|
On Thu, Feb 4, 2021 at 09:54 PM, Reinier Gerritsen wrote:
I had 3 broken units a few months ago (the first version of the SAA). Replaced
the switches and they were back to life. On one unit I installed esd
protection diodes. But since I have my V2plus4, I don't use the old analyzers
anymore. I still have to modify all the other analyzers. TVS diode part
number: ESD101B102ELE6327XTMA1, a 5.5V TVS bi-directional diode from Infineon,
0.1 pF capacitance. Diodes are cheap. Switches are also cheap at Aliexpress.
Replacing the switch is not for the faint hearted. It is a tiny package with a
ground pad under the device. Hot air soldering is your only option. A good
stereo microscope helps.
Someone should make add on cards with those protection diodes. I wont solder them myself, 0603 are hard enough and sometimes even a 1206 between two SMA connector PCB pins.
|
thanks to Reiner info I bought the tvs diodes ESD101B102ELE6327XTMA1 with the idea of ??making me a protection for the nanovna with a pair of sma ports.
But when I got the tvs I felt lost and scared, I had never seen anything smaller to deal with.
However, I remembered that in one of my compulsive attacks I bought 35g of 60-40 paste and a portable hot air gun. On a pcb crumb I made a groove, two soldering pads, I tried to solder. A first tvs splashed somewhere in the lab while I was holding it in tweezers. With a second I was luckier and I think I soldered it. I assembled everything, on two sma ports, with also a 560 kohm resistor in parallel.
I tested the door with the vna to see how it felt with the measurements. I have tried it between 1 and 150 Mhz and I would say that it behaves well, moreover that I am currently interested in the 2 30 MHz range because I want to build a multiband octopus. I add some photos, hoping it will be useful (reference is calibration).
Ciao everybody.
gab
|
In the HackRF, the majority of the circuit needs to be protected from excursions outside the 3.2v rail voltages. The two diodes near the input achieve this, allowing excursions from the mid-point (set by the 10K resistors) to below -0.7v or above 3.2+0.7, a peak excursion of 2.3v from the centre in either direction.
The LNA needs additional protection when it is switched into the circuit, which is what the diode to ground achieves.
This use of these limiter diodes is a bit off-label, more like you'd use traditional diode clamps. I couldn't use inductors because it's not possible to find inductors that actually act like inductors over the range from 1MHz-6GHz.
I did not have the budget to prototype this and destroy multiple HackRFs in pursuit of exact data on RF immunity, but there have been zero failure reports from any users of the first batch of 1000 units sold. Time will tell, but compared with a typical failure rate of over 30% LNAs being killed within the first year, things are going well so far.
|
Has this been implemented for a fix to the issue of protection? I would like to update my unit with the fix.Thank you very much,BillW4MXT
|
On Fri, Feb 5, 2021 at 03:30 PM, Clifford Heath wrote:
When I redesigned the HackRF antenna protection, I removed the useless TVS, a
bi-directional 15V type (that will only activate after everything in the
HackRF is already dead), and instead used Skyworks Limiter Diodes
SMP1330-085LF.
In the case of the HackRF, I need to protect the LNA from any excursions above
13dBm, but the transmitter and mixer inputs are a bit more robust, so I used
three of these diodes.
Clifford - Just took a look at your schematic. --- I see you reverse biased one of the diodes to Vaa (3V) and the other to ground. Why not both to ground? ---- It look like you used 10 k resistors instead of RF chokes like the Skyworks app note. Any particular reason? ---- How much RF power can you now safely inject without damage using this new circuit. Thanks - Roger
|
When I redesigned the HackRF antenna protection, I removed the useless TVS, a bi-directional 15V type (that will only activate after everything in the HackRF is already dead), and instead used Skyworks Limiter Diodes SMP1330-085LF.
These are low-capacitance PIN diodes, but are designed with an unusual characteristic: after a forward pulse greater than 0.7v has turned them on, they briefly conduct in both directions with between 1 and 2 ohms impedance. The diode itself can absorb up to 30dBm safely, but the impedance discontinuity reflects up to 50dBm back toward the source, which protects the antenna input amplifiers.
In the case of the HackRF, I need to protect the LNA from any excursions above 13dBm, but the transmitter and mixer inputs are a bit more robust, so I used three of these diodes.
|
Hi, the most likly reason, that the protection diode is not alredy present on the vna, is cost or possible problems against I good measure?
No, cost is very low, jut a few cents. During the design phase ESD strikes were applied directly at the input, it passed the tests. However, after my report of broken units, the designer (Gabriel) tested again, now with at least 50 cm of cable connected and he was able to reproduce the problem. This happened to be my usual setup too. Gabriel concluded that the input voltage protection of the switch is sufficient, but that the current kills it. (current due to discharge of the cable capacitance). He suggested insering a small resistor of 5 Ohm to limit the current to the RF switch's protection diodes. A series resistor probably has some influence on the source impedance. A 0.1 pF protection diode does not. The non-linearity of the diode is a fraction of the 0.1pF. DC levels are zero anyway so you can consider the diode to be of no influence. I did not see any difference with or without diodes.
|
Hi, the most likly reason, that the protection diode is not alredy present on the vna, is cost or possible problems against I good measure?
|
On 2/5/21 6:04 AM, Manfred Mornhinweg wrote: Jim,
Standard 0.405" coax (RG-8, RG-213) is about 40 pF/meter, so a 100 ft/30meter run is 1200pF. You figure is far too low! 100pF per meter comes much closer to reality.
Manfred You're right.. I was looking at the wrong table..
|
Jim, Standard 0.405" coax (RG-8, RG-213) is about 40 pF/meter, so a 100 ft/30meter run is 1200pF. You figure is far too low! 100pF per meter comes much closer to reality. Manfred
|
Ken,
indeed you have to use the bidirectional type. Unidirectional ones conduct like any diode in the forward sense. They are pretty much like improved zener diodes. Bidirectional ones are two of them back-to-back in the same package.
When selecting one you need to be very careful about capacitance. Those large enough to survive the discharge pulse that happens when you connect an antenna system that has a charge on it, usually have capacitances of several hundred pF, and thus are useless at RF. Small ones for RF use exist, with capacitances even lower than 1pF, but they have very limited pulse power handling ability and may fail when connecting a large antenna charged to a significant voltage.
|
Yes, that makes sense. The diode I used is a bi-directional type with a clamping voltage of 5.5V. Static charge build-up is limited to 5.5V which should be harmless. In my case I'm developing UHF RFID antennas (860-930 MHz). That involves lots of touching the antenna, adding or removing copper tape for frequency tuning and impedance matching. The risk of ESD is high, I blew up 2 units and a customer of mine also 1. For me this is the only serious flaw of the SAA2 analyzer series. Next models will get ESD protection according to the designer. By the way, I did not add a bleeder resistor, might do that to make it even more fool proof.
Reinier
Op 4-2-2021 om 22:56 schreef Ken Sejkora:
toggle quoted message
Show quoted text
Greetings all,
Interesting discussion. When I Googled TVS (Transient Voltage Suppression) diode, it appears there are two varieties, unidirectional and bidirectional. I would assume a ‘bidirectional’ TVS diode effectively does not have a polarity and would dissipate static charge in either direction. It make perfect sense that wind/dust/snow/etc. blowing across an antenna could induce a static charge across the ‘capacitor’ represented by a coax cable, so the “polarity” of the center-conductor versus the coax shield could change depending on the specific conditions. If a unidirectional TVS diode was connected “backwards” across the coax connector as referenced to the coax itself, wouldn’t it represent an ineffective drain of the static charge, and potentially result in damaging the NanoVNA? If TVS diode polarity is important, I would think a bidirectional TVS diode would be the preferred device, followed by a high-value resistor, to bleed off the static charge.
Is my logic flawed? Please enlighten me. Thanks.
Ken, WB?OCV
From: Reinier Gerritsen
Sent: Thursday, February 4, 2021 03:18 PM
To: [email protected]
Subject: Re: [nanovna-users] SAA2N problem
It's kind of expensive (because SMA connectors aren't free), but one
could probably make a little protection board with a SMA jack one end
and SMA plug on the other, with the TVS diode on the single
microstripline trace in between.? Sort of a dual "connector saver" and
"VNA saver".? The parasitics of the board would "calibrate out" for
the most part.? If you had a steady hand, you might be able to build
one out of just the two connectors, if you get the kind with the posts
- solder the posts together and somehow put the diode in between. I've
done this for making a T or for oddball loads, but it's not something
you'd be proud of.
Soldering the diodes directly on the pcb is easy and you can never
forget them...
See pictures (follow the trace from connector to series capacitor to TVS
diode to resistive pad (3 resistors). The other port has the TVS diode
directly at the input (protects the capacitor too)
|
Greetings all, Interesting discussion. When I Googled TVS (Transient Voltage Suppression) diode, it appears there are two varieties, unidirectional and bidirectional. I would assume a ‘bidirectional’ TVS diode effectively does not have a polarity and would dissipate static charge in either direction. It make perfect sense that wind/dust/snow/etc. blowing across an antenna could induce a static charge across the ‘capacitor’ represented by a coax cable, so the “polarity” of the center-conductor versus the coax shield could change depending on the specific conditions. If a unidirectional TVS diode was connected “backwards” across the coax connector as referenced to the coax itself, wouldn’t it represent an ineffective drain of the static charge, and potentially result in damaging the NanoVNA? If TVS diode polarity is important, I would think a bidirectional TVS diode would be the preferred device, followed by a high-value resistor, to bleed off the static charge. Is my logic flawed? Please enlighten me. Thanks. Ken, WB?OCV From: Reinier Gerritsen Sent: Thursday, February 4, 2021 03:18 PM To: [email protected]Subject: Re: [nanovna-users] SAA2N problem It's kind of expensive (because SMA connectors aren't free), but one
could probably make a little protection board with a SMA jack one end
and SMA plug on the other, with the TVS diode on the single
microstripline trace in between.? Sort of a dual "connector saver" and
"VNA saver".? The parasitics of the board would "calibrate out" for
the most part.? If you had a steady hand, you might be able to build
one out of just the two connectors, if you get the kind with the posts
- solder the posts together and somehow put the diode in between. I've
done this for making a T or for oddball loads, but it's not something
you'd be proud of.
Soldering the diodes directly on the pcb is easy and you can never forget them... See pictures (follow the trace from connector to series capacitor to TVS diode to resistive pad (3 resistors). The other port has the TVS diode directly at the input (protects the capacitor too)
|
Hi Dave
The charge only accumulates if it is connected to an antenna that is not a DC short.? ?i.e.? long wire, gama matched Yagi, 1/4 wave vertical etc.
Not a problem with a folded Dipole, loop, or the antenna has most balun designs.
We use to connect an NE-2 neon lamp across the coax connector to a long wire antenna.? ? ? ?Takes about 80 volts to light up an NE-2.? ? ? ? ? ? ? ?Brisk breeze when the humidity was low and that neon light would flash every few seconds.? ??
This is why most preamps have a 10K resistor across their inputs to bleed off that change.
OK, in the shack, but for outside antennas I would not connect my VNA's to an antenna that is not a DC short.
Kent WA5VJB? ?Antenna Editor CQ Magazine.
toggle quoted message
Show quoted text
On Thursday, February 4, 2021, 02:56:03 PM CST, David Eckhardt <davearea51a@...> wrote: A coaxial cable is a long cylindrical capacitor.? Being by design low loss, it can accumulate and store a charge for quite some time.? When connecting any longer piece of coaxial cable or other low-loss transmission line, I always make it a habit before connecting it to anything expensive or valued by pressing my finger across the end to discharge any accumulated charge. Sometimes, there is nothing, but the first time you get 'bit', you will become a believer.? Transmission charge is highly likely to accumulate during a thunder storm, wind-blown dust orsand, and even snow fall. Dave - W?LEV On Thu, Feb 4, 2021 at 1:27 PM schweppe <schweppe@...> wrote: When looking at the front end of the nanovna you see a 10?F capacitor in
series with the MXD8641–SP4T Switch. This switch can resist 100V in
electrostatic discharge sensitivity testing (Machine Model). This test
is done with a loaded 200pF capacitor and a 0.5 ?H inductor in series.
A typical koaxcable (RG213 od RG58) with 100pF/m and 20m length will
have 20nF of capacitance. If this koax is loaded to 200V by
electrostatic, it will damage the MXD8641. Also if you measure an
outdoor antenna and there is electrostatic in the air, you may damage
your measuring equipment also.
It may be a good thing to protect your nanovna with ESD-diodes when
measuring long koaxcables and/or outdoor antennas.
73, DK5DN
Am 04.02.2021 um 10:30 schrieb Stephen Laurence:
Does this mean that a long length of coax can store/return dangerous (to the vna) voltages, for example like a Tesla coil, or was the coax connected
to something at the other end?
Should we all be installing esd diodes to the port 1 input? Port 2
already has a 14db attenuator.? Is there a limit to the length we can
measure?
Steve L
-- *Dave - W?LEV* *Just Let Darwin Work*
|
A coaxial cable is a long cylindrical capacitor. Being by design low loss,
it can accumulate and store a charge for quite some time. When connecting
any longer piece of coaxial cable or other low-loss transmission line, I
always make it a habit before connecting it to anything expensive or valued
by pressing my finger across the end to discharge any accumulated charge.
Sometimes, there is nothing, but the first time you get 'bit', you will
become a believer. Transmission charge is highly likely to accumulate
during a thunder storm, wind-blown dust orsand, and even snow fall.
Dave - W?LEV
toggle quoted message
Show quoted text
On Thu, Feb 4, 2021 at 1:27 PM schweppe <schweppe@...> wrote: When looking at the front end of the nanovna you see a 10?F capacitor in
series with the MXD8641–SP4T Switch. This switch can resist 100V in
electrostatic discharge sensitivity testing (Machine Model). This test
is done with a loaded 200pF capacitor and a 0.5 ?H inductor in series.
A typical koaxcable (RG213 od RG58) with 100pF/m and 20m length will
have 20nF of capacitance. If this koax is loaded to 200V by
electrostatic, it will damage the MXD8641. Also if you measure an
outdoor antenna and there is electrostatic in the air, you may damage
your measuring equipment also.
It may be a good thing to protect your nanovna with ESD-diodes when
measuring long koaxcables and/or outdoor antennas.
73, DK5DN
Am 04.02.2021 um 10:30 schrieb Stephen Laurence:
Does this mean that a long length of coax can store/return dangerous (to the vna) voltages, for example like a Tesla coil, or was the coax connected
to something at the other end?
Should we all be installing esd diodes to the port 1 input? Port 2
already has a 14db attenuator. Is there a limit to the length we can
measure?
Steve L
--
*Dave - W?LEV*
*Just Let Darwin Work*
|
I was measuring what turned out to be about 325 foot run of coax. About 200 feet of lmr400 and 125 feet of rg213. Think I will have to find some diodes to solder on the pcb. I don't think I have any surface mount stuff handy. I have mostly larger components on hand :-)
Thanks again!
|
It's kind of expensive (because SMA connectors aren't free), but one could probably make a little protection board with a SMA jack one end and SMA plug on the other, with the TVS diode on the single microstripline trace in between.? Sort of a dual "connector saver" and "VNA saver".? The parasitics of the board would "calibrate out" for the most part.? If you had a steady hand, you might be able to build one out of just the two connectors, if you get the kind with the posts - solder the posts together and somehow put the diode in between. I've done this for making a T or for oddball loads, but it's not something you'd be proud of.
Soldering the diodes directly on the pcb is easy and you can never forget them... See pictures (follow the trace from connector to series capacitor to TVS diode to resistive pad (3 resistors). The other port has the TVS diode directly at the input (protects the capacitor too)
|
On 2/4/21 12:00 PM, Thomas Kerns wrote: Thanks for the help. I don't remember exactly what I was doing at the moment it quit, but I had been measuring a long run of coax, and I had also measured the swr of an outdoor antenna (a 160 meter dipole). That is a lot of wire in the air, with the potential to pick up static, I suppose. I wonder the best way to protect my vna in the future.
I see someone mentioned a drain resistor. Would this be something I would do only when testing? ie make a coax pigtail with a drain resistor across from the shield to the center? would that affect measurements? It's a good practice in general to "permanently" have a leakage path to ground.? A 100k or 1 meg resistor in a Coax T is one way. If you get a big transient, it will probably fail, so that makes an ohmmeter a useful diagnostic tool <grin> Some people use a RF choke that has high Z at the operating frequency, but lower DC resistance to bleed the charge faster. ?Standard 0.405" coax (RG-8, RG-213) is about 40 pF/meter, so a 100 ft/30meter run is 1200pF.
|
On 2/4/21 11:54 AM, Reinier Gerritsen wrote: I had 3 broken units a few months ago (the first version of the SAA). Replaced the switches and they were back to life. On one unit I installed esd protection diodes. But since I have my V2plus4, I don't use the old analyzers anymore. I still have to modify all the other analyzers. TVS diode part number: ESD101B102ELE6327XTMA1, a 5.5V TVS bi-directional diode from Infineon, 0.1 pF capacitance. Diodes are cheap. Switches are also cheap at Aliexpress. Replacing the switch is not for the faint hearted. It is a tiny package with a ground pad under the device. Hot air soldering is your only option. A good stereo microscope helps. It's kind of expensive (because SMA connectors aren't free), but one could probably make a little protection board with a SMA jack one end and SMA plug on the other, with the TVS diode on the single microstripline trace in between.? Sort of a dual "connector saver" and "VNA saver".? The parasitics of the board would "calibrate out" for the most part.? If you had a steady hand, you might be able to build one out of just the two connectors, if you get the kind with the posts - solder the posts together and somehow put the diode in between. I've done this for making a T or for oddball loads, but it's not something you'd be proud of.
|
Thanks for the help. I don't remember exactly what I was doing at the moment it quit, but I had been measuring a long run of coax, and I had also measured the swr of an outdoor antenna (a 160 meter dipole). That is a lot of wire in the air, with the potential to pick up static, I suppose. I wonder the best way to protect my vna in the future.
I see someone mentioned a drain resistor. Would this be something I would do only when testing? ie make a coax pigtail with a drain resistor across from the shield to the center? would that affect measurements?
|
Gab IU0 OQM
Roger Need
