@M0YNG FT8 start/stop envelope shaping and the Gaussian slides between tones have all but eliminated the old switching transient wideband splats, but I find it extraordinary that ops still don't have a clue about how to set levels. It ain't rocket science. Turn off any audio gain control, compression or processing. Cut any ALC connection to amplifiers. Switch it off. Disable it. Woteva. Monitor your TX power. Turn the FT8 power slider in WSJT-X down very low. Next, (1/3)
@M0YNG ...turn the power slider up until you reach the output level you want to achieve. Make sure that if you back off by 3dB that the PA output drops 2-3dB. If not, get a bigger PA or use less power. 1dB compression isn't going to ruin the inter-tone slides.
Make sure you have Fake-It or real split working enabled so you never transmit tones that are less than 1500 Hz, so even if your TX audio chain is truly terrible, you won't transmit in-band audio harmonics.
@M0YNG If your radio can't even support Fake-it, then either never operate with a tone less than 1500 Hz or get a radio that can do split or Fake-it. Make sure your radio has linear phase response to at least the maximum tone you intend transmitting, and that the amplitude response doesn't fall off a cliff anywhere, so you get imbalances between tones.
@M0YNG It's very basic stuff, but I still see folks who don't have a clue why they are sending multiple copies of their FT8 tones at 2/3/4 times the correct spacing because they haven't enabled split or fake and they are transmitting ludicrously high levels into a terrible audio transmit chain at less than half the audio bandwidth. I also see start-up splats where some n00b is using ALC to set the transmit level, when there's a flippin slider on the actual WSJT-X page to set the level. Sheesh.
First attempt at a carbon-fibre-filled PLA support arm (one of a pair) for the milled-from-plate offset parabolic 122/134/248 GHz mmWave antenna dish. The left end attaches either side of a block bolted to the threaded holes behind the dish. The face is inclined at 18.9 degrees so points at the horizon with the bottom edge horizontal. The four mounting holes at the right end bolt to the enclosure containing the radar chip, focus rail, control electronics and feedhorn. Bit of Voronoi for bling.
This is roughly how the arms are used, but with the dish inclined at 18.9 degrees to the vertical rather than pointing at the zenith. Dish is 230 x 250 mm, made from 3/4 inch aluminium plate. The heaxagonal pockets are milled out to save weight but retain stiffness. The dish surface is accurate to about 20 micrometres, with 4 micrometre ridges from the spiral machining using an 8mm ball-ended carbide mill at 11500 rpm and 2800 mm/min on my SYIL X5 CNC mill #hamradio
Mysterious Things appearing on my Bambu X1C. Spiders to support hyperbolic Cassegrain subreflectors in the very deep Edmund Optics 18 inch parabolic dishes with f/d 0.25, so the focal point is at the same height as the dish rim. Total pain to illuminate unless you set it up as a Cassegrain, with the feedhorn phase centre right at the dish face centre orifice. These are for 76,122,134 and 248 GHz #hamradio
And this is what the printed Thing is for. Three carbon fibre rods flare outwards from the three sockets and go through metal sleeves to a support ring behind the dish. There's an M12x0.5 mm thread inside the hyperbolic sub for fine focus adjustment. One turn is a fifth of a wavelength at 122 GHz #hamradio
Here's one of the smaller Edmund parabolic dishes. This is the 12 inch/300 mm version. These take a 48 mm diameter Cassegrain sub, the 18 inch/450 mm versions take a 68 mm sub and the 24 inch/600 mm versions take a 98 mm sub and need rather stronger mountings, 6mm carbon fibre tube instead of 4mm pultruded CF rod. Shiny! #hamradio
I paid DHL ludicrous money for an urgent delivery from Italy. It arrived in the UK very rapidly, cleared customs (bargain £11 ransom fee) and was out for delivery by 8am. It was 12 miles away when it was supposed to be delivered at 10 AM. Six hours later and it's now 13 miles away. DHL: "Sorry, we're running late". That makes two of us pal. #dhl
Update: the order eventually arrived SEVEN hours late. That means I missed the last shipping time so the Things might not be with the customer before the Friday deadline. Super service from the supplier in Italy, but it took longer to get here from Leeds than it did to get from Italy to Leeds, including customs clearance. If I'd know this, I wouldn't have paid for the premium service. I have another order on the way from TNT, let's see how THAT goes.
@jmorris Working on a full range of items for sale, SMA, 2.92, N and TNC to waveguide transitions, waveguide to waveguide tapered transitions, Cassegrain and Gregorian subreflectors for multiple dishes and frequencies, directional couplers, waveguide and coaxial filters, Edmund Optics dish mounts and focus slides and hopefully some high precision machined dishes for mmWave, parabolic but also axially-displaced ellipse types, lots of feedhorns and lenses. No time to sleep or make Youtube vids!
When I'm designing a Cassegrain antenna for mmWave radio, the main dish is a parabola, but the deep ones from Edmund Optics are very hard to illuminate because they have a focal length to diameter ratio of 0.25, which is WAAAAAY less than most dishes. Solution is to put a hyperboloidal subreflector with its rear focus coincident with the prime focus of the parabola, and the front focus of the hyp is where you put the feedhorn. I made the rear so it can take a very fine adjustment thread
When I create the hyperbolic curve, it is described the usual mathematical way, with just the values of a and b from the standard equation of a vertical hyperbola. Problem is that Fusion CAD describes conic curves in terms of the rho value and the height and width of the tangent intercepts. I guessed there would be a conversion function. HAHAHAHAHA nope. So I sat and did a load of calculus and trig and algebra, resorting to Wolfram Alpha at one point. All sorted, and here is the first reflector
Minor improvements in surface finish on the 122 GHz Cassegrain subreflectors. Using a larger corner-radius end mill. Seven pockets milled, just for the sheer joy of non-symmetry. Still a lot of work to do on pocket finish, deburring/chamfering/radiusing the ribs. My Fusion360 licence doesn't include the 3D debur option. Boo. Investigating whether I can use trace with a sketch path, or design-in some fillets, but that might push the cycle time from "waaay too long" towards "srsly dude?"
@clacke@drahardja@lispi314 I wrote code in the early 1980s that took full notice that two-digit year fields were a liability and that some folks who were in the datasets were possibly *born in the 1800s* and that the dataset might endure past the end of the 1900s or require forward calculations into the then-distant 2000s. Agreed there were some systems that had so little memory that date compression was needed, but we used binary fields for time, not truncated text dates.
@drahardja Every time some cretin mentioned that Y2K was a hoax, the entire team would respond and suggest that they take their uninformed opinions and go elsewhere. Even 20 years later, we were easily triggered and responded with overwhelming rage. I've almost moved on and now just add them to the idiot pile and treat them with due caution. Two years of our lives wasted on fixing the work of idiots. Marvellous. See also C/C++ coders who didn't understand bounds checking. Idiots everywhere!
Radio ham since 1973. DC to light. Youtuber (Machining and Microwaves) Hacker, machinist, maker of Things, Infosec CTO, basketmaker, carpenter, Chihuahua herder, sceptic. East Yorks UK