开云体育

Forth is a niche language created for writing embedded software on very limited, bare HW. In particular, it is intended to allow you to write and test peripheral control software interactively. The STM32F072CBT6 is 2-4x the memory size of the machine Chuck Moore created it on to control the Kitt Peak radio telescope dishes and many times faster.

With so many people working on variations of the original C code. I'm not sure there is much point to my getting involved with that. Also, I think the use of threads is a major mistake from a software engineering perspective.

I have major issues with bugs and even more with code crashing, ever. Once you allow multiple independent processes to operate in the same address space, unless you use a segmented architecture al la Multics and the Intel x86 and 432, you leave yourself open to one process corrupting another.

In any case, I'm just an old man amusing himself. I've been fooling around comparing the TDR response of my 11801/SD-24 using a bare 3.5 mm connector and the nanoVNA open. There is as much difference between the two channels in both cases as there is between the bare 3.5 mm and the open.

Not quite sure how to interpret it yet. The trace calculated differences are much larger than the display suggests. That's symptomatic of small phase differences, but it's been a long time since I dealt with that and I don't remember yet how you handle it.

Have Fun!
Reg

I tested different SMA connectors with hand connection with no wrench and don't find any noticeable difference.
Probably because NanoVNA works with less than 1 GHz.

Does somebody have any confirmation that using wrenches improves measurement with NanoVNA?

On Fri, Sep 20, 2019 at 03:46 PM, Andy G0FTD wrote:

2 - Wouldn't a diode introduce capacitance across Ch0 ? Or inductive lead
effects at VHF/UHF and ruin any measurements ?

Yes, limiter diode will ruin measurement. There are two issues with diode:
- it add parasitic reactance
- it add non-linear distortions of the signal

There is no simple way to protect RF device with no deterioration of it's dynamic range.

For antenna gain measurement you're needs to place receive antenna at least half wavelength away from transmitting antenna. Otherwise your measurement will be affected by near field of antennas.

Actually the more complete way to calculate minimum distance between antennas is the following:

R = 2 * D^2 / lambda,
where:
R - minimum distance between antennas in meters
D - maximum dimension of your antenna in meters
lambda - wavelength in meters (lambda = 300 / frequency_in_MHz)

If possible place antennas at more longer distance, for example at least on 2-3 lambda. It will reduce near field influence.

Also calibrate NanoVNA through cables which you're using to connect antennas. It will reduce cable influence.

On Fri, 20 Sep 2019 at 13:57, Andy G0FTD via Groups.Io <punkbiscuit=
[email protected]> wrote:

One other thought.

During a discussion yesterday on this group, measurments were taken as to
what the OUTPUT levels were from Ch0.

They were 200mv peak to peak IIRC.

As such, I think I'd want to assume that this would be the maximum level
that any of the ports could handle and would want
to account for.

0.6v (600mv) for a basic diode to start clamping sounds a bit risky to me.

If I'm missing something here then I'm willing to learn ;-)

73 de Andy

The output level of a source is not an indication of the damage threshold.
My HP 8720D can output 10 dBm, but the damage threshold on the test ports
is I believe 30 dBm, so 100x higher.

--

Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@...

Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales.
Company number 08914892.
Registered office:
Stokes Hall Lodge,
Burnham Rd,
Althorne,
Chelmsford,
Essex,
CM3 6DT,
United Kingdom

I found some finger "wrenches" here:


and a torque limiting "wrench" here:

On Wed, 18 Sep 2019 at 21:39, Rune Broberg <mihtjel@...> wrote:

Hi David,
I don't necessarily intend to break new ground
here, and I haven't yet looked at what's the industry norm for averaged
display.

Do you have suggestions?

--
Rune / 5Q5R

I will have to look for it, but a description of the method used by HP was
posted on the Keysight forum by Dr. Joel Dunsmore. I recall him saying it
is very memory efficient as it does not require one to store all previous
values That was important when early versions of the 8753 were developed,
but far less so now.

The averaging must be done as vectors, not scalers. I will try to find out
how the professionals do this.

Dave


--
Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@...

Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales.
Company number 08914892.
Registered office:
Stokes Hall Lodge,
Burnham Rd,
Althorne,
Chelmsford,
Essex,
CM3 6DT,
United Kingdom

I've laser-cut some from 3mm acrylic sheet. Black for 0 ohms, green for 50,
clear for open.

John

Hi Jim,
If you could send me the calibration file, I'll have a look at it.

Mihtjel@...

Thanks for taking the time to report the issue!

--
Rune / 5Q5R

Hello Reg,
you can also print them by your own 3D printer.
See

I printed them with PLA in 102% size to fit tight.

73, Rudi DL5Fa

On Thu, Sep 19, 2019 at 02:43 AM, <hugen@...> wrote:

I am discussing with the AA6KL9 to build a new firmware using the
STM32F303CCT6. If all goes well, it can be provided later. To replace the mcu
directly on the current hardware, you need to add a 1.5K resistor to the VCC
on the USBDP Pin. The DFU mode of the STM32F303 requires an external 8M
crystal to use USB, but STM32F072 does not need it, so you need to use SWD to
update the firmware. You need to prepare an ST-Link or DAP-Link.
BTW: The new STM32F072-based NanoVNA firmware already supports TDR.

Thank you!

ok, 8M crystal is normally not a problem for me, but for most nanovna owners F303 is then not real 1:1 replacement. We should probably, if we need more memory, stick on crystal-less USB STM32 (AN4879) , with pullup support, with DFU support (AN3155), LQFP48 package and ChibiOS support. Afaik STM32L433CCT6 should be good replacement, only pullup for USB necessary, and some ADC / clock / I2S changes to code, but it does need newer ChibiOS, which then does not compile with nanovna. Anyway, for myself i will probably solder crystal and use F303 (ADC/clock changes necessary), if i not manage to fix newer ChibiOS compile Errors with STM32L4x series.

tinhead

Bryan,
The lab I worked at had spectrum analyzers and receivers that cost more than my first house. To protect the equipment our laboratory manager purchased a set of HP 11867A limiters ( I'm guessing you could buy 5 or 6 nanoVNA's for what he paid for one of those). I believe the limiter was just a couple of well matched, low capacitance, fast switching diodes packaged in a Type N adapter. It turned out that the limiters were a waste of money for our application because diodes are non-linear devices and sometimes would mix with the signal we were measuring and generate phantom signals. For tuning an antenna that extra noise might not be a concern, but for compliance testing it was a deal breaker. We eventually ordered some Type N adapter RF fuse holders that had less than a dB of loss up to 1.5 GHz. The holders had fusible links that could be replaced in-house without having to send the equipment out for expensive front end repairs. I was the beneficiary of their use more times than I care to admit.

Never tried either solution with our network analyzers as I certainly would have been fired on the spot for attempting to use an HP8753 VNA in the scenario you are envisioning. Makes you wonder what kind of front end protection expensive antenna analyzers use.

Yeah, Reg - Python is the future....!

Actually, Python is close to becoming the #1 programming language - educators, engineers and scientists have all created extensions.
Talk to forum member Rune about Python.

Also from today's news, Microsoft has a YouTube series for beginners:


From the article:
"A new video series, titled 'Python for Beginners', hosted by Microsoft Senior Program Manager, Christopher Harrison, and Business Development Manager at AI Gaming, Susan Ibach, has been launched by the firm. Comprising of a total of 44 videos, this series - as indicated by the name - isn't an in-depth study of Python, but rather, it just provides a foundation for the programming language. The course material is noted to be quick and concise, in order to encourage learners to start specifically looking into the areas that interest them as soon as possible."

Cheers,
Larry

Folks,

Thanks for all the comments. I notice that the diagram provided by erik does not indicate any protection. Thanks for that diagram. The 50 ohm bridge mentioned above does provide some help but only a small amount. ESD events are often in the thousands of volts so even a few dB of attenuation is not enough. I have previously blown up a VNAs input switch when I touched an antenna connected to the VNA in my workshop so this is a real concern for me. Yes, it was repairable. I got to learn a lot about soldering very tiny surface mount components in that case.

Back-to-back zener diodes are available for this very purpose with very low capacitance (0.06 pf) for example the PGB010603 made by Littlefuse. I have attached a data sheet for others to consider. The spec sheet says it is good up to about 9 kv for a contact discharge and about 15 kv for an air discharge. Furthermore, the response time is under 1 ns which is about 1 cycle at 900 MHz. I suggest the mixer in the nanovna won't handle a typical ESD event without some help. I purchased some of these devices for about $0.50 each so they are not a big investment. Please be careful interpreting the data sheet. You might conclude these diodes don't help, but they have in my case.

I hope that when we do an open calibration any stray capacitance such as 0.06 pf is measured and removed. 0.06 pf at 900 MHz is a reactance of about 2950 ohms which should not have a major impact on most measurements anyway.

I hope folks find this information useful.

--
Bryan, WA5VAH

On Fri, 20 Sep 2019 at 16:18, Reginald Beardsley via Groups.Io <pulaskite=
[email protected]> wrote:

Dave and Ohan,
I'm rather tempted to load Mecrisp on the nanoVNA and try my hand at
serious work in forth for the first time in almost 40 years.

Have Fun!
Reg

The problem with forth is that nobody else is likely to want or be able to
contribute. There are many languages that have niche areas - Lisp, Haskell
etc. But the number of people knowledgeable in these languages is a small
proportion of the people able to write software.

Dave

--
Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@...

Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales.
Company number 08914892.
Registered office:
Stokes Hall Lodge,
Burnham Rd,
Althorne,
Chelmsford,
Essex,
CM3 6DT,
United Kingdom

Thanks for posting that detailed test report...fascinating to see what the ultimate ceiling might be. I recall reading that the mixer chip alone has an upper spec. limit of 500 MHz, so the fact that we can get ANYTHING >1 GHz is impressive.

Sometimes people ask if a nanoVNA can be hacked to work above 1500MHz.
So I tried. The firmware was modified so it theoretically can go up to 2.1GHz (7/9 overtone mode)
But is it usable with the supplied plastic cables?
Measuring a 2GHz cavity filter the S21 dynamic range is 10dB with a noise of about 5dB so with 1/16 exponential averaging using TAPR VNA PC software you get confirmation the cavity filter is 20MHz wide at 2GHz but due to the long measurement time (about 2 minutes) to get rid of the noise, tuning is impossible.
I could calibrate (1000 points from 0.5MHz till 2.1GHZ but also 101 point in the device) and the difference between S, O and L is visible at the right place of the Smith chart but the noise is at least 10% and sometimes much more and the stability is terrible.
Measuring a 1.9GHz low-pass filter was impossible. Both the S11 and the S21 where basically useless above 1.5GHz, probably due to leakage, unbalance in the bridge and other parasitic capacitances.
To be sure I repeated the tests with semi rigid SMA cables instead of the plastic SMA cables but there was no visible difference.
Further analysis revealed the bridge has still 10dB directivity above 1.5GHz and the S21 dynamic range is still about 30dB but due to leakage, impedance mismatches, bridge imperfections and temperature instability it will be difficult to use this.
Can you tune a 2GHz antenna? Although it is possible to see the L/S/O on the smith chart on the right position it was impossible to determine the input impedance of the 2GHz cavity filter so it may be difficult

I had the same problem and to fix it give the file a shorter name and then upload it.
It then worked for me.
73, Trip

--
"When the people fear the government, there is tyranny. When the government fears the people, there is liberty." ~~ Thomas Jefferson

Hi, Folks,
My whole NanoVNA process started because I went out on eBay looking for an antenna analyzer for ham VHF/UHF use and got my white model because it intrigued me at what was being done at that price point. No prior personal experience with VNAs, although I've been aware of them for years. Big learning curve to get started, but it did almost everything I was looking for once I could see SWR and a Smith Chart. It's been well worth the $60, and I'm getting a VNA education to boot reading here.

I say almost because I would like to be able to get some values to at least compare HT Antenna A to HT Antenna B for gain. To get started, I put a test antenna on CH0 and a reference antenna on CH1. I can tell the CH1 antenna is "seeing" the CH0 antenna because logmag dB values change with/without something on CH0, and I can also see that dB values go up/down as antenna distances change.

What I'm looking for is suggestions on how to do this the right way, or at least a better way. Now I'm using a scanner-type antenna for reference, it's about 30cm/12" away from the test antenna. I'm looking for suggestions on ref antenna type/spacing, and also a calibration protocol to make this choice useful. Nothing really fancy, I just want to get some data so when someone shows up with what they think is an antenna that's cooler that what they (or someone else) already has, I want to have some numbers to avoid a food fight and/or be able to objectively evaluat/comapre the two. Suggestions? Issues?

Appreciate the input...

Ken Buscho
KB6KOB

Rune,

Thanks so much for doing this. Did notice one minor bug. When calibrating out to 1.5 GHz and loading the calibration file in, when starting a sweep it will crash. If only going out to 900 MHz it works fine. Windows 10 system using your provided exe.

Thanks again for all your effort on this.

Jim K.