1. Home
  2. Electronics Design
  3. Sensing 300A on a SMD PCB
  4. Page 3

Sensing 300A on a SMD PCB

te:

om

it?

for

o 8

e

gs,

any

tion.

.

re

a

ns,

.

w

at

in

r

y
g

IBM basically created the transistor business. They developed their own pla nar process, then needed more capacity, so they set TI up in the transistor business--they gave them all the fab equipment and everything. That line ran for, like, 30 years. Even in 1987, IBM was making a quarter of the wor ld's semiconductors. (Probably by value rather than by pound.)

Classically, Fishkill made logic and packaging, and Burlington made DRAM. In my time in Manufacturing Research, the DRAM generations were Antelope (1 Mb), Gazelle (4Mb), Luna (16Mb), Oberon (64 Mb) and Titan (256 Mb). I buil t measurement tools for Luna and Oberon, as well as retrofitting Perkin-Elm er Censor G-line wafer steppers with an improved alignment system so that I BM could squeeze another logic generation out of them before going to I-lin e (365 nm). That was ATX-4 ECL.

The bricks were (and I think still are) made in East Fishkill NY. They're sintered in a 30-foot-long belt oven filled with hot hydrogen. I've talked about it here before--it's like the throne room of the Great Oz.

The original alumina/refractory metal ones were in continuous use for over

20 years without a single field failure. The problem was that alumina has an epsilon of 9.5 or something like that, so propagation speed was slow and capacitance high. That didn't matter so much with 100-MHz clocks and ECL drivers, but it sure did with CMOS, so they went to a copper/glass-ceramic system. That was a disaster--in the old bricks, the metal had lower CTE th an the ceramic, and so was under compression in use. With the copper/glass bricks, the copper was in tension, so they had a lot of reliability proble ms in development, to the point that it looked like it was going to fail.

Then some bright spark figured out that if you gave the brick a really big thermal shock, you could crack the copper loose from the glass matrix, so i t wasn't in tension any more. Otherwise they'd never have been able to shi p it, I don't think.

There were giants in those days. ;)

Cheers

Phil Hobbs

Reply to
Phil Hobbs
Loading thread data ...

Who would have thought, when the CD4000 was introduced, that CMOS would win the supercomputer CPU business?

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

Did he copy that from Holec, or did Holec copy that from him?

joe

Reply to
Joe Hey

Stupid? The delta-I is pretty small, sure. At least the first ones were phase-controlled power supplies (SCRs modulating the conduction angle of the three-phase 400Hz input). I think they replaced them with switchers in the 3090s and after.

All made in East Fishkill NY. Burlington was all MOS.

Well, almost. There was a case of a $100M worth of Intel memory that went missing. ;-) I knew the rep that got the commission on it.

Reply to
krw

rote:

res wrapped around the bus bars.

976 - used the C-core as the core of a Royer inverter, that drove the core into saturation in alternating directions. When the current in the inverter winding added to the current in the bus bar, the C-core saturated at a low er current than when it opposed. Monitoring the DC content of the current r unning into the inverter was a remarkably precise measure of the current in the bus-bar, divided by the turns ratio.

of turns. IIRR (and this was told to me verbally, not revealed on paper, s ome forty years ago) 300A in the bus-bar equated to 100mA asymmetry in the drive currents.

s fundamentally accurate if you didn't build them into a null-sensing syste m.

or any local magnetic fields.

No idea. The most likely explanation is that Holec got him to do the design - he seems to have been doing a certain amount of private subcontract desi gn at the time, from contacts he mostly made via his golf club.

It's a while since I was in contact, but he wasn't much older than I am and may well be around to answer the question.

Playing golf is one of the many things I wouldn't do to get work. Hockey wa s fine - but didn't get me anything but the occasional bit of editorial wor k - but I did enjoy the game. Golf has always struck me as a silly idea. My parents tried it, but didn't like it, and I was never remotely interested.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

snip

Billiards is best. All forms.

Reply to
DecadentLinuxUserNumeroUno

275 Amps is a lot for a PCB. Better use the .090 thickness stuff and a lot of well placed large diameter PTHs.

The form factor for a precise 300 Amp precision shunt resistor is HUGE!

Maybe better to read the input line and extrapolate what is making it into the circuit.

Reply to
DecadentLinuxUserNumeroUno

Modern technique (multiphase) and available commodity components ($) may find that there are very few traces that actually carry 300A.

Figure out your topology first, then figure out what your actual problems are likely to be.

RL

Reply to
legg

Polyphase is certainly one of the arrows in the quiver, plus it looks likely there's to be a requirement for redundancy.

300A in the PCB itself looks silly--there's not much point in making FR-4 bus-bars.

Cheers, James Arthur

Reply to
dagmargoodboat

}snip{

Then he must have been drunk to give the idea away, because a certain Holec engineer named Lisser (IIRC) got the name there that he invented it and got a patent (I didn't check) on it. You didn't mention the word hysteresis though, which seems to play a more or less important role in Lisser's implementation.

joe

Reply to
Joe Hey

My favourite being snooker.

joe

Reply to
Joe Hey

formatting link

Holec NV was created in the Netherlands in 1962. They probably wouldn't hav e been interested in providing current monitoring equipment for the London Underground in the early 1970's. Lisser - whoever he was - may well have go t a patent for the same technique, which he might have invented independent ly (which happens rather often). Colin Hunter wouldn't have thought about p atenting his scheme. Kent Instruments did patent stuff at the time - I came under pressure to write a patent query for a pulse-width modulation circui t I put together there around 1975, but couldn't find a form of words that made it sound innovative enough to be worth filing.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

Epoxy resin compounded to fire retardation standard 4 isn't electrically co nductive, and doesn't lend itself to the construction of buss-bars.

There are electroforming techniques that can put a lot of copper onto print ed circuit board traces. "One ounce copper" is only 35 micron thick, but yo u can plate it up more or less indefinitely, and the technique is used for printed circuit transformer windings.

In theory, you could laminate FR-4 epoxy-resin bonded fibre-glass sheets on to fairly thick copper sheet, etch away some of the copper to give you the conductor pattern you wanted, and laminate up a printed circuit board where the copper layers were thicker than the layers of epoxy-bonded glass fibre .

I haven't talked to anybody who has ever done it.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

Try shooting pocket billiard's 2.25" balls on a snooker table. Very damn difficult.

Reply to
DecadentLinuxUserNumeroUno

Then there's old school:

formatting link
formatting link

Solderable bussbars made of .04in tinned copper were commodity for some time, the only issues being their termination or inclusion in inductive structures.

It's sometimes ludicrous - the amount of time, effort and money that gets funnelled into making an assembly pass through an automated soldering process, for something that will never see large volume automated assemmbly, or even compete price-wise if it did. Just saying......

RL

Reply to
legg

formatting link

-Lasse

Reply to
Lasse Langwadt Christensen

Gives a whole new meaning to "printed wiring board".

Reply to
DecadentLinuxUserNumeroUno

onto fairly thick copper sheet, etch away some of the copper to give you th e conductor pattern you wanted, and laminate up a printed circuit board whe re the copper layers were thicker than the layers of epoxy-bonded glass fib re.

The etching process relies somewhat on the copper being thin. If you used 4 mm Cu, track edges would all be undereaten by over 4mm.

NT

Reply to
tabbypurr

Use a laser to cut 2mm copper traces then laminate that onto a matching milled PCB "layer".

Those sign/nameplate lasers in the consumer channel would like work fine..

Then you could dip the trace into etch media before laminating it onto a PCB and get 'rounded' track 'edges'.

Reply to
DecadentLinuxUserNumeroUno

snip

formatting link
engraver-cutter

Reply to
DecadentLinuxUserNumeroUno

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.