引用: 60user22 发表于 2019-1-7 09:23
尽管可以使用3-D E＆amp; M模拟器来模拟开口端连接器，但由于许多不明确的物理结构，其有用性受到限制。
例如，对于公连接器，公引脚的作用类似于天线。
raidation特性对联轴器螺母的位置和长度非常敏感。
Hi Ken,
thank you for replying.
> {quote:title=kenwong wrote:}{quote}
> Although it is possible to model an open end connector using 3-D E&M simulators, its usefulness is limited because of many ill defined physical structures. For the male connector for example, the male pin acts like an antenna.
I was thinking of shielding the male pin, by screwing a female socket on, after removing the female pin and any PTFE insulation from the female socket. That will stop most if not all of the radiation, but does not get around the fact the pin shape is not well very defined by the MIL standard for an N connector.
> The raidation characteristics is highly sensitive to the coupling nut's position and length.
As I say, one can eliminate that by screwing a female socket on, with all the inner parts removed. That provides an inexpensive open which should not radiate.
> The shape of the male pin tip is not well define and yet it has a first order impact on the radiation characteritics. Resonances within operating frequencies were also observed. For Type-N, the male pin is partially shielded but sets behind the outer conductor's mating plane. This set back creates any set of variables such as step impedance change. If the connector's dimensions are well control, it can be modeled and used below the resonance region.
I've noticed the effect of the nut on an open male SMA far more than open male N, since the SMA male pin protudes into a region of space close to the nut, whereas even a totally unterminated N does provide a degree of shielding. But for my purposes, I am able to at least remove the effects of the nut, by screwing in a female N with all the inner parts removed.
> An easier way to obtain the electrical characteristics is to do an accurate 1-port calibration, using the connector of interest as the test port connector. With error correction ON, measured the response of the open ended connector. That was how the FieldFox connector model generated.
I tried that, but for some reason I got what I think were totally unrealistic results. Do you see anything wrong with my logic? This is what I did.
1) Assumed the thin part of the male pin which inserts into the female is a contant diameter. I know it tapers at the end, but that was ignored. Of course, this introduces some error, which you did not have since your standards were well defined.
2) Looked at MIL-C-39012C standard which shows the thin part of the male pin extends to a distance of between 0.208" to 0.211" behind the reference plane. Of course, this variability gives a problem, but I assume a mean value of 0.2095" behind the reference plane.
3) Assumed that the offset delay would be -17.775 ps, which is the case for a length of 0.2095" It would be negative, as the pin is behind the reference plane.
4) Did a one-port cal on the VNA, then set a port extension of -17.775 ps.
5) Measured the phase shift as a function of frequency from 50 MHz to 6 GHz.
6) Calculated the capacitance from the phase shift, assuming a resistance of 50 Ohms.
Anyway, after doing that, I calculated a fringing capacitance varying from 211 fF at 50 MHz rising to 240 fF at 6 GHz, which seems a bit unrealistic to me. Perhaps I should have assumed a resistance greater than 50 Ohms, since the thin part of the pin would produce an offset Zo of > 50 Ohms. That would result in a reduced value for the capacitance, but the change with frequency would still be about 20%, which does not seem to be the case with any cal kit.
> My paper deal with modeling of calibration standards that are well defined physically. The open standards are shielded opens with a supported center conductor and with a defined offset length. The fringing capacitance can be modeled nicely using today's 3-D E&M simulators. Back then, I had to use a comination of simulation tools including proprietary ones.
OK, so you did not need to determine the offset delay, as you got this precisely from your measurements - something impossible with any standard connector.
> Many papers had been publushed on the open capacitance topic, specifically on open ended coaxial line dielectric probes. The Waveguide Handbook by Marcuvitz, section 4.16 deals with coaxial line radiating into semi-infinite space. These models are more applicable for open ended female connectors.
I found a lot of the papers, and found the maths too heavy! But they all seem to assume a groundplane infinite in extent, which is easy to make on a dielectric probe, but is not the case on an RF connector. Again, I thought an EM simulator might give me more insight into what is going on.
I think the basic problem is I'm trying to do something more complex and less well defined than you. For your paper, you knew precisely the offset Zo and offset delay from purely physical measurements, so you only used the EM simulator to find capacitance. I'm trying to model a connector when neither the offset Zo or offset delay are precisely known.
At the moment my two VNAs are on the dining table, much to the annoyance of my wife. I'm in the middle of decorating the house, so don't have easy access to my equipment. I will have to look at this again some time.
Thank you for your help Ken. If you have any comments, feel free to add them.
Dave
引用: 60user22 发表于 2019-1-7 09:23
尽管可以使用3-D E＆amp; M模拟器来模拟开口端连接器，但由于许多不明确的物理结构，其有用性受到限制。
例如，对于公连接器，公引脚的作用类似于天线。
raidation特性对联轴器螺母的位置和长度非常敏感。
Hi Ken,
thank you for replying.
> {quote:title=kenwong wrote:}{quote}
> Although it is possible to model an open end connector using 3-D E&M simulators, its usefulness is limited because of many ill defined physical structures. For the male connector for example, the male pin acts like an antenna.
I was thinking of shielding the male pin, by screwing a female socket on, after removing the female pin and any PTFE insulation from the female socket. That will stop most if not all of the radiation, but does not get around the fact the pin shape is not well very defined by the MIL standard for an N connector.
> The raidation characteristics is highly sensitive to the coupling nut's position and length.
As I say, one can eliminate that by screwing a female socket on, with all the inner parts removed. That provides an inexpensive open which should not radiate.
> The shape of the male pin tip is not well define and yet it has a first order impact on the radiation characteritics. Resonances within operating frequencies were also observed. For Type-N, the male pin is partially shielded but sets behind the outer conductor's mating plane. This set back creates any set of variables such as step impedance change. If the connector's dimensions are well control, it can be modeled and used below the resonance region.
I've noticed the effect of the nut on an open male SMA far more than open male N, since the SMA male pin protudes into a region of space close to the nut, whereas even a totally unterminated N does provide a degree of shielding. But for my purposes, I am able to at least remove the effects of the nut, by screwing in a female N with all the inner parts removed.
> An easier way to obtain the electrical characteristics is to do an accurate 1-port calibration, using the connector of interest as the test port connector. With error correction ON, measured the response of the open ended connector. That was how the FieldFox connector model generated.
I tried that, but for some reason I got what I think were totally unrealistic results. Do you see anything wrong with my logic? This is what I did.
1) Assumed the thin part of the male pin which inserts into the female is a contant diameter. I know it tapers at the end, but that was ignored. Of course, this introduces some error, which you did not have since your standards were well defined.
2) Looked at MIL-C-39012C standard which shows the thin part of the male pin extends to a distance of between 0.208" to 0.211" behind the reference plane. Of course, this variability gives a problem, but I assume a mean value of 0.2095" behind the reference plane.
3) Assumed that the offset delay would be -17.775 ps, which is the case for a length of 0.2095" It would be negative, as the pin is behind the reference plane.
4) Did a one-port cal on the VNA, then set a port extension of -17.775 ps.
5) Measured the phase shift as a function of frequency from 50 MHz to 6 GHz.
6) Calculated the capacitance from the phase shift, assuming a resistance of 50 Ohms.
Anyway, after doing that, I calculated a fringing capacitance varying from 211 fF at 50 MHz rising to 240 fF at 6 GHz, which seems a bit unrealistic to me. Perhaps I should have assumed a resistance greater than 50 Ohms, since the thin part of the pin would produce an offset Zo of > 50 Ohms. That would result in a reduced value for the capacitance, but the change with frequency would still be about 20%, which does not seem to be the case with any cal kit.
> My paper deal with modeling of calibration standards that are well defined physically. The open standards are shielded opens with a supported center conductor and with a defined offset length. The fringing capacitance can be modeled nicely using today's 3-D E&M simulators. Back then, I had to use a comination of simulation tools including proprietary ones.
OK, so you did not need to determine the offset delay, as you got this precisely from your measurements - something impossible with any standard connector.
> Many papers had been publushed on the open capacitance topic, specifically on open ended coaxial line dielectric probes. The Waveguide Handbook by Marcuvitz, section 4.16 deals with coaxial line radiating into semi-infinite space. These models are more applicable for open ended female connectors.
I found a lot of the papers, and found the maths too heavy! But they all seem to assume a groundplane infinite in extent, which is easy to make on a dielectric probe, but is not the case on an RF connector. Again, I thought an EM simulator might give me more insight into what is going on.
I think the basic problem is I'm trying to do something more complex and less well defined than you. For your paper, you knew precisely the offset Zo and offset delay from purely physical measurements, so you only used the EM simulator to find capacitance. I'm trying to model a connector when neither the offset Zo or offset delay are precisely known.
At the moment my two VNAs are on the dining table, much to the annoyance of my wife. I'm in the middle of decorating the house, so don't have easy access to my equipment. I will have to look at this again some time.
Thank you for your help Ken. If you have any comments, feel free to add them.
Dave
举报
