Beware of cheap underperforming clones

As of 2022 there are many badly performing clones on the market. V2/3GHz NanoVNA uses parts like ADF4350 and AD8342 which are costly and clones have been cutting costs by using salvaged or reject parts.

See official store and look for V2 Plus4/V2 Plus4 Pro versions only to avoid getting a bad clone. We have stopped selling V2.2 versions since October 2020, so all V2 hardware that are not Plus or Plus4 are not made by us and we can not guarantee performance.

NanoVNA V2 Forum

Note: this page is a mirror of https://groups.io/g/NanoVNAV2.
Click here to join and see most recent posts.

CH0 has gone noisy?


moleonacid 2020/11/11 13:16

Hello All,

I've had my V2Plus for about a month now and have been getting on very well with it.  I took it out to work on an antenna today and the screen looked very "noisy" at startup - no ramps on the phase, Smith Chart all over the place etc.  I have tried clearing, recalibrating, saving to slot 0 all no no avail.  There seems to be be some sense to it - the noise centers around the centre of the chart with the 50Ohm load attached, but it is really unusable noisy.  Trying to measure anything on the demo board just yields nothing meaningful.

I've tried removing the screen on CH0 and checking for shorts.  The centre pin soldered connection, doesn't look amazing, but it doesn't look dry either.

I've attached a couple of photos to illustrate what I mean.  I see there is another thread on the go with similar symptoms.  I'd certainly welcome anyone's thought on this one.

Many Thanks,

Dave.

Siegfried Jackstien 2020/11/12 06:22

i guess you have a defective cable and calibrated the device with that
defective cable

why i do guess that way?? hmmm on second picture you have ch1 open ...
s21 leaking (noise) should be at -80 or -90 db but it is at -36 db

so the calibration went wrong

dg9bfc sigi

Am 11.11.2020 um 21:16 schrieb moleonacid@gmail.com:

moleonacid 2020/11/12 00:41

Thanks for the reply, Sigi

I don’t think it’s that.  I get the same result using two different cables. I also get the same result with the calibration kit straight onto ch0. I can replicate it further by calibrating against the SOLT elements of the demo kit with completely different cables (I know it’s not calibration grade, but should at least be close enough to rule out the cables).  In short no combination of cables or calibration elements makes any difference.  The 50Ohm load checks out fine at DC and the whole calibration kit behaves as it should from 1.8 to 175MHz on my MFJ 259b antenna analyser.

Many Thanks,
Dave M0MYA

Reinier Gerritsen 2020/11/12 10:32

I have access to 6 units SAA2 (first model). 3 of them show similar
behaviour. No response to no matter what you connect to channel 0. In my
case they are showing an open on the Smith chart
I suspect the RF switch (ecal) directly at the input of ch0. It probably
died due to esd. I already received a bunch of new switches, just had no
time to try to fix it.
Gabriel, does the new plus4 model also use this switch directly at the
input?

Reinier

Op 12-11-2020 om 09:41 schreef moleonacid@gmail.com:

Stephen Laurence 2020/11/12 02:33

Dear Reinier,

Three identical failures of SAA2 models, as you suggest possibly by esd, is worrying. Port 2 has a 14db attenuator, but port 1 goes straight into the switch via a dc blocking capacitor.

Is there any benefit to adding a 10k smd ( or lower value) resistor across the input, somewhere, before the blocking capacitor, without affecting vna performance? It would provide some esd protection. Esd vulnerability must be greater when working on aerials etc. An alternative would be to put a low value (3db) sma attenuator on port1, when “outside”.

would Gabriel have any views on this?

Steve L

moleonacid 2020/11/12 05:34

Thank you both for your thoughts.  It’s a shame that my nanovna is cooked, but hopefully it might contribute to a fix so that future versions are not susceptible.

I did find the part number from the schematic, but they are difficult to source in small batches, and I don’t think I’ve a hope of replacing something with four sides and a pretty tiny pitch!

Thanks again,
Dave.

Reinier Gerritsen 2020/11/12 15:11

I sourced the switches from Aliexpress, 50 pcs for €7,65 in total,
including free shipping. I just removed the part and placed a new one.
Unfortunately that did not fix the problem. But now ch0 is stuck at the
center of the smith chart (50 Ohm).
Desoldering is easy with a hot air gun and a small nozzle. Soldering the
new one on the other hand is not that easy. I'll retry. Note this part
is in a QFN package, with a ground pad at the back of the part that is
not accessible with a solder iron.

Reinier

Op 12-11-2020 om 14:34 schreef moleonacid@gmail.com:

Jim Lux 2020/11/12 06:20

On 11/12/20 2:33 AM, Stephen Laurence wrote:
> Dear Reinier,
>
> Three identical failures of SAA2 models, as you suggest possibly by esd,
> is worrying. Port 2 has a 14db attenuator, but port 1 goes straight into
> the switch via a dc blocking capacitor.
>
> Is there any benefit to adding a 10k smd ( or lower value) resistor
> across the input, somewhere, before the blocking capacitor, without
> affecting vna performance? It would provide /some /esd protection. Esd
> vulnerability must be greater when working on aerials etc. An
> alternative would be to put a low value (3db) sma attenuator on port1,
> when “outside”.

3dB pad is one approach.
On board, clamp diodes, perhaps. The parasitic C from the diodes would
"calibrate out"
The MXD8641 is rated as a 1000V HBM for ESD, which isn't super
sensitive, but in a device that hooks up to coax cables, which can store
a lot of charge, some sort of transient protection would be a good idea.


Reinier Gerritsen 2020/11/12 16:13

A little rework brought it back to life! So now I'm sure it was the
input switch at CH0. Still, 3 out of 6 broke down within a few months,
that's a horrible MTBF. I'll try adding a low capacitance ESD
suppression diode, maybe a ESD101-B1-02 will do (0.1 to 0.2 pF).

Reinier


Op 12-11-2020 om 15:11 schreef Reinier Gerritsen:
> I sourced the switches from Aliexpress, 50 pcs for €7,65 in total,
> including free shipping. I just removed the part and placed a new one.
> Unfortunately that did not fix the problem. But now ch0 is stuck at
> the center of the smith chart (50 Ohm).
> Desoldering is easy with a hot air gun and a small nozzle. Soldering
> the new one on the other hand is not that easy. I'll retry. Note this
> part is in a QFN package, with a ground pad at the back of the part
> that is not accessible with a solder iron.
>
> Reinier
>
> Op 12-11-2020 om 14:34 schreef moleonacid@gmail.com:
>> Thank you both for your thoughts.  It’s a shame that my nanovna is
>> cooked, but hopefully it might contribute to a fix so that future
>> versions are not susceptible.
>>
>> I did find the part number from the schematic, but they are difficult
>> to source in small batches, and I don’t think I’ve a hope of
>> replacing something with four sides and a pretty tiny pitch!
>>
>> Thanks again,
>> Dave.
>
>

--
IDcircuits - RFID & Electronics
Het Halster 40
6581 JL Malden
The Netherlands
+31 633702492
www.idcircuits.com
r.gerritsen@idcircuits.com

Siegfried Jackstien 2020/11/12 15:23

i guess switch was stuck at the load position (input selection did not
work) ...

good that you found it ...

3 out of 6 ??? do you buy analyzers in half a dozen numbers??

maybe make a few screenshots with the defective ones so other users know
when it looks "like this" then replace the switch

well done

dg9bfc sigi

ps i am sure you do not need all 47 left??

Am 12.11.2020 um 15:13 schrieb Reinier Gerritsen:

Reinier Gerritsen 2020/11/12 16:49

Bought 2, then package got stuck due to Covid, ordered new ones.
Original order was resent after some months by HCXQS group (Tindie). And
a customer of mine also ordered a few. He had one failure, I had two.
All the same symptoms. It happpened at the same time when Microsoft did
a major Windows update, messing up with the COM port driver. Took a
while to find out I had both a hardware and a software failure...
Attached two pictures of the defective units with input connected to
output (1 m RG316 cable). Channel 0 is open.

Op 12-11-2020 om 16:23 schreef Siegfried Jackstien:
>
> i guess switch was stuck at the load position (input selection did not
> work) ...
>
> good that you found it ...
>
> 3 out of 6 ??? do you buy analyzers in half a dozen numbers??
>
> maybe make a few screenshots with the defective ones so other users
> know when it looks "like this" then replace the switch
>
> well done
>
> dg9bfc sigi
>
> ps i am sure you do not need all 47 left??
>
> Am 12.11.2020 um 15:13 schrieb Reinier Gerritsen:
>> A little rework brought it back to life! So now I'm sure it was the
>> input switch at CH0. Still, 3 out of 6 broke down within a few
>> months, that's a horrible MTBF. I'll try adding a low capacitance ESD
>> suppression diode, maybe a ESD101-B1-02 will do (0.1 to 0.2 pF).
>>
>> Reinier
>>
>>
>> Op 12-11-2020 om 15:11 schreef Reinier Gerritsen:
>>> I sourced the switches from Aliexpress, 50 pcs for €7,65 in total,
>>> including free shipping. I just removed the part and placed a new
>>> one. Unfortunately that did not fix the problem. But now ch0 is
>>> stuck at the center of the smith chart (50 Ohm).
>>> Desoldering is easy with a hot air gun and a small nozzle. Soldering
>>> the new one on the other hand is not that easy. I'll retry. Note
>>> this part is in a QFN package, with a ground pad at the back of the
>>> part that is not accessible with a solder iron.
>>>
>>> Reinier
>>>
>>> Op 12-11-2020 om 14:34 schreef moleonacid@gmail.com:
>>>> Thank you both for your thoughts.  It’s a shame that my nanovna is
>>>> cooked, but hopefully it might contribute to a fix so that future
>>>> versions are not susceptible.
>>>>
>>>> I did find the part number from the schematic, but they are
>>>> difficult to source in small batches, and I don’t think I’ve a hope
>>>> of replacing something with four sides and a pretty tiny pitch!
>>>>
>>>> Thanks again,
>>>> Dave.
>>>
>>
>

--
IDcircuits - RFID & Electronics
Het Halster 40
6581 JL Malden
The Netherlands
+31 633702492
www.idcircuits.com
r.gerritsen@idcircuits.com

Siegfried Jackstien 2020/11/12 16:52

hmmm ... after your repaired it it was stuck to load (guessing switching
lines not well soldered) ... now what if in the defective one not the
chip is toasted but also switching line not well soldered?? (or output
of chip to the rest of circuit) ??

did you try to just reflow them?? ... is it for sure the chip and not
maybe the coupling cap and/or the socket??

just guessing what may cause such a fault ... i more think of mechanical
stress as esd but ok if it did havven on a (well charged) coax ... (britzel)

hmmmm ... if we would put a very high value inductance in parallel to
output ... that would have thousands of k ohms on rf ... not?? (but
provide a dc path to ground) ... just thinking

dg9bfc

Am 12.11.2020 um 15:49 schrieb Reinier Gerritsen:

Reinier Gerritsen 2020/11/12 18:01

First attempt was "stuck to load", second attempt was back to normal
operation. Some of the pins did not make contact. It's a little pita to
solder them...

Reinier

Op 12-11-2020 om 17:52 schreef Siegfried Jackstien:

Stephen Laurence 2020/11/12 09:30

No. A high inductance would fail its task because-

1. It would wreck performance at the top end
2. The high inductance would *prevent* a quick discharge so it would still burn out the input.

A 10k resistor would be better on both counts. But a 3db attenuator would be better still and easily fitted when working on aerials.

On second thoughts, if shunting the input with a low db attenuator would not materially affect the vna operation, then a 1k or even 300ohm shunt resistor would be better. A 300ohm resistor shunting a 50ohm input impedance is hardly going to change input impedance at all.

Perhaps we should all be fitting them on Port1 (ch 0).

Steve L

Jim Lux 2020/11/12 10:32

On 11/12/20 9:30 AM, Stephen Laurence wrote:
> No. A high inductance would fail its task because-
>
>  1. It would wreck performance at the top end
>  2. The high inductance would *prevent *a quick discharge so it would
> still burn out the input.
>
> A 10k resistor would be better on both counts. But a 3db attenuator
> would be better still and easily fitted when working on aerials.
>
> On second thoughts, if shunting the input with a low db attenuator would
> not materially affect the vna operation, then a 1k or even 300ohm shunt
> resistor would be better. A 300ohm resistor shunting a 50ohm input
> impedance is hardly going to change input impedance at all.
>
> Perhaps we should all be fitting them on Port1 (ch 0).
>

The problem is that your source is basically a charged capacitor with
low series R.

Let's say you have a 30m of coax with 40pf/meter charged to 50V. That's
1.2nF. If you hook it to 300 ohms, the voltage is still 50V at the
beginning of the discharge. Will it get low enough before the part is
destroyed? You can put a series R on the input, and between the series
R and the shunt R you can sort of limit it, but remember, at best, you
get Rshunt/(Rseries+Rshunt) and for reasonable values for RF work,
you're not going to get much voltage divider action.

It's not like a 1k resistor limiting the current into protection diodes
on a High Z input (like a line receiver)

Human body model is 100 pF in series with 1500 ohms. But that's just a
"test model" and not necessarily representative of real life.

In any case, if the HBM is charged to 1000V (the rating of the switch
part) and the input impedance of the switch is 50 ohms (does it
terminate an input that is "switched away from"?) the voltage divider is
0.032, so the 1000V is now 32V, which is a plausible number for the part
to withstand, since the energy is small (1000V in 100pF is 0.05 millijoule)

But, is your environment such that you're limited to 1000V on a human?

my example of a 50V charge on 30m of coax is pretty plausible. And
there's no series 1500 ohms like in the HBM. That's why the data sheet
says MM (machine model, very low series R) is <100V.

I'd look at clamp diodes, - a VNA can reject any IMD from the diodes
operating in their square law region. Or a 3dB pad (although do look at
the analysis for a charged capacitor - I'm not sure that the voltage
divider ratio is enough to save you.

Jim Lux 2020/11/12 11:16

On 11/12/20 10:32 AM, Jim Lux wrote:
> On 11/12/20 9:30 AM, Stephen Laurence wrote:
>> No. A high inductance would fail its task because-
>>
>>   1. It would wreck performance at the top end
>>   2. The high inductance would *prevent *a quick discharge so it would
>> still burn out the input.
>>
>> A 10k resistor would be better on both counts. But a 3db attenuator
>> would be better still and easily fitted when working on aerials.
>>
>> On second thoughts, if shunting the input with a low db attenuator
>> would not materially affect the vna operation, then a 1k or even
>> 300ohm shunt resistor would be better. A 300ohm resistor shunting a
>> 50ohm input impedance is hardly going to change input impedance at all.
>>
>> Perhaps we should all be fitting them on Port1 (ch 0).
>>
>
> The problem is that your source is basically a charged capacitor with
> low series R.
>
> Let's say you have a 30m of coax with 40pf/meter charged to 50V. That's
>  1.2nF.  If you hook it to 300 ohms, the voltage is still 50V at the
> beginning of the discharge. Will it get low enough before the part is
> destroyed?  You can put a series R on the input, and between the series
> R and the shunt R you can sort of limit it, but remember, at best, you
> get Rshunt/(Rseries+Rshunt) and for reasonable values for RF work,
> you're not going to get much voltage divider action.
>
> It's not like a 1k resistor limiting the current into protection diodes
> on a High Z input (like a line receiver)
>
> Human body model is 100 pF in series with 1500 ohms. But that's just a
> "test model" and not necessarily representative of real life.
>
> In any case, if the HBM is charged to 1000V (the rating of the switch
> part) and the input impedance of the switch is 50 ohms (does it
> terminate an input that is "switched away from"?) the voltage divider is
> 0.032, so the 1000V is now 32V, which is a plausible number for the part
> to withstand, since the energy is small (1000V in 100pF is 0.05 millijoule)
>
> But, is your environment such that you're limited to 1000V on a human?
>
> my example of a 50V charge on 30m of coax is pretty plausible. And
> there's no series 1500 ohms like in the HBM.  That's why the data sheet
> says MM (machine model, very low series R) is <100V.
>

I looked it up. MM is 200 pF discharged through 0.5 nH (the inductance
of 0.5 mm of wire, give or take)


> I'd look at clamp diodes, - a VNA can reject any IMD from the diodes
> operating in their square law region.  Or a 3dB pad (although do look at
> the analysis for a charged capacitor - I'm not sure that the voltage
> divider ratio is enough to save you.
>

correction, it's not IMD, it's just harmonic distortion - there's only
one signal. But it's not a trivial problem. A shunt diode, even far
below "fully turned on" is a fine harmonic generator

Stephen Laurence 2020/11/12 11:30

Er......

Perhaps I (as a retired doctor) should shut up.

However there can never be absolute protection in any sphere of life. When I was working, I could never give an absolute guarantee my patient would wake up from their anaesthetic.

Back-to-back protection diodes might be excellent protection, but, the degredation of performance of a £50-£100 device may not be justified compared to the less effective protection of a 300ohm resistor. Most users of this forum probably would not blink if they took the family for a “cheap meal out” for the cost of buying one of these vna devices. I pour the cost of 1-2 vnas into my car every time I fill up (ok, it is a gas guzzler).

Although I have never owned, or even seen one of the fancy professional vnas used in the field for aerial erection and maintenance, it would not surprise me if there are not expendable write-offs such as connectors, leads etc, every time they are used, costing more than a nanovna.

Returning to aspects of my original profession, I wonder how many of us do a risk analysis of their aerial testing and setting up, regarding their own wellbeing (tripping, falling off ladders etc) and cost of hospitalisation, loss of income etc, compared to the cost of their Vna.

Moan over.

PS. I have not *yet* blown one of my four nanovna devices yet, despite gutting them to experiment regarding improved performance.

Steve L. G7PSZ

moleonacid 2020/11/12 11:45

Good info here, and some hope!  Thanks for the Ali Express tip, and congratulations on resurrecting your dead analyser.

Dave.

Jim Lux 2020/11/12 11:50

On 11/12/20 11:30 AM, Stephen Laurence wrote:
> Er......
>
> Perhaps I (as a retired doctor) should shut up.
>
> However there can never be absolute protection in any sphere of life.
> When I was working, I could never give an /absolute /guarantee my
> patient would wake up from their anaesthetic.
>
> Back-to-back protection diodes might be excellent protection, but, the
> degredation of performance of a £50-£100 device may not be justified
> compared to the less effective protection of a 300ohm resistor. Most
> users of this forum probably would not blink if they took the family for
> a “cheap meal out” for the cost of buying one of these vna devices. I
> pour the cost of 1-2 vnas into my car every time I fill up (ok, it is a
> gas guzzler).

The problem is that a 300 ohm resistor won't protect it. It does nothing
to reduce the peak voltage seen by the RF switch when the ESD first hits.

The diodes are cheap - maybe the next rev of the board will include
them. This is an ideal application - the parasitic C from the diode will
be calibrated out, and the nonlinear junction effects are filtered out
because the VNA is a narrow band tuned receiver.


>
> Although I have never owned, or even seen one of the fancy professional
> vnas used in the field for aerial erection and maintenance, it would not
> surprise me if there are not expendable write-offs such as connectors,
> leads etc, every time they are used, costing more than a nanovna.

Generally not - They're designed to be robust to typical things like DC
voltages, ESD, etc. Hook up line voltage, and it's toast, but that's
not super common.

And, in commercial practice, it's standard to ground the coax center pin
to the shield and to something else before connecting it.

One thing you will see is "connector savers" - a plug/jack combo that
you put on the thing, so that if you bang it and damage the threads, you
don't have to replace the connector (which is expensive).

The damage I've seen has been things like:
1) hooking up an instrument to an amplifier that normally has a DC
blocking capacitor on the output. 15 or 24V will cook the input of the
instrument, sometimes.

2) inadvertently putting high power in - wrong switch position, wrong
cable, hit the wrong button - the diode won't save you when it's 100 Watts

3) Something in the breadboard on the bench shorts, putting DC on the
instrument input (oops, that ezhook came loose when I brushed the wire,
and it shorted 15VDC to the RF output.

4) Mechanical damage - dropping it, dropping something on it, setting it
down on the ground and getting a rock stuck in the connector, not seeing
it, then mating it, driving the rock further in.

5) Mechanical damage from mating with the wrong connector (K connectors,
I'm looking at you!) which has the same thread and size, but has a
different center pin configuration)

6) Ungrounded bench, person is live with line voltage (due to leakage
capacitance from line to chassis), but doesn't realize it and touches
circuit, putting a (high impedance) 120V 60Hz into the circuit. Not
enough to feel the shock (around 1-5 mA), but certainly enough to cook
the instrument input.

In 40 years, I think I've broken or seen someone else break things just
about every way possible.

Siegfried Jackstien 2020/11/12 20:03

maybe just add a short cicuited connector to your vna case

before antenna with longer cable used ... discharge coax

dg9bfc sigi

ps double female and short from cal kit can be used

Am 12.11.2020 um 19:50 schrieb Jim Lux:

Clifford Heath 2020/11/13 13:49

My "Hardened" version of the HackRF uses Skyworks SMP1330-085LF Limiter diodes. In general it was not the RF switches that fail, since those are made to pass 28dBm (almost a watt!) but the more sensitive receiver components, especially the LNA. These diodes clamp around 0.7V, going quickly to approximately 2-3 ohms resistance, which reflects the incident RF power back at the source. So despite not handling pulses of more than a watt, they can reflect pulses up to 100W (50dBm). Read the data sheets and application notes for more insight. They're only characterised to 2GHz, but (in the HackRF case) seem to cause little reduction in sensitivity even above 5gHz.

As Jim says, there may be IMD if you have higher-level signals in the vicinity of your measurement. Because the SAA2 does not have a selective receiver (narrow bandpass) like a proper VNA, this will actually affect your measurements. I'm not convinced that this "VNA can reject any IMD from the diodes operating in their square law region". Just avoid relying on it in the presence of ambient RF.

Stephen Laurence 2020/11/14 00:47

There are two components to this thread, it seems..

-1 Apparent failure due to ESD on hf/vhf aerials, hopefully tested during clement weather. Only a madman would fiddle with his aerials before an impending thunderstorm or even during it (unlike Benjamin Franklin in 1752).

- 2 Protection of our test gear to all possible mishaps, inside as well as outside the shack.

I certainly have used back to back diodes on the input to some of my test gear, but I do not believe a length of coax/300pf capacitor charged to 1000 volts would cause a high voltage when discharged across a 1k resistor (unless it was a wirewound one). In my experience with 45 years of petrol cars, using capacitor discharge units since 1980, a 30kv spike is nothing when there is a bit of damp, dirt etc to produce a relatively high resistance “ short” anywhere on its way to the spark plug. And no, I was not stupid enough to look at the waveform with my oscilloscope, although I did make current probes on the spark plug leads consisting of pickup coil on a split toroid glued inside a clothes peg (just like an rf current pickup) to help diagnose a mechanical fault. A second pickup on cylinder 1 lead gave triggering of the scope.

Steve L

Reinier Gerritsen 2020/11/14 12:41

In my case it was the RF switch. I do not use long cables, just 1 meter.
I'm developing 860-930 MHz antennas for RFID applications, touching the
antenna all the time when adding or removing copper tape for tuning. I'm
pretty sure ESD killed the switches.
Two devices survived a couple of months, one just a few days. Amazing
that all 3 units failed within 2 days, at 2 different locations 150 km
apart and right after a major Windows update.
I hope that adding protection diodes (ESD101B102ELE6327XTMA1) will
prevent further failures. These diodes have a capacitance of just 0.1 pF
(max 0.2 pF). The Skyworks SMP1330-085LF is not suited for this job, it
does not survive ESD. Its purpose it to protect receiver inputs from
transmitters. Their ESD rating is just 500V, the capacitance is 0.5 pF.

Reinier

Op 13-11-2020 om 22:49 schreef Clifford Heath:

moleonacid 2020/11/14 03:48

>
> -1 Apparent failure due to ESD on hf/vhf aerials, hopefully tested during
> clement weather. Only a madman would fiddle with his aerials before an
> impending thunderstorm or even during it (unlike Benjamin Franklin in
> 1752).

For me, the kicker seems to have been while tweaking a Moxon Rectangle cut for 144MHz (a folded 2 element beam). It was about 4m AGL fed by 7m of good quality RG58U.   Weather was typically British (grey with a hint of drizzle), but no sign of thunderstorms!

50 new switches ordered from AliExpress - we'll see how it goes replacing it once they make their way to me.

Dave.

Jim Lux 2020/11/14 06:15

On 11/14/20 12:47 AM, Stephen Laurence wrote:
> There are two components to this thread, it seems..
>
>   -1 Apparent failure due to ESD on hf/vhf aerials, hopefully tested
> during clement weather. Only a madman would fiddle with his aerials
> before an impending thunderstorm or even during it (unlike Benjamin
> Franklin in 1752).
>
>   - 2 Protection of our test gear to all possible mishaps, inside as
> well as outside the shack.
>
> I certainly have used back to back diodes on the input to some of my
> test gear, but I do not believe a length of coax/300pf capacitor charged
> to 1000 volts would cause a high voltage when discharged across a 1k
> resistor (unless it was a wirewound one).

Of course it will have HV - The voltage will be Vstart * exp(-t/RC) like
any RC discharging

At the beginning of the discharge, the voltage across the resistor will
be the full voltage on the capacitor.


In my experience with 45 years
> of petrol cars, using capacitor discharge units since 1980, a 30kv spike
> is nothing when there is a bit of damp, dirt etc to produce a relatively
> /high resistance “/short” anywhere on its way to the spark plug.

Different situation - There you have
a) resistive spark plug wires
b) transformer with signficant leakage inductance
c) transformer with significant series resistance in the secondary
d) it doesn't take much voltage reduction to misfire. If it's 20kV
instead of 40kV, the plug misfires.


And no,

Jim Lux 2020/11/14 06:25

On 11/14/20 3:41 AM, Reinier Gerritsen wrote:
> In my case it was the RF switch. I do not use long cables, just 1 meter.
> I'm developing 860-930 MHz antennas for RFID applications, touching the
> antenna all the time when adding or removing copper tape for tuning. I'm
> pretty sure ESD killed the switches.
> Two devices survived a couple of months, one just a few days. Amazing
> that all 3 units failed within 2 days, at 2 different locations 150 km
> apart and right after a major Windows update.
> I hope that adding protection diodes (ESD101B102ELE6327XTMA1) will
> prevent further failures. These diodes have a capacitance of just 0.1 pF
> (max 0.2 pF). The Skyworks SMP1330-085LF is not suited for this job, it
> does not survive ESD. Its purpose it to protect receiver inputs from
> transmitters. Their ESD rating is just 500V, the capacitance is 0.5 pF.
>

That's actually the Max, the Min is 250V.

Wow, that's actually lower than the RF switch which is HBM 1000V.

n2msqrp 2020/11/14 10:07

My antennas have a transorb and static dissipation resistor at the coax connection. It is a homebrew transient suppressor.

I would recommend installing a similar protection circuit; especially when using test equipment.

Mike N2MS

Stephen Laurence 2020/11/14 08:12

Well, my experience with car ignition systems seems to have got me nowhere. As I have said before, nothing can be guaranteed, but a basic discharge path in the form of a resistor seems potentially less detrimental to vna performance and must give a degree of protection.

When I have to dissassemble one of my V2 devices, I will see where best to add a sm resistor.  In the meantime, I will be careful, and avoid nylon carpets, make sure my soldering iron is grounded, not fiddle with long lengths of coax without a weather forecast, or measuring port1 through a 3db attenuator etc.

Out of my four V2 devices, I still have four of them without a failure.

Steve L

Neal Pollack 2020/11/14 09:09

I don't know why so many people treat delicate lab type test equipment and
semiconductors,
like it was general purpose consumer electronics like a mobile phone. If
your VNA was designed
for general consumer mishandling like a cell phone, the VNA would also cost
at least $750.
For less than $100, you are getting relatively raw circuitry with very
little expensive protection
and hardening.

Have you guys ever heard about simple basic ESD protection? Every single
commercial & aerospace lab that
deals with delicate instruments does, as do many hobbyists.

Hint:

https://www.amazon.com/Bertech-Wrist-Strap-Grounding-0-093/dp/B00HFQPR8M/

https://www.amazon.com/FEITA-Monitor-Constant-Wristband-Workstation/dp/B07WC3GCQZ/?th=1

Of course, you could choose to ignore it all and keep blowing up sensitive
devices.

Semiconductor devices today are made with ever decreasing small layers and
processes.
It sometimes only takes a few hundred volts, a level of static you will not
see or hear or feel,
in order to puncture a silicon layer or cause gradual latent damage that
shows up later.
At work, even in "good" humidity and weather, we don't even let you within
2 feet of a work bench
without an ESD wrist strap plugged into a ground alarm.

I have used the above grounding ESD mat and ESD wrist strap alarm for a
long time.
I don't suffer frustration from marginal or blown out devices.
Feel free to choose "proper" or "frustrating". No one will criticise you
:-)

Cheers,

Neal

Rey Machado Interte 2020/11/14 18:29

Well said Neal.

Rey Machado
G.M. Fiji and the Pacific Islands
Caleb Brett
Mobile +6421946156

________________________________

Reinier Gerritsen 2020/11/14 19:41

Sure, I know how to prevent ESD damage. You should take precautions when
handling bare parts or assembled pcbs. But when working on passive
antennas, there's little damage to be expected. I never had an esd
related problem with my measurement gear. I had countless hours using
oscilloscopes, my DG8SAQ VNWA and various SDR receivers. Never a
problem. And all the sudden I have 3 broken SAA2 and from the forum I
know of at least 2 other similar problems. There is just zero protection
right from the SMA input to the RF mux. It was probably overseen by the
developers. Maybe they live in an area where the climate is always warm
and humid and esd exists in theory only. Here in the Netherlands we have
to heat our houses and that leads to dry air. Especially in winter time,
the indoor humidity can be really low and that gives huge static
build-ups and corresponding huge discharges. By the way, the European CE
requirements includes ESD immunity. The SAA2 obliously would fail big
time. A € 0.10 transient voltage suppressor would probably fix it.
I really hope there will be improved devices soon that include esd
protection.

Reinier


Op 14-11-2020 om 18:09 schreef Neal Pollack:

Jim Lux 2020/11/14 10:52

On 11/14/20 10:41 AM, Reinier Gerritsen wrote:
> Sure, I know how to prevent ESD damage. You should take precautions when
> handling bare parts or assembled pcbs. But when working on passive
> antennas, there's little damage to be expected. I never had an esd
> related problem with my measurement gear. I had countless hours using
> oscilloscopes, my DG8SAQ VNWA and various SDR receivers. Never a
> problem. And all the sudden I have 3 broken SAA2 and from the forum I
> know of at least 2 other similar problems. There is just zero protection
> right from the SMA input to the RF mux. It was probably overseen by the
> developers. Maybe they live in an area where the climate is always warm
> and humid and esd exists in theory only. Here in the Netherlands we have
> to heat our houses and that leads to dry air. Especially in winter time,
> the indoor humidity can be really low and that gives huge static
> build-ups and corresponding huge discharges. By the way, the European CE
> requirements includes ESD immunity. The SAA2 obliously would fail big
> time. A € 0.10 transient voltage suppressor would probably fix it.
> I really hope there will be improved devices soon that include esd
> protection.
>

Likewise here in SoCal, where we mop the floor in the lab to try and get
the humidity up above 20%.

Transient protection is one of those things that is an afterthought when
bringing out the first prototype. One reason these things are
inexpensive is that they *are* basically prototypes that are still in an
iterative design process. The next version will be better.

In a non-ESD-rich environment, you can probably use it without damage
for quite a while - just normal handling of the boards would tend to
discharge the human body capacitance, to the circuit ground, and picking
up the load/short/open, likewise.

It might be that when you put it in an enclosure, it actually becomes
more vulnerable, because now it's insulated from the charge on your body
- just casually handling a "caseless" one like the original NanoVNA is
probably enough to make sure that the potential difference between
sensitive pins and the rest of the circuit is small.

Something to think about when setting up a test setup, connected to
things outdoors, where I won't be touching the coax in the process of
connecting it.

Neal Pollack 2020/11/14 11:02

I understand what you are saying.
I have a commercial antenna analyzer with far less frequency range,  it does not blow up from ESD when connected to the outdoor large antenna's on windy days, but it costs $1,000. USD.

It is likely that proper ESD input protection that will not degrade a 3 GHz frequency response is beyond the cost of a $75 USD NanoVNA.  I do wish the designers would stop focusing on sub-$100 price point, and give us a better machine for a fair price.   I would gladly pay more for better input protection, faster sweep speed, larger number of sample point (means more RAM and faster CPU), and better dynamic range.   In fact, for that I would be happy to pay $300.  Or even $375.   I think the market is driving requests and we are starting to see an evolution of design competition and features creeping upward, which is good.   Hopefully they notice all the threads about input protection.  No argument, it could be a lot better.    When you don't add input ESD protection devices on the PCB, the last thing you would want to do is use it for an outdoor antenna analyzer, since any amount of mild breeze/wind is going to generate insane static potential very quickly.

Cheers,

Neal

Stephen Laurence 2020/11/14 13:03

Well, I am following all this with interest...(and some added care of my four V2 vnas).

Neal, you must have also been lured by the $60 trap to invest in one of these devices, otherwise you probably would not be reading these posts.

I am sure the designers of the vna concept are also following this thread, and your wish for a $60 device with the robustness of your $1000 unit will be realised soon. In the meanwhile, the attraction of a device costing so little has convinced many tens of thousands to invest, experiment, learn and maybe throw away in favour of buying and using a more professinal piece of kit with the knowledge they have gained.

I went through such a cycle in the 1990’s when I bought a kit of parts to make a simple 0-120mhz spectrum analyser. I was then stimulated to buy three(!) Hewlet Packard 141 boat anchor professional analysers, which I hardly ever used as they were huge, heavy and took ages to warm up. Anybody want them? Otherwise they are heading for the rubbish tip. I have the Tinysa instead, for the price of half a tank of petrol for the car. If I blow it, I will buy another.

Steve L

Siegfried Jackstien 2020/11/14 21:18

now ask bill gates if he pays you the switches ... it was sure the
windows update (just kidding a bit) :-)

greetz sigi dg9bfc

Am 14.11.2020 um 11:41 schrieb Reinier Gerritsen:

Clifford Heath 2020/11/14 15:30

Nevertheless we've used these Skyworks limiter diodes extensively in mobile foxhunting applications for years with no failures (they were common before). Bashing through forests and climbing onto the car roof to change antennas in the middle of the night, driving the Yagi right adjacent to transmit antennas with 100 watts or more up the spout... no failures. So I'm pretty confident that it works.

Peter KA6Z 2020/11/14 16:22

Fascinating issue. There's a "get-what-you-pay-for" aspect to the VNA2 ESD issue as others have highlighted. What we save by paying $70 we could lose with further costs of time and components with a problem like this that could have been avoided with further engineering that would have made the price $150-$200. So there's no free lunch but good this group can help preserve our $70 investment.

In the meantime I'm sitting here with a used 20 year-old Tektronix oscilloscope bought for ~5% of its original cost, and it keeps working regardless of what I or whoever owned it before me throw/threw at it. But even at 5% it still cost $500! And if I look at the schematic my eyes glaze over at the engineering that went into it to justify the price. Expect it will last another 20 years, while by then I may be on by n'th NanoVNA .... !

Neal Pollack 2020/11/14 17:21

We sure, there is no free lunch. But for lower frequency hobby use, below
3 GHz, I would really rather pay
$100 and own a few spares, than visit Keysight, bend over and squeel, and
do a 30 year mortgage.
The Keysight I use at work costs right about $200,000 USD, and then the
calibrator costs another $25,000 USD.
Each SMA port test cable is $3,250.00 (Four of them) (Google up Keysight
PSA-X prices, and weep....)
Oh, and it seems to have no issue with ESD.
But if you cross-thread or dent a front SMA connector, they will gladly
sell you another one for $3,495.00

[Very clearly, when getting started, Keysight must have met for lunch
with the Ferrari family, and the De Beers
diamond business (Oppenheimer family) to learn how to separate a fool from
his money.]

At home I don't work with anything over 1.5 GHz, and mostly use a VNA for
filters and antenna's and tank circuits.
The VNA V2 is JUST FINE for that, even if I have to buy a new one each
year. :-) :-)

Cheers,

Neal

On Sat, Nov 14, 2020 at 4:22 PM Peter KA6Z via groups.io <p.knazko=
yahoo.com@groups.io> wrote:

Peter KA6Z 2020/11/14 19:04

Hi Neal -- It was just a way to illustrate in my awkward way that engineering costs money. As you said, Keysight will have solved any ESD issue for that $200k. Woo hoo. But obviously, I use my VNA2 because I don't have the cash for the next step of VNA which seems to go straight to >$1000. So it's good we have each other, you know, to get us through the challenges of under-engineered equipment like the nanoVNAs, at least until I find a 20 year old Tektronix VNA for under $100.

To reply to this topic, join https://groups.io/g/NanoVNAV2

View this thread on groups.io