low MHz connector design consultant
in [Design]
|
From: JosephKK on 2 May 2010 00:18 On Fri, 30 Apr 2010 06:54:29 -0500, "mook johnson" <mook(a)mook.net> wrote: > >"John Larkin" <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote in message >news:ql9kt5tg7s7e5q7pb460gcdig4r00jjmrj(a)4ax.com... >> On Thu, 29 Apr 2010 18:30:41 -0500, "mook johnson" <mook(a)mook.net> >> wrote: >> >>>Gents, >>> >>>I looking for a consultant that can assist in the design of a custom >>>military style connector with controlled characteristic impedance and >>>insertion loss between two terminals in the 1MHz - 20MHz frequency range. >>> >>>This connector is special because is the application so an off the shelf >>>component will not work. >>> >>>Any leads where I can start looking for such a consultant? >>> >>>thanks >>> >>> >> >> There are so many military connectors, including ones for wild >> environments, hermetic, etc, some standard part might work. >> >> 20 MHz isn't very demanding. Most any mil connector will be "matched" >> to any impedance at 20 MHz. Wavelength is 15 meters! >> >> John > > >The connector I need goes into environments that far exceed military >applications but the physical concept is similar. We have an in-house >connector company that makes these connectors for us but they have never had >to deal with never greater than >200KHz signals before. We tried using our >regular connectors for this application and the impedance mismatch and >insertion loss were extreme at 5 - 20MHz. The Zo mismatched by 50% and was >not stable above 1MHz and had several resonant modes. Insertion loss >something on the order is 6dB/connector and we'll have a couple dozen in >series on this line. The number of series connection is the is the rub and >it is unavoidable, non negotiable. > >There are some commercial plastic connectors that tested very well (just to >validate our test setup) with good impedance match/stability (+/- 5%) and >low insertion loss (.1dB/connector) but they won't take the environment. > >I'm looking for a consultant that can provide either of the following > >1) model a connector that is already designed but not made (basically review >the in-house company proposed design) and simulate the high frequency >response of the design. > >2) Give direction for the connector design based on constraints of material >choices, physical size and geometry to meet the desired electrical signal >characteristics while withstanding the environmental conditions. > >The cut and try approach based on simple equations has a long cycle time. >I'm looking to improve my chances of getting it right the first time. > > What is the problem with the environment? Temperature? Corrosives? Vacuum? Vibration / shock? other? Combination?
From: Paul Keinanen on 2 May 2010 02:19 On Sat, 01 May 2010 15:16:31 -0700, John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>>> >> On 30 abr, 13:54, "mook johnson" <m...(a)mook.net> wrote: >>>> >> > The connector I need goes into environments that far exceed >>>> >> > military applications but the physical concept is similar. �We >>>> >> > have an in-house connector company that makes these connectors >>>> >> > for us but they have never had to deal with never greater than >>>> >> > >200KHz signals before. �We tried using our regular connectors >>>> >> > for this application and the impedance mismatch and insertion >>>> >> > loss were extreme at 5 - 20MHz. �The Zo mismatched by 50% and >>>> >> > was not stable above 1MHz and had several resonant modes. >>>> >> > Insertion loss something on the order is 6dB/connector and we'll >>>> >> > have a couple dozen in series on this line. �The number of ----------------^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ >>>> >> > series connection is the is the rub and it is unavoidable, non >>>> >> > negotiable. How did you measure the mismatch and losses. Did you really measure these for a single connector (with a matched load on the other side) or did you measured it with a string of connectors on a bus ? >How he can get 6 dB of loss at low MHz is a mystery to me. While the OP did not specify so, this sounds very much like the problems with RS-485 multidrop networks on a single trunk line (such as Profibus-DP). At speeds up to 20 Mbit/s, it is essential that the branches (drop line/stubs) to the slave stations from the trunk line is only a few millimeters to avoid any extra capacitive loading. On a typical Profibus-DP slave, the RS-485 transceiver is very close to the connector, thus the stub length is the connector internal length plus a 2-5 mm, provided that the incoming trunk wire and the outgoing wire to the next slave is soldered directly to the same connector pin. This is quite acceptable at 1.5 Mbit/s, since connectors and cables of varying quality will work OK. However, the stub stray capacitance will cause problems at 12 Mbit/s and series inductances are used in the plug on the trunk line (wire from previous node, series inductance, connector terminal solder point, series inductance, wire to the next slave). With several long uncompensated branch lines hanging from the trunk line, the measurements from one end of the trunk line, would behave just as the OP described. Regarding connectors for harsh environment, look for instance at some variants of the M12 (IEC 60947-5-2) connector (up to IP68), which is used e.g. in Profibus-DP or CAN networks in dirty industrial environments. I do not quite understand the point of making 1/2, 1/4 etc. unit load RS-485 transceivers that will allow 64, 128 slaves on the single line. At least at higher speeds, the connector losses and reliability issues with several dozens of connectors can easily make such systems quite unreliable. It is often better to divide the systems in several segments and use some RS-485 repeaters to drive the segments, even if a single master is used.
From: Baron on 2 May 2010 10:05 tm Inscribed thus: > > "Baron" <baron.nospam(a)linuxmaniac.nospam.net> wrote in message > news:hri41l$ple$2(a)news.eternal-september.org... >> tm Inscribed thus: >> >>> >>> "Baron" <baron.nospam(a)linuxmaniac.nospam.net> wrote in message >>> news:hri2ce$910$1(a)news.eternal-september.org... >>>> mpm Inscribed thus: >>>> >>>>> On Apr 30, 10:58 pm, "k...(a)att.bizzzzzzzzzzzz" >>>> >>>> For co-axial cables its a function of the ratio of the diameters of >>>> the conductors modified by the dielectric constant of the >>>> insulator, or for twin conductor the spacing between them and the >>>> dielectric constant of the insulator. >>>> ie (impedance = (138 / e^(1/2)) * log (D/d)) >>>> >>> >>> >>> It also assumes a match at both ends of the line. >>> >>> >>> Tom >> >> ??? > > Sure. Take a piece of coax, any impedance, 1/4 wavelength long, and > short one end. What do you see at the other end? > > Now match both ends. What impedance do you see at any point of the > line? > > > > > Tom That has absolutely nothing to do with characteristic impedance ! -- Best Regards: Baron.
From: Baron on 2 May 2010 10:08 John Larkin Inscribed thus: > On Sat, 01 May 2010 21:19:52 +0100, Baron > <baron.nospam(a)linuxmaniac.nospam.net> wrote: > >>mpm Inscribed thus: >> >>> On Apr 30, 10:58 pm, "k...(a)att.bizzzzzzzzzzzz" >>> <k...(a)att.bizzzzzzzzzzzz> wrote: >>>> On Fri, 30 Apr 2010 18:39:21 -0700 (PDT), mpm <mpmill...(a)aol.com> >>>> wrote: >>>> >On Apr 30, 4:31 pm, Wimpie <wimabc...(a)tetech.nl> wrote: >>>> >> On 30 abr, 13:54, "mook johnson" <m...(a)mook.net> wrote: >>>> >>>> >> > "John Larkin" <jjlar...(a)highNOTlandTHIStechnologyPART.com> >>>> >> > wrote in message >>>> >>>> >> >news:ql9kt5tg7s7e5q7pb460gcdig4r00jjmrj(a)4ax.com... >>>> >>>> >> > > On Thu, 29 Apr 2010 18:30:41 -0500, "mook johnson" >>>> >> > > <m...(a)mook.net> wrote: >>>> >>>> >> > >>Gents, >>>> >>>> >> > >>I looking for a consultant that can assist in the design of >>>> >> > >>a custom military style connector with controlled >>>> >> > >>characteristic impedance and insertion loss between two >>>> >> > >>terminals in the 1MHz - 20MHz frequency range. >>>> >>>> >> > >>This connector is special because is the application so an >>>> >> > >>off the shelf component will not work. >>>> >>>> >> > >>Any leads where I can start looking for such a consultant? >>>> >>>> >> > >>thanks >>>> >>>> >> > > There are so many military connectors, including ones for >>>> >> > > wild environments, hermetic, etc, some standard part might >>>> >> > > work. >>>> >>>> >> > > 20 MHz isn't very demanding. Most any mil connector will be >>>> >> > > "matched" to any impedance at 20 MHz. Wavelength is 15 >>>> >> > > meters! >>>> >>>> >> > > John >>>> >>>> >> > The connector I need goes into environments that far exceed >>>> >> > military applications but the physical concept is similar. We >>>> >> > have an in-house connector company that makes these connectors >>>> >> > for us but they have never had to deal with never greater than >>>> >> > >200KHz signals before. We tried using our regular connectors >>>> >> > for this application and the impedance mismatch and insertion >>>> >> > loss were extreme at 5 - 20MHz. The Zo mismatched by 50% and >>>> >> > was not stable above 1MHz and had several resonant modes. >>>> >> > Insertion loss something on the order is 6dB/connector and >>>> >> > we'll have a couple dozen in series on this line. The number >>>> >> > of series connection is the is the rub and it is unavoidable, >>>> >> > non negotiable. >>>> >>>> >> > There are some commercial plastic connectors that tested very >>>> >> > well (just to validate our test setup) with good impedance >>>> >> > match/stability (+/- 5%) and low insertion loss >>>> >> > (.1dB/connector) but they won't take the environment. >>>> >>>> >> > I'm looking for a consultant that can provide either of the >>>> >> > following >>>> >>>> >> > 1) model a connector that is already designed but not made >>>> >> > (basically review the in-house company proposed design) and >>>> >> > simulate the high frequency response of the design. >>>> >>>> >> > 2) Give direction for the connector design based on >>>> >> > constraints of material choices, physical size and geometry to >>>> >> > meet the desired electrical signal characteristics while >>>> >> > withstanding the environmental conditions. >>>> >>>> >> > The cut and try approach based on simple equations has a long >>>> >> > cycle time. I'm looking to improve my chances of getting it >>>> >> > right the first time. >>>> >>>> >> Hello Mook, >>>> >>>> >> When you have already a company that knows mechanical design and >>>> >> reliability issues, you only need the EM-field guy. When you >>>> >> can keep the mechanical guys and the EM-field guy in one room, >>>> >> this should converge to a solution fast. >>>> >>>> >> One familiar with the concept of characteristic impedance, >>>> >> complex propagation constant (contains both complex epsilon and >>>> >> permeability) and know how to measure the material constants can >>>> >> do the job. >>>> >>>> >> What about: required Return Loss, characteristic impedance, >>>> >> Insertion loss, propagation delay, size, cross section, >>>> >> (coaxial, square, symmetrical, etc), peak and average power, >>>> >> etc? >>>> >>>> >> Kind regards, >>>> >>>> >> Wim >>>> >> PA3DJSwww.tetech.nl >>>> >> when you delete abc first, PM will reach me- Hide quoted text - >>>> >>>> >> - Show quoted text - >>>> >>>> >I have a question for you: >>>> >>>> >Do you believe that the characteristic impedance is dependent on >>>> >the length of the coax? >>>> >(For the ratio of transverse electric field to transverse magnetic >>>> >field launched on a transmission line of infinite length.) >>>> >>>> "infinite length" <> "independent of length"- Hide quoted text - >>>> >>>> - Show quoted text - >>> >>> Not sure I understand your cryptic post. >>> >>> That is how characteristic impedance is classically defined, and I'm >>> asking him if he believes that characteristic impedance is dependent >>> on length? >> >>For co-axial cables its a function of the ratio of the diameters of >>the conductors modified by the dielectric constant of the insulator, >>or for twin conductor the spacing between them and the dielectric >>constant of the insulator. >>ie (impedance = (138 / e^(1/2)) * log (D/d)) > > That's the simplified lossless mid-frequency model. At low > frequencies, Zo goes up because of resistive losses. At very high > frequencies, you get moding and complex whoopie-doos. None of that > matters if the coax or connectors are a tiny fraction of a wavelength, > as in the OP's case; they devolve to a small lumped capacitance. > > How he can get 6 dB of loss at low MHz is a mystery to me. > > John I concur ! 6db is +/-75%. -- Best Regards: Baron.
From: Baron on 2 May 2010 10:11
Joerg Inscribed thus: > John Larkin wrote: >> On Sat, 01 May 2010 21:19:52 +0100, Baron >> <baron.nospam(a)linuxmaniac.nospam.net> wrote: >> >>> mpm Inscribed thus: >>> >>>> On Apr 30, 10:58 pm, "k...(a)att.bizzzzzzzzzzzz" >>>> <k...(a)att.bizzzzzzzzzzzz> wrote: >>>>> On Fri, 30 Apr 2010 18:39:21 -0700 (PDT), mpm <mpmill...(a)aol.com> >>>>> wrote: >>>>>> On Apr 30, 4:31 pm, Wimpie <wimabc...(a)tetech.nl> wrote: >>>>>>> On 30 abr, 13:54, "mook johnson" <m...(a)mook.net> wrote: >>>>>>>> "John Larkin" <jjlar...(a)highNOTlandTHIStechnologyPART.com> >>>>>>>> wrote in message >>>>>>>> news:ql9kt5tg7s7e5q7pb460gcdig4r00jjmrj(a)4ax.com... >>>>>>>>> On Thu, 29 Apr 2010 18:30:41 -0500, "mook johnson" >>>>>>>>> <m...(a)mook.net> wrote: >>>>>>>>>> Gents, >>>>>>>>>> I looking for a consultant that can assist in the design of a >>>>>>>>>> custom military style connector with controlled >>>>>>>>>> characteristic impedance and insertion loss between two >>>>>>>>>> terminals in the 1MHz - 20MHz frequency range. >>>>>>>>>> This connector is special because is the application so an >>>>>>>>>> off the shelf component will not work. >>>>>>>>>> Any leads where I can start looking for such a consultant? >>>>>>>>>> thanks >>>>>>>>> There are so many military connectors, including ones for wild >>>>>>>>> environments, hermetic, etc, some standard part might work. >>>>>>>>> 20 MHz isn't very demanding. Most any mil connector will be >>>>>>>>> "matched" to any impedance at 20 MHz. Wavelength is 15 meters! >>>>>>>>> John >>>>>>>> The connector I need goes into environments that far exceed >>>>>>>> military applications but the physical concept is similar. We >>>>>>>> have an in-house connector company that makes these connectors >>>>>>>> for us but they have never had to deal with never greater than >>>>>>>>> 200KHz signals before. We tried using our regular connectors >>>>>>>> for this application and the impedance mismatch and insertion >>>>>>>> loss were extreme at 5 - 20MHz. The Zo mismatched by 50% and >>>>>>>> was not stable above 1MHz and had several resonant modes. >>>>>>>> Insertion loss something on the order is 6dB/connector and >>>>>>>> we'll >>>>>>>> have a couple dozen in series on this line. The number of >>>>>>>> series connection is the is the rub and it is unavoidable, non >>>>>>>> negotiable. >>>>>>>> There are some commercial plastic connectors that tested very >>>>>>>> well (just to validate our test setup) with good impedance >>>>>>>> match/stability (+/- 5%) and low insertion loss >>>>>>>> (.1dB/connector) but they won't take the environment. >>>>>>>> I'm looking for a consultant that can provide either of the >>>>>>>> following >>>>>>>> 1) model a connector that is already designed but not made >>>>>>>> (basically review the in-house company proposed design) and >>>>>>>> simulate the high frequency response of the design. >>>>>>>> 2) Give direction for the connector design based on constraints >>>>>>>> of material choices, physical size and geometry to meet the >>>>>>>> desired electrical signal characteristics while withstanding >>>>>>>> the environmental conditions. >>>>>>>> The cut and try approach based on simple equations has a long >>>>>>>> cycle time. I'm looking to improve my chances of getting it >>>>>>>> right the first time. >>>>>>> Hello Mook, >>>>>>> When you have already a company that knows mechanical design and >>>>>>> reliability issues, you only need the EM-field guy. When you >>>>>>> can keep the mechanical guys and the EM-field guy in one room, >>>>>>> this should converge to a solution fast. >>>>>>> One familiar with the concept of characteristic impedance, >>>>>>> complex propagation constant (contains both complex epsilon and >>>>>>> permeability) and know how to measure the material constants can >>>>>>> do the job. >>>>>>> What about: required Return Loss, characteristic impedance, >>>>>>> Insertion loss, propagation delay, size, cross section, >>>>>>> (coaxial, square, symmetrical, etc), peak and average power, >>>>>>> etc? Kind regards, >>>>>>> Wim >>>>>>> PA3DJSwww.tetech.nl >>>>>>> when you delete abc first, PM will reach me- Hide quoted text - >>>>>>> - Show quoted text - >>>>>> I have a question for you: >>>>>> Do you believe that the characteristic impedance is dependent on >>>>>> the length of the coax? >>>>>> (For the ratio of transverse electric field to transverse >>>>>> magnetic field launched on a transmission line of infinite >>>>>> length.) >>>>> "infinite length" <> "independent of length"- Hide quoted text - >>>>> >>>>> - Show quoted text - >>>> Not sure I understand your cryptic post. >>>> >>>> That is how characteristic impedance is classically defined, and >>>> I'm asking him if he believes that characteristic impedance is >>>> dependent on length? >>> For co-axial cables its a function of the ratio of the diameters of >>> the conductors modified by the dielectric constant of the insulator, >>> or for twin conductor the spacing between them and the dielectric >>> constant of the insulator. >>> ie (impedance = (138 / e^(1/2)) * log (D/d)) >> >> That's the simplified lossless mid-frequency model. At low >> frequencies, Zo goes up because of resistive losses. At very high >> frequencies, you get moding and complex whoopie-doos. None of that >> matters if the coax or connectors are a tiny fraction of a >> wavelength, as in the OP's case; they devolve to a small lumped >> capacitance. >> >> How he can get 6 dB of loss at low MHz is a mystery to me. >> > > Send in a kilowatt and see what starts to smoke? > > I know, I know, that was facetious. Profound apologies :-) > I've seen the enamel burn off a coil at that power level. :-) -- Best Regards: Baron. |