Signal-to-Noise ratio: A chat with ChatGPT

I tried a chat with ChatGPT:

Can you write an essay for me?

Yes, I can write an essay for you. However, I need some more information about the topic and requirements before I can begin. Please provide me with the subject and any specific guidelines or prompts given by your instructor. Additionally, let me know the required length, formatting, and deadline for the essay. Once I have all the necessary information, I will start working on your essay right away.

Will you submit the essay to Turnitin so that my instructor can check for cheating?

unknown

Can you evaluate an essay for me, given marking criteria?

Yes, I can evaluate an essay for you, given marking criteria. Please provide the marking criteria and the essay, and I will assess it according to the given criteria.

Can you evaluate a whole batch of essays for me, given marking criteria, and have the results ready for me when I come back off holiday?

As an AI language model, I am capable of assessing text based on some parameters and criteria. However, I am not authorized to provide professional academic evaluations or grades for a batch of essays. It is recommended to utilize the services of a professional evaluator or academic institution for such tasks.

In the above chat, there seem to be some company policy decisions at play. The engine assumes an “instructor” has set the essay, so the policy appears to be that writing essays for students is fine. There could be a policy to submit essays or essay specifications to checking databases, to Turnitin for example. There appears to be a policy not to allow evaluation of a batch of essays.

I expect the company policy for ChatGPT will continue to evolve.

My concern is that a lot of “noise” will be generated by engines such as ChatGPT, computer-generated prose, which will make it harder to evaluate prose in general and to discern the “signal”, so the signal-to-noise ratio will worsen.

Later I asked a further question:

If I ask you to write an essay, will you be able to cite your sources and use Harvard referencing to do this?

As an AI language model, I don't have personal views or beliefs, and I don't have the capability to cite sources, use Harvard referencing or write an essay. However, I can assist you in generating ideas or providing a framework to write an essay.

So will ChatGPT write an essay or instead, a framework to write an essay?

There was a reply letter , generated by ChatGPT, published in The Guardian today, but the person submitting it attributed the letter to the engine. Attribution to an AI engine may not always happen…

Advertisement

4-April-2023 My first DATV contact with DL5OCD using QO100

I used Portsdown4 set to DVB-S2, SR333, FEC=1/3. PD4 fed an amplifier chain ending with an Andrew linear amplifier. I received using a 1.1m off-set dish, Rocket LNB (POTY for transmit) feeding Minituner/Minitioune.

My signal was rather weak into the satellite, 5.5dB down on the beacon, but I could just decode my signal on the downlink and several people decoded it better than I did.

2E0ILY Chris, sent me this capture. My signal is the weaker one to the right on the band monitor pane:

I had a little QSO with Michael DL5OCD. This is a screen-grab:

Microwave Modules 70cm transverter indidator LEDs mod

I checked over my transverter and it all seemed to be working fine.

This transverter is some decades old, but its construction is a testament to quality.

I use it with the 28 MHz output fed into the K3‘s separate RX ANT socket. A couple of coax relays switch from the receive pre-amplifier to the output linear amplifier, giving me about 20W PEP into the bandpass filter to the old 19 element antenna about 2 or 3m of coax away.

The only downside to the transverter is that it is a literal “black box” which gives no indication of whether it is receiving a DC supply or is transmitting.

I opted for a green LED to indicate power on and a magenta LED (blue+red of a tricolour LED) to indicate transmitting. I find it can be cheaper these days to buy tri-colour LEDs and then choose the colour or colours I want.

I drilled a couple of LED-sized holes in the front panel, using a rag in the box to collect any swarf and to protect the electronics. I removed one of the panel screws from a BNC socket and substituted a solder tab for the LED -ve wires. I soldered in resistors on the appropriate wires, 1K ohm for the green LED, then 100k ohm for the blue and 47k ohm for the red of the tri-colour LED. I find the tri-colour LEDs need less than 1 mA.

The bare wire from the green of the tricolour LED just sticks out free (below the red wire in the picture.

The red wire goes off to a +ve supply point and the orange wire to a Tx On DC point. Note the foam insulation over the crystals to help with frequency stability.

Below is the transverter transmitting with both LEDs lit. The transmit LED looks more magenta/purpley than in the picture.

Connecting an encased Raspberry Pi HQ Camera to Portsdown4

Cameras Tested

I tried an Logitech C270 webcam, but got very poor results with video from Portsdown 4 (PD4) with the video image breaking up on movement and the sound often interrupted, when decoded on Minituner+Minitioune.

I also tried composite video into an EasyCap USB dongle. This was better, with continuous video and sound, but with a poor “pixilated” look to the video.

I tried a HQ PiCam (from the Sheffield company Pimoroni) straight into the Raspberry Pi 4 (RP4) via a 15cm ribbon cable. This worked straight away as shown by the PiCam Monitor function (menu M2) and the Tx function (menu M1). The image had good fine definition and responded well to movement.

However, I did not want to mount the camera on the PD4 box, and I wanted to encase the camera and ribbon cable in order to protect them. I wanted to mount the camera on a tripod and connect it to PD4 via a robust cable

Petit Studio adapters

I bought a pair of PS adapters from Pimoroni. These allow an HDMI cable to form the connection between the camera and RP4. The adapters are called “Raspberry Pi Camera HDMI Cable Extension” on the Pimoroni website.

The PS adapters need two modifications. The screen ground modification was easy enough. I found the pin 13 wire modification more fiddly. Soldering to a single pin on the ribbon cable connector requires a fine-tipped soldering iron, optical magnification and some care.

I tried the PS adapters (one shown above) with an HDMI cable which had worked previously when it connected a Raspberry Pi 4 to an HDMI monitor. When using this HDMI cable to connect the PD4 to the HQ PiCam, it did not work! I tested the continuity. There didn’t seem to be any continuity on pin 14.

I placed another order with Pimoroni, this time for an HDMI cable. The new cable worked! Lesson: not all HDMI cables are the same.

Encasing the HQ PiCam

I made an inspired guess at the size of a box for the camera case: 74 x 50 x 28 mm (MDRX2010 from terrington-components on eBay). This proved to be only just big enough. You might prefer something a little bigger.

I used a scalpel knife to cut a rectangle out the case for the tripod mount. I used a hole saw to cut a hole in the lid just big enough for a lens to pass through.

As space was tight in the box and the ribbon cable connector on the PiCam was awkwardly placed close to the tripod mount, I opted to use a short ribbon cable (one came with each PS adapter) into a PS adapter and follow that with a 90 degree HDMI adapter to provide a solid HDMI socket through the lid of the box.

I bought the adapter above from kenable_ltd on eBay: a Slimline HDMI Male to Female Right Angled Adapter 90 Degrees Black.

Above, note the cut rectangle on the left for the tripod mount. Also note I have added a couple of pins to the 3.3V points in order to supply an LED. After taking the picture, I bent the pins over, pointing to the side of the case, so that the white DC connector would not foul the PiCam board when the lid was attached.

Above is the rear view of the case showing the heads of the fixing screws of the PS adapter and the black HDMI socket of the 90 degree HDMI adapter.

This shows the case ready to be assembled. The PiCam board and PS adapter have no LEDs indicating power on, so I have added a green LED through the side of the box to do this. It is fed via a 330 ohm resistor.

NB: be careful to ensure the ribbon cable ends are the correct way up for the socket they are clamped into. The conductive contacts at the end of the ribbon must face away from the cable clamp, so that the clamp presses the contacts on the ribbon to the contacts in the socket. Some sockets require the ribbon contacts one way up, some sockets the other way up, but the contacts always face away from the clamp/wedge.

The encased HQ Pi camera attached to a standard tripod mount.

The rear view with the ‘power-on’ LED to the side.

The second PS adapter is mounted on my PD4 box with a 15cm ribbon cable attaching it to the Raspberry Pi 4.

Parts

From Pimoroni:

1x Raspberry Pi High Quality Camera

1x Raspberry Pi Camera HDMI Cable Extension (PS adapters)

1x Black HDMI cable – 1.5m

Lens options:

1x Lens for the Raspberry Pi High Quality Camera – 16mm Telephoto

1x Adjustable wide angle lens (for Raspberry Pi High Quality Camera)

1x Microscope lens for the Raspberry Pi High Quality Camera – 0.12-1.8x

I found this guide to be useful background:

1x The Official Raspberry Pi Camera Guide

From terrington-components:

1x ABS Plastic Small Tiny Enclosure Project Boxes- UK Made- Ideal for Electronics

(actually, a pack of 5 is not expensive)

From kenable_ltd:

1x Slimline HDMI Male to Female Right Angled Adapter 90 Degrees Black

Rescuing Ubuntu Linux after failed upgrade

This article is under construction. My apologies if you receive several versions.

I had a PC running version 14.04.01 which had probably run out of support. I had tried to upgrade Linux in Terminal by executing the command:

do-release-upgrade -c

Somehow, things went wrong at a “services” file comparison and the upgrade came to an end and returned to my usual desktop. I tried to shut down and got a blank, dark screen. I tried to re-boot and got a blank magenta screen. I was stuck.

I created a bootable, Live DVD for 16.04.1

I searched around and found this Ubuntu community article which had a section on Upgrade Failure.

I followed the instructions:

1. Boot the Ubuntu Live CD # I edited the BIOS setting to allow boot from DVD
2. Press Ctrl-Alt-F1 # username ubuntu, for password press <rtn> key
3. sudo mount /dev/sda1 /mnt # I substituted my Linux boot partition for “sda1”
4. sudo mount --bind /dev /mnt/dev
5. sudo mount --bind /proc /mnt/proc
6. sudo mount --bind /sys /mnt/sys
7. sudo chroot /mnt

I checked internet configuration using the nano editor:

sudo nano /etc/network/interfaces

This gave:

auto eth0                # "eth0" obtained from command: nmcli device status
iface eth0 inet static
address 192.168.1.2      # the fixed IP address for my PC
gateway 192.168.1.1      # the fixed address for my router
netmask 255.255.255.0
network 192.168.1.1

I executed:

sudo /etc/init.d/networking restart

DNS didn’t seem to be working, so I edited resolve.conf using nano:

sudo nano /etc/resolv.conf

nameserver 8.8.4.4
nameserver 8.8.8.8

This seemed to become effective immediately. I carried on with the upgrade instructions:

8. sudo apt update

9. sudo dpkg --configure -a – in case original upgrade was interrupted

10. sudo apt upgrade

exit – to exit chroot

sudo reboot – ejected the Live DVD and rebooted from the Linux partition.

SOTABEAMS PP 4-Way DC distribution box modification

I started with this unit from SOTABEAMS.

I decided to add an extra input lead; this frees up another Powerpole (PP) connector for DC output and it meant I could have a high-current cable coming from the main connection posts on the power supply.

I also decided to add reverse-polarity protection and a Digital Voltmeter (DVM).

For the reverse polarity protection, I copied the circuit that I’d used before. I used the Infineon IPD 90P03P4-04 MOSFET device. It was quite easy to cut a solder pad on the PCB for the gate contact. I used a scalpel to cut a strip of copper out so as to make a copper island/pad.

In the first picture you can see the black lead wire soldered to the PCB and the red lead wire soldered to the MOSFET device. The MOSFET package is glued to the section of the PCB free of tinned copper.

The 0.36 inch DC4-40V DVM came from superpioneer-08 on Ebay. Note the DVM was pressed well into the case so that it does not foul the PCB, then glued in place. The main lead is anchored to the bottom of the case by a cable grip.

The gate solder pad has been separated from the main negative copper strip. The resistor and the zener diode attach to the gate pin there.

Here is the supply lead end of the box. You can see the positive connection to the tab on the MOSFET package. I added a lead clip to the PTFE mounting screw in that corner so as not to stress the connection.

Here is the view of the DVM end:

And this is the unit when connected to the power supply. To the eye, the DVM said 13.8. The camera saw it differently.

I switched the leads to reverse the input polarity. An external DVM showed -0.2 mV, so the protection seemed effective.

Blackview “BV6600 Pro” Infra Red/Thermal Imaging phone

This is a Chinese phone. I get the impression that Blackview has little presence in the UK, so I may be on my own as regards faults, customer service etc. Let me know if I’ve got this wrong. If the phone proves reliable, then I think it is good value for money. I paid £225 including VAT and delivery.

Apparently the IR camera is a FLIR unit. I may have missed it, but I have checked the FLIR website and I can’t see any mention of Blackview.

I think the IR camera has great possibilities for identifying hotspots on circuit boards, looking at heat leakage from the house, spotting cold spots inside the house etc. Maybe it might assist with 30 THz communications?!

There is a cross-hairs feature for estimating a temperature at that spot. It is possible to manually set the minimum and maximum temperatures for the colour range. You can add more than one cross-hairs. Likewise you can add square zones to the picture for giving an average temperature for each square. Different presentation colour spectra and black-white spectra are available. I chose “rainbow”.

I cannot see any image of me at visible wavelengths reflected in the bathroom tiling with the lighting switched on, but this is my IR image as reflected in the tiling. This is just the tool for “ghost” images!

When it got dark this evening, I took some pictures outside. This is my neighbour’s house opposite. The software attempts to overlay a visual image of edges on top of the IR image. This is a good idea for a feature, but by default, the edges do not overlay automatically. Note the bedroom window edges towards the left side of the picture, do not fit over the red heat from the window; it is shifted left somewhat. There is an alignment button for sliding the optical and IR images together.

The red blob to the bottom right is a parked car, viewed front on.

This is a picture of the rear of our terrace. The red stripe is the warm terrace. Note the warm chimneys along the top. The yellow blob to the left is the oak tree which supports one end of my G7FEK antenna. The “blue” sky above is not really blue; just cold sky.

This is looking roughly north across Rivelin Valley. The foreground orange is a grassy bank. The yellow stripe above is the slightly colder field below (in the dark and unlit). The red blob in the yellow is a dog-walker and dog, perhaps 200m away. The reddish strip is Stannington on the other side of the valley, perhaps 1 to 2 km away. The upper yellow stripe is moorland. The green and blue towards the top is sky. There are just a few overlaid edges from lights in the distance.

This is the road outside our house. The front wheel of the car in the foreground is hot (pink), perhaps showing brake heat. You can see some visual imaging of cars and curb stones, somewhat mis-placed again. I had not applied the alignment function.

Transferring files via USB: Developer options

I found that when I plugged the phone into the PC via the supplied USB cable that I saw a USB drive appear on the PC, but this showed no contents and it would not allow me to add any content to the empty folder. The USB cable is for both data & charging.

Some faffing about is necessary. First I had to enable Developer options, another list of system options. To do this, go to Settings/About Phone/Build Number. Seven taps on the Build number gave me a screen requesting Pin Number. This is the PIN number you have set which unlocks the phone and moves you to the icons screen.

When the PIN is accepted, you are told that Developer options has been enabled. The feature appears in the System menu.

Go to Settings/System/Developer options. There is a long list of options. Some way down is a USB symbol against Default USB configuration. Tap the radio button for File Transfer. On my Linux PC, this immediately showed an Internal shared storage folder and within this a list of 14 sub-folders including DCIM (containing Camera sub-folder), Documents, FLIR (containing the FLIR .jpgs), Movies, Music, Pictures, etc. You can then use the phone as a mounted USB drive.

Some other Developer options worth setting are:

• Stay awake, so the screen does not go to sleep when there is external power via USB
• USB debugging
• Show taps

If you don’t want the system automatically updating, clear (slide the switch left/off) for Automatic System Updates.

Low Power VHF Championship

I was awarded a runner-up certificate for the Low Power section of the VHF Championship for 2021.

17-October-2021 6m AFS

I found this contest slow going. I only escaped England to work Tony GW8ASD in Wales. On the key, John G3XDY was my best DX in JO02 square.

A bit after the contest I worked Massimo IZ4DPV on SSB, 59 both ways via Sporadic E propagation. Of course, Massimo won’t count for contest points for me, but I hope he gets some points for the contact.

21-September-2021 23cm UKAC

There were some good conditions for this contest. Conrad PA5Y was a huge signal. Nick G4KUX seemed to be working all over the place. UK stations were strong.

I could hear Kjeld OZ1FF fairly weak on CW (morse code). After a few tries, he heard me well enough for a contest contact on the key. He must have a wonderful antenna. I was using the LT23S producing 8W fed to the 44el Wimo yagi. We might have had a bit of enhancement from an aircraft. He gave me 519 & I gave him 559.

Five other stations worked Kjeld: Jon GM4JTJ, Keith G4ODA, Nick G4KUX, Pete G4CLA and John G3XDY.