For interest sake, I ran the new test...

Yes, that is good
--

开云体育

Eric:
The test on my unit seems to start at 29dB of attenuation.

charlie

On Tue, Jan 19, 2021 at 11:53 PM, Erik Kaashoek wrote:

Version v1.1-77

Added some pictures to the wiki page as measurement examples


?
--

New FW released
Version v1.1-77

Changes:
- MEASURE/MORE/LINEAR added

The linearity test requires an input signal between -40dB and -25dB and a single marker positioned on that signal.
Activating the linearity test will step the built-in attenuator through all attenuation values and draw a green line on the screen at the measured signal level.
If all goes well the line should be flat within +/- 1dB with an occasional measurement deviating due to interference/noise

The test also helps in getting confidence about the 1dB compression point of the tinySA. This can be measured by applying a signal of 0dB and activating the MEASURE/MORE/LINEAR test
The compression starts when attenuation is below 5dB and is visible in the green line being curved down a bit at the right side.

--

It would help us if you could install the a new FW. V1.1-77 and run an additional test

Connect high and low with SMA cable
Set CAL OUTPUT to 30MHz
Start MEASURE/MORE/LINEAR

A green line should appear and the attenuation will start at 30dB and step by step decrease till zero
When completed (will take some time) with a good attenuator the screen should look like this
The line does not have to be perfectly flat.


Please post a picture of the screen with your tinySA


--

?Received Tiny SA from R&L last week. After calibration only attenuator test 11 failed.
Did manual 30mhz attenuator test as suggested.
Came with v 1.1-52 Dec 14 2020. Not updated

1 db step failed at 4 db at 28.9db
Started with 25.4db fro 0 to 3db.
Noise floor did creep up as value increased.

Was hoping a newer version may fix it but not tried yet. Generated some test signals from an a calibrated Aeroflex service monitor? and mos levels appeared within 1 db or so at different levels, I think I had manual atten at 0 when I did most of those. Only at higher levels -10 to -5db was accuracy on screen varying 5db or more and at very weal levels near -100db and at frequency extremes of high and low inputs.
I can swap with vendor if needed.

? ?Thank You
? ?Tom

On Tue, Jan 19, 2021 at 09:54 AM, Erik Kaashoek wrote:

... There is already such an attenuator linearity test in the code. But disabled at the moment.
It steps through all 64 attenuator positions (0.5dB per step) ...

While playing around with some code to also step through the 64 attenuator positions,? I noticed that 0.5 dB steps can be entered from the on-screen keypad but only only 1 dB steps can be programmed using the 'attenuate' command (i.e. attenuate 15.5 displays Atten: 15dB on the tinySA status display.? Is that an intentional limitation??

- Herb

I did some thinking here, plus some math and testing, then I found an excellent youtube from W0QE describing all that was asked here :) The only reason NOT to check his presentation would be the time - two videos (Part 1 and Part 2) are 30 minutes each, and are well worth time spent!




For some of you too busy to check them out, here is the digest version :) Even if we assume that my digest will tell you everything you need to know, still go and check Larry's posts!!!

Let's assume you want your tap to cover some extreme cases - from 1W to 1500W. You also want to get signal from the tap that you can take to your instrument (tinySA for example) that's 50ohm. This will drive the need attenuate your signal by 40dB (too much and your low power will be to small, too little and your legal limit will fry your instrument). Without too much explanations (see Larry's Part 1), you will end up with resistor divider 2500ohm/50ohm. Given that 1500W@50ohm->270Vrms,?your 2500ohm will need to be 30W. Too much :)
If you scale down your expectations to use tap for up to 100W, you will deal with 70Vrms, and resistor will be 2W, more manageable.
NOTE: all the numbers are for measuring where SWR is very low. Increase SWR, voltage goes up.

Larry is showing examples where you can use Cap divider (or inductive) so you eliminated power dissipation only related to resistors, but both have quite frequency dependent "behavior".

In Part 2 he explains inductive coupling (the current transformer), impact of stray capacitances, sizing, ... I especially liked the part with "how to minimize stray capacitance" and different design approaches to ""extend the shield"

He also touched on impact of SWR, and that applies to all cases (R/L/C/inductive).

Through both parts he is EXTENSIVELLY using SimSmith - if not familiar, don't be scared, he's explaining what's being done.

If you like his videos, check other of his stuff, i just finished his video on designing CMC (Common Mode Choke) and finally understood some of stuff that was always in front of me, but never clicked!

I just need to find more hours in a day and to check all that's on his youtube channel :)

So, to SUMARIZE:

• resistive taps are quite OK, but come with significant challenges related to needed power rating of a resistor used if you are constrained with 50ohm output?
• capacitive and inductive dividers are not practical
• inductive coupling, when done correctly, is the best design to cover wide range of power and frequency
• neither (divider/coupling) should be used to measure power if SWR is not very low

?

There is already such an attenuator linearity test in the code. But disabled at the moment.
It steps through all 64 attenuator positions (0.5dB per step) and if all is well it should draw a perfect straight line on the display.
Will see if I can re-enable it.

The attenuator is solid state so it should not deteriorate over time. The early FW used in manufacturing did not test all attenuator positions as I was unaware of the less than zero fall-off rate of the attenuator.
In the current production all attenuators are tested to eliminate these? bad attenuators but there could be post production stress caused by low temperatures during air shipment.

On Mon, Jan 18, 2021 at 11:46 PM, Erik Kaashoek wrote:

It seems your attenuator is bad.
You can verify this by enabling the cal output at 30MHz (MODE/Cal Output/30MHz) and putting the tinySA in low input mode with center 30MHz and span 10MHz.?
Set LEVEL/ATTENUATE/MANUAL 0 (0 dB attenuation) and write down the level of the 30MHz signal
Then increase the attenuation in 1dB steps and see how the noise floor moves up and the level of the 30MHz signal stays the same.
If the 30MHz signal level suddenly makes a jump up or down when changing the attenuation write down the attenuation levels before and after the jump and post your observation.

Erik,
? To take the drudgery out of manually entering 31 attenuator values, would it be possible to either add a module to TinySA-App (Settings?) that performs an extended attenuator test, as outlined by yourself above, or write a separate executable that performs an extended attenuator test?? An extended attenuator test would also help hugen to detect a device with a borderline bad attenuator in the QA process before shipping it.

? Does the manufacturer? specify the number of operations the attenuator can be switched (MTBF?) before it needs to be replaced?

? I can throw something together for myself in C#, but my programming skills are not good enough that I would trust publicly releasing an executable.

- Herb

Thanks.
The attenuator seems bad. Suggest you contact your supplier for resolution while referring to this communication.?

8DB. the other picture is 7DB

what attenuation is the picture with the marker at -29.5dBm?
--

OK Erik.?
I think I did the test correctly.?

On 2021-01-14, at 10:26, Carsten Bormann via ?<cabocabo@...> wrote:

Banggood now also sends clones (at least from their CZ warehouse).

I’ll try to get a refund; if the clone is illegal, I cannot send it back (because that would be illegal!).

I’m still in the process of trying to get a refund from banggood.

Thank you for the help.? I will follow up with Zeenko store.

Clear.
You can contact Maggie from the Zeenko store on AliExpress and ask for a solution, refer to these messages as evidence.
--

I went through the 1dB steps between 0 and 30 and this seems like the biggest change:

Good suggestion
--