jlarkin@highlandsniptechnology.com wrote:> On Tue, 19 May 2020 20:36:48 -0700 (PDT), Michael Terrell > <terrell.michael.a@gmail.com> wrote: > >>On Tuesday, May 19, 2020 at 10:08:09 AM UTC-4, John Larkin wrote: >>> >>> On Tue, 19 May 2020 00:18:10 -0700 (PDT), Michael Terrell wrote: >>> >>> >On Tuesday, May 19, 2020 at 2:46:25 AM UTC-4, Cydrome Leader wrote: wrote: >>> >> >>> >> > On Monday, May 18, 2020 at 7:41:34 PM UTC-4, John Larkin wrote: >>> >> >> ST makes a nice little LDO, super-low dropout with an aux Vbias >>> >> >> supply. Saves me from rolling my own with an opamp and a mosfet. >>> >> >> >>> >> >> It's an ST1L08. >>> >> >> >>> >> >> So why is the data sheet file en.DM00123507.pdf ? >>> >> > >>> >> > >>> >> > To track revisions. >>> >> >>> >> Not buying that as good reason to obfuscate document names. Part numbers >>> >> for components are not random numbers only trackable with a database >>> >> lookup. There's no reason document names have to be completely contextless >>> >> to be "trackable". >>> >> >>> >> Let's see how a smart company that's been able to adapt over the years and >>> >> still remain relevant does it. >>> >> >>> >> Texas Instruments has employees with IQs over 43 maintaining their >>> >> website. I clicked around for a random part, the OPA1622 audio amp. >>> >> >>> >> The URL to the datasheet has a clear meaning, even if you don't click on >>> >> it >>> >> >>> >> http://www.ti.com/document-viewer/OPA1622/datasheet >>> >> >>> >> If you want to save the PDF, you get this really cool file name by default >>> >> >>> >> opa1622.pdf >>> >> >>> >> wow, how did they do it? >>> >> >>> >> The first link even has a clear revision history list link in case you >>> >> want to know what happened prior to November 2015. There appear to be a >>> >> document control number at the top of the PDF too. >>> >> >>> >> So yeah, short story is ST is just stupid. >>> > >>> > So, it's better to have all the datasheets by part number? Try keeping multiple versions from one vendor, plus second sources, all with the same file name. Some jellybean parts have dozens of versions of a datasheet. >>> >>> Easy. LM9999_revA.pdf. >>> >>> I rename the meaningless data sheets. >>> >>> > >>> > The head of Engineering at Microdyne was pushing for a new inventory system. His concept was that the part number for a 1K resistor would be 1002. He threw a real hissy fit when I pointed out that we stocked 14 different 1K resistors, and other components had a value of 1002. He wasn't there very long before they fired him. >>> >>> We had a series of very intense/interesting meetings about this. We >>> settled on a telephone-number format, 123-4567 for all parts. We have >>> a document SACRED.TXT that explains all the rules. >>> >>> > >>> > All of our documentation had form numbers, unrelated to the product model numbers. They were filed by model number, but under ISO9001 we had to have traceability for all revisions. One board had 14 versions, and 14 test procedures. Every one ended up with over 100 lines that had to be marked N/A. I wrote a new procedure to cover every version of the board. It had 14 datasheets, one per version with no extra lines, plus a spare to allow for you to add new versions. The first page clearly stated to only use the specified datasheet for the version being tested. It reduced both the test time, and the paperwork to be stored. There are reasons to use other than obvious file naming. I am currently going through thousands of PDF files for test equipment. I am adding the actual document numbers to the file names, not just 'HP3325' since the A and B versions are different, but the manuals were revised multiple times. I have 180 GB of files, over 32,000 of them to sort, rename and >>compare. >>> >Some are duplicates, some are incomplete but I save every Electronics manual I come across. >>> >>> I know of one big company that assigns a 12-digit number to anything. >>> Parts. Products. Drawings. Trucks. Buildings. Employees. >>> >>> I've worked for companies that just assigned the next sequential >>> number to anything. If you know the PCB number, you had to look in the >>> log book to find the schematic. The numbers were totally unrelated. >> >> >>There were still reference to some trucks in the Microdyne inventory, along with all the test equipment. The founders of the company mad the Cal Lab keep their records in the database rather than their own system. It was on a Prime mini computer with about fifty terminals scattered all over the complex. Those trucks were gone, but they had been used to install their C-band equipment in head ends, and businesses before they shut down that product line. > > > In our system, if you know any drawing or product part number, you > automatically know all the related ones. > > We refer to people by their names.Haha. I tried to post data on a company system as my employee ID instead of name. It caused lots of confusion and got some folks bent out of shape.
why do they do this?
Started by ●May 18, 2020
Reply by ●May 21, 20202020-05-21
Reply by ●May 21, 20202020-05-21
Michael Terrell <terrell.michael.a@gmail.com> wrote:> On Wednesday, May 20, 2020 at 12:09:51 PM UTC-4, John Larkin wrote: >> >> In our system, if you know any drawing or product part number, you >> automatically know all the related ones. >> >> We refer to people by their names > > > Some boards and modules were used in multiple product lines, so your > system wouldn't have worked at Microdyne. It was a different market, > requiring different methods. For instance: The 700 and 1620-base > models shared 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, Some 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 > repaired.That's impressive. What was the design life of this equipment in the first place? I get excited if 30 year old test equipment fires up, without an actual fire. I'm even more shocked if CFL bulbs can last a year.> Completed units had custom build list, per the contract and all the > test data for a unit was stored long term. The ISO inspectors spent > most of their time looking at the files, since a record for one unit > could be a half inch thick. That was why I pushed to change the test > procedures to streamline them.It reduced the paperwork by about 25%. > Some were rewritten from scratch, because the designer had the steps > out of order, wasting test time. You had to do the same step several > times before they were updated. A test fixture I redesigned educed > the test time from 7.5 hours to 18 minutes and gave a more repeatable > result. I know that you dislike trimpots, but they were used to > adjust gan in many circuits. Even with 1% resistors and capacitors, > they were often out of spec since 14, 1% components were used per > video filter.So it NASA sitting on faded out dot matrix printouts of test data nobody can read anymore? How do you file away test data from ages ago? The last open frame linear power supplies I got still had the goofy folded up dot matrix "report" stuffed inside of them. I can only imagine the test rack they use for that stuff is so old it has a black crinkle finish and giant bakelite handwheels and some Okidata printer jammed in there.
Reply by ●May 21, 20202020-05-21
John Larkin <jlarkin@highland_atwork_technology.com> wrote:> On Wed, 20 May 2020 14:07:53 -0700 (PDT), Michael Terrell > <terrell.michael.a@gmail.com> wrote: > >>On Wednesday, May 20, 2020 at 12:09:51 PM UTC-4, John Larkin wrote: >>> >>> In our system, if you know any drawing or product part number, you >>> automatically know all the related ones. >>> >>> We refer to people by their names >> >> >> Some boards and modules were used in multiple product lines, so your system wouldn't have worked at Microdyne. It was a different market, requiring different methods. For instance: The 700 and 1620-base models shared 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, Some 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 repaired. > > We reuse boards, and assemblies, in multiple prducts. If there are > different versions, each has its own dash number and associated BOM. > That's standard mil practice.do you have to relabel parts to give them the correct (for the end user) part number? Sometimes there's goofiness with OEMed parts and trying to re-order them. I've gotten parts with say a different brand printed on them.> > In the aircraft business, xxxxx was a drawing and xxxxx-1 was a thing, > and xxxxx-2 was its mirror image thing. Odds and evens were mirrors, > like wings maybe, without requiring two drawings. We don't mirror > parts, so for us -1 and -2 are just assembly versions of some sort.Were there ever "funny" issues with mirror image parts, like the left part ended up with english thread fasteners or something dumb like that?>> Completed units had custom build list, per the contract and all the test data for a unit was stored long term. The ISO inspectors spent most of their time looking at the files, since a record for one unit could be a half inch thick. That was why I pushed to change the test procedures to streamline them.It reduced the paperwork by about 25%. Some were rewritten from scratch, because the designer had the steps out of order, wasting test time. You had to do the same step several times before they were updated. A test fixture I redesigned educed the test time from 7.5 hours to 18 minutes and gave a more repeatable result. I know that you dislike trimpots, but they were used to adjust gan in many circuits. Even with 1% resistors and capacitors, they were often out of spec since 14, 1% components were used per video filter. > > Most of our testing is automated now, and test reports get pushed up > to a server, as both PDF and JSON files. We can easily extract > statistics from the JSON files.How did this work say 20 or 30 years ago?
Reply by ●May 21, 20202020-05-21
Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:> On 2020-05-19 09:50, jlarkin@highlandsniptechnology.com wrote: >> On Tue, 19 May 2020 12:50:44 +0530, Pimpom <nobody@nowhere.com> wrote: >> >>> On 5/19/2020 9:31 AM, jlarkin@highlandsniptechnology.com wrote: >>>> On Mon, 18 May 2020 17:57:06 -0700 (PDT), >>>> bloggs.fredbloggs.fred@gmail.com wrote: >>>> >>>>> On Monday, May 18, 2020 at 7:41:34 PM UTC-4, John Larkin wrote: >>>>>> ST makes a nice little LDO, super-low dropout with an aux Vbias >>>>>> supply. Saves me from rolling my own with an opamp and a mosfet. >>>>>> >>>>>> It's an ST1L08. >>>>>> >>>>>> So why is the data sheet file en.DM00123507.pdf ? >>>>> >>>>> Who cares, it's a crappy regulator. And the lying bastards with their fake dropout specs while conveniently omitting the fact that Vbias must be greater than Vout + 1.5V. >>>> >>>> Lying? It's all over the data sheet. It's how they get the millivolts >>>> of dropout. I do that when I make my own super-LDOs, power an opamp >>>> from some higher voltage and over-drive an nfet follower down to >>>> milliohms of Rds-on. >>>> >>>>> The GND current at no load of 35uA, sucks , as does that showy 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 getting 800mA out of it with any kind voltage headroom without using a liquid nitrogen drip. >>>>> >>>> >>>> I'm dropping a switched 1.8 to 1.5. That's 0.3 volts. Times 800 mA >>>> would be 0.24 watts dissipated. Actually, I don't need that much >>>> current to run a couple DRAM chips. >>>> >>> >>> Not directly comparable to the ST1L08 but the Holtek HT75xx-1 >>> series is nice. Max Vin 30V, 100mA, 2.5uA ground current, 25mV >>> drop-out. 16 different fixed output voltages from 2.1V to 12V >>> with 3% tolerance. As usual with products originating in the >>> East, the datasheet is rather sparse about details, but I've used >>> them and they do what I want. >> >> Is it stable with low ESR caps? We use polymers or ceramics mostly. >> >> We need so many goofy voltages that we usually buy adjustable >> regulators for stock. The board that I'm doing now has a 24-channel >> analog mux to BIST the power supplies, using the dreadful Xilinx >> 1-volt XADC that's inside their FPGAs. Free and worth it. >> >>> >>>> You sure are in a bad mood lately. >>>> >>> I've noticed that lately with some regulars here, including a few >>> who normally exhibit decent manners. >> >> Well, some never show any sign of manners. They are repulsive but >> you've got to feel sorry for them, stuck being around themselves all >> day. >> >> There's a basically perfect -1 correlation between being obnoxious and >> designing electronics. > > Well, now that Jim Thompson is apparently no longer with us. :( > > He was a bit of a statistical outlier.Was that the guy who designed all the 2 digit LM series chips? What happened to him?
Reply by ●May 21, 20202020-05-21
On Thursday, May 21, 2020 at 2:44:49 PM UTC+10, Cydrome Leader wrote:> Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote: > > On 2020-05-19 09:50, jlarkin@highlandsniptechnology.com wrote: > >> On Tue, 19 May 2020 12:50:44 +0530, Pimpom <nobody@nowhere.com> wrote: > >> > >>> On 5/19/2020 9:31 AM, jlarkin@highlandsniptechnology.com wrote: > >>>> On Mon, 18 May 2020 17:57:06 -0700 (PDT), > >>>> bloggs.fredbloggs.fred@gmail.com wrote: > >>>> > >>>>> On Monday, May 18, 2020 at 7:41:34 PM UTC-4, John Larkin wrote: > >>>>>> ST makes a nice little LDO, super-low dropout with an aux Vbias > >>>>>> supply. Saves me from rolling my own with an opamp and a mosfet. > >>>>>> > >>>>>> It's an ST1L08. > >>>>>> > >>>>>> So why is the data sheet file en.DM00123507.pdf ? > >>>>> > >>>>> Who cares, it's a crappy regulator. And the lying bastards with their fake dropout specs while conveniently omitting the fact that Vbias must be greater than Vout + 1.5V. > >>>> > >>>> Lying? It's all over the data sheet. It's how they get the millivolts > >>>> of dropout. I do that when I make my own super-LDOs, power an opamp > >>>> from some higher voltage and over-drive an nfet follower down to > >>>> milliohms of Rds-on. > >>>> > >>>>> The GND current at no load of 35uA, sucks , as does that showy 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 getting 800mA out of it with any kind voltage headroom without using a liquid nitrogen drip. > >>>>> > >>>> > >>>> I'm dropping a switched 1.8 to 1.5. That's 0.3 volts. Times 800 mA > >>>> would be 0.24 watts dissipated. Actually, I don't need that much > >>>> current to run a couple DRAM chips. > >>>> > >>> > >>> Not directly comparable to the ST1L08 but the Holtek HT75xx-1 > >>> series is nice. Max Vin 30V, 100mA, 2.5uA ground current, 25mV > >>> drop-out. 16 different fixed output voltages from 2.1V to 12V > >>> with 3% tolerance. As usual with products originating in the > >>> East, the datasheet is rather sparse about details, but I've used > >>> them and they do what I want. > >> > >> Is it stable with low ESR caps? We use polymers or ceramics mostly. > >> > >> We need so many goofy voltages that we usually buy adjustable > >> regulators for stock. The board that I'm doing now has a 24-channel > >> analog mux to BIST the power supplies, using the dreadful Xilinx > >> 1-volt XADC that's inside their FPGAs. Free and worth it. > >> > >>> > >>>> You sure are in a bad mood lately. > >>>> > >>> I've noticed that lately with some regulars here, including a few > >>> who normally exhibit decent manners. > >> > >> Well, some never show any sign of manners. They are repulsive but > >> you've got to feel sorry for them, stuck being around themselves all > >> day. > >> > >> There's a basically perfect -1 correlation between being obnoxious and > >> designing electronics. > > > > Well, now that Jim Thompson is apparently no longer with us. :( > > > > He was a bit of a statistical outlier. > > Was that the guy who designed all the 2 digit LM series chips? What > happened to him?You are thinking of Bob Widlar, and Wikipedia will tell you all you need to know. He seems to have drunk to much and died relatively young - aged 53 - in 1991. https://en.wikipedia.org/wiki/Bob_Widlar Jim Thompson designed a bunch of chip for Motorola early on, and I had the dubious pleasure of getting a few of them to work. It wasn't easy, and every last one of them got replaced by something that had been designed to do much the same job either a bit better or bit more easily. The MC14046 comes to mind. It's Jim's TTL MC4024 and MC4044 combined and executed in CMOS by somebody else. It's still a conceptual mess but works well enough to have lasted. Jim's multiplier chips were much the same as the multiplier chips Barry Gilbert put together for Analog Devices, but every last one of Barry's parts was easier to use than Jim's, and Barry kept on tinkering with them to make the better - faster or more accurate - for some years. I met Barry once when I was working in the UK and he was going around talking to customers, essentially about new products that we weren't going to be buying. Jim was an adequate journey man designer, and Bob and Barry were master craftsmen. Bob seems to nave been a real virtuoso. -- Bill Sloman, Sydney
Reply by ●May 21, 20202020-05-21
On 2020-05-21 00:44, Cydrome Leader wrote:> Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote: >> On 2020-05-19 09:50, jlarkin@highlandsniptechnology.com wrote: >>> On Tue, 19 May 2020 12:50:44 +0530, Pimpom <nobody@nowhere.com> wrote: >>> >>>> On 5/19/2020 9:31 AM, jlarkin@highlandsniptechnology.com wrote: >>>>> On Mon, 18 May 2020 17:57:06 -0700 (PDT), >>>>> bloggs.fredbloggs.fred@gmail.com wrote: >>>>> >>>>>> On Monday, May 18, 2020 at 7:41:34 PM UTC-4, John Larkin wrote: >>>>>>> ST makes a nice little LDO, super-low dropout with an aux Vbias >>>>>>> supply. Saves me from rolling my own with an opamp and a mosfet. >>>>>>> >>>>>>> It's an ST1L08. >>>>>>> >>>>>>> So why is the data sheet file en.DM00123507.pdf ? >>>>>> >>>>>> Who cares, it's a crappy regulator. And the lying bastards with their fake dropout specs while conveniently omitting the fact that Vbias must be greater than Vout + 1.5V. >>>>> >>>>> Lying? It's all over the data sheet. It's how they get the millivolts >>>>> of dropout. I do that when I make my own super-LDOs, power an opamp >>>>> from some higher voltage and over-drive an nfet follower down to >>>>> milliohms of Rds-on. >>>>> >>>>>> The GND current at no load of 35uA, sucks , as does that showy 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 getting 800mA out of it with any kind voltage headroom without using a liquid nitrogen drip. >>>>>> >>>>> >>>>> I'm dropping a switched 1.8 to 1.5. That's 0.3 volts. Times 800 mA >>>>> would be 0.24 watts dissipated. Actually, I don't need that much >>>>> current to run a couple DRAM chips. >>>>> >>>> >>>> Not directly comparable to the ST1L08 but the Holtek HT75xx-1 >>>> series is nice. Max Vin 30V, 100mA, 2.5uA ground current, 25mV >>>> drop-out. 16 different fixed output voltages from 2.1V to 12V >>>> with 3% tolerance. As usual with products originating in the >>>> East, the datasheet is rather sparse about details, but I've used >>>> them and they do what I want. >>> >>> Is it stable with low ESR caps? We use polymers or ceramics mostly. >>> >>> We need so many goofy voltages that we usually buy adjustable >>> regulators for stock. The board that I'm doing now has a 24-channel >>> analog mux to BIST the power supplies, using the dreadful Xilinx >>> 1-volt XADC that's inside their FPGAs. Free and worth it. >>> >>>> >>>>> You sure are in a bad mood lately. >>>>> >>>> I've noticed that lately with some regulars here, including a few >>>> who normally exhibit decent manners. >>> >>> Well, some never show any sign of manners. They are repulsive but >>> you've got to feel sorry for them, stuck being around themselves all >>> day. >>> >>> There's a basically perfect -1 correlation between being obnoxious and >>> designing electronics. >> >> Well, now that Jim Thompson is apparently no longer with us. :( >> >> He was a bit of a statistical outlier. > > Was that the guy who designed all the 2 digit LM series chips? What > happened to him? >He was more in the Motorola orbit back then, IIRC, but he was certainly the most accomplished IC designer in the group. He disappeared a couple of years ago, after having been diagnosed with early-stage pancreatic cancer during tests for something else. He said that he thought he had beaten it, but then just disappeared from the group. His website subsequently disappeared as well. Various folks have looked for an obituary, but nothing turned up. (His full name was James Elbert Thompson, and he lived most recently in Queen Creek AZ.) He was a bit of a chingada but designed a lot of interesting electronics and really helped keep the local leftists in check. 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 http://electrooptical.net http://hobbs-eo.com
Reply by ●May 21, 20202020-05-21
On Thursday, May 21, 2020 at 4:26:50 PM UTC+10, Phil Hobbs wrote:> On 2020-05-21 00:44, Cydrome Leader wrote: > > Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote: > >> On 2020-05-19 09:50, jlarkin@highlandsniptechnology.com wrote: > >>> On Tue, 19 May 2020 12:50:44 +0530, Pimpom <nobody@nowhere.com> wrote: > >>> > >>>> On 5/19/2020 9:31 AM, jlarkin@highlandsniptechnology.com wrote: > >>>>> On Mon, 18 May 2020 17:57:06 -0700 (PDT), > >>>>> bloggs.fredbloggs.fred@gmail.com wrote: > >>>>> > >>>>>> On Monday, May 18, 2020 at 7:41:34 PM UTC-4, John Larkin wrote:<snip>> >> Well, now that Jim Thompson is apparently no longer with us. :( > >> > >> He was a bit of a statistical outlier. > > > > Was that the guy who designed all the 2 digit LM series chips? What > > happened to him? > > > > He was more in the Motorola orbit back then, IIRC, but he was certainly > the most accomplished IC designer in the group.The only one who did it for a livng.> He disappeared a couple > of years ago, after having been diagnosed with early-stage pancreatic > cancer during tests for something else. He said that he thought he had > beaten it, but then just disappeared from the group. His website > subsequently disappeared as well. > > Various folks have looked for an obituary, but nothing turned up. (His > full name was James Elbert Thompson, and he lived most recently in Queen > Creek AZ.) He was a bit of a chingada but designed a lot of interesting > electronics and really helped keep the local leftists in check.I didn't much like it when he found out my postal address in the Netherland's and threatened to send a couple of hit men to keep me in check. His apologists claimed that he was just joking, but he persisted in his claim to have to reported me to the FBI for dangerously anti-American attitudes. I'm certainly more left of centre than Jim was - you'd have to rabid to get further to the right - and I'm almost certainly to the left of Phil Hobbs. This doesn't make me a leftist outside of the US. where they are plenty of peole who are much further to the left. As for keeping me - or anybody else - "in check", Phil is fooling himself. In chess this means putting enough pressure on the king to force some kind of counter move to get it out of check, and I've not experienced anything like that. I once - unwisely - took up John Larkin when he claimed that he couldn't find enough competent help, and suggested that I might be able to do some remote consulting for him. John Larkin then behaved rather badly by passing on the e-mail to Jim, who proceeded to crow - here - about persuading John not to follow up. He didn't have anything to crow about. John's passing on the e-mail to Jim would have been quite enough to put me right off the idea, anyway. You have to trust the people you work with to behave more or less properly, and John clearly couldn't be trusted. -- Bill Sloman, Sydney
Reply by ●May 21, 20202020-05-21
On Wednesday, May 20, 2020 at 6:17:59 PM UTC-4, John Larkin wrote:> On Wed, 20 May 2020 14:07:53 -0700 (PDT), Michael Terrell wrote: > > >On Wednesday, May 20, 2020 at 12:09:51 PM UTC-4, John Larkin wrote: > >> > >> In our system, if you know any drawing or product part number, you > >> automatically know all the related ones. > >> > >> We refer to people by their names > > > > > > Some boards and modules were used in multiple product lines, so your system wouldn't have worked at Microdyne. It was a different market, requiring different methods. For instance: The 700 and 1620-base models shared 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, Some 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 repaired. > > We reuse boards, and assemblies, in multiple prducts. If there are > different versions, each has its own dash number and associated BOM. > That's standard mil practice. > > In the aircraft business, xxxxx was a drawing and xxxxx-1 was a thing, > and xxxxx-2 was its mirror image thing. Odds and evens were mirrors, > like wings maybe, without requiring two drawings. We don't mirror > parts, so for us -1 and -2 are just assembly versions of some sort. > > > > > Completed units had custom build list, per the contract and all the test data for a unit was stored long term. The ISO inspectors spent most of their time looking at the files, since a record for one unit could be a half inch thick. That was why I pushed to change the test procedures to streamline them.It reduced the paperwork by about 25%. Some were rewritten from scratch, because the designer had the steps out of order, wasting test time. You had to do the same step several times before they were updated. A test fixture I redesigned educed the test time from 7.5 hours to 18 minutes and gave a more repeatable result. I know that you dislike trimpots, but they were used to adjust gan in many circuits. Even with 1% resistors and capacitors, they were often out of spec since 14, 1% components were used per video filter. > > Most of our testing is automated now, and test reports get pushed up > to a server, as both PDF and JSON files. We can easily extract > statistics from the JSON files.We were just starting to automate testing, but that was almost 19 years ago. Documentation and board or model number matched the blank board, plus the BOM and test procedure for different versions but with over 20 modules in a unit, plus a separate plug in tuner for older models (depending on which microwave band) still complicated issues. Add that the company had been around since 1968 which was before most small businesses used computers, the system was deeply embedded in the daily operations. They had changed their part numbering system, once and it was a nightmare. Every BOM had to be rewritten, on a typewriter, then all old copies destroyed. On top of that, when they decided to close their original Rockville Maryland plant, the employees shredded every document in the vault, plus all working copies, leaving the engineers the task of reverse engineering the last complete units and to recreate the mountain of data.
Reply by ●May 21, 20202020-05-21
On Thursday, May 21, 2020 at 12:32:42 AM UTC-4, Cydrome Leader wrote:> > So it NASA sitting on faded out dot matrix printouts of test data nobody > can read anymore? How do you file away test data from ages ago? > > The last open frame linear power supplies I got still had the goofy folded > up dot matrix "report" stuffed inside of them. I can only imagine the test > rack they use for that stuff is so old it has a black crinkle finish and > giant bakelite handwheels and some Okidata printer jammed in there.The files at Microdyne were hand written, in black ink on photocopies using toner so they don't fade. They were stored in a locked file room, with no lights when no one was in the room, and they filled long rows of file cabinets.
Reply by ●May 21, 20202020-05-21
On Thu, 21 May 2020 01:55:31 -0700 (PDT), Michael Terrell <terrell.michael.a@gmail.com> wrote:>On Wednesday, May 20, 2020 at 6:17:59 PM UTC-4, John Larkin wrote: >> On Wed, 20 May 2020 14:07:53 -0700 (PDT), Michael Terrell wrote: >> >> >On Wednesday, May 20, 2020 at 12:09:51 PM UTC-4, John Larkin wrote: >> >> >> >> In our system, if you know any drawing or product part number, you >> >> automatically know all the related ones. >> >> >> >> We refer to people by their names >> > >> > >> > Some boards and modules were used in multiple product lines, so your system wouldn't have worked at Microdyne. It was a different market, requiring different methods. For instance: The 700 and 1620-base models shared 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, Some 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 repaired. >> >> We reuse boards, and assemblies, in multiple prducts. If there are >> different versions, each has its own dash number and associated BOM. >> That's standard mil practice. >> >> In the aircraft business, xxxxx was a drawing and xxxxx-1 was a thing, >> and xxxxx-2 was its mirror image thing. Odds and evens were mirrors, >> like wings maybe, without requiring two drawings. We don't mirror >> parts, so for us -1 and -2 are just assembly versions of some sort. >> >> > >> > Completed units had custom build list, per the contract and all the test data for a unit was stored long term. The ISO inspectors spent most of their time looking at the files, since a record for one unit could be a half inch thick. That was why I pushed to change the test procedures to streamline them.It reduced the paperwork by about 25%. Some were rewritten from scratch, because the designer had the steps out of order, wasting test time. You had to do the same step several times before they were updated. A test fixture I redesigned educed the test time from 7.5 hours to 18 minutes and gave a more repeatable result. I know that you dislike trimpots, but they were used to adjust gan in many circuits. Even with 1% resistors and capacitors, they were often out of spec since 14, 1% components were used per video filter. >> >> Most of our testing is automated now, and test reports get pushed up >> to a server, as both PDF and JSON files. We can easily extract >> statistics from the JSON files. > >We were just starting to automate testing, but that was almost 19 years ago. Documentation and board or model number matched the blank board, plus the BOM and test procedure for different versions but with over 20 modules in a unit, plus a separate plug in tuner for older models (depending on which microwave band) still complicated issues. Add that the company had been around since 1968 which was before most small businesses used computers, the system was deeply embedded in the daily operations. They had changed their part numbering system, once and it was a nightmare. Every BOM had to be rewritten, on a typewriter, then all old copies destroyed. > > On top of that, when they decided to close their original Rockville Maryland plant, the employees shredded every document in the vault, plus all working copies, leaving the engineers the task of reverse engineering the last complete units and to recreate the mountain of data.Some companies have a "configuration control" department, whose job is to control all the documents and their relationship, and make sure the right stuff gets manufactured and remembered. All documents have to be submitted to them. Makes sense; keeping this stuff organized is non-trivial, especially when mistakes can kill people. -- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
