Tindie vs Amazon

Amazon have been an interesting place to sell, and have sold quite a few things. But oddly the main thing that sells is the Faikin boards. I don't think my other products are not sellable, indeed, I have had direct sales to local A&A customers for a load of other projects, just not so much on Amazon. I generally have a policy of Tindie £ price less 10% for direct sales to A&A customers.

Amazon are making EU sales simple - Amazon buy from us and sell in EU so they sort local VAT. With Tindie the recipient gets a VAT bill on import, sorry.

I started selling on Tindie as well some time ago, and oddly the same is true - most sales are the Faikin. The main reason for listing on Tindie as well was to cover countries Amazon do not do. Well, that is how it started, but now it seems we are selling to people in EU where Amazon do sell. The main reason looks like the extras (a cable, and 3D resin case). These are extremely easy to list on Tindie, and provide a simple up-sell on the original product. They are only a few $ extra but sell well.

Amazon, on the other hand, do not seem to have an easy way to add options or link products that I can see. I will ask for help, though any interaction with seller support needs a deep breath and a lot of patience. But at present I still cannot even list a product - Amazon are actively making it impossible for me to list things, in spite of reassurances. They are insisting I am on the brand registry (something I have tried three times), and they are taking forever to say yes or no. It is odd, it is a simple thing - do I have a trademark? Well yes, there is it with my name on UK trademark web site. How does it take more than a few minutes?!

Well, being able to list the cables and case clearly is helping as Tindie sales are creeping up, and I think this month is the first month that net pay out from Tindie is more than Amazon. That is a milestone I feel.

3D resin printed case, snap together

http://tindie.uk/ if you are interested.

Scammers

OK, not too many details, sorry. I mean privacy and shit, and remote tiny tiny chance not a scam, LOL

We are always aware of scammers, and very occasionally we fall foul of them, but we are always working on improving processes and checks.

If, somehow, this is not a scam, then sorry - but a story of how not to look like a scam.

Because of a recent scam (and you have to love the gall of the guy calling up to complain his number does not work and he cannot log in) we have been extra vigilant and are all looking closely at an order that came in today.

As I say, not too many details, sorry.

So someone orders a telephone number, and we ask for payment up front. This is normal as we can only get close to checking an order is genuine by leveraging the UK banking system to get someone to send us money, even if only £1. But they refuse to send with order, though eventually, after hassling, they send payment via Wise (Both Wise and Revolut are listed on our web site as don't send money from them as we have no way to verify or refund).

So they then provide bank details for DD, and a new (company) name, and we check it matches bank details, but really no way to check it is actually them.

So what next? Well check more - this has magically changed from an individual to a company. But not a UK company. Not something we can check in any way.

Maybe check address...

Well, street view is cool, and this is it. This is not close to the company name. Maybe they work from the flat? But that would be [number]A or some such, surely.

So, really, if you are in fact running a non UK company by the name you say, from a flat over a Chinese Restaurant in Belfast, but cannot send any payment from a UK bank account but can quote one that matches your stated company name, well, sorry, we may be at the point of saying no.

And basically, we can say no. We can say we do not want your business.

I am really sorry if that really is you living over that Chinese. Get a UK bank account from which you can send money, and we can talk.

To be clear, I started my business decades ago from a house in Bracknell, and that was registered office address. But it was registered office address, and could be checked, and I could send money from a UK bank account matching my company name. So yes, a new business, and a small business, no problem. We do not want to discriminate. But scammers can piss off, sorry.

That is not how import tariffs work!

It seems Trump has started charging tariffs on small packages now, and as US does not have capacity to handle collecting the tariffs, so is expecting tariffs collected at source!

That really is not how it works. But...

It sounds like Royal Mail are going to do a pre-paid tariff option (I am amazed they are bothering). Sadly this probably means I'll end up creating a load of Tindie carriage rates to US for each tariff amount, as I don't think Tindie do a shipping rate that includes a percentage of goods price (maybe they will - I have asked them).

Just to be 100% clear to anyone in the USA who is confused, this means YOU ARE STILL PAYING THE TARIFF, it just becomes part of the shipping cost to the USA.

I feel sorry for RM trying to keep up with the changes, and we all feel sorry for people living in the USA.

https://www.royalmail.com/usapersonalupdates

First resin print failure

It had to happen, for some reason a print did not release.

This is another area where FDM and resin are very different. With FDM it keeps going and you end up with a lot of spaghetti and goo. With resin the print keeps moving up but the failed layer stays on the release film, and just gets re-cured as the print go - you end up with one aggregate layer.

So, poor out the resin (need funnel and filter) in to a bottle. Remove stuck layer, that was easy. And ready to go.

Well, not quite, for some reason it kept saying foreign material detected. I cleaned out the tray, both sides, and started again and OK. In fact, it seems once can remove failed print layer with a fingernail (in a glove) without pouring out resin even.

I have noticed a scratch on the film though, so have more coming and will have he fun of replacing the file. That may be a job for tomorrow.

I have no idea why it did not release. The print was the same as previous.

Update: sorted release film, seemed good, but had another failure - which highlights a feature of resin. With FDM a failed item would wreck the whole print, the spaghetti sticks to the nozzle, and you end up with a total failed print. With resin, one can have a partial failure, like this.

Prototype 3D

This is why I got the printer...

One of the challenges with any engineering is that no matter how much to check the design, some issue are only noticed when you finally make the actual item, and have it in your hands. This is even more true for 3D models.

I can order 3D prints from China, with a turnaround of around a week. I could print designs on my FDM printer, but not the same - especially small details like the interlocking lips on the case designs (those on the right of the image are for a Faikin). The precision needed for these is not possible on FDM.

But even simple things can catch out out. The white 1 gang back box cover on the left took a week, and on arrival I see the problems - not just the square corners that do not match the back box I have, positioning nipples in the ring are wrong so the back does not lock in the right place, and the back of the ring is 0.5mm thinner which shows on the actual print. Now I can prototype I can test with a turn around of half an hour. I can get the designs right, and then order some from China. I have white resin coming soon as well.

What is fun is that having spotted the errors in the white 1 gang plate, I fixed the design, and printed a prototype for this post. I have just checked and my fix is not correct - I see why now, and have another fix to prototype.

As you can see this saves a lot of time, and some money.

Getting the hang of it (3D resin)

I am getting the hang of it. I am sticking with the washer washable resin for now. It will be an extra step of messy to use the standard resin, and I don't have any spray bottles of isopropyl alcohol left at the moment.

The Mercury washing/curing machine was somehow jamming the stirrer, which is why it did not seem to work. It is OK now, and does a good job cleaning - which makes it much easier. It also seems I can clean with isopropyl alcohol, which dries quickly before curing and will be the same for standard resin.

I did manage to spill resin all over the machine, which was annoying, and some is still leaking out of the bottom days later! I am, however, very reluctant to dismantle it. It is working fine.

I have the air purifier now, which links in to the back of the printer with a hose, and keeps the odour down. Again, standard resin may be a different matter.

So far the printing side is click and print with no problems. The model supports are indeed a lot simpler than for FDM - just needing the smallest contact. Basically it looks like bridging is much cleaner on resin printing just needing an anchor each end where as on FDM the bridge will sag. With resin it is upside-down.

The detail is impressive. The fact that the resolution of FDM is the same as resin hides the ability for resin to produce detail that is impossible for FDM. The extruded filament on FDM has a shape and per layer texture and minimum size that make the detail impossible. The teeth on this model highlight this nicely.

Moving from FDM to resin 3D

I have done 3D printing for a long time, since the days of 3D printers being kits and burning out stepper motor drivers.

My latest 3D printer was Lulzbot Workhorse, and it is pretty good. The Bamboo ones are, as I understand it, way more impressive than that even.

But this is FDM - using reels of filament, melted and deposited. It works well. But there is another whole 3D printing world with resin printing.

My new printer is an Elegoo Saturn 4 Ultra 16K - a resin printer.

Print size

Downside number 1 is the print size, way smaller. FDM can do all sorts of large sizes, my Lulzbot is 279x279x284mm which is quite big.

The resin printer is 211.68 × 118.37 × 220 mm. The reason is the print bed has an LCD or some such that prints a layer at a time. Where as the FDM has X/Y/Z motors and is limited only by physical size. Bigger LCDs are expensive. Longer bits of metal to make a larger FDM printer are less so. 

Resolution

Upside number 1 is resolution, and this is where it gets crazy. FDM is good, but resin printers are crazy. For Z axis the Lulzbot can go down to 50µm, but the Elegoo starts at that, and can, AFAIK go down to 20µm or perhaps lower. As for X/Y, both do well, with Lulzbot down to 10µm, but the filament is a 0.5mm (500µm) nozzle. The resin can print down to pixel at 14x19µm.

The result is resin printing sharp points and groves, and features really well where FDM is 0.5mm nozzle limited, and printing with way lower layer size. This allows sharp and exact features for resin printing.

Also the print is per layer textured, the layer has a bulge in centre of filament track. Not so with resin. Vertically resin is smooth and clean. So even at same 50µm layer, resin printing is way cleaner.

Level surfaces

Another difference I was amazed to see for resin was the top edge flat surface of a print was level, exactly level, I mean shiny level and smooth. FDM has ridges and texture. The bottom not so different as both print against a base plate and tend to be as smooth as that plate.

Levelling for the start of print is important and the resin printer seems better.

Print speed

This is more of an interesting one - FDM prints depend on time to trace the print layer by layer. So more detail in any layer takes longer. Adding support takes longer.

With resin a layer takes a number of seconds, end of story - no matter what.

So a biggie is if I have a small thing to print, for FDM, if I print 5 (assuming they fit on build area) is 5x longer, or worse. For resin it is same time as same height. Uses more resin but not more time.

Similarly the complexity of a design, the detail, the supports, all matter for FDM print time, but no issue for resin.

All that matters for resin is height and layer thickness. I.e. how many layers. It is a very different way to consider complexity.

Smell

The resin printing has a smell, but very minor. I have an air filter on order, but to be honest I am not sure I need that.

Sound

Resin printing is way quieter. I like this. Really way better.

Sticky

With FDM, print finish, take off bed, done.

With resin, messier. I am working on the best way to handle this, but you remove print, it drips, take to tub for cleaning (or maybe tap for water cleaned resin), put in curing for a few minutes, remove from bed. It is more work, and more chance of stuff dripping. More chance of "mess".

Changing material means pouring resin back from tray, cleaning and new resin, maybe changing fluid for cleaning (water/alcohol), way more hassle than a filament change on FDM.

So yes, a bit more, but working on it - not sure it is a lot more hassle in the long run. The results are amazing.

Prototype

The main reason for this is prototype designs - I can order bulk resin prints from China, but they take a week. This allows way quicker design refining before ordering, and quick one-off prints.

That alone makes it worthwhile.

Interesting new scam "worry free purchase"

Ordered something online and this is on they list this...

It has a delete icon, but do it, and it re-appears.

Update: They credited the £11.99 - yay!

Seems a third party service.

Seems mandatory.

So ordered the goods, and then emailed (seller, and supplier of 3rd party service) and said under various UK distance selling directives I wish to cancel the "worry-free purchase" service as I am more than happy with my purchase to be covered by UK statutory provisions, which seem to cover all they cover on that service.

We'll see if I get a refund or I have to claw back on card.

Extra irony, their "worry-free" purchase service is causing me "worry" I would not have on any normal purchase.

OK this has worried me...

NFC reader and alarm/door entry system

A long time ago I made a door entry and alarm system. Actually my first was approx 1989 using a mag head from a Sony walkman on a block of wood on the door and a wire wrap 6502. My latest attempt a few years ago is a lot more sophisticated.

It is good, indeed, I would say it is very good. It has a lot of off line working built in but backed by internet management/control system, so designed for power and internet failure, for hours, or even days.

The NFC door control uses AES on DESFire cards, so no plain text even on RF, and challenge both ways, so super secure. It even has different keys for each card. But it scales to any number of cards, with different access levels encoded in the card so as to work off line if needed (for a configurable time).

Like I say, it is good, and we use it, and a hack spare uses it (maybe two) and several small offices.

But it is tricky to sell.

Insurance and alarm companies have stuff tied up

Perhaps the biggest issue is that insurance companies and alarm companies have some industry standards and something of a private club. Yes, we could join, I am sure, but only if I wanted to be come an alarm company. Open source stuff is not going to get in to the club no matter how good it is. This is a shame as proper (Galaxy) alarm systems can be crap by comparison, as we know because we had one, installed by a certified installer, and were robbed!

My understanding is that one of the rules (I heard from an installer) was you can't have some external indication of armed or not. This ment staff did not understand they failed to arm. It beeps in various incomprehensible ways, and shows stuff on the keypad (which you cannot read from outside when using the fob to arm), so they assumed armed when not.

My system can show such, but we go for subtle - internal lights go off when armed. I mean you would turn off anyway, but staff can tell instantly that the lobby light not going off means not alarmed. The light switch does not allow manual turn off. Simple steps but has meant that on the odd occasion of not arming (e.g. fire exit open) staff knew. This is what caused the problem on Galaxy.

I also allow a forced alarm for such cases as a last resort - where the open fire exit becomes a sensor - someone closing it would set off the alarm. This means I can set forced alarm on timer, and be alarmed if staff have missed the problem somehow and still have a working alarm (all PIRs, etc). The Galaxy, when it does not arm, does not arm, end of story. It is a compromise I suspect an official system would not allow.

So basically you need to check (and record the call) insurance are happy with no proper alarm system, etc. Seems many are, and may not even charge more, but having it on record should mean you are covered.

Monitoring

Another issue is monitoring of the alarm. We had that once at an office, and it was such a pain. And one time when we did not know then code word, they said the police would be called. The police did not turn up so I was stuck there for 4 hours on a Sunday waiting for police. Also police won't come if you have too many false alarms.

My system monitors, but messages people, several people, like me, people with access to the CCTV who can confirm it is a false alarm, and if not call the police explaining they can see the burglary in progress on CCTV. That will, I expect, get a way better response than an alarm company calling.

Also much easier to integrate to external systems - staff in office logging for fire list, etc.

Locks

Another issue is locks, and the typical internal door lock is a mag lock, which is really easy to defeat. I have found much better locks and would recommend using them - Abloy locks. These work without the alarm and use a proper euro profile lock, and can be set to open from inside regardless, but also have a load of sensors (key used, handle used, lock in/out, etc).

Opening from inside also avoids the messy "break glass" and "exit button" you typically need.

So my system has to handle everything from a simple maglock and exit button, to the Abloy with something like 5 inputs - which it does nicely now.

It costs more, but in the end it is worth it.

Professional kit

With all of the above, an alarm or door entry system can work, with insurance confirming OK, etc.

But the kit is not as professional, or is it.

When I started I had single sided copper clad boards milled and hand soldered. I have moved on to proper PCBs made in China. I have moved on through a load of connector types to WAGO PCB connectors which are just simple to use for an installer. I have added per input/output LED status on the PCB.

But still, the case is a messy 3D print on my printer. Well, now we moved on finally with high resolution, smooth, clean, 3D resin prints from China.

The wiring and connector to the NFC reader was also a concern, but ironically the leads I got made in China for Faikin boards are perfect for this, and again, professional, so making it all easy to install and just more professional.

I finally feel like I have stuff that is professional looking.

The NFC sensor on a door. It looks proper now.

I have even now designed one for a 1 gang UK pattress box.

So now I plan to list these boards and cases on Tindie soon. The NFC reader is ideal for a hobbyist working on any NFC stuff.

Ordering 3D prints

I have a 3D printer, it is pretty good, but I decided to try ordering some 3D prints from JLC.

Price

The most obvious issue is price, and actually, it is good. I have 2 part 3D printed cases for my Faikin boards for $0.47, but the postage is actually more, making $1.52 delivered, (£1.13+VAT). This is actually pretty good.

Quality

This is perhaps more important - could I have printed myself better - well NO, really NO! The quality is amazing. Resin prints are always good compared to FDM, and these are post processed with sanding.

The precision is also amazing. These designs are made with exact edges, no margins. I thought this was a good test. Well they fit smoothly, and just stick, but not very firmly. This means the dimensions of the print are, well, pretty damn exact to tiny fractions of a millimetre.

The result is quite impressive.

Improvements

My case design are auto generated from the PCB, so as to make sure they fit exactly. I have now done a load of work to tweak these, adding 0.1mm more to the edges that lock and a tiny angle to snap to place. So will be trying those. Will be interesting to see if I can make it a tighter locking edge now.

But, for now, a tiny spot of glue makes a nice case, so selling these on Tindie.

Trying 0201

I decided to take the step to move some of my PCB designs to using 0201 components.

This is basically accepting that I do not even try to solder these things now. A few years ago I started milling boards and hand soldering 0805, and then 0603. I can just manage 0402 if I am very careful. But no way I could do 0201s. These days I design boards and have them made and assembled and shipped. 

It seems JLCPCB can do 0201, so why not?

What's an 0201

Components have various size designations, sizes liked 0805, 0603, 0402, 0201, 01005 are used for standard chip type resistors and capacitors. An 0402 is 0.04" by 0.02" which is 1mm by 0.5mm. An 0201 is 0.6mm by 0.3mm which is tiny.

This picture gives you an idea - a ball point pen shown for stale (a banana would not fit), and the diagonal square things (LEDs) are 1mm by 1mm. You can see ten 0201 resistors and an 0201 capacitor.

So yes, grains of sand. The 01005 things (not doing those yet) are like dust, and there are some that are smaller!

Pros and cons

With 0402 it is possible to run a track under it, i.e. between the two pads, but with 0201, not so much. That said, I rarely ever do a track under an 0402 anyway. Sometimes larger components.

Obviously some components have to be bigger, and you end up with this crazy mix for capacitors - but that has always been the case - larger caps, higher voltages, mean physically bigger caps. But resistors are less of a problem usually - yes sometimes you need bigger ones, but mostly they can be 0201s.

The main advantage is the space saving - my PCBs are often nice and small, so this is handy.

Now, I know my colleagues doing PCB design also need 0201 caps for decoupling as they need to be physically small and close to components. Thankfully I am not doing stuff that high speed with my ESP32 modules, well, yet.

So far they are working - in that JLC seem more than capable of placing 0201.

Footprint

KiCAD has standard footprints, but interestingly I found this fun article, so trying their suggestions. Some test boards to make sure no production issues - surprisingly JLC has not specific recommendations.

GS1, scam?

You will have noticed codes on products you buy, with a barcode, these are product codes. Also known initially as UPC (Universal Product Codes), and then for Europe, EAN (European Article Numbers), and now GTIN (Global Trade Identification Number).

These are 13 digits (12+checksum), and allocated by a company, GS1.

At the beginning the UPCs were allocated on a one off basis to companies for a one off fee, but this changed and now they are allocated to companies on the basis of an ongoing rental.

Rental makes sense.

It is 12 digits, but this is some whole blocks to each organisation so not evenly spread out, and ultimately they will run out, so a system to manage these makes sense. Rental makes sense on the basis that companies will only rent as many as they need, will be encouraged to recycle from discontinued products, and the whole blocks could be re-allocated to new companies once a company no longer needs them or goes bust, etc. Obviously re-use of codes needs a sensible waiting period, and GS1 even had recommendations on that for companies recycling numbers.

Except!

Things have changed, in that GS1 no longer recommends re-cycling numbers because many platforms stick them to a product and do not update/delete that record.

What is extra odd, when querying this, I found GS1 do not re-allocated lapsed blocks to new companies.

This means GTINs are allocated as a one-off operation to companies - never recycled to new companies and not even expected to be recycled within that company!

So rental is a scam!

Rental for one-off allocation makes no sense. If the allocations really are forever, then the pricing should be for a block to be allocated. Ongoing rental is a scam as if you stop renting the numbers stay allocated. Indeed, discontinued products continue to cost you if rental.

We only have 100, and the price has doubled this year. We do discontinue products, and we tried to recycle (several weeks or Amazon support is failing to do this). So at some point we will be renting a significant number of dead codes, and it will be worth getting a new 100 block, re-allocating new numbers for current products, and stopping rental of the old 100 block.

What is interesting is that platforms like Amazon do seem to lock in a GTIN, but also they seem to not care if it is your GTIN unless there is a conflict. So if Company A got a block of numbers, paid the year, did not use them, and then ceased. Company B could use those numbers on a platform like Amazon as no chance of a clash.

Interestingly GS1 have replied to my various emails right up until I said the above, and they said Company B could face fines. I asked for legal basis for fines. GTINs are not covered by a contract with Company B (not that fines are allowed in a contract). GTINs are not protected by trademark, copyright, or patent or any other legal framework I am aware of. I mean I may have missed something, so I am happy for them to enlighten me - and asked as much - but no reply.

It seems to me, in my honest opinion, a rental arrangement for a permanently allocated resource is a scam, simple as that. If it is permanently allocated it should be a one-off fee for the allocation.

That is just my view, obviously.

The printer that just worked (and other fairytales)

I am impressed with the Canon TC-21 A1 printer. Don't get me wrong.

But it seems that Canon doing something very stupid! I have had this with many printers before.

It is telling the printer the type of paper!

Please make it simple!

Firstly there seems to be no standard such as what satin or coated actually means, or even terms like heavy/light weight.

But it does seem the printer considers the paper type (and notably the weight, and hence thickness) in deciding when the roll ends, and somehow I must have had it wrong as it ended several metres before it really did with no option to say just bloody print - I know what I am doing.

But there is one standard and that is gsm (grams per square metre), and the paper is marked with it, so why not make the paper settings on the printer also show the gsm - that way I stand some chance. Another idea may be to allow me to set the length of the roll, as that is also printed on it - or at least show the lengths for each option.

Doing the right thing

So, it is a Canon printer, I'll do the right thing and use Canon ink, and Canon rolls of paper, what could go wrong.

This is the paper.

You can see it is Canon, and is 130 gsm, 610mm wide, and 30m long, and described as "Premium Paper FSC". The FCS is just a certificate not related to type of paper though.

I have some options.

There are many more options, but they get quite specific. The only one marked "Premium" (well "Prem") looks like 80gsm perhaps. So not that. I am assuming "Coated Paper" for now, but I really have no clue at all. The options are not clear and none use the exact wording on the roll of paper itself or state 130gsm.

If you sell a printer and it has a list of paper types, and you sell paper and they have specific names for each type, why the hell not use the same terms/names in both places, please, Canon!

Quality product?

But it gets worse. The roll has to be installed with the paper pushed properly to the right hand end stop that fits in the core or the roll. It checks this (good) and even has little diagrams showing you how to fit, and lock the end stop in the core.

Except...

The core sticks out, so no way to push the end stop up to the paper edge. I tried several times, and no joy.

Only fix was a sharp knife, and finally it works.

This is official Canon paper for the printer and does not work in the printer. Really not that impressed.

Reseller?

The reseller has taken this seriously, and has pointed me to some reference information that may help.

Age verification

The Online Safety Act is in force to block porn sites accessed in the UK now. You have to prove your age.

There is even a petition to repeal and rework it. Do sign, but we all doubt it will help. Maybe if it gets to millions.

Just to be clear - this legislation does not just impact porn sites, or just adult sites, but millions of sites and services, and there are millions more that may be in scope. This is not something where one can say that compliance is a "cost of doing business" as the vast majority of sites and services in scope are not businesses. They do not have money to comply, or even to get legal advice to find out if they have to comply - get it wrong and they face huge fines. That is the crux of the petition.

Let's stick to porn sites for now.

This is a huge invasion of privacy and a largely pointless exercise as there is no real way to stop teenagers that want to access porn from doing so. In my opinion a better approach is education, and especially on the nature of porn as fantasy and fiction so young people do not get the wrong idea about healthy sexual relationships. Blocking will not work, in my view, but it creates a lot of problems.

• It does not just impact kids, it impacts everyone.
• The legislation has huge overreach causing a lot harmless sites to shutdown to avoid the burdens and risk involved. It is not even clear when it applies (what of a shared diary with my wife and nobody else? That seems in scope of risk assessments, at least, as we can each post user content the other sees, and perhaps even AV if anything we add is racy).
• It creates a norm of proving your ID, or camera access, in order to access many web sites (not just porn sites), so opening the floodgates for scammers. Even if some sites have less intrusive means (see SMS below) there will be scammer sites that insist on camera access.
• Even when not scammers it creates the risk of a huge databases of sexual preferences linked to real identities being leaked.
• Teenagers will find ways around it, and even have to help adults to do so (irony!).
• It is questionable as to the extent that porn is actually harmful in the first place, especially with associated education.
• Obviously VPNs are a way to bypass as the restrictions are country specific.

So, let's look at what has happened.

I have done a few checks, and the AV falls in to a few categories as to how it works. This is "legit" AV, scammers may be more creative... Actually I have only checked one site which seems to use "age>>go". Some other sites start by insisting on a sign up to the site and creating a login before they do any more checks, which seems intrusive.

But these are some of the "age>>go" choices...

• A selfie - i.e. allow video/camera access on your device (can you see how that can be abused), and confirm some facial expressions (open mouth). Apparently there are on-line images with expression settings to which you can easily point your camera in order to circumvent this and that is just some games, not even a site set up for this purpose, yet.
• ID upload, like wow - how can that be abused, but also selfie to match ID. No idea if that copes well with edited images in the ID. I was not going to upload an ID, sorry.
• An SMS check, sends a code and they confirm the mobile operator has no age restriction.
• A credit card check. I have not tried this, but they do know kids can have cards? Maybe kids cards are debit not credit cards and that matters somehow. It claims to be a zero value "active card check" - does that show on all card apps? i.e. borrowing a parent's card may work, and leave no trace... Again, I was not going to provide a credit card - but you can see how scam sites will abuse this.

SMS

I looked specifically at the SMS, which concerns me for several reasons. This is, however, by far the least intrusive - as no camera or images or actual ID, just a mobile number.

They take a number and send an SMS with a code to enter, and then do a check with the operator to confirm the number has no age restrictions. This may be an issue in itself - the privacy policy for mobile services can be vague, but sharing whether you have age restrictions with a third party, for a number, is not a clearly identified thing that I can see. So may, in itself, be a GDPR issue.

What they do not immediately say is they then want an email address to which they can send a code. This too is a GDPR issue, as having confirmed you (a) control the number (can get SMS), and (b) the operator confirms no age restrictions, they have no legitimate interest in knowing an email address, and no option to not provide one that works. And this was a "legit" AV site. Scammers will do way more.

What is interesting is the email address has a "remember me" option - but not clear what for. Well, the answer is that you can then verify using "login", i.e. enter the email address and get emailed a code. So the use of the mobile number has now made the email verified with no further need to use the mobile number.

Back of the bike sheds!

This is one of the concerns I had with any age verification system.

So let's assume that..

• Some teenager happens to have access to a mobile with SMS and no age restriction for some reason, or
• A sixth former that is 18 has legitimate mobile SMS with no age restriction, or
• Some guy in a dodgy trench coat has legitimate mobile SMS with no age restriction.

Can they sell (or just give) AV access to horny teenagers?

(Just to be clear, A&A numbers fail to get this to work, the SMS works, but then says you do not have access. This is no surprise as we have no system to allow some third party to check if our SIMs have age restrictions.)

Obviously they can simply provide the code sent to their mobile, and code emailed to them, to their customer to allow them access.

But actually it is even simpler.

Using the mobile number for the first step, and their customer's email address for the second step, the customer tells them the emailed code, or the supplier can tell them the mobile code, either way, but use the customer's email address. Now the customer's email is considered verified, and can be used to login in future without the need for the mobile number. It just needs access to an email address.

By using a domain and mail forwarding the customer's email can be hidden as well, allowing for some ongoing income as the supplier can revoke the mail forwarding at any time.

So yes, this now creates an opportunity for people to exploit others - even adults that want access without giving up any details! Of course those doing the exploiting can be scammy as well, they know the email address, and can even see how often it is used if they wanted.

Testing

I used a mobile (Three data SIM with no age restriction - I am an adult after all) and an email address (one of my @fuck.me.uk addresses) to get access to a dodgy site, yay! But also I can then login using just the email address.

I then did the same, using the same mobile number, but a different email address. This also worked, and both email addresses can now simply login using the email address. I can now forward the second email address to someone else and they can simply login. This has the advantage for them that the site and AV service do not have their details (mobile or real email). No, I am not going to send to a child, obviously.

Now, I do not know if they permanently allow the login or ever re-validate using SMS. It is not even clear how long a site grants access from a login (though clearly at least a day, from my testing).

More data collection

Another issue here is that it allows access to a site to be correlated. With NAT and incognito browsing it is harder to link multiple accesses to be the same person (though browser fingerprinting may allow this). But if there is a login of some sort - or some auth code from the AV service, it can allow all accesses to be linked together, even if not knowing the actual personal identity. With common AV systems it could allow multiple site's accesses to be correlated now without even the need for working cross site cookies / pixels, etc.

Update:

What is interesting is that age>>go have dropped ID check as a verification, and then a bit later dropped credit card check as well. They only have selfie and SMS now, but still - once an email address is validated you only need that working email address!

Update:

Oddly it has changed to Selfie and credit card now. There is shit going down behind the scenes.

Bloody Amazon

Once again, weeks of seller support tickets on Amazon.

The problem - re-use of an EAN - a simple matter.

What is an EAN?

A European Article Number, known now as a Global Trade Item Number (GTIN-13), is a code used on a barcode on a product. They are issued in blocks by GS1. We have a block. We assign to products. UPC (Universal Product code) is the same system.

Reusing an EAN?

When a product is discontinued, there is no reason for that EAN to stay assigned to the discontinued product and so it can be re-used. Or can it?

GS1 did have stuff on re-use of EANs, and time frames for discontinued products before reuse. But apparently now they recognise that platforms assign an EAN more permanently to a product record and they may not be recyclable. So they have changed policy on this!

This is interesting - EAN/UPC used to be assigned for a one-off fee and that was it, but GS1 assign on a rental basis.This year they seem to have doubled the price, even. In theory when you stop the contract, they can assign the block to someone else. But based on this new policy, they cannot - so why do we have a rental? The idea being one could get a block, use it, cease it, but know it can never be re-assigned to keep using for free. Well not quite, there was a contract which continues to say you cannot use once ceased. OK, but a different legal entity could use the codes now, knowing they will never be re-assigned to someone else. Yes, I asked GS1 this. No reply yet.

Basically the idea of GS1 codes being rented only works if they can be reassigned. If they can no longer be reassigned, then rental makes no sense. Also, they will run out with such a policy (which is why rental came in, AFAIK).

Just to be clear, GS1 retain all rights to the numbers they allocate, but I asked, and they could not say, what rights they are! They are not covered by copyright, trademark, patent or any legal framework of which I am aware. They have no rights that I can see apart from contract (which is only with contracting parties).

Amazon

So, Amazon use the EAN barcode, yay. We had a product, with an EAN, sold some, discontinued, and some time later I re-assigned the EAN two a new product. One would expect Amazon to have a process to handle this. It is not an odd thing to do AFAIK.

But I cannot make a new listing, as the description does not match that of the old, deleted, listing.

The issue is the total ineptitude of Amazon seller support...

• Told yes, I am entitled to re-assign an EAN as a GS1 holder (good)
• Told the existing ASIN+EAN cannot be deleted, so tough, so no new listing
• Told the EAN cannot be removed from the existing ASIN, so no new listing
• Told the EAN can be removed from the ASIN, but I have to report a violation (i.e. someone misusing our EAN). So I did that.
• Told eventually (many times over many days) the EAN has been removed from the ASIN, so should work to make a new listing (it did not).
• Told that an EAN cannot be re-assigned, tough. I asked if they lied before or are lying now.
• Now told the ASIN can be deleted, and that will fix, but I have to re-make the listing and then do some spreadsheet update to delete the ASIN. That is going to be fun, and no clue if it will fix.

In my view this should be simple - we prove we hold the GS1 allocation (easy), we state the (unused) ASIN+EAN is no longer valid - Amazon delete it (or remove EAN) - we make a new listing.

To be clear, if they had a system that an EAN was always stuck to a description, and were actually consistent in that, and GS1 agreed (which they sort of do), then that would be annoying but not as bad as this - they keep giving hope it can be (or even, has been) fixed, and then changing their damn mind.

But no, Amazon seller support is, without fail, a battle at every step of the way, every fucking time.

Update

I have someone with windows and Excel to try this... They made the file to upload. Thanks. But...

• They say "Create listings with a spreadsheet in any format - Al will convert it for you." Nice.
• They say "Accepted file formats: Excel, TSV" nice
• I load an excel file and they say "File Type: Inventory Loader File (Automatically detected)"
• They then say "Please upload a tab-delimited text file (file format txt or .tsv).  This feed does not support the type of file you uploaded. If you're using Excel, please convert it to a tab-separate file by following these instructions."

So yes, upload any file AI will sort, or upload excel file, and we recognise you uploaded an excel file that is an inventory loader file, but a final FUCK YOU you have to load text or TSV.

I mean what that actual fuck. We loaded a TSV, it may have worked, watch this space!

Update:

Did not work - zapped the listing but left the same problem, so I have scrapped the labels, and allocated a new GTIN/EAN.

Except that does not work - new unhelpful error at the last stage, no reference to any existing ASIN, no mention of an existing description, just "does not match the product you are trying to list". I tried loads of EANs and no joy.

WTF Amazon!

Update:

They want a video call to prove my ID (I have had to do ID for Amazon many times, so why?) and to see 50 of each product, and open and show brand on them. I don't have 50 of each product here. They are mental!

Updated regulator

I do a lot of circuits, and they all use the same basic design for power supply. OK, technically not quite, I have one for battery and one for USB+DC. But the latter is a basic buck regulator.

I revise the design from time to time and newer and better chips come out and as I learn more.

The latest design is using a TI buck regulator based design using a TPS562246. This is a big step up from the MD8942. The key difference is the old design was 600mA and the new design is 2A.

Do I need 2A?

For almost all of my boards the answer is no, the ESP32 can peak to 500mA with WiFi apparently. 600mA is close but enough. But some boards have other peripherals and you soon find you really want a supply rated over 600mA - even adding just 10 of the small WS2812 diodes can hit 100mA total when fully lit and some of my boards have more than that. Some have GPS modules and a lot more.

Why TI?

They are a well known and competent brand with good data sheets.

Will it make any difference?

The main difference is many of my designs would take a DC input 5V to 35V, and now 5V to 17V. I think for almost all cases this is not an issue. Being able to run off 12V DC or some 13.5V or similar battery based 12V, is what is needed (and I do all sorts of stuff at that level).

I think it will result is less ripple, and allow more peripherals within the power budget.

Fun?

As always, I am learning a lot - finding the right components, the inductor with low enough resistance, caps that work in the required temperature range, careful PCB layout. It has indeed been fun.

So, yes, many of my designs will undergo an upgrade over time.

IR cameras

I have an IR camera, well technically I have three now. One was a bit of fun, a FLIR attachment for my iPhone, which is somewhere in the loft. One was actually for work, a Fluke PTi20, and now a Hikmicro pocket2. The Fluke actually melted its own USB charging port (if only they had a thermal camera in checking that design!).

Why do I have these?

Working on circuit design I have to consider the power and components - which work within specified temperature ranges. A thermal camera is extremely useful for testing and confirming which components get how hot under load, etc. Even things like the thickness of PCB tracks can matter. So that is why I have one.

I am currently working on a new regulator design for my boards, one that allows 2A at 3.3V. I do not need that much for any of the simple designs, but some have a lot more peripherals, and some have 3.3V power pads to connect and power other devices from the board, so making a generic 2A capable design - this means testing under load. One of my first test boards has an underrated diode which is getting hot under high load, and highlighted my error. I don't have the full picture yet as I am awaiting the macro lens (a nice feature of the Hikmicro camera).

Gripes!

The biggest annoyance is that thermal cameras are all very low resolution (or alternatively very very expensive). For example, the Fluke is 120 x 90 pixels (really). The Hikmicro is 256 x 192. But I guess this is a limitation of the technology, and we have to live with it. Compared to the normal cameras I have this is quite an amazingly low resolution.

Tell us the resolution!

One of the biggest gripes is so many devices do not say the resolution. Many say the screen resolution. They will also say the normal camera resolution, but the IR camera resolution is buried in the small print if at all. Some are even more devious, saying the enhanced IR resolution which is actually upscaled from a lower resolution sensor. Please just own up and make clear the specification.

Transferring files

Another gripe is some cameras (notably the Fluke) make it really hard to just get the images. It had WiFi and even a dedicated send button, but that only worked if you have a cloud account and only sends to the cloud and you then have to log in to get the images. Maddening.

The Hikmicro seems to be saner. It has an app, and WiFi, but also has simple USB bulk storage to allow direct access to the images and videos (yes it does video too 25Hz). So at least that is sorted. 

Show us the image

The image stored is a fucking screenshot! I mean literally, it is the screen with any image blending and overlayed text and icons and scaled to the screen size. This was the same on both the Fluke and Hikmicro. And is just really annoying. The Hikmicro has an option to also save the visible image separately but no option to save the clean IR image.

On the Hikmicro it even removes the [MENU] overlay to take the screenshot and puts it back afterwards!

Why why why? The IR is low enough resolution, why lose loads of pixels covering with a colour scale and overlay text. I mean I can just about see the logic of a marker for a temperature point (optionally), and the hikmicro does make it optional but if not set you don't get the temperature reading at all!

But why not save the IR image clean, in a larger image with the colour scale and text overlays all off to the side (and ideally in some meta data too). This is simple in software, and there really is no reason not to do that.

The other annoyance is it saves in jpeg, a lossy format. FFS this is such low resolution there is no reason not to use a lossless format - just use png even.

If the camera had an option to also save the clean IR image, that would be good enough, but it does not seem to have a way to do that.

More

The macro lens has arrived, and it excellent, but IR only (blocks visual), which was a surprise. Even so, the results are impressive, and useful.

Wish list: Power/battery management IC

I mentioned I am playing with battery management. It is clear some interesting ICs out there. One of the problems is that there are so many ICs with so many variations, even if one just looks at one manufacturer like TI (who do some cool ICs).

So I thought I would try and explain what would be an ideal IC for some of my projects. From what I can tell this is all available in various ways with various ICs but not as one simple IC. I am not actually asking for much, honest. I have some of this pretty closely now but in a lot more components than is ideal.

There are some monster large QFN power management ICs out there as well, which seem overkill, and not sure I have yet found one with the "button" aspect.

So here is the list of what I would really like.

1. A small IC, probably not actually BGA, but small, a QFN package or even small SOP style if few enough pins. It does not need a lot of pins for what I am talking about. TI have some funky QFN style packages with larger pads for the power, that would be fine. Ideally not special handling at PCB manufacturer (hence not BGA).
2. A 3.7V LiPo connection for one LiPo cell.
3. A 5V power input connection, i.e. from USB.
4. A regulated 3.3V output - ideally 1A but at least 500mA. Can be switched off.
5. A simple push button input (see below). Maybe also button state output pin.
6. An I2C interface (see below).
7. Perhaps even a "reset" output, that rises once power has stabilised, drops before power off - this saves a couple more passives.
8. Perhaps even a LED output (I don't really need this).

Now the idea is simple, this provides both regulation for 3.3V working of a device, running from 5V and/or 3.7V LiPo, but also provides battery charging for the LiPo from 5V. There may need to be a pin to define charge rate, and even a pin for voltage output, but a fixed 3.3V saves more passives.

The regulated output could be a simple LDO type thing, or perhaps a synchronous step-down with integral FETs and small external passives. Minimal passives is ideal. Needs to run off 5V or 3.7V, i.e. work when there is no LiPo.

The idea of the button is simple. The device would have off or on modes. When off, and just on LiPo, its current needs to be a few uA at most, and no 3.3V output. But the button is used to turn on. The button could also have a last ditch "hold for 10 seconds" power off to allow it to be used as a system reset. It makes sense to turn "on" when powered up, and when 5V applied, as processor can decide to turn off if not needed.

I2C would allow checking state of battery, and 5V. Ideally an ADC for battery voltage, maybe even for 5V supply voltage too. Needs a way to check button state so it can be used when on for other functions, or perhaps a pin output to mirror the button. Needs a way to tell the device to go to "off" mode. A pin output mirroring button is probably best as this allows for processor (e.g. ESP32) in low power sleep when this chip is "on" and still use button to "wake up". There could be some special power regulation modes for extra low power but still "on" perhaps to allow this type of working as well as a proper "off".

I can't help feeling this fits the needs of a wide range of small battery operated devices.

If anyone knows of devices that tick some of these boxes, please do let me know.

Oh, and reverse polarity protection and reporting, thermal shutdown, etc.

Some feedback I have had, thank you all.

• https://www.ti.com/product/BQ25155
• https://www.ti.com/product/BQ25186
• https://www.ti.com/product/BQ21088

Faikin board ATE

The Faikin boards are a development board, a component to fit in to your Daikin air-con as your own project.

However, I do try and make sure they work before shipping, which means some testing. This means flashing some code and doing some tests.

Basically, there is always a small risk of badly placed, badly soldered, or faulty components on any board. It is rare, often less than 1 in 500 boards. But it is a risk.

TC2030-USB

The key component is a TC2030 lead - it means a simple pattern of pads and holes on the PCB (which essentially has no cost), and a suitable lead (which costs a bit more).

Oddly TC2030 seems to be one of the very few sprung pin header leads you can get for this. I would have expected way more, but they do seem pretty common and standard, and they just work.

The lead allows me to power the board and connect to USB to flash the processor module (ESP32-S3-MINI-1-N4R2). The programming batch process is pretty slick. I end up doing several in parallel and putting in anti-static bags, all pretty slick now. Yes, manual labour which I am not used to, but I am getting good at it. Amazingly the process of then labelling the bags takes pretty much the same amount of time.

The flashed code then operates the LED on board so I know it is all flashed and working. Simple. Basically, to get as far as an LED means partition table, boot loader, and code, all working (all signed and checked) to get as far as the sequence for a WS2812 LED. So this is not just some GPIO happens to be low, this is proof of code running cleanly - a good test.

But this only tests the USB, power, reset, and ESP32-S3 module and flash and RAM. Not all of the other electronics for Tx and Rx with the Daikin. These extra bits have some components because they have to level shift to 5V open drain logic with pull ups as used by Daikin.

TC2030-USB+

Can I test these remaining components? Well yes, I can...

The trick is to take the final pins that go to the Daikin for Tx and Rx, and feed them on to two spare pins on the TC2030 spare pads.

Then I make a custom lead that has USB but shorts those two pads.

This means my code can do a loop back test on the final Tx/Rx from the Faikin board, and indicate red not green on the LED if it is not seeing a loop back.

This tests almost all of then components (there are a couple of ESD diodes which it doesn't test other than that they are not a short - but that works as a visual inspection mostly).

Always valuable lessons learned as I go.

Battery life

I am doing well learning my way with battery based circuits. But it has taken some time to get the hang of it.

ESP32 deep sleep

The first thing is the ESP32 can go deep sleep, with controls of some GPIOs that can have set states and be used to wake from sleep, or a timer. Using this with a simple LDO regulator from a 3.7V LiPo I know I can make a device that lasts for several weeks on a charge from a pretty small battery because the Watchy does this. The nice thing is this is deep sleep most of the time, with timed wake to update display every minute. It works, and I have made similar circuits myself.

Leaks

However, some of my designs are not as long lasting, maybe a few days of idle/off time. The reason is that the idle current needs to be very very low, a few uA. Deep sleep on the ESP32 is that low, which is ideal.

Well, except you can set GPIOs in some states and that could use current. If careful though, e.g. week pull on a GPIO input to a button which is not normally pressed, you can indeed keep the current that low.

But there is one annoyance that has led to standby times of more like a day, and that is the nice WS2812 style status LED I use. Its idle current when off is still typical 350uA, which, by comparison to the CPU is huge! There can then be other devices that use power, and even the LDO will leak some.

The solution was a simple pair of FETs to switch the LED off, though for a single status LED I could probably power directly from a GPIO, which may be a neater solution.

This has led to many days standby - well probably indeed the few weeks as I measured on the Watchy.

More LEDs

Once example where this was a lot more of a problem, leading to standby times of more like a few hours at most, was the IronMan glove with 88 LEDs, so idle was more like 30mA. That was a mistake, and one reason I wanted to improve things a lot.

Point of load / power distribution ICs

Whilst a simple pair of FETs works well for controlling power to LEDs, there are some devices really intended for this, which are very cheap, and really quite clever. An example is the TI TPS22916. It is tiny (0.78mm x 0.78mm), has power in and out, GND, and a control input, and that is it.

Its job is to switch power input to output, simple as that, but it has a slew rate to reduce input current surges, and the control input is a simple logic level, active high, that will switch from as little as 1V, so will work directly from a GPIO, no other external components at all. It also has thermal cut out.

This makes it much smaller and neater than a dual FET, plus two resistors, and better and safer (the slew rate and thermal cut out). It also has a standby of typical 0.1uA, so basically nothing.

This makes it ideal to control the LEDs, even when only one status LED.

But there is more

I think I can do better. The idle ESP32 has some current and the LDO will have some even when idle. What if I could eliminate almost all power usage when off.

This is my plan...

I'll explain...

This idea is the LiPo (the connector at the top) feeds in to two of these switch ICs. It could be just one if all I wanted was an on button. This idea is the button links the battery to the switch IC control - the switch IC even has a built in pull down so no extra component needed. When button is pressed power comes on.

Then a GPIO (PWR) is driven from the processor - if I only had one switch IC it would be linked to the button to logically hold it pressed. This keeps the power on until the processor commits suicide by lowering that GPIO level.

By using two switch ICs I can still keep things on from a GPIO, but by linking the button to a GPIO input, I can still detect the button pressed when on. Button press does nothing for power when power is already on, but the CPU can tell the button is pressed, and use that, including using it as a power off switch if wanted.

The concept is that when logically off (and this means no wake up on timer, properly off), the only leaking current is of the order of 0.1uA (x2) through the switch ICs. Not even leakage via the LDO or the deep sleep processor applies. This should allow the battery to last a long time, e.g. even a tiny 150mAh should last decades in idle, with the batteries own internal leakage being more of an issue.

So, ordering some sample boards to test.

Update:

The new boards have arrived, and the good news is that the power on/off works as expected. It was a tad fiddly making power down work, needed held GPIO in open drain mode, but it worked.

The only snag is that the button does not work on USB only with no battery. It connects the input to the battery, and there is no battery. My hope was the charger circuit would be providing some voltage on the battery pins that would allow the button to work, and indeed, at one point it seemed to, but no, this is not the case. So I need a new way for the button to work. The main idea is a two poll button, which means one side to a GPIO for button working logic, and the other to the power control for power on. The good news is that can make it work using just one power controller and not two. Now to hunt for a nice two poll button!

Tax, duty, VAT, GST, Tariffs on import?

I sell stuff on Tindie. (see www.tindie.uk).

Now, when you get it, if not in UK, your courier may charge you some amount of tax, duty, VAT, GST, tariff, or admin fee on receipt. Pain in the arse, I know.

So could we sort this as seller?

Well, basically, we have no way to know what tax, duty, VAT, GST, tariffs, or admin fees might apply. It is not even a simple per-country thing - it depends what you are selling, and the country, and in some cases even matters what region in the country where you are located. If you are registered for VAT/GST in your country you may be able to avoid some charges on import. It is simply impossible for us to work this out, even if we wanted to.

And, of course, in some countries (sorry USA), the tariff may change from day to day at the whim of a mad man, so really we would have no clue. 

Even if we knew?

Even if we knew the exact amount you are liable to pay, I have no easy way to "pre-pay" that as sender. Some couriers clearly do this (my experience being EU based VAT, and even some US shippers to UK, surprisingly), but Royal Mail, which we use, do not seem to have any way - we just have a customs form to complete on sending, and you end up paying the extras on receipt - we have not way to pre-pay your import costs.

Even if we could pre-pay?

But even if we could work out what is due, and even if we have a way to pre-pay it for you, what then?

Well simple, it adds to the cost for us, so we would simply add to the price. Indeed, that could mean more, as setting our price higher to cover such charges probably impacts the charges as they are often a percentage of sale price. So actually, if we could work out what was due, and pre-pay it, it may mean we end up charging you even more that if you just paid it on import.

So, tough.

Simple answer is we quote a price, we charge a price, and you get to pay your government any import duty, tax, VAT, GST, tariffs or whatever when it arrives. Sorry.

Canon CT-21

So, I like it. But there is some learning...

Colour

One aspect is setting colour options, and ensuring it knows the paper type, else you get too dark or too blue, or whatever. This is a general thing anyway with colour printing, but on my PRO-1000 I only printed on the same proper photo paper all the time, and so it always just worked. On the CT-21 I have a roll of A1+ "coated" paper. It needs slightly different settings.

Paper Size

This is where it gets fun...

The paper is A1+ width, i.e. 610mm (A1 is 594mm). Fair enough, it means you can print over size, and cut down with bleed over the edge of the print. The catch is you need a big cutter or are good are cutting against a straight edge.

Just to be clear, A1+ is not the max size, it will print up to 18m long from a roll - a banner, which is just crazy!

The print driver essentially allows basic size options, for example, for A1, it can be A1 with 5mm margins, A1 borderless, or A1 oversize borderless. The A1 is great if you are putting in a frame with more than 5mm edges. A1 borderless is great if the frame has little or no border, or if cutting to A1 exactly. Oversize A1 is great for that bleed over the edge, but still has margins of a few mm on the A1+ paper each side and at the ends of the print.

Then there is a "scale to borderless roll" which goes to around 609mm wide, a slight white edge each side. It also cuts the length in to the edge of the print by a few mm. This means you get an over size print with two cut edges and two very narrow white edges.

Confused? Well, yes, too many options.

What I did was make the oversize with print close as you can, and ended up with nice over size A1. The snag is they don't fit in the clip frames so needed cutting, on all four sides.

Now, it can, of course, do A2 on the same roll, sideways - well A2+ (all the same issues, just on the other two edges).

But A3 and smaller I can use the sheet feeder and plain or photo paper, etc. However it does not do borderless at all on the sheet feed, always at least 5mm borders, so a tad more limiting.

Three sided borderless

There is yet another option - 3 sided borderless, and I realise now that this is actually what I need to use.

It prints the image the size you selected, so typically A1 exactly (to then go in a frame, but no scaling or cropping involved from the original artwork). But prints it as far to one edge as it can, and then cuts the start and end. This means you have two cut edges, one very close to edge, and one margin. The margin means only having to cut (or fold over) that one side of four. This makes it very easy to put in a clip frame, or only one cut to use with no frame.

This also means I could print smaller sizes, A3, A4, etc. These are extra easy as the (now quite large) margin fits in my guillotine. It also means all the same (roll) paper with the same colour settings, and edge to edge, unlike using the sheet feeder.

This is an example of A1 three sided borderless - only one side needs cutting or folding over. This is 840mm left to right.

Some more of my videos

I have done a couple of videos on my circuit boards.

The fun one, I hope, is the "shopping channel" style one.

But also a more instructive one on making circuit boards.

And one on the Faikin boards specifically.

Canon PRO-1000 vs CT-21

I have had a Canon PRO-1000 for some time, it is an amazing printer. It is a must for someone producing professional photographic image prints. I cannot stress how good quality it is.

But I have just replaced with a Canon CT-21. So I felt it's worth explaining why, and what difference it makes.

Do I need a photo printer?

The first point is whether I need a photo printer. It is my only colour printer, and I do print photos for the family that get framed and go on walls, and so on. But I print stuff like a menu with a colour logo on it, etc, which seems overkill to use a PRO-1000.

Basically, I probably do not need a professional photo printer, really. It is just my usual overkill.

What is my use now?

Given the ability to print up to A2, I have actually been using it for posters for the pub. They are stunning.

Is A2 big enough?

The catch is, posters at a pub, may want more, like A1. So I decided to get the TC-21, which is an A1 poster printer that takes roll paper feed. If someone accidentally orders A0 clip frames at the pub, I may be in trouble.

What of TM-255?

I considered the TM-255 - it uses 5 inks, not 4, but has a base, and is just so big. It looks great, but I simply did not have space. The TC-21 fits on my bench. The ™-255 is also high volume and stupidly quick, but the TC-21 is not a problem for what I need in terms of print speed.

Key points in favour of the PRO-1000

• The photo print quality is just amazing, cannot be faulted.
• It does edge to edge printing on most paper sizes (oddly not A5), which is very cool. Even A2.

Key points in favour of TC-21

• It does A1+. Indeed, it will do 610mm wide and 18m long if you want!!!
• It does roll paper feed (which I am using for A1+ or A2+), but it can take narrower paper roll.
• It does sheet feed (no need to take out rolls feed, just manual sheet feed via top loader), all the way down to A6. Very handy.
• Easy colour refill - the PRO-1000 is easy, but this is only 4 inks.

Some downsides for PRO-1000

• It uses 12 inks, this is a feature in terms of quality but a pain in terms of costs and hassle with ink. It also gets through "maintenance kits" which also cost.
• If you do not use for a while, weeks, it will waste a lot of ink and take ages to start up. But it will then print perfectly. I don't know if the TC-21 does this.

Some downsides of the TC-21

• It is bigger, check dimensions, but wider by quite a bit. Well, obviously it has to be.
• It does not do edge to edge well - it can get very close on roll feed, maybe 1mm each side, and cut with no margin top/bottom. But not at all on sheet feed. This is perhaps its main lack of feature.
• It is only CMYK not 12 inks, but the photo printing is still good, and good enough by far for some family photos.

I do not know running costs yet, but the results are pretty good.

IR LED controllers

I ordered a couple of LED controller remotes.

Now to work out how they work.

The first (big one) is simple 32 bit NEC coded (i.e. address, inverted address, code, inverted code) with address 00. Laid out as follows for the code:-

5C 5D 41 40

58 59 45 44

54 55 49 48

50 51 4D 4C

1C 1D 1E 1F

18 19 1A 1B

14 15 16 17

10 11 12 13

0C 0D 0E 0F

08 09 0A 0B

04 05 06 07

As you see, this has some pattern to it, but why not literally a proper grid or sequence, why this. But OK, I can cope. I plan to make my LED controller boards have IR receivers and understand this in some sane way - with extra keys like DIY1 do device automation to Home Assistant.

But what of the smaller one?

Well this is special, it is 32 bits, just like an NEC remote, but the first bytes are 00 EF not address and inverted address. I mean, it is distinctive, and consistent on all keys, but not quite standard. The code is still code and inverted code though. The codes are then:

00 01 02 03

04 05 06 07

08 09 0A 0B

0C 0D 0E 0F

10 11 12 13

14 15 16 17

This is way more logical key codes.

Simpler

For completeness, this is a nice simple IR from Amazon.

Which is address 00 and codes:

45 46 47

44 40 43

07 15 09

16 19 0D

   18

08 1C 5A

   52

Which, as you can see, is all over the place!