On a sunny day (Mon, 30 Oct 2023 09:04:03 -0700) it happened John Larkin <jl@997PotHill.com> wrote in <egkvjih16kc0bu4crm5vk6g1cu32s1rge2@4ax.com>:>The HP5370 time-interval counter had/has 20 ps picosecond resolution >and about a 30 ps RMS noise floor. The new Keysight equivalent is, as >I recall, about twice as good on jitter. After 40 years! > >The 5370 is astounding. It does amazing measurements with an MC6800 >CPU, less compute power than a toaster has nowadays. The user >interface is wonderful too. > >The Keysight 53200 will not display an N-sample running jitter >measurement! The 5370 did that.I just keep wondering, if I do light speed divided by time I get 300,000,000 * 10^-12 = 0.000300 meter .3 mm in vacuum as distance the signal travels in a pico second, a lot less in cables, even if you bended a cable it would vary significally, Temperature changes changing cable length would create pico second delay changes too. How did you measure that 1 ps, seems just sales crap to me? Video digitizers have jitter depending on clock frequency.. I have worked with those. No wonder they cannot get fussion or was it fusion to work. ;-) Seems you need a 1 THz reference ? and likely locked to some 10 MHz atomic Rubidium thing? even than 10^12 / 10^7 = 10^5 increase in jitter and drift of the rub-it-in-dium.. Drift over 1000 seconds???
Tww
Started by ●October 24, 2023
Reply by ●October 31, 20232023-10-31
Reply by ●October 31, 20232023-10-31
On 31/10/2023 4:01 am, Jan Panteltje wrote:> On a sunny day (Mon, 30 Oct 2023 09:04:03 -0700) it happened John Larkin > <jl@997PotHill.com> wrote in <egkvjih16kc0bu4crm5vk6g1cu32s1rge2@4ax.com>: > >> The HP5370 time-interval counter had/has 20 ps picosecond resolution >> and about a 30 ps RMS noise floor. The new Keysight equivalent is, as >> I recall, about twice as good on jitter. After 40 years! >> >> The 5370 is astounding. It does amazing measurements with an MC6800 >> CPU, less compute power than a toaster has nowadays. The user >> interface is wonderful too. >> >> The Keysight 53200 will not display an N-sample running jitter >> measurement! The 5370 did that. > > I just keep wondering, if I do light speed divided by time I get > 300,000,000 * 10^-12 = 0.000300 meter > .3 mm in vacuum as distance the signal travels in a pico second, a lot less in cables, > even if you bended a cable it would vary significally, > Temperature changes changing cable length would create pico second delay changes too. > > How did you measure that 1 ps, seems just sales crap to me? > Video digitizers have jitter depending on clock frequency.. > I have worked with those. > > No wonder they cannot get fussion or was it fusion to work. > ;-) > > Seems you need a 1 THz reference ? and likely locked to some 10 MHz atomic Rubidium thing? > even than 10^12 / 10^7 = 10^5 increase in jitter and drift of the rub-it-in-dium.. > Drift over 1000 seconds??? > > >This HP Journal is great reading; <https://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1978-08.pdf> piglet
Reply by ●October 31, 20232023-10-31
On Tuesday, October 31, 2023 at 6:51:41 PM UTC+11, piglet wrote:> On 31/10/2023 4:01 am, Jan Panteltje wrote: > > On a sunny day (Mon, 30 Oct 2023 09:04:03 -0700) it happened John Larkin > > <j...@997PotHill.com> wrote in <egkvjih16kc0bu4cr...@4ax.com>: > > > >> The HP5370 time-interval counter had/has 20 ps picosecond resolution > >> and about a 30 ps RMS noise floor. The new Keysight equivalent is, as > >> I recall, about twice as good on jitter. After 40 years! > >> > >> The 5370 is astounding. It does amazing measurements with an MC6800 > >> CPU, less compute power than a toaster has nowadays. The user > >> interface is wonderful too. > >> > >> The Keysight 53200 will not display an N-sample running jitter > >> measurement! The 5370 did that. > > > > I just keep wondering, if I do light speed divided by time I get > > 300,000,000 * 10^-12 = 0.000300 meter > > .3 mm in vacuum as distance the signal travels in a pico second, a lot less in cables, > > even if you bended a cable it would vary significally, > > Temperature changes changing cable length would create pico second delay changes too. > > > > How did you measure that 1 ps, seems just sales crap to me? > > Video digitizers have jitter depending on clock frequency.. > > I have worked with those. > > > > No wonder they cannot get fussion or was it fusion to work. > > ;-) > > > > Seems you need a 1 THz reference ? and likely locked to some 10 MHz atomic Rubidium thing? > > even than 10^12 / 10^7 = 10^5 increase in jitter and drift of the rub-it-in-dium.. > > Drift over 1000 seconds??? > > > > > > > This HP Journal is great reading; > > <https://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1978-08.pdf>But it's from 1978. When we tackled much the same problem at Cambridge Instruments in 1988 we had more options. What the HP article doesn't seem to mention is the minimum delay available. Our system couldn't produce an output edge earlier than about 40nsec after the input edge - it took that long to work out where the input edge was vis-a-vis our continuously generated 800MHz clocked edges and subtract that that offset from the delay from the appropriate clock edge to the output edge. It was bit more complicated but offered slightly better performance. The main driver was the boss's insistance on being able to offer 10psec granularity - which was bit silly when positioning a 500psec wide pulse, but the boss sold the machine himself. and knew what sales spiels worked. -- Bill Sloman, Sydney
Reply by ●October 31, 20232023-10-31
Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:>> How do you handle ic's such as tssop, soic, tsop, etc. With tweezers, >> there is a risk of mangling the leads. The vacuum approach does not >> touch the leads. > > Never recall having a problem with that. I normally hold them by the > ends, where there aren't any leads. Of course those parts are big > enough that you can just place them with the tweezers and then press > them down on the board (usually with the closed jaws) while soldering. > > I still lose parts, of course, but not nearly as many as I used to. > > Cheers > > Phil HobbsTweezers don't help with Quad Flat Package (QFP). Vacuum pickup handles all packages, needs only one hand, and loses fewer parts. -- MRM
Reply by ●October 31, 20232023-10-31
On Mon, 30 Oct 2023 21:37:08 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:>On 2023-10-30 21:31, Mike Monett VE3BTI wrote: >> Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote: >> >>> I cordially dislike vacuum pens, though I know some folks like them a >>> lot. I have one, but rarely use it. You have to change tips to go >>> from doing chips to doing small parts, or else your 0402s get sucked >>> right up the hose. A pain. >>> >>> Dumping a couple of parts out of a tape is pretty simple--the issue is >>> holding them in the right position and orientation long enough to get at >>> least one lead soldered down. > >> >> How do you handle ic's such as tssop, soic, tsop, etc. With tweezers, there >> is a risk of mangling the leads. The vacuum approach does not touch the >> leads. > >Never recall having a problem with that. I normally hold them by the >ends, where there aren't any leads. Of course those parts are big >enough that you can just place them with the tweezers and then press >them down on the board (usually with the closed jaws) while soldering. > >I still lose parts, of course, but not nearly as many as I used to. > >Cheers > >Phil HobbsDr B, the guy who saved my retinas, says to get some Stortz surgical tweezers on Ebay. Amazon has similar stuff.
Reply by ●October 31, 20232023-10-31
On 31/10/2023 12:32 pm, Anthony William Sloman wrote:> On Tuesday, October 31, 2023 at 6:51:41 PM UTC+11, piglet wrote: >> On 31/10/2023 4:01 am, Jan Panteltje wrote: >>> On a sunny day (Mon, 30 Oct 2023 09:04:03 -0700) it happened John Larkin >>> <j...@997PotHill.com> wrote in <egkvjih16kc0bu4cr...@4ax.com>: >>> >>>> The HP5370 time-interval counter had/has 20 ps picosecond resolution >>>> and about a 30 ps RMS noise floor. The new Keysight equivalent is, as >>>> I recall, about twice as good on jitter. After 40 years! >>>> >>>> The 5370 is astounding. It does amazing measurements with an MC6800 >>>> CPU, less compute power than a toaster has nowadays. The user >>>> interface is wonderful too. >>>> >>>> The Keysight 53200 will not display an N-sample running jitter >>>> measurement! The 5370 did that. >>> >>> I just keep wondering, if I do light speed divided by time I get >>> 300,000,000 * 10^-12 = 0.000300 meter >>> .3 mm in vacuum as distance the signal travels in a pico second, a lot less in cables, >>> even if you bended a cable it would vary significally, >>> Temperature changes changing cable length would create pico second delay changes too. >>> >>> How did you measure that 1 ps, seems just sales crap to me? >>> Video digitizers have jitter depending on clock frequency.. >>> I have worked with those. >>> >>> No wonder they cannot get fussion or was it fusion to work. >>> ;-) >>> >>> Seems you need a 1 THz reference ? and likely locked to some 10 MHz atomic Rubidium thing? >>> even than 10^12 / 10^7 = 10^5 increase in jitter and drift of the rub-it-in-dium.. >>> Drift over 1000 seconds??? >>> >>> >>> >> This HP Journal is great reading; >> >> <https://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1978-08.pdf> > > But it's from 1978. When we tackled much the same problem at Cambridge Instruments in 1988 we had more options. > > What the HP article doesn't seem to mention is the minimum delay available. Our system couldn't produce an output edge earlier than about 40nsec after the input edge - it took that long to work out where the input edge was vis-a-vis our continuously generated 800MHz clocked edges and subtract that that offset from the delay from the appropriate clock edge to the output edge. > > It was bit more complicated but offered slightly better performance. The main driver was the boss's insistance on being able to offer 10psec granularity - which was bit silly when positioning a 500psec wide pulse, but the boss sold the machine himself. and knew what sales spiels worked. >Yes, but Jan Panteltje was asking how one can get 20ps resolution without needing a 50GHz timebase. piglet
Reply by ●October 31, 20232023-10-31
On 2023-10-31 12:13, Mike Monett VE3BTI wrote:> Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote: > >>> How do you handle ic's such as tssop, soic, tsop, etc. With tweezers, >>> there is a risk of mangling the leads. The vacuum approach does not >>> touch the leads. >> >> Never recall having a problem with that. I normally hold them by the >> ends, where there aren't any leads. Of course those parts are big >> enough that you can just place them with the tweezers and then press >> them down on the board (usually with the closed jaws) while soldering. >> >> I still lose parts, of course, but not nearly as many as I used to.> > Tweezers don't help with Quad Flat Package (QFP). > > Vacuum pickup handles all packages, needs only one hand, and loses fewer > parts. >Sure, if you like the vacuum approach, go for it--you're far from alone in that. I've used larger ones for wafer handling, and they're great. We try not to use QFNs because of iterated problems with soldering.(JLCPCB is great, but for some reason gives us problems with de-wets on QFN packages.) QFNs are also unreliable in vibration environments, which we encounter a fair amount. 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 ●October 31, 20232023-10-31
piglet <erichpwagner@hotmail.com> wrote:> On 31/10/2023 12:32 pm, Anthony William Sloman wrote: >> On Tuesday, October 31, 2023 at 6:51:41 PM UTC+11, piglet wrote: >>> On 31/10/2023 4:01 am, Jan Panteltje wrote: >>>> On a sunny day (Mon, 30 Oct 2023 09:04:03 -0700) it happened John Larkin >>>> <j...@997PotHill.com> wrote in <egkvjih16kc0bu4cr...@4ax.com>: >>>> >>>>> The HP5370 time-interval counter had/has 20 ps picosecond resolution >>>>> and about a 30 ps RMS noise floor. The new Keysight equivalent is, as >>>>> I recall, about twice as good on jitter. After 40 years! >>>>> >>>>> The 5370 is astounding. It does amazing measurements with an MC6800 >>>>> CPU, less compute power than a toaster has nowadays. The user >>>>> interface is wonderful too. >>>>> >>>>> The Keysight 53200 will not display an N-sample running jitter >>>>> measurement! The 5370 did that. >>>> >>>> I just keep wondering, if I do light speed divided by time I get >>>> 300,000,000 * 10^-12 = 0.000300 meter >>>> .3 mm in vacuum as distance the signal travels in a pico second, a lot less in cables, >>>> even if you bended a cable it would vary significally, >>>> Temperature changes changing cable length would create pico second delay changes too. >>>> >>>> How did you measure that 1 ps, seems just sales crap to me? >>>> Video digitizers have jitter depending on clock frequency.. >>>> I have worked with those. >>>> >>>> No wonder they cannot get fussion or was it fusion to work. >>>> ;-) >>>> >>>> Seems you need a 1 THz reference ? and likely locked to some 10 MHz >>>> atomic Rubidium thing? >>>> even than 10^12 / 10^7 = 10^5 increase in jitter and drift of the rub-it-in-dium.. >>>> Drift over 1000 seconds??? >>>> >>>> >>>> >>> This HP Journal is great reading; >>> >>> <https://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1978-08.pdf> >> >> But it's from 1978. When we tackled much the same problem at Cambridge >> Instruments in 1988 we had more options. >> >> What the HP article doesn't seem to mention is the minimum delay >> available. Our system couldn't produce an output edge earlier than about >> 40nsec after the input edge - it took that long to work out where the >> input edge was vis-a-vis our continuously generated 800MHz clocked >> edges and subtract that that offset from the delay from the appropriate >> clock edge to the output edge. >> >> It was bit more complicated but offered slightly better performance. The >> main driver was the boss's insistance on being able to offer 10psec >> granularity - which was bit silly when positioning a 500psec wide pulse, >> but the boss sold the machine himself. and knew what sales spiels worked. >> > > Yes, but Jan Panteltje was asking how one can get 20ps resolution > without needing a 50GHz timebase. > > piglet > >By not limiting oneself to simple 1/0 clocked quantization. Since the starting phase of the oscillator is known, ideally you get the start time to 1/(2^N * 12) radians rms with one B-bit sample. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics
Reply by ●October 31, 20232023-10-31
Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:> We try not to use QFNs because of iterated problems with > soldering.(JLCPCB is great, but for some reason gives us problems with > de-wets on QFN packages.) QFNs are also unreliable in vibration > environments, which we encounter a fair amount. > > Cheers > > Phil HobbsGood info. Thanks Why is vibration a problem? Seems it has more legs to attach to the pcb. -- MRM
Reply by ●October 31, 20232023-10-31
Mike Monett VE3BTI <spamme@not.com> wrote:> Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote: > >> We try not to use QFNs because of iterated problems with >> soldering.(JLCPCB is great, but for some reason gives us problems with >> de-wets on QFN packages.) QFNs are also unreliable in vibration >> environments, which we encounter a fair amount. >> >> Cheers >> >> Phil Hobbs > > Good info. Thanks > > Why is vibration a problem? Seems it has more legs to attach to the pcb. > >I believe that it’s that the flexible leads relieve the stress on the solder joints, extending their fatigue life. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics
