-
Why 26.5GHz?
Posted by
tkamiya
on 10 Jan, 2019 14:14
-
Some spectrum analyzers as well as frequency counters are rated for 26.5GHz. Why .5? It would make much more sense if it was 25 or 30GHz, but 26.5? Is there a historical or technical reason for this odd choice of bandwidth?
-
#1 Reply
Posted by
Jeroen3
on 10 Jan, 2019 14:41
-
Strange instrument specifications or "limitation" are sometime caused by laws.
-
#2 Reply
Posted by
Eric_S
on 10 Jan, 2019 14:47
-
Is it tied to the K-band, perhaps?
https://en.wikipedia.org/wiki/K_band_(IEEE)
The IEEE K band is a portion of the radio spectrum in the microwave range of frequencies from 18 to 27 gigahertz (GHz). The range of frequencies in the center of the K band between 18 and 26.5 GHz is absorbed by water vapor in the atmosphere due to its resonance peak at 22.24 GHz, 1.35 cm. Therefore these frequencies experience high atmospheric attenuation and cannot be used for long distance applications. For this reason the original K band has been split into three bands, Ka band, K-band, and Ku band as detailed below.
-
#3 Reply
Posted by
tsman
on 10 Jan, 2019 14:49
-
Probably because the lower limit of the Ka band is 26.5GHz. If you want to do something in the Ka band like satellite links then they want you to pay extra for the fancier models.
-
-
It might be because of connectors. I think you need to go to K from 3.5mm.
-
#5 Reply
Posted by
tkamiya
on 10 Jan, 2019 15:15
-
Hum..... ANY of them could be a reason. All very convincing!
-
#6 Reply
Posted by
tcottle
on 10 Jan, 2019 15:29
-
The upper limit of WR-42 waveguide is 26.5GHz
-
#7 Reply
Posted by
DaJMasta
on 10 Jan, 2019 16:04
-
It might be because of connectors. I think you need to go to K from 3.5mm.
This was always my go-to explanation, but there could be any number of potential ones. Manufacturers will often make special tight-tolerance variants of connectors that can run well beyond a typical connector's spec, letting them connect to cheaper interfaces while letting them advertise the high bandwidth. I guess it sells more interface connectors too!
-
-
yea but then you end up with pain in the ass 'metrology grade' connectors which are not labeled any different then regular connectors and unless someone knows what they are exactly you can destroy them with a lesser connector.
I mean it still happens but if you try to push a SMA to 30 GHz with precision manufacturing processes and someone connects a regular SMA to it, its going to cause many more problems then connecting a lab grade 3.5mm to a metrology grade 3.5mm (which might just cause a small reflection or whatever, rather then completely destroying performance).
I think it keeps the chaos down vs making very very different connectors that look identical. But you do wonder how much more expensive it would be to just make a better standard then SMA..........
But for the sake of argument imagine if someone made like a 40GHz BNC connector with some kind of invisible change. They would probably go to sleep with it at night around their neck in order to protect it from the rest of the workplace. It would be like a riced out honda. No one would ever put premium fuel in that.
-
#9 Reply
Posted by
Wolfgang
on 10 Jan, 2019 17:28
-
What I know this strange number comes from early SMA coaxial cabling and connectors that ensured a mode free operation up to this frequency (with some safety margin, of course).
-
-
when I used the nice knurled 3.5mm and apc-7 the slow connectors started to make me feel sad. The N-connector makes me feel like I am doing something on a car now.
-
-
when I used the nice knurled 3.5mm and apc-7 the slow connectors started to make me feel sad. The N-connector makes me feel like I am doing something on a car now.
A car? More of a cart! The N connectors which I want to eliminate at my working place feel like something that ought to be touched with hooves!
-
#12 Reply
Posted by
Howardlong
on 11 Jan, 2019 22:11
-
-
#13 Reply
Posted by
jjoonathan
on 11 Jan, 2019 22:52
-
I bet it's the cutoff frequency of the waveguide formed by the cheapest section of pipe available at the nearest hardware store to the MIT rad lab during WWII, or something along those lines.
-
-
lol
-
#15 Reply
Posted by
TheSteve
on 03 Mar, 2019 02:14
-
I was testing a pair of ebay special Gore cables today for insertion loss and phase distortion. They were supposed to be 2.1 meter SMA to N cables but are really 2.4 meters and one is 3.5 mm to N. They are right on the money with respect to loss at 18 GHz but thought I'd post this picture which shows the N connectors "moding" at 22.9 GHz. Doesn't really answer the mystical "why 26.5 GHz" value but does show why N is typically only spec'd to 18 GHz.
-
#16 Reply
Posted by
Wolfgang
on 03 Mar, 2019 13:26
-
Some spectrum analyzers as well as frequency counters are rated for 26.5GHz. Why .5? It would make much more sense if it was 25 or 30GHz, but 26.5? Is there a historical or technical reason for this odd choice of bandwidth?
Its not odd. Its the moding limit of SMA cables and connectors. Beyond this frequency there are several propagation modes possible, with different attenuations and velocities. If you have the equipment, you can try what happens if you connect a SMA cable (via adapters) to a V or W signal source and you then run more than 26.5GHz thru it. The cable attenuation becomes really irregular.
-
#17 Reply
Posted by
HalFET
on 03 Mar, 2019 14:35
-
I bet it's the cutoff frequency of the waveguide formed by the cheapest section of pipe available at the nearest hardware store to the MIT rad lab during WWII, or something along those lines.
This is honestly more likely than any other option posted so far.
-
#18 Reply
Posted by
tkamiya
on 03 Mar, 2019 21:32
-
I read a while back, this "dip" caused by N connector or it going into "mode" operation is known to Hewlett Packard engineers, yet they decided to offer N connector as standard on most 26.5GHz spectrum analyzers. Apparently, they absorb discrepancy by wide enough specification for the device. (In other words, even with this dip, it's within tolerance) The article also said it always appear at the same place so you just "consider it" as part of interpreting the data.
-
#19 Reply
Posted by
tefe
on 04 Mar, 2019 00:50
-
so...why 9k? why 9k as beginning but actually from 0Hz? almost all spectrum analyzers come from here.
-
#20 Reply
Posted by
TheSteve
on 04 Mar, 2019 01:52
-
so...why 9k? why 9k as beginning but actually from 0Hz? almost all spectrum analyzers come from here.
That's easier to answer. Most SA's that start at 9 kHz have a DC block (capacitor) in series with the input. That limits the lowest usable frequency but makes the analyzer front end more robust.
-
-
Keysight's MXA series don't have a DC block (mixer will blow up with DC applied) but still start at 9 kHz. The PXA series go down to 3 Hz if I recall. I thought this was more a limitation of the mixer or perhaps the phase noise from the analyzer's LO.
-
#22 Reply
Posted by
tefe
on 06 Mar, 2019 00:43
-
Thanks.
Maybe I mean, why almost all and exactly 9k? Could anyone design a 8k DC block?
Was this limitation set default by HP and followed by?
There are many EMI limitations starting from 9k, but I beleive this is the result.
-
#23 Reply
Posted by
jjoonathan
on 06 Mar, 2019 00:57
-
Sure, but it would inject a bit more charge into the protection diodes on each DC transient, it would require a bit more plate closeness/area (design difficulty), it would probably require spinning a new series of RF baubles (they're typically shared across instruments), and at the end of it all, your fancy expensive microwave analyzer would still have mediocre performance (noise floor, phase noise, etc) compared to a dedicated low-frequency instrument.
-
#24 Reply
Posted by
TheSteve
on 06 Mar, 2019 01:30
-
Most work below 9 kHz but the signal amplitude will drop(unless they don't display below 9). Mine starts at 0 even though it has a DC block and is spec'd to start at 5 kHz.