Choosing a good HDMI adapter

[whacky]

Continuing the discussion from 1090 MHz interference from HDMI adapter?:
With home use of USB C-HDMI adapter on the rise, I’d like to share my findings on this subject. (Very) long story short, I ended up buying two adapters (C2 and C3) in attempts to replace the old one my office IT furnished (C1). All adapters are not created equal.

Now, I get some verifiable results. The tests are conducted during day time when “natural” noise is high. Each adapter is turned on for 20 minutes (as marked in colored time band), then off for 10 minutes before testing the next. They all connect to the same monitor and the same USB C port. The adapter is placed 1 m lower (39") than the magnet-base wire antenna and 0.5 m off its axis (19"). The antenna is connected into FA blue dongle without filter.

As the above charts show, on-cycles of C1 and C2 each raises noise floor by 6 dB, but that of C3 shows no measurable difference from background noise. (Aircraft count appears to have a corresponding counter move.) So, the last purchase finally landed a device that does a good job controlling RF emission.

Now, to the selections. After I verified that the old unit (C1) may have a float metal case, I tried many ways to “ground” it. After failing to make repeatable improvement, I bought the first replacement, C2. In a way it feels sturdier (and verifiably heavier). My first test resulted in a measurable, but smaller noise increase. But later tests showed it to be just as noisy as the old one. Needless to say I was disappointed and disillusioned - or misdisillusioned, abandoning the hope that less noisy adapters exist.

If all adapters are equally bad, I reason, I ought to move it away from the antenna. Although I cannot move my workstation or antenna, I can add 2 m (6.5’) to axial distance by connecting adapter input via the USB C charge cable - in addition to the built-in 5-cm (2") USB C pigtail. This will nearly double the distance between the adapter and the antenna, reducing noise intensity by 6 dB. This, however, requires a USB C extension, not just the male-to-male charge cable.

I was almost ready to order a USB C male-to-female adapter when the unique design of this TNP adapter (C3) popped into my eyes: It has no pigtail-style USB C input cable.

Instead, it sports a female USB C port for input (marked as “TO HOST” on the body). It also comes with a 5-cm detachable (and concealable!) male-to-male USB C cable (as shown stacked on top of the body in the above picture). In product description, the included stub is depicted as the input cable, like the following:

But this adapter’s design allows alternative cabling: Use the longer charge cable as input, and use the stub to connect to charger, as illustrated below.

In other words, even if C3 doesn’t offer better RF suppression, I have a plan to reduce noise received by the antenna. In other words, no-pigtail design is still more desirable. it is super sweet that this unit seems to solve the interference without using alternative cabling.

Note:

1. The laptop used in this test is a 13" Macbook Pro, not the 15" Macbook Pro from work used in previous testing. In previous tests, noise floor was raised by 10 dB, not 6 dB.
2. The test interval is merely 20 minutes, and performed during high noise time. There is a non-zero chance that the result is a fluke.

All should become clearer when work returns after New Year.

[obj]

If any of these are USB3 devices it’s probably also worth checking if they’ll run at USB2 speeds (USB3 signalling tends to generate a lot more noise than USB2 signalling)

[whacky]

At least the two purchased devices are USB 3. How do I make them run at USB 2 speeds? (USB C ports will automatically attempt USB 3, I believe.)

[whacky]

Well, before reconnecting the adapter to work laptop, I spent days and nights trying to figure out how to build Spektrum on MacOS and Pi or use it with preinstalled Processing without having to build myself. Pi is still not working but I finally figured out MacOS build so I can confirm the difference with actual scans. (The scans also help reveal some inner workings of the blue stick.)

IMPORTANT: Because the the blue stick has wideband preamp in front of a narrowband filter, the following scans are skewed by that filter, rendering them less useful to assess preamp saturation, which appears to be the problem i am combating. I wish I had an orange dongle (or a generic stick) for this task. Also keep in mind that the scans does not reflect RF spectrum received at the antenna. The result is the product of RF spectrum and the transmission function of that filter. (Actually, the result is also altered by the transmission function of the antenna.)

1. The setup
   In this setup, the receiver is directly plugged into Macbook Pro using a USB A to USB C adapter. Power is also plugged directly, as adapter C1 doesn’t have a power passthrough. Adapter C2 is used for a different purpose so the comparision is only between C1 and C3.
   In Spektrum, default gain is 10 but very little can be seen at that level so I opened up all the way to 500. (I guess that’s the equivalent of 50 in dump1090.)

2. Wideband survey
   The first tests are full range 24 MHz - 1,800 MHz. The baseline clearly reveals the effect of the built-in ADS-B filter.
   

   
   
   This baseline is saved as reference in other conditions (blue line). The next two shots show really broad and severe leaks. Not only does C1 raises noise level in the pass band of the ADS-B filter, but noise is seriously elevated even in the suppressed lower side.
   
   
   
   
   The next two are taken with C3 connected, with and without monitor. They show little difference from the baseline inside or outside the filter’s passband.
   
   
   

3. ADS-B vincinity
   To further examine the characteristics, I narrowed the range to 940 MHz ~ 1,240 MHz. I also increased bin to 200 kHz to speed up scan. (Did not seem to have effect.) In addition, as I discovered Spektrum’s median hold (aqua blue), median gives a better gauge than “Video average” whatever that does.
   

   
   
   This baseline reveals the passband of the blue dongle’s filter as ±20 MHz.
   
   
   
   
   
   
   (Keep in mind that the built-in filter is designed to suppress signals outside of the passband. Judging on the baseline, it does a good job in that. But once C1 is powered up, even the apparent passband widened. This is in addition to a sharp rise on lower frequencies.)

These scans reveal some details of leakage from the “good” adapter C3. But overall, it is much smaller. (Also, C2 and C3 does not leak till an HDMI cable is attached; I believe that C3 also does not leak till the monitor is attached.)

While these results are promising, I get additional frustrations after work resumes because C3 appears “temperamental” when attached to my work laptop. Noise can change wildly when my laptop move slightly; it can reach the same level as the “bad” adapter. It seems to have the least noise when the input cable is bent at 90º against the connector. The strangest of all, there seems to be a correlation with how much activity I have on my laptop when there is no physical disturbance. If I leave the laptop for a time, noise can decrease. But when I do certain things - I haven’t deciphered which, noise increases. (This makes me suspect that this device is a spy for my boss.) Without Pi Spektrum, it is rather difficult to do live scan when the adapter is attached to work laptop. I’ll keep trying to build on Pi.

[abcd567]

@whacky

I spent days and nights trying to figure out how to build Spektrum on MacOS and Pi or use it with preinstalled Processing without having to build myself. Pi is still not working but I finally figured out MacOS build

Can you please reveal the “secret” of making Spektrum work on Mac?

I dont have Mac, in fact never had a Mac, but your revealations will help lot of users who do have a Mac.

[whacky]

@abcd567

Can you please reveal the “secret”

It’s been my plan all along to append to your great guide. It will just take some time. (This update took me 3 days to write )