why do they do this?

Got to ask here, if I didn't years ago. How much different in behavior are analog circuits layed out with discrete components from the final IC itself? Let's talk pre computer simulations. For example, microwave and RF boards are constructed different from audio circuit boards, but if you look at a schematic, it's just a bunch of discretes wired together.

Reply to
Cydrome Leader
Loading thread data ...

I came across something terrible once while servicing an older machine. It was "standard" octal socket/DIN mount timers or temp controllers that had been tampered with by the manufacturer of the final product. The intent was to render the use of off the shelf replacements inoperable. It took some time to figure this out. Never seen a dick move like that before. I went ahead and un-f***ed the rest of the parts and rewired the sockets to allow replacement of parts in the future. Sort of surprised they didn't wire a NC relay contact across the power supply terminals to be honest. I broke one of the controllers open to see why the new counterparts did not work.

If a customer sounds qualified, do you ever give tips in troubleshooting/recalibration or anything like that?

There should be callouts, but sometimes folks have to learn the hard way. I recently came some not-working test equipment that somebody tried on 208 volts instead of 120 volts. Guess they forgot about that fuse block voltage selector.

Did anybody keep copies in filing cabinet or microfilm or someting weird like that?

Reply to
Cydrome Leader

lol. Coded squelch? How are those things matched or activated between transmitter and reciver?

Reply to
Cydrome Leader

Cydrome Leader wrote in news:raaked$in1$ snipped-for-privacy@reader1.panix.com:

Document management standards are now ISO 9001 At least for any company with any brains and some contractors (The US Gov) require it.

Reply to
DecadentLinuxUserNumeroUno

:

eir fake dropout specs while conveniently omitting the fact that Vbias must be greater than Vout + 1.5V.

ts

B PSRR at 100 Hz. Battery operation usually doesn't care a whole lot about PSRR. And the thermal impedance specs are so bad, you just try getting 800m A out of it with any kind voltage headroom without using a liquid nitrogen drip.

d

That's a bit of a laugh. Suggesting that JT kept anyone in check with his delusional ranting is like suggesting Larkin tells the straight story of th e coronavirus by his in depth analysis of the data and his knowledge of epi demiology.

JT may have done interesting design work, but when it came to politics he w as a raving maniac.

--
  Rick C. 

  --- Get 1,000 miles of free Supercharging 
 Click to see the full signature
Reply to
Ricketty C

rote:

:

as

et.

h their fake dropout specs while conveniently omitting the fact that Vbias must be greater than Vout + 1.5V.

ivolts

pamp

o
80dB PSRR at 100 Hz. Battery operation usually doesn't care a whole lot ab out PSRR. And the thermal impedance specs are so bad, you just try getting 800mA out of it with any kind voltage headroom without using a liquid nitro gen drip.

mA

h
d

y.

el

w

all

s and

at

He was imperfect enough to look remarkably like a psychopath.

mes

he was close to not being with us any more

e it

over at that point

ate the design, and then had another designer lay out the chip

e

My impression is integrated circuit designs don't look much like discrete c omponent designs.

Lot's more current sources, and many fewer resistors for a start. At high f requencies, the fact that the connections are lot shorter (and less inducti ve) makes a big difference. Bob Widlar and Barry Gilbert were famously good at exploiting the advantages these sorts of difference offer.

If you look at the printed circuit boards, microwave and RF boards are cons tructed with the critical connections laid out as constant impedance transm ission lines routed over (or between) solid ground planes (mostly buried in multilayer boards). Buried ground planes do show up in some audio circuits - they do a good job of shielding sensitive bits from noise radiated from the higher current bits, and any digital signals running around - but you c an get by without them.

Looking at the schematic isn't all that educational - essentially it's a wa y of representing the net-list (which pins on which components are connecte d to which other pins). It leaves out stray capacitances, lead inductances, mutual inductances between leads, heat dissipation, component dimensions a nd the rest of the stuff that constrains printed circuit layout.

Back when I was designing circuits, the schematic got passed to the draftin g shop with several pages of notes on how it was going to have to be laid o ut.

The good layout draftsmen didn't need much, but they weren't all good. Once a printed circuit manufacturer didn't like my board stack-up - they feared that the board would warp - and the draftsman okayed their version.

A whole lot of constant impedance transmission ended up with with the wrong impedance. It took us a couple weeks to wake up.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

Anything they want, including schematics and advice and replacement loaners. But few customers can repair and recalibrate a product, because that generally needs a rack full of computer and test gear. We have sold a few test racks to customers who want to support the things themselves.

It really annoys me when somebody charges 100x the real value for a replacement part that their design fried. Car dealerships sell low-end cars at a loss and make their profit on parts and service, like a $1000 door wiring harness. Kinda like ink cartriges.

Ancient history! It's great that most gear now runs on most any AC or DC voltage.

The paper's all on file, and we rarely have to look it up.

--
John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 
 Click to see the full signature
Reply to
jlarkin

Or those tire pressure sensors. A $30 retail part that some places charg e $2400+ to replace. All they have to do is break down the tire from the ri m, and exchange the part. One failed on my Caravan before it was two years old, and my dad had one fail on a Ford pickup truck before it was six month s old. I can buy a complete aftermarket system for under $20 that has the t ransponder in a metal valve cap. The OEM use the same receiver for the elec tronic key. The rest is in the slopware.

Reply to
Michael Terrell

so how did they come up with designs that worked? Take the first op amps for example. The behavior of the silicon isn't going to match a breadboard with some parts on it. Was it just a completely iterative process back in the day to get anything to work?

Was any of this salvagable, or was it on to rev b?

Anyone have stories of most expensive screw ups with board design or assemble?

Reply to
Cydrome Leader

It's sort of sad wiring harnesses haven't been figured out yet. I was reading about some maker that used some sort of soybean based wired insulation that critters would eat up. Who the hell came up with compostable wire or whatever the heck it was supposed to be?

yes and no. Most throw away type internal switching power supplies are not serviceable and it might be hard to find replacement down the road. I actually hit a roadblock with a plain old filament+isolation transformer. I don't recall the generic part #, but they were a dime a dozen 100 years ago, but not now and I can't find a suitable replacement, even if I mount the new transformer somewhere else in the chassis. Sort of annoying. May have to give in an call the manufacturer to see how much they want, assuming they have spares. The 6.3 winding voltage seems too high so I suspect shorted turns, but can't really confirm that easily.

Reply to
Cydrome Leader

u

your system wouldn't have worked at Microdyne. It was a different market, r equiring different methods. For instance: The 700 and 1620-base models shar ed a lot of boards, and the front panel/embedded controller was also used i n the custom system built for NOAA to control their 100 foot dishes that tr ack their LEO birds. Each board or model had it's own base model number, an d some had over 20 versions because customers wanted different options, Som e of their first products were still in use 24/7 for over 30 years at NASA, tracking deep space satellites. They had never been turned off, or repaire d.

.

er)

to

n
a

It

ad

I

to

I

ot

It's not just soybean based plastics. The military often finds out the har d way when some critter decides a material is din-din. That's why there is a fungus spec in all military electronics now. I believe they had some ai rplane accidents because of bio-degradation of petroleum based plastics.

--
  Rick C. 

  --+ Get 1,000 miles of free Supercharging 
 Click to see the full signature
Reply to
Ricketty C

BITD they had kits of prototype parts from the process they were going to use, and wired them up as breadboards. Some issues with that include:

(1) temperature tracking. Discrete parts don't track well no matter what you do. Monolithic parts are a great deal better.

(2) Strays. Stray inductance on-chip is nearly always negligible, and stray capacitance is probably 2 orders of magnitude less, even in a

10-um process. (Capacitance tends to go like the linear dimension.)

(3) Strays (2). Mutual capacitance is very different in a monolithic design vs. a breadboard, and of course there's RFI and stuff to worry about. (It's a lot worse now than in 1970.)

Despite these issues, early IC designers managed to make some pretty neat parts.

Cheers

Phil Hobbs

Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510

formatting link
formatting link

Reply to
Phil Hobbs

Shorted turns: look for smoke.

Most likely just poor regulation... add some more load on the 6.3 and see if it gets sensible. Phil A knows mains transformers and will no-doubt be around to correct me if I am mistaken.

--
  Jasen.
Reply to
Jasen Betts

I just wanted to add to this thread...

Way back, I used to fix beepers for a living. I recall that Panasonic pagers had ridiculously long part numbers. (Panasonic was never a big name in the beeper business - and I think we onl y carried one model - a UHF band tone-only?)

Anyway, all Panasonic part numbers had the same length. But too many characters for preventing human mis-key on order entry. That is, unless the part numbers had error correction built in - they were certainly long enough for that.

For example: 690004766AC6335E3E22010 would be typical. No hypens, no spaces, no breaks. And they would list them in the service manuals in one long non-alphabetize d list (sorted according to component number: R1, R2, C1, U1, etc..)

And the next part you might need to order would be something like:

609007664C6A635F3E22110

Stupid. Stupid. Stupid. !!

Personally, I am not of fan of trying to encode product information (featur es, configurations, options, etc..) into part numbers. That leads to too m any problems, especially if the overall numbering plan is not very, very we ll though-out ahead of time. Exceptions and overlaps are bound to occur, a nd then what? The whole system is crap. Better to have a relational datab ase and just look it up.

Reply to
mpm

e:

ote:

Vbias

osfet.

with their fake dropout specs while conveniently omitting the fact that Vbi as must be greater than Vout + 1.5V.

illivolts

n opamp

n to

owy 80dB PSRR at 100 Hz. Battery operation usually doesn't care a whole lot about PSRR. And the thermal impedance specs are so bad, you just try getti ng 800mA out of it with any kind voltage headroom without using a liquid ni trogen drip.

800 mA

much

V

used

stly.

annel

nx

few

but

es all

ious and

nd

ve at

on

times

as

ss he was close to not being with us any more

like it

ok over at that point

mulate the design, and then had another designer lay out the chip

are

te component designs.

gh frequencies, the fact that the connections are lot shorter (and less ind uctive) makes a big difference. Bob Widlar and Barry Gilbert were famously good at exploiting the advantages these sorts of difference offer.

e

a

d

o

y

It got sort of cobbled into sort of working with lot's of sub-minature Filo tex coaxial cable (about 1.1 mm OD IIRR). The other thing that messed up th e board was Gigabit Logic's failure to put the input capacitance of their l ogic on the data sheet - on rev b every last clock input had it's own priva te driver.

The other thing that messed it up was that it was supposed to clock a long

100k ECL synchronous counter at 200MHz, which would have been doable with c areful design - as I had spelled out in the specification. The guy who did the detailed design of the board was entirely capable of managing that, but was also smart enough to know that the machine was never going to get into production - it didn't - and didn't bother.

If we'd done a design review on the schematic before releasing for printed circuit layout I might well have caught it, but our idiot project manager d idn't want to slow down the process by the day that would have taken.

We had a similar drop-off on the digital signal process board which took mu ch longer to find (but was much easier to cope with when it did get noticed ).

I was sufficiently peeved by the time the project got cancelled - after thr ee years work when the machine was working - to take away three years worth of weekly reports. About fifteen years later I got around to scanning them .

You could read them here

formatting link

if you were a glutton for punishment. The reports are accessible - in one y ear chunks - at the bottom of the page. There's also a link there to a docu ment that fills in some of the immediate background.

It will make more sense if your read my "short history of voltage contrast" link at the top of the page, which ends with links to same set of document s.

I think we were paid about a thousand UK pounds for the board that got scre wed up - it was a six layer board, with the two outer layers made of isocya nate bonded Teflon (PTFE) cloth, rather than than the standard epoxy-bonded glass fibre. The printed circuit house had had to buy in an expensive chun k of the PTFE based board material before they could make the board, and d idn't expect to sell any of it to anybody else.

We then stuck about five hundred UK pounds worth of GaAs integrated circuit s from GigaBit Logic onto the board. That meant that wed spent more than tw o thousand US dollars on the board. It was a triple extended Eurocard, whic h was pretty big for a rack-mounted board plugging into a backplane.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

e:

you

o your system wouldn't have worked at Microdyne. It was a different market, requiring different methods. For instance: The 700 and 1620-base models sh ared a lot of boards, and the front panel/embedded controller was also used in the custom system built for NOAA to control their 100 foot dishes that track their LEO birds. Each board or model had it's own base model number, and some had over 20 versions because customers wanted different options, S ome of their first products were still in use 24/7 for over 30 years at NAS A, tracking deep space satellites. They had never been turned off, or repai red.

e

OM.

user)

g to

on

o

s a

ey

e. It

had

nt

ok

. I

s to

't

t. I

not

e
s
d

ard way when some critter decides a material is din-din. That's why there is a fungus spec in all military electronics now. I believe they had some airplane accidents because of bio-degradation of petroleum based plastics.

One of the more famous examples was the WW2 Mosquito figher-bomber, which m ade brilliant use of urea-formaldehyde bonded layers of wood.

It was extreme;y effective in the temperate zone, but fungi got at the str ucture in the tropics, and the planes exhibited a nasty habit of coming apa rt in midair in those theatres.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

rote:

:

as

et.

h their fake dropout specs while conveniently omitting the fact that Vbias must be greater than Vout + 1.5V.

ivolts

pamp

o
80dB PSRR at 100 Hz. Battery operation usually doesn't care a whole lot ab out PSRR. And the thermal impedance specs are so bad, you just try getting 800mA out of it with any kind voltage headroom without using a liquid nitro gen drip.

mA

h
d

y.

el

w

all

s and

at

mes

he was close to not being with us any more

e it

over at that point

ate the design, and then had another designer lay out the chip

e
F

It can be very simelar

In our case, to keep NRE low, we needed to use almost only predefined block s. When you team up with a fab, they supply all datasheets and simulation m odels for standard blocks. So, opamps, comparators, digital blocks, voltage references, PWM circuits and individual transistors

The blocks are proven in the technology and simulation vs real life is 100% spot on. The tools are very expensive, but they model the chip to every de tail. Thus, experienced ASIC designers never need to do new revisions.

If you wander off, and do custom blocks, it gets a lot more expensive

So you if you find a discrete opamp that matches your ASIC block, so can do proof of concept in discrete design If not, simulations is the way forward

And by the way, prototypes with MOSIS is very cheap, you just need to wait a loooooong time from design to the first chips comes in

Cheers

Klaus

Reply to
Klaus Kragelund

Interesting read- nearly completed with 1989. Got some more observations questions, if you don't mind.

How were low production SMD boards populated back then? Semi-manual or automated?

The references to 17A -5.2v ECL boards was amusing too. I was reading somewhere that some folks didn't use convention regulated power supplies for ECL boards as the load never changed. Is there any truth to this?

What were the special mu-metal tanks and covers for?

The 0.4hz cursor update speed on a display is funny, in comparison for the hunt for 100ps timing on other parts of the device.

There was reference to some type of glue logic (7400 series?) LS vs HC where the HC series stuff used too much power. Wasn't the CMOS stuff supposed to be lower power, or was this only below certain frequencies? I recall some other boards where the power hungry "S" series stuff was still used in place of LS series parts.

- in

I wonder what the board houses yield was, 0.008" traces is still pretty small. How large were these boards? The AMD 0.1% yield on another component was pretty funny too. There was an issue at work where a process (not product) had a 5% pass rate. My joke was the systems that passed validation were simply done right, by accident.

OK, so triple extended euro card was the size of the fancy board? Need to look that up.

How hard would all this be to contruct using modern technology? I find it fascinating that we still fall back to plain silicon and don't require exotic materials for speeds you were dealing with.

Is this simply because components can be made smaller now? The spec sheets for USB3 transceivers have many values in uS and even nS, way faster than even fancy chips back then. They're plain old silicon and work fine on even crappy circuit boards, and cost nothing. It's quite amazing.

Reply to
Cydrome Leader

Hmm. What about equivalent, full sized components?

Take the example the 74LS00 nand gate, TI has a schematic

formatting link

Are there equivalent transistors and diodes used in that schematic I can run out and buy, to make an as close to 100% electrically compatible, drop in replacement?

I'm not even sure what the first transitor is with A and B inputs is even called. What is it?

Reply to
Cydrome Leader

I'll re-open the thing soon and post more data. The transformer itself is really small, maybe rated 15VA so the regulation and efficiency will be pretty terrible to start with. I think I was measuring 7.5 VAC out, with the load of a single tube, this seemed wrong, but I don't work with tubes. Will have to snip the leads off to really test further.

Reply to
Cydrome Leader

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.