Amp Receiver

They say a picture is worth 1000 words. Well A video must be worth 1000 pictures then. I have to admit, i really do not enjoy making videos for youtube. I do it reluctantly and only when I really have something worth showing off. I really do not care about building an audience and being famous, I just like to do what I want, when I want and not feel compelled to make anyone else happy other than myself. I am not entertainment, I am just a guy who is documenting his journey, nothing more. Even this blog, its not a how to, its not a guide, its often not even correct, does not work or is fundamentally flawed. That is what happens when you home brew. You make mistakes, things do not work and you have to trouble shoot and trouble shooting means learning something. A good day for me, is a day I learn something.

Well, this is the PCB. I have named this receiver OpAmpia. Because its just a bunch of op-amps. There are 3 high speed op-amps on this board, acting as RF Amp and IF Amps and then on the other board another 5 op-amps doing audio agc and preamp and audio filtering. So all up, we have 8 op-amps, hence the name.

This is a revision of what I actually built. I have redesigned the bandpass filter and will be using TOKO styled canned inductors. The higher Q of the inductors should see much less loss than the SMD inductors I used on the test board.

Same goes with the IF strip, the IF filter is now not plugin, but will use all SMD components, I have a bazillion SMD crystals and so its time to start using them. Actually in everything I am moving away from through hole components where I can. SMD is just so easy to use you can always find a useful part that is cheap. Take the Gainsil op-amps I used, they cost like 30 cents each. Try and find a highspeed op-amp in though hole for that price.

For alignment I used the function generator set to 7mhz and with both VFO’s showing no offsets tuned things until they sounded great. Having 2 VFO’s really helps here. Now i can just do some simple math and remove the offest and display the actual RX frequency. A little bit of coding and we will be in business.

That’s all from me. Another month and I might have this as a complete and working transceiver. LOL Who are we kidding 😉



The Fine Art Of Making Power

Make no mistakes, I think real amp designers are witch doctors and that amp design is more voodoo than art. But this has never stopped me from having a go and seeing what I can come up with and what I can learn.

Today I hit a new milestone, I raised peak power to 50W. 50W might not sound like much to those who have gone out and bought their kilowatt to power cheat with, or for the guy who bought a amp pallet off the interwebs and assembled a kit. But for me, someone who has looked at lots of schematics, tried to understand each design decision the designer made and why they made it, then went and captured the schematic, laid out the PCB and had boards made, 50W is something of an achievement.

And don’t get me wrong here, 50W is what WA2EBY was getting out of 2 IRF510 and I am using 4 of them to make that same amount of power and still struggling to contain the magic smoke within the epoxy that is covering the silicon.

I was hoping for a little more out, 70W or there abouts would have been really nice. I do have room to push the amp a lot harder than I currently can. I can give it more drive, more bias and more voltage, with the current limitations being 5w drive from a CW transmitter i home brewed, 2V gate bias, and 30V and 10A from the lab power supply.

Along the way many parts were broken. I call this pile My Shame. Actually its not as bad as it looks, the first 4 fets to go were Ebay specials from China. I think they are fakes, they were never happy and just ran into uncontrolled oscillation.

The next 4 were all my fault. I gave it all the amps and they gave up the ghost after about 40 seconds of full key down madness.

This is the current mess of what is my work bench. Plenty of crap on the bench. A certain old guy home brew legend would say my bench is to clean LOL

Nothing like burned flux to get the soldering fan boys to go wild. There is nothing fundamentally wrong with the board, its is a little bit pedantic though. Its window of oportunity is quite narrow. Its kind of all or nothing. I had to change the input transformer and its ratio, the bifilar inductor feeding power needs to be changed as it gets to hot. Everything else is ok. Mechanically its easy to change out the fets on detonation. That is a plus.

Using the CW transmitter I built a while back to drive it.

This is the schematic. I probably should learn about temperature compensated biasing to make it a little more bullet proof. Other than watching the duty cycle and keeping it to say 50% or less, its ok and I have not harmed anything yet. It really could do with a bigger heatsink and some forced cooling to really crank it up. But for now, I am happy that it works.

I did key it up on 40m and sent some CW with it. Listening on KiwiSDRs i could hear myself in VK2,3 and 4. Not that, that is a challenge, I can do that with 5w. But i know its working and sounds fine. So that’s the end of this journey. I am calling this one won and done and now its on to the next thing.

Better add this image as well. Fiddy Whats. Oh one last thing, 5w in and 50w out, is 10dB gain. It is what it is.


40m WSPR Transmitter

Following on from playing with WSPR last week, I grabbed out on of CW TX boards i have here, reworked the gain in each stage to account for the greater signal input from the SI5351a and ended up with 8W out. Keydown for the last 30min and all the important bits have remained steady at 55 Deg C. So it should be more than upto the job of high duty cycle TX from WSPR.

The Transmitter board.

Signal from the SI5351A VFO board.

What the scope says is happening into a 50ohm load. And here are a few more pictures, because why not 🙂

Setup as transmitting currently into my 40m antenna.

A close up of the swarez meter. Showing 6W but its closer to 5W according to the scope.

And this was the reports after the very first wspr cycle. It works rather well.


ESP32 With Dual VFO’s

Hey look, its a home brew radio post. HAHAHA. So i have settled into this whole locked down for corona virus nightmare and have sorted out most things and have a game plan for how life will go on for the next 6 months of lock down and no work. And because i have everything in order, i have slowly been getting the enthusiasm to get into building some of the stuff i have in the pipeline and seeing if it works.

I have shelved the tuner for now, i really am not sure about it and it or how to correct its problems, which are both design and code and have started on the second iteration of the universal control board. First time round all the audio stuff worked just fine, more or less, but there were some design issues in the micro controller side of things. Namely, i screwed up things pretty bad and destroyed the 2 older ESP32 Dev Boards i had here which meant redoing footprints and all the other exciting things.

Obviously not everything has gone smoothly this time either, turns out that i laid out the board with to use an L7833 3.3v regulator as I am using a L7805 5v regulator as well, you know bog standard parts everyone uses. But guess what, the TO220 3.3v regulators I had were 1117A’s and they have a different pinout. Turns out, getting an L7833 regulator in Australia is not as simple as it seems, I had to order them from Element 14 for way to much plus way to much in postage. But, it got the parts and well got to getting this thing working. Lesson learned.

On the board, we have an ESP32 micro controller, the little red daughter board is a shift register module, ESP32 is 3.3v logic and most of the crap hanging off it is 5V logic, thus the need for shift registers on the I2C lines. The fuse is there from another lesson learned from the previous version of this board where I turned the solder on the micro into lava HAHAHA. We also have 2 headers for LCD screens, 2 rotary encoders, SI3531A module, yeah still using modules because the 2 times i have tried to roll my own using bare components, they never worked, and finally a real time clock. There is also footprints for 5 buttons, one of which is a reset button for the micro, 2 banks of 5 pins for band switching and pads for a thermocouple for monitoring heatsink temp of the final PA.

A bit blurry, but this is the initial fire up to make sure that I had things working right. Not much use spending hours writing code if there is an issue on the board.

Here I am starting to make some progress on the software, which is pretty much a ground up rewrite of one of the basic VFO codes out there, I think if i recall right was a very early version of a VFO by Jason Mildrum NT7S

And finally, this is where I am at with this. All the buttons work, the rotary encoders work and both the displays work. The one thing i do not like about just about all the VFO codes out there is that they all follow the radix math method for setting the vfo increment value. I want to have a 500hz increment, and actually, when i start to build things for 2m FM i will want an 12.5Khz increment and the like. Now using radix math you only get to have things go up in orders of magnitude, 1, 10, 100 etc. So I changed how that all works and simplified things alot, by combining the increment and displaying into the one function that gets checks once per loop. I can now set my increments to whatever value I like to suit my purpose.

Finally before i close down this epic post, on Jason Mildrums etherkit github some people have reported issues with ESP32 and the SI5351A library. I am not having any problems at all here. Everything works as expected and there are no I2C glitches. Well, that was a manuscript, thanks for reading and I will catch you next time.




Corona Virus Update

Corona virus, whats it mean to me and my home brewing? That is an interesting question and one i am asking myself daily. Sure it looks like I might have more time on my hands with the chance of being locked down growing daily. So that is the positive. The negative is I am also watching the economy tanking and along with it my income. I am a contractor, our clients are all pulling work so that means I will not be earning much at all. This is ok, I have some savings and will survive, but it means that I will lock down my budget to essentials and so I cannot see me starting any new projects for quite sometime.

However, I do have 4 or 5 projects already going on at various levels of progress, so i have plenty to do for now and I have a lot of parts so that is not going to be an issue. I think what will happen is that once i have worked through the projects in the cue, I will end up going back to doing manhattan builds with through hole parts. I have a lot of board and parts, so I am good for a couple years at least. About all it will really mean is that I wont be buying any fancy PCBs at all.

Oh and I have a lot of kits here also, so I can bring them out when things get desperate and have something to build. All in all, i will be ok, but if things get really tight, I might have to shut down my blog, or move back to cheaper VPS hosting rather than having dedicated server hosting. One stupid flu and the world is turning to shit.

Stay classy everyone.