One trick I've used with big machines is to replace the cutting tool with a spring-loaded pencil (and the thing you're cutting with a piece of paper) so you can dry run it (for example, lay the parts on the paper to make sure they fit and look good). Worst case you snap a pencil rather than crashing an end mill into a vise during a rapid (non-cutting move).
--sp
--
Best regards,
Spehro Pefhany
Amazon link for AoE 3rd Edition: http://tinyurl.com/ntrpwu8
Microchip link for 2015 Masters in Phoenix: http://tinyurl.com/l7g2k48
FR4 is fine in the 50 ps ballpark; just keep things short, which happens naturally in fast stuff.
On a microstrip trace, most of the current is on the bottom of the trace, because the dielectric constant of FR4 is almost 5x that of air. So the top plating doesn't matter. What's worse than FR4 dielectric loss is the fact that normal PC boards have the bottom side of the copper crudded up with "black oxide" to improve adhesion of the copper to the epoxy. That multiplies the skin effect losses. Peel up a trace and see; it's disgusting.
For seriously high frequencies, you can buy low-loss low-Er dielectrics with microinch-smooth finish on both sides of the copper. The traces will peel off easily with an x-acto knife. That's what works well for RF stuff, like microstrip filters and LNAs, where skin losses cost noise figure.
We did one board that cleverly had microwave laminate for layer 1, then FR4 for the other layers. It curled up like a potato chip. If you laid it on a desk and gave it a good twirl, it would spin for a full minute.
A pantographic cutter might be cool. I could trace a sketch.
As noted, we have a Tormach nc mill, a cute little Sherline nc mill, and we used to have an official nc PCB routing machine, but we got rid of it.
I did another iteration on my pulse driver circuit, which is beginning to look like it might work, in about 10 minutes of Sharpie marking and Dremeling.
I'm not adverse to nc machining when it makes sense. I have written three compilers for nc machines, mostly big to gigantic stuff. The Whitney punch press would shake the building every hit; that was fun.
Ok, let's see if the diode can handle some RF power. I would guesstimate the contact surface area of the common 7/16 DIN connector at about 800 sq-mm: That's probably off by an order of magnitude in either direction. However, with variable contact surface area, alignment problems, air gap clearances, and my lousy math, the errors might cancel each other and magically produce a usable guess(tm).
The junction is 0.01 sq-microns or 1*10^-8 sq-mm. Assuming uniform current density, uniform contact surface, and only one diode, the ratio of diode area to contact area is: 1*10^-8 / 800 = 1.3*10^-13
A typical cellular xmitter belches about 40 watts per antenna. Into
50 ohms, that's: I = (P/R)^0.5 = (40/50)^0.5 = 0.9 Amps of RF current.
The portion of that which goes through the diode is: 0.9A * 1.3*10^-13 = 1.2*10^-13 A = 0.12 picoamps
Power dissipated in the junction is about: P = I^2 * R = 1.2*10^-13 * 100 = 1.4*10^-24 watts P = 1.4 yoctowatts which is a very small amount of power that I don't believe is sufficient to blow up the diode.
But to be sure, will the self forming nickel-oxide diodes survive 1.4 yoctowatts and make my life miserable by mixing RF signals and creating intermod, or will the power be sufficient to vaporize the diode(s) and save me from a weekend of excessive worry?
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
Mine survived bias up to about 0.2V pretty consistently, so that's 400 uW DC. However, the resistance goes up exponentially with thickness past about 20 angstroms. Mine were right around 25 angstroms, based on curve fitting the junction parameters (which actually fit amazingly well when you do it right). So just a few atomic layers difference results in the thin places hogging all the current. You might well be off by 5 or 6 orders of magnitude in your calculated junction area.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
Did you use a magnifier to do that? I'm extremely nearsighted so I just take off my specs and put on shop glasses. And a mask for glass, though I haven't done a board in a while.
Is OSP finish ever considered as an alternative to ENIG for such RF stuff?
--sp
--
Best regards,
Spehro Pefhany
Amazon link for AoE 3rd Edition: http://tinyurl.com/ntrpwu8
Microchip link for 2015 Masters in Phoenix: http://tinyurl.com/l7g2k48
RF stuff on such materials "never" get solder mask, OSP, or any other coating (won't stick to teflon, interferes with ribbon bonding). Most of the time the boards are stored in sealed bags or in dry boxes - standard practice for the general class of applications. Worst case, if boards have been exposed to open air for a long time or have been inappropriately handled, they get a wash in a lab grade detergent and some time in a plasma cleaner (Ar, sometimes with a small amount of forming gas (NH4 aka ammonia)) to restore solderability, minimize epoxy bleed, improve ribbon bond adhesion...
I've run TDR, 30 ps rise time, on microstrip traces that had sections of solder mask removed. I couldn't see any difference.
Maybe I could find a good microstrip board and try removing solder mask some places, and removing the ENIG down to bare copper in other places. I'm guessing that won't show up either.
I can see what is probably the glass fiber weave in the FR4.
--
John Larkin Highland Technology, Inc
picosecond timing precision measurement
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
Up to about 500 MHz, we would solder mask everything. The danger from external contamination was far greater than anything the solder mask could bring to the table. This is for marine radios, where salt water contamination is common and where a solder mask really helps. Also, no conformal coatings except where demanded by the customer or where some errant designer decided to use high impedances in a marine environment. Bad idea.
The choice of cleaner varied by the type of contamination, but we just washed the PCBs with a mild detergent in a washing machine and rinsed them with deionized water. At the time, we used wave soldering, with water soluable flux. Vapor or IR reflow soldering will probably require different techniques.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
I've done PIN switches, filters, couplers, preamps and such on Duroid
5880 and 6002. Most of these devices included hard substrates like BeO or AlN where bare die and other RF components were attached. Duroid or AlO3 was used for baseband circuitry - switch drivers, regulators, modulators and to route RF signals to connectors. In several cases I changed the ENIG to a Pd barrier with a thin Au coat to reduce losses due to Ni's magnetic permeability. With the hard substrates die and component attach uses a succession of solders with different melting points - AuSn to start (semiconductors), AgSn for passives, and PbSn for bonding the whole assembly to the housing. After that, there's no more soldering - interface is through ribbon or ball bonding (soft gold is absolutely required for that) or conductive epoxy. Most of these devices were put into a hermetic housing but some went into a gasketed housing (a degree of repairability) so any protective coating like solder mask served no purpose.
FR4 and solder mask are fine for lumped circuitry up 500 MHz at a few watts but when the power gets above 2-300 WCW FR4 is simply too lossy - it melts and become even lossier, burns and you end up with an ash tray. At higher power levels a teflon/ceramic or glass fiber material with tightly controlled dielectric properties is required. The 5880 class of materials is very low loss and is preferred until you get to something like 1 kW cw at 500 MHz - it gets too hot and solder melts. The alternative is the 6002 material which has much higher thermal conductivity - 1200 WCW at 1 GHz for example.
Losses due to solder mask are probably insignificant (it just isn't where the E fields are). Losses due to Ni are - a PIN diode switch at
500 MHz might see 0.5 db loss which will drop to 0.35 db without nickel
- a useful improvement in S/N and power handling when loss budgets are tight (4% here, 4% there - it adds up).
We've used a Parylene coating on finished circuitry where water accidents could occur. Not something you do in you own shop but the coating is very thin, durable and gets into small spaces. Not sure how well it works at really high impedance (giga ohm) but for normal electronics that might get splashed, it covers everything important. I think if I were doing stuff for chemical labs, it would be my first choice (I hated scrubbing green stuff off expensive electronics back when I did that stuff). I've used the krylon trick but it doesn't get under things. I also tried dipping or spraying with LPS-1 - good for routine humidity and mild lab fumes but not for a direct splash.
Most metallic magnetic things quit working at high frequencies. Ferrites are better than, say, Metglas in the 100s of MHz. I wonder when the nickel quits being megnatic.
Right. We don't gold plate the bottoms of the traces!
Losses due to Ni are - a PIN diode switch at
I work in time domain, mostly low power stuff, usually around 1 ns edges but sometimes down to 40 psec. Some fast serial stuff, like 1 GHz SERDES and PCIexpress. Ordinary boards, FR4/ENIG, seem to work fine. But I wouldn't notice the equivalent of 0.5 dB loss.
--
John Larkin Highland Technology, Inc
picosecond timing precision measurement
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.