开云体育

On 7/21/21 10:35 AM, Manfred Mornhinweg wrote:

Well, I did the test for you! :-)

I used my small clothespin-based test jig, clipping a calibration strap and a 49.9? 1% 1206-size resistor to it, calibrating from 10kHz to 900MHz, then measuring a 100? 5% 1/8W carbon film resistor just like yours. Photos and test results are attached, showing the calibration resistor, the clothespin holding it down, and the resistor being tested.

As you can see, the resistance is pretty flat up to a little above 500MHz. Inductance seems essentially zero, in fact some slight capacitance appears first. Of course the resistor must have some inductance, but since its pins are clipped to the test jig in such positions that some of the jig's copper ends out of circuit, the resistor's inductance is replacing some of the jig's inductance and that of the calibration strap and resistor. In practice this means that the resistor has pretty much the same inductance as a board trace of the same length.

There is a resonance at 600MHz. I would need to analyze where this is coming from. But it's not very strong. And keep in mind that I used a 1206 size calibration resistor, and a simple piece of wire, for calibration. Using a physically smaller calibration resistor, and a flat shorting strap, would maintain accuracy to a higher frequency.

Based on this test, I would say that 1/8W leaded 100? carbon film resistors are really very good at RF. SMDs are better, but the leaded parts are very usable into the hundreds of MHz. And clip leads aren't! :-)

We always knew clipleads are bad, the question is really "how bad is it"

Of course, 100? is inside the sweet range. Do this with a 1? resistor, or a 100k? one, and the results will be very much worse.

Very much so.?? But for folks building 50 ohm (or 75 ohm) loads, that 1/8 watt carbon film is probably going to be just fine.

Larry & DiSlord,

Here is what I did:
I did another "clearconfig 1234," re-installed FW .64, and did another "clearconfig 1234," via serial connection. Then I did a "touch cal" and saved. I can freely navigate with the stylus, but vna freezes as soon as I do ANYTHING with the switch or lever ..... Same as before.

Then I removed top bezel and bottom panel. Traces on top of board are covered by the display and can't be inspected. Chips on bottom of board were carefully inspected with magnifier. Nothing found that might be a short, or anything else suspicious.

Before replacing the panels, I fired up the vna. At that point the lever was working, albeit very erratically. I was able to move the markers, but jogging the lever one step to the left (or right) caused the marker to first move one step then reverse direction and continue moving in the opposite direction. After a few more jogs of the lever, the vna would freeze.

If the original FW was not working, I would suspect a possible hairline crack. But the original FW works perfectly. At this point, I think it is time for me to just use the nanovna-H with the stylus. I was thinking of upgrading to an H4 anyway.

Thank you for working with me on this issue.

Jack K1VT

Personally I wouldn't risk trying to solder wires directly to the CPU. It looks like you have one of the older units (not necessarily a clone). I believe the P3 connector was added on the 3.4 revision of the PCB. I have the 3.3 and it doesn't have the port either. The version number is printed underneath the battery on mine. You are probably better off getting the H4 unit which will make life a lot easier to add the bluetooth module to it as there will be more room to work with as well.

That is just a simple pull-up resistor to Vdd and will not cause a lock-up. It was changed to 1K to fix erratic control of the jog sw.
It can be fixed very simply by placing a 1.2K chip in parallel with the 10K. I did that with my DFU jumper mod so my 2.8" Nano goes into DFU the same way the 4" unit does.
All the original NanoVNA and H units earlier than 3.4 had a 10K and none of them have a lock-up issue, so it may be a 'sensitive' or even counterfeit uP chip. I think the jog sw was changed to interrupt control from polling in the original firmware (correct me if I'm wrong) and there may be an issue with the interrupts on the chip.

Jack should check the Vdd line to ensure it is 3.3 volts and not higher or lower as that would cause issues.
Jack should also try installing FW from edy555 (January 2021) and Hugen (June 2021) and see if there is still an issue.

Larry,
remote repair is a difficult thing because if I see the thing in question directly I will look at more options.
For v3.3, R5 = 10k, this changed to v3.4, it changed to R5 = 1k.
I didn't look back on the cause, but it may be the cause.

73, Gyula HA3HZ
--
*** If you are not part of the solution, then you are the problem. ( ) ***

Well, I did the test for you! :-)

I used my small clothespin-based test jig, clipping a calibration strap and a 49.9? 1% 1206-size resistor to it, calibrating from 10kHz to 900MHz, then measuring a 100? 5% 1/8W carbon film resistor just like yours. Photos and test results are attached, showing the calibration resistor, the clothespin holding it down, and the resistor being tested.

As you can see, the resistance is pretty flat up to a little above 500MHz. Inductance seems essentially zero, in fact some slight capacitance appears first. Of course the resistor must have some inductance, but since its pins are clipped to the test jig in such positions that some of the jig's copper ends out of circuit, the resistor's inductance is replacing some of the jig's inductance and that of the calibration strap and resistor. In practice this means that the resistor has pretty much the same inductance as a board trace of the same length.

There is a resonance at 600MHz. I would need to analyze where this is coming from. But it's not very strong. And keep in mind that I used a 1206 size calibration resistor, and a simple piece of wire, for calibration. Using a physically smaller calibration resistor, and a flat shorting strap, would maintain accuracy to a higher frequency.

Based on this test, I would say that 1/8W leaded 100? carbon film resistors are really very good at RF. SMDs are better, but the leaded parts are very usable into the hundreds of MHz. And clip leads aren't! :-)

Of course, 100? is inside the sweet range. Do this with a 1? resistor, or a 100k? one, and the results will be very much worse.

Jack,
Grab a magnifying glass and check the board for any small shorts between the pins of the micro or DSP chips. As the firmware has matured, some of the formerly unused pins are now being used.? Hopefully, you don't have a wonky uP chip.

I assume you've done a "clearconfig 1234" before and after flashing?
It's not a HW design issue - otherwise we would have seen a lot of forum messages regarding this.
...Larry

On 7/20/21 5:53 PM, KENT BRITAIN wrote:

Without a balun, the coax shield is also part of the antenna.
Got any Ferrite beads?? ?Big ones that you can slide over the coax, or?better yet some of the clip on ones often used on AC power cords?
Clip one or two of those on the coax near the antenna and I bet both plots will be very similar.

Also, are you physically next to the antenna when measuring at the antenna? At 350MHz you are a really big interfering structure.

-kb, another Kent.

On Wed, Jul 21, 2021 at 06:23 PM, Jack Mandelman wrote:

My hardware version is H v.3.3.

I searched for the v3.3 hardware schematic and found a photo of it.
If you help with debugging, I can provide that.

73, Gyula HA3HZ
--
*** If you are not part of the solution, then you are the problem. ( ) ***

I don't really understand why the hang occurs. Let's take the steps. In normal mode, a marker is selected on the screen (on the screen it is > at the value of the marker, see screenshot and red square), which means that with the lever you will move the marker left-right, and the menu should be called up with the central button. If the marker is on the left side of the screen and you click to the right, what happens? Does the device just freeze? Or will the marker run to the right and freeze there?

On 7/21/21 9:12 AM, Manfred Mornhinweg wrote:

That resistor seems to be an 1/8 watt carbon film type.

The lion's share of the inductance you measured is in those clip leads. The resistor most likely has less than 10nH, if you measure it with short leads, as it would normally be mounted on a circuit board.

And I would suspect that the rising resistance is more due to your lead inductance approaching resonance, than to real resistance increase with frequency in the resistor.

You would really need a much better test jig to measure such a resistor in a reliable way. Such as a small piece of circuit board with a gap in the middle, where you do the calibration by soldering a shorting strap over the gap, leaving it open, and soldering a small (0805 or smaller) 50? SMD resistor across the gap, then soldering your test resistor across the gap, with short wires, all at the exact same location.

Yeah, I've been thinking about an easily fabricated jig.? This was more an experiment to see just what happens when you don't do anything special.

I'm not sure about resonance - I guess I could sweep wider and see if it gets that characteristic resonant bump.? It's not exact a series or parallel resonant circuit.. there's C between the leads, L in the leads and the resistor, and R in the resistor.? I guess it could be modeled as a C in parallel with a series R and L.

The X term is always positive and gets bigger as frequency goes up, so it's not like there's a C that's cancelling the L (as would be the case for resonance)

And it's pretty close to linear with frequency - 25MHz is 34 ohms, 50 MHz is 71 ohms, which is almost exactly twice, which is what you would expect for a pure inductance. (which the clip leads, being the dominant inductance, almost certainly are)



But, 100 MHz is still 3 meter wavelength, so it's not looking much like an antenna, at least.


Maybe tonight, I'll try taping the leads down, close together, which minimizes the L, and doing a short/open measurement.



This is one of those little experiments that a cheap VNA enables.

Firmware 0.2.3 (original FW) works perfectly.

The following FW versions that I tried have the freeze problem:
0.3.0
1.0.26
1.0.45
1.0.64

My hardware version is H v.3.3. Could hdw v.3.3 have problems with the more recent FW releases? If I want to use the latest FW, I will have to live with using the stylus to navigate the menus.

Jack K1VT

That resistor seems to be an 1/8 watt carbon film type.

The lion's share of the inductance you measured is in those clip leads. The resistor most likely has less than 10nH, if you measure it with short leads, as it would normally be mounted on a circuit board.

And I would suspect that the rising resistance is more due to your lead inductance approaching resonance, than to real resistance increase with frequency in the resistor.

You would really need a much better test jig to measure such a resistor in a reliable way. Such as a small piece of circuit board with a gap in the middle, where you do the calibration by soldering a shorting strap over the gap, leaving it open, and soldering a small (0805 or smaller) 50? SMD resistor across the gap, then soldering your test resistor across the gap, with short wires, all at the exact same location.

One problem I have noticed in configuring the HC-05 is that if you type in the AT command and any parameters too slow, it doesn't work sometimes.

AT command expects "fixed" format , my guess would be missing / wrong "timeout " or command length option or check sum etc. . But I would have to check the EIA / AT standard to make sure - so do not quote me.

On 7/21/21 7:46 AM, kk7xo via groups.io wrote:

From the picture it looks like a carbon film resistor.
When I measured leaded resistors in the past using clip leads, I received significantly different results just by moving the clip leads closer to the resistor or further away on the resistor's leads.
de KK7XO

Sure - that will change the L (farther away, more L, about 1 nH/mm) and, as well, it would increase the area of the loop formed by the leads (bigger L, as well).? And less C between the leads.

As for the kind of resistor - it's more than 40 years old, probaby Mepco/Electra brand (since I bought a bunch of boxes of resistors in the early 80s, and that's probably from those), loose in a compartmented tray of leaded resistors on my bench. I *might* have the paperwork from the order buried, but I'm not going to go look for it <grin>.

OK, getting off subject after I said it is too early to start talking abut software....

However, I do have reason to ask about configuring the I/O pins.
I am still thinking about hooking up HC-05 to MY nanoVNA which does not have dedicated P3 pins.
To do that I need some room to add wires directly to the CPU and if such room is not available near CPU pins 30/31 I'll will be looking for other pins.

BTW
I agree that EIA can be intimidating , but after all it is all about options, no magic involved.

I am not sure I share concerns about I/O speed, my first encounter with the EIA was @ whopping 110 bauds.

Years ago I got professionally involved communicating with a device of unknown options - so one of my "for hire " C code was to
run thru all of the options until intelligent / readable response was received

..and if the communication is too slow , then there is more time for cup of coffee...

On 7/21/21 7:32 AM, DiSlord wrote:

serial interface can be implemented in software mode, just use pin bitbang. But this mode slow, need lot or CPU resources (all tx, rx time need control this pins)
you can use software mode only if you CPU not do anything else, only read and send data (on slow speed), and this really stupid idea not use hardware.

Most cpu have hardware serial modules, not possible select any wanted pin for it.

Exactly.. for the particular microcontroller, USART1 connects to those pins, and no others.

In a "perfect world" there would be a detailed interface control document that says what Tx and Rx (in this specific case) means, what the voltages are, what the currents are, timing parameters, control registers, etc.? For now, though, it's left to the user to read and understand the schematic, and potentially, the part datasheet, to get all that info.

That's why it costs $50 and not $5000 <grin>

I've gone down the rathole of lots of configuration options, and it turns out that

a) most of them are never used

b) it takes more work to explain them, and have configuration files, than to just not make it configurable.

One nice point here is that there are open source versions of the firmware, so if you're "ambitious" and "motivated" you are "free" (in the speech, not the beer, sense) to implement something else. I could imagine someone implementing an RS485 style multidrop interface to use multiple VNAs on one control line, and not having to change the board layout.

From the picture it looks like a carbon film resistor.
When I measured leaded resistors in the past using clip leads, I received significantly different results just by moving the clip leads closer to the resistor or further away on the resistor's leads.
de KK7XO

serial interface can be implemented in software mode, just use pin bitbang. But this mode slow, need lot or CPU resources (all tx, rx time need control this pins)
you can use software mode only if you CPU not do anything else, only read and send data (on slow speed), and this really stupid idea not use hardware.

Most cpu have hardware serial modules, not possible select any wanted pin for it.

OK Larry I will do so one of these days

Op 21-7-2021 om 16:08 schreef Larry Rothman:

--
Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware.