65C51 up to his old tricks again

[cb3rob]

yuk. the 28L92 (like pretty much all things NXP) is smd-only. but then again we could buy a machine to bend the pins down and weld a piece to it but that'd make it even more inferior no lolol.

[cb3rob]

it seems to do some more things like detecting framing flag and addressing too (kinda odd for a serial port to do that internally). but the packaging sucks. sure we can solder pins to the bottom of an smd device (Rather weld them to prevent them coming off again in the solder bath and put some metal clamps over them to keep that crap in place no matter what but we'd rather exclude manufacturers that don't adhere to our specification (dip -40 - +125C) if we have to take their chips and practically decap them and put them in packaging that does conform to our requirements we might as well make the whole god damn chip ourselves (what would wdc charge for a license anyway getting so sick and tired of all this smd garbage poluting the market.

[Chromatix]

Well, that's where the SPI devices like the MAX3100 come in. They're cheap and compact in DIP-14 format.

And you can still get hold of vintage M6850s and R65C51s if you're dead set of a through-hole parallel-interface solution.

[cb3rob]

coming to think of it in this -specific- case the broken 65c51's might as well work as we need to read the bytes we wrote to the canbus driver back anyway to see if there was no collision (it's half duplex and both the tx and rx driver are connected to the bus independently so if anything else transmits at the same time it gets corrupted)... so doing that the rx buffer full flag can be used instead of tx register empty... but still they're essentially broken chips. as long as the application is half duplex and you get to see your own data back (so on a wire, not on a radio) i guess they can still be used.

[Chromatix]

There's also the option of using the PLCC package version of the 28L92, in a through-hole PLCC socket, if you don't feel like soldering up a QFP package.

[cb3rob]

considering that the 65C51 literally is cheaper than a lightswitch maybe it would be better if they simply asked 50 bucks for the things but made sure they work instead. lol. hmm. those PLCC sockets usually grow bad contacts over a decade or 2 plus the plastics get brittle. PLCC in a socket is still better than 'anything smd' ever was.. but it doesn't have the proven 40+ years of uninterrupted functionality dip packages have. (at least, without sockets and using lead solder they do (heck we'd use chassis mounting like in vacume tube radios (which are proven to work for well over 100 years if we could get chips that mounted like that lol.

[cb3rob]

essentially you're pusing matte-tin (rohs) against copper plated springs. that is gonna corrode at some point. if you simply solder it with pbsn you flush their primer off and replace it with lead. which won't whisker or corrode. but repriming them with pbsn won't work if you wanna put them in a socket again afterwards

[cb3rob]

thing is manufacturers should just make what we want them to make. which is the proper stuff. not the cheap stuff. the market currently works the wrong way around with 'nxp' deciding for us which package types we should use. (don't give a rats ass about the price either... it should simply be made to last

[BigDumbDinosaur]

BigEd wrote:

Never will be fixed, in my opinion.

I second that. There's absolutely no financial gain to be had in developing new masks to fix a UART that is fundamentally a 1970s design. My guess is WDC will sit on them for a few years, scrap them and take a write-down.

[BigDumbDinosaur]

cb3rob wrote:

yuk. the 28L92 (like pretty much all things NXP) is smd-only. but then again we could buy a machine to bend the pins down and weld a piece to it but that'd make it even more inferior no lolol.

With the lone exception of the 28L91, all NXP UARTs are available in PLCC. PLCC sockets are laid out on a 100×100 mil grid, which is hobby-friendly. There are also wire-wrap PLCC sockets.

[cb3rob]

PLCC sockets... what is it they do again? ah yes. they grow green edges where the contacts were supposed to be and need reseating.(and when you try that usually the entire side breaks off cuz the plastic has dried out) (probably not in the first 5 years but definately after 20 or so still better than anything smd. but not good enough.

[BigDumbDinosaur]

cb3rob wrote:

thing is manufacturers should just make what we want them to make. which is the proper stuff. not the cheap stuff. the market currently works the wrong way around with 'nxp' deciding for us which package types we should use. (don't give a rats ass about the price either... it should simply be made to last

Being that most production electronics involves automatic pick-and-place, followed by reflowing, SMD makes perfect sense. Also, the required board real estate is much less than using DIP packages, which makes for denser layout and lower PCB costs. Component manufacturers sell what the industry wants, not what hobbyists want.

Quote:

PLCC sockets... what is it they do again? ah yes. they grow green edges where the contacts were supposed to be and need reseating.(and when you try that usually the entire side breaks off cuz the plastic has dried out) (probably not in the first 5 years but definately after 20 or so still better than anything smd. but not good enough.

I've been working with digital electronics for 50 years and have yet to see a PLCC socket do what you are describing. Perhaps you should invest in better sockets if that is what is happening to you.

[BillO]

cbmeeks wrote:

Hmm. I may have spoke too soon.

You may have to add some short delays on the terminal software at the EOL (if processing is done after a line is entered) or between characters to allow time for the polling routine on the 6502 to find the character and read the receive register.

[cb3rob]

as for sockets doing what i'm describing: try going through some old equipment that's been around the block you'll find plenty and they don't even need to be anywhere near a leaking battery. some moist suffices to screw up joints between package pins and sockets... then try to get the plcc chip -out- of the socket without breaking the 20 year old socket. either way. better techniques were invented. that better technique is called 'through hole' and 'DIP'. so why bother with anything but that. "muh pick and place robot won't handle dip that easily'. well then hire a room full of chinese kids to do it by hand or get a better pick and place robot. but having equipment break prematurely (80 years sounds acceptable, 10 years does not) is not an option.

as for wdc fixing the 65c61... what's the problem anyway. the 65C51S worked INCLUDING the parity generator/checker no? they have the plans for that, they had previous production runs of that, so just scrap the N and only make the S again. (or else reproduce the harris 65C51AE (with the 'A' for alternate behaviour as it's better anyway as it doesn't cut off bytes halfway through when the flow control kicks in - that's something harris licensed from wdc no? so just make that one themselves (a bit faster than 4mhz would be nice they already HAD working chips for decades... they only screwed it up later on. it's a simple matter of sending the old plans back to the factory and telling them to make those.

[cb3rob]

not exactly 'hobbyist' considering that i'm considering to buy a cmos chip production line just to get the stuff we actually NEED no ? can't exactly explain to a customer that their missile went missing somewhere. lol. "where did it go? oh probably into that house over by the big smoke cloud"... probably some lose contact due to some smd crap falling off the pcb by itself after the moist under it froze at that altitude "oops".