fortunantly I never had the "throw it over the wall" experience
I have always sat with the layout guy for parts placement, marked up his copy of the schematic with colored pencil annotations as we worked
usually had a very early review where he fully expected some significant rework
I have always sat with the layout guy for parts placement, marked up his copy of the schematic with colored pencil annotations as we worked
usually had a very early review where he fully expected some significant rework
Straight from the guy next desk: yes, connector torque is critical for calibrating microwave equipment. The inherent discontinuity in connector mating can be minimized this way. To little torque and the mating surfaces are to far away, resulting in reflections, to much torque and the mating surfaces hit, the connector bends and is then permanently damaged. You should see this guy, he's never mating a connector before gauging it with a special micrometer tool (and he's at less that 40GHz, I'm told that at 110GHz this process becomes a special form of insanity). He has a full kit of torque wrenches, all sizes.
A man after my own heart. I went further and dragged the mechanical designers up onto rooftops, preferably in the rain to show that, just because they could assemble one of these in a nice warm office didn't make it easy for the guys on site.
In Minneapolis in the winter all the site teams quit to work for telcos who still used equipment huts.
Yep - it needs to be right, but a stuffing great spanner wielded by someone used to installing sub station transformer is as wrong as it gets... Usual result is the whole assembly spinning until until the cables shear...
FR4 is fine for anything that doesn't require a known dielectric - like canoes 😀 Just about every different manufacturer's FR4 has a different value... Depends what you are doing! PCB microwave antennas need a very predictable substrate, if you want the best possible link budget...
Or a stilson 🙂
Serves the designers of the 7/16 right for leaving somewhere a big spanner can go 🙂 . Of course audiophiles will be happy to know it's one area where connector plating material really does make a difference.
If you are buying thousands a month oddly the supplier is very happy to control the layup and give you want you want. We spent about 5 years getting the suppliers under control mind. G200 was too much of a price jump for the bean counters. Happy days.
Yes, general purpose FR4 is good enough for most of you. However, the CTC Blowtorch used Teflon and I am not sorry that I chose it.
However, there are different 'types' of FR4, and some are OK for Parasound, etc.
However, there are different 'types' of FR4, and some are OK for Parasound, etc.
Attachments
Last edited:
Relative permittivity and low dielectric loss, constant across the board and across batches. The typical Rogers 4350 material is a glass-reinforced hydrocarbon and ceramic dielectric (that's all I know about).
As far as I know the FR4 is glass/epoxy and the G20 materials were glass or paper/ phenolic material. Can't tell you the differences in dialectic properties but the G20 cost much more as far as I remember it. We used these materials in aerospace for other than electronic uses.
Microwave PCBs
My understanding was that it was all done using software since every element (trace, shape even corner) becomes a circuit element. Kind of goes from autoplace and route are more and more meaningless as frequencies go up to suddenly the equivalent simulator is fundamental to even functioning in millimeter wave stuff.
Say goodbye to lumped component models.
A physicist I know had been used to working at low RF frequencies. When he was hired in at TRW he tried to make an attenuator with flat response out to about a GHz. He soldered it together out of little resistors (this was at best the dawn of surface mount) and put it in a little box. It was terrible. He began to appreciate the significant divergence that occurred around that frequency and above.
It makes a difference even at high RF freq. The torque wrench will break or click once when you are at the correct torque setting of the wrench. You need to match the wench to the connector to get the right torques for each connector type. Without the proper tools, you will damage connector mating surfaces and get bad data (and may not know it for awhile).
Microwave pcb material (the grey color material) is very stable in its electrical characteristics but must also be stable in dimension with temp changes. Physical dimensional changes of minute amounts lead to bad results.
THx-RNMarsh
Last edited:
Of course all boards are different, but the rules are not. I would assume chess games are all different to some extend, but not the rules. There are specific strategies for a specific situation, valid for any game and any PCB.
I recently did a switching converter for the first time in my life so I read up all about switching converter PCB design. What do you know - it's the same sensible things as in analog audio!
Proof of the fact that it is not creativity but just a trick to master the unwritten rules: as you say, with lots of experience you get better at it. Just the opposite to creativity.
I grant you the last word.
Jan
The RF types I used to work with and try and herd on occasion in the early 90s were an odd bunch even by my standards. Years later we joked that it was an aspergers research project not R&D. There was a hard core of them that used to say that 'anything under 3GHz was DC'. These days a mobile phone has a 5GHz (albeit flea power) radio in it. It's then you step back and realise what a boon PCs have been to radio engineering and what amazing steps have been taken in integrated silicon processes.
My understanding is that some advanced BiCMOS processes are at the point where uC, DAC and RF power stage can all be one one chip, albeit sub watt levels up to significant GHz. 2.4GHz is boys stuff now.
Progress 🙂
We worked with a group trying to define the shapes as components, they were drafting their boards on an 80s CAD package that let them adjust the component geometries on screen as they laid out the boards. I would presume that these days there are bespoke packages, as said we mere mortals don't get involved with the high GHz stuff.
Yeah, 15GHz is well integrated these days. Even R/C models use 5GHz for control now!! While modelling is needed if you are doing anything more than a bit of stripline, the software isn't megabucks anymore. But a lot of skill is still needed in the original visualisation. Machines can't do that for you; my colleague at an EMC lab still sees a lot of bad layouts at any frequency...
We're involved in a lot of SDR (software defined radio) - that's another fascinating technology.
- Status
- Not open for further replies.