Time for something new...

Latest project is a time server.

So why? Well, they exist - you can buy a really nice LeoNTP server, with impressive specs. We see response times of 0.1ms, and it claims 100,000 requests a second. They also have a PPS output, and can do a calibrated 10MHz output apparently (I can't do that).

Can I get close using an ESP32? Well, sort of.

Challenges

It should be simple, in theory - a GPS module, capture CPU cycle count on PPS interrupt and use to get clock rate. Capture NMEA to get time for PPS. Capture cycle count on NTP packet and use reference cycle count, clock rate, and reference time to know exact time of day to fill in NTP reply. yay!

Of course it is never entirely that simple. I did all this and got a working system, but I could do better. My latency as measured on a FireBrick was 2.5ms. I did averaging of the PPS intervals to get a more consistent clock rate, and often the standard deviation on that was below 10ns! But not always. I then did a best of last 5 seconds in terms of a PPS interval to use as a reference - that way the odd delayed interrupt had no impact. That seems to work.

One thing I wanted was PPS interrupt at higher priority than Ethernet, but all GPIOs on an ESP32 are the same interrupt source. Bugger. Do a search and you see plenty of people pissed off about this. I found a fix, make the PPS a PCNT (Pulse count) which has an interrupt (count to 1) which can be set separately to the Ethernet interrupt. There is a trick to remember.

The hardware

The PCB design is not that hard now I have cracked Ethernet. A main PCB, with USB-C, DC input, PoE, and Ethernet.

I have since made an even more compact design.

This then connects to a GPS module, which you can hang out the window (or better, fit in a Stevenson screen). Link with 5 core cable (solid cat5 is ideal), and it can handle a few metres.

Then put both in a nice 3D printed case. See https://shop.revk.uk/ if you want to buy. £60 not £600!

Making it faster

The fact my response times where around 2.5ms was not ideal. I wanted better, but how.

First bodge was hook in to the Ethernet driver receive code and check for NTP packets as they come in, and do a direct reply. I got latency down to 1ms, yay!

But I can do better :-)

Scrap the Ethernet controller altogether. Write my own custom low latency driver. Dedicate CPU1 to PPS and Ethernet only. My custom driver can...

• Be reactive only - reply immediately to ARP, ND, and NTP.
• Being reactive, no need for mutex wrap on access to Ethernet chip.
• Work in interrupt (yes, nasty, but dedicated CPU) so not task switching.
• Know state of registers to avoid read-modify-write as used a lot in standard driver code.
• Do read fifo and write fifo with no clean-up between

And guess what - latency down to 0.1ms - bang on what I wanted. Indeed I have seen 0.077ms even.

Now, the logic is fun - it makes the Ethernet no use for anything but NTP. So I have made it (a) optional, and (b) normal Ethernet for first 2 mins so you can access it via Ethernet if needed (assuming you can control power/PoE to reset).

But once switched to low latency NTP only Ethernet, you have to use the WiFi for any access, management, MQTT, and so on.

Bugs

I found bugs in the Ethernet chip (KSZ8851SNL) which does not check IPv6 UDP checksums correctly!

I also found it almost impossible to convince the Ethernet chip to give me unicast, multicast, and broadcast packets - in spite of a lot of reading the data sheet and trial and error. It is now in promiscuous and relying on the switch to protect it. Even that makes no sense - you have to set "Allow any packet" and "Invert the input filter" which to mean means "allow no packets!". I may yet find a working setting. I have already spent a day on this.

I also found it used edge triggered interrupt with a check in task for ISR set as well on timer, I changed to level triggered and that seems to work without hanging.

I also found it nearly impossible to cleanly decommission the Ethernet driver in ESP IDF. I managed to take over interrupts and kill the task, and that was all. Anything more thorough was a nightmare. But that works.

What is it good for?

Well some people are putting in the UK NTP pool, it seems to meet the requirements well and work at a level similar to the rest of the pool.

But it is mainly aimed as being the main server in any business / office. Include with your pool NTP, but being local it will win.

I do not know if it would get close to even 10,000 requests a second, but that is not needed for a typical business, even a large one.

Shipping to EU from UK

We are selling some stuff to EU, and all over the world now (see shop.revk.uk).

This is all a bit new to us, as we have traditionally only dealt with UK customers. This started with circuit boards stocked at Amazon, and Amazon just sorted the EU VAT (paying us the VAT exclusive amount as an export) for sales in EU - very slick.

Moving on (after Amazon screwed us over) we started selling on Tindie, and Tindie do not do anything special for EU at all. So all our parcels hit customs, and the recipient had to pay VAT (and admin fee). This is the default if you do nothing special when shipping to EU.

However, when Tindie screwed up badly (will they ever recover?) we started selling on Lectronz. I have to say that they have been great. This led us to start to understand how it should be done.

IOSS

The Import One Stop Shop system in EU means we could get an IOSS number for sending small packages (value up to €150) to EU with VAT paid - this means the customer has a much quicker and more seamless delivery with no hassle from customs. We would have to file returns for this, but pay in one go for all EU.

However, it is even easier when using an on-line marketplace like Lectronz, as they handle all the VAT - working out the right rate and charging the customer and paying the VAT. All we have to do is quote their IOSS number as part of shipping.

We integrated the Lectronz API and the Royal Mail Pro Shipping API so staff can just click "print postage" and out pops the postage label and customs label, simple.

The EU has gone to some effort to make this slick and easy for small parcels going in to the EU. Well done EU.

The new €3 duty

As of 1st of July the EU made things difficult - the exact opposite of the slick and easy IOSS process - adding a new €3 import duty on small parcels value up to €150. This is separate from IOSS, so even with IOSS and prepaid VAT, the recipient gets hassled by customs for €3 duty!

Lectronz moved quickly, and added the feature to their systems to work out the €3 duty, and add to the price, to pay us, so we can post "Duty Paid". I updated the API code and set to send duty paid via Royal Mail Pro Shipping API.

Yay, all sorted, all slick again, albeit €3 more expensive for the customer (thanks EU!).

Well no!

This started off OK, with a few parcels, but then we hit an error where Royal Mail will not do Duty Paid to Latvia. Arg, why not? Lectronz were very helpful sorting a refund of €3 so we send not duty paid and explained to customer.

So I decided to check the rate card from Royal Mail. The rates to EU are a mess anyway - they have a base price and per-kg price. The base price ranges from £4.45 to £7.70 which is not too mad, but the per-kg ranges from £0.80 to £10.85 which is mental.

The surprise is that duty paid is not available for Belgium, Bulgaria, Croatia, Czechia, Denmark, Estonia, Finland, Greece, Hungary, Italy, Latvia, Lithuania, Poland, Romania, Slovakia, Slovenia, and even Spain.

This is crazy - with all EU countries charging this new €3 duty - why the hell do Royal Mail not handle this.

The other oddity is that the extra cost for sending duty paid, where possible, varies wildly - from a more reasonable 50p to nearly £5 in one case. It is daft for a customer to pay an extra £5 to handle a €3 duty.

What have I done about it?

I have done a shipping rate for all Europe countries individually. Yes, tedious.

Where duty paid postage is available I have listed that (with the extra cost) so customers have a choice.

It would help if Lectronz allowed me to code a per-kg rate for some countries - I have asked.

We still do IOSS for EU where applicable.

So yes, sorted for now. And it means a sting of countries have the hassle of paying duty on receipt, and a few others make sense to do that anyway as the extra cost is daft.

Update

And today we have...

• Shipment to Romania saying IOSS not allowed, but we have shipped to Romania with IOSS before, and it is on the rate card.
• Shipment to Germany with no IOSS and no duty paid, a simple DDU but over €150 and they say NDX not allowed. Well NDX is simple "non document". We tried HV which is "Non documented high value" in case over €150 was the issue, but still says NDX not allowed (yes, I checked the JSON, we set HV).

Arrrrg!

Update

Now seems there is a PDDP with non PDDP service to do the €3 duty and IOSS at once somehow. Trying to make sense of what Royal Mail have sent. If so, it may be simple, but this is mental.

Update

Nope, they sent me a link to a document saying MP7 (DDU normally) with DDP and IOSS is available, but an email saying MPR (DDP but only to some EU countries), but then sent the original PDF again which says MP7 and DDP.

They clearly have no clue!!!

That CO alarm is giving me a headache

OK, yes that is an old joke, but I do have a slight mystery over the CO alarm.

This first happened last year, and drove me round the bend. A chirp every 49 seconds. This is normally a smoke alarm with low battery.

The problem is that we have 8 alarms in the house, and trying to work out which is chirping is not as easy as it sounds, especially when it is 2am, as it always is with these tings.

I actually ended up replacing every battery and still had a chirp. I then remembered there is a smoke alarm in the loft, and replaced that, only then to realise it was the CO alarm  in the loft! I went to replace that and found it is mains only, and Ei3018 CO alarm. Annoyingly it continued to chirp for some time once removed from the power.

I actually ended up buying a new one, and has been fine for over a year.

Then this week, it happens again. Thankfully I remember the loft this time. What is extra odd is that when I opened the loft hatch, the chirping stopped!

The next night it started again and did not stop. So I removed it, and waited. I put back in place next day.

The next night it started again, so removed, and new one ordered.

But I decided to actually read the manual, and it is odd.

It did not alarm!

Now, a key thing here is, it did not alarm. I would know, I have been in the house all the time, it is linked to all the other alarms, and to a relay input to my alarm/monitoring system as well. It did not do an alarm, honest.

But the manual says it has a memory mode, where, for 24 hours after an alarm, it will chirp. There is however a problem with this.

1. It had not alarmed!
2. It stopped chirping when I opened the loft hatch, so was clearly not doing a 24 hour chirp.
3. It is a single chirp - the manual explains it does 2 or 3 chirps, etc, for different types of alarm.

However, reading further, the manual does have a single chirp every 48 seconds. This is for "AC mains off or low battery backup", or (with green LED) low battery backup. I do not think it had an LED on.

So it does indeed sounds like the backup battery is depleted and the action is "replace alarm".

But this is just over a year for a device that should last over 10 years, arrrg!

I wonder why?

ESP IDF v6

I do not wish to speak ill of such an awesome project, but, this is head banging against wall stuff.

ESP IDF v6 is out, and there is a migration guide, and a lot of breaking changes. I appreciate these happen, and appreciate a guide to help.

But today I have hit some major stuff with no simple answer.

SHA1

Regardless of security issues, SHA1 is a function needed by some protocols and applications, and generating an SHA1 should be a simple common standard function. ESP32 even has hardware to help AFAIK.

It was in mbedTLS, but no, now it is not, and is some other API, but the ESP IDF does not actually list the functions.

But there is an esp TLS thing with a simple SHA1, except you have to initialise a whole TLS subsystem to use it. I just want to simply do an SHA1 of a buffer. That is all. I do not want to create some huge config object and init a whole subsystem. At this point a local SHA1 coded function may be in order.

This should not be complicated.

I ended up, for this application, changing to esp_rom_md5, which is just stupid. It will be fine for this application, security is not the issue, just a sane hash, but something as basic as a simple SHA1 function should not be this hard.

HTTPS

I have an application that does a lot of https client requests. This used to work well.

Now it breaks it horrid ways - the https client itself may break, or other things break that need memory. Even a simple SPI bus init fails after a few https fetches. My best guess is a memory leak. Long ago https client was not good, and was fixed, but now it seems seriously broken once again. I know this is hard work for an embedded system.

Work around, and I do not like, but files are signed, is use http for now for this application. Will be back to https as soon as this is sorted.

Arrrrg!

I'll keep at it.

A new UPS scam, it seems.

I think I am seeing a new scam.

Background

When an item is delivered to UK from overseas, we, as recipient, may have to pay VAT, and occasionally duty, as the importer. It is a legal requirement.

Yes, as a business we have "postponed VAT accounting", and even the possibility of a "Duty Deferment Account", and DHL get some credit here for handling both very well, with no admin fees. UPS do not get any credit for this at all.

But as a consumer there are two ways this goes down.

1. They refuse to deliver unless you pay first, or they demand payment on the doorstep. The scam is they want an admin fee that was not pre-agreed with them, and not part of any contract. And you have no choice if you want the parcel. It is a scam as it breaks pretty much all consumer contract protection laws, and is admin that is normal and so should be part of what they charge the sender, IMHO.
2. They bill you later, and try and charge the admin fee as well. Usually paying the legally required VAT/Duty and NOT paying the admin fee, can work. They do not like it, but I do not think they have any legally enforceable right to their admin fee. Even so it is time consuming and hassle, and I really need to publish an admin fee I will deduct from such payments and argue that is as valid as theirs.

So yes, un-agreed admin fees to recipient are a scam. That is my view anyway.

Note: Royal Mail have a law allowing them to charge an admin fee, couriers do not. There fact there is a law especially for this - kind of proves it would not be legal without such a law.

A new scam

I am now seeing what I assume is a new scam. This time by UPS. Yes, I believe this is a scam.

This relates to a shipment with Duty/VAT pre-paid by sender. So no charge to recipient. No legally required payment by recipient. Sender PAID to get parcel to recipient duty/VAT pre-paid.

In this case a parcel ordered on Amazon UK (no clue non UK shipper). And Amazon do generally handle everything pre-paid Duty/VAT. They are actually really good at that, and for shipments to EU are "deemed supplier" and handle local VAT and all sorts. Very neat.

The item had zero VAT (condensed milk, but declared as tomato sauce!).

But UPS decided to send an invoice (after delivery) for £6.65+VAT (£7.98) for an "entry prep fee".

Did the recipient ask them to do any entry prep? No, obviously not. That is a normal part of delivery, something they are charging the sender for. So this is a case of charging the admin fee even when there is ZERO VAT or Duty to pay.

It is not a lot, but I bet a lot of people pay, and UPS must handle millions of parcels. This is a big scam, and needs to be reported.

I think this is time to report this fraud to the police.

PCB designs, Ethernet, and PoE

First off, I am working on adding Ethernet to my ESP32S3 designs. I am going for an KSZ8851SNL SPI Ethernet MAC+PHY, mainly because the ESP IDF has a module for this - so it should be a couple of lines of code (LOL).

I am also going for a Silvertel PoE module.

At this stage I have designed, and ordered, a board to allow me to prove the concept. It is somewhat more expensive than I hoped as I am using a large mag-jack, but I should be able to improved that. At this stage I want to prove it works, both Ethernet and PoE. I should have boards late next week.

But this has led me to thinking of a couple of changes to designs of some of my boards.

Faikout Remote

The Faikout is a small board you fit in a Daikin aircon. It works with Home Assistant, and that is how most people use it, but it also has a simple web control page via WiFi.

The Faikout Remote is designed for situations where this hi tech approach is not ideal, e.g. my guest room, and people without HA. It has a display and joystick button to show temperature, settings and allow control of the aircon via the Faikout. It actually connects to the Faikout over BLE, so can be used when no HA, or even without any WiFi.

Is is USB-C powered so easy to put on a wall with a simple USB lead. It has a temperature sensor, so can work as the temperature reference for Faikout auto mode, but can also work with small battery powered BLE temperature sensors if you want the reference elsewhere, e.g. by a bed.

The problem is I have gone overkill on this. The design also has a load of environmental monitoring sensors, including an expensive CO₂ sensor. This makes it unnecessarily expensive. So I plan to simplify things.

New Faikout Remote

The new Faikout Remote will have the display and joystick (though a more robust joystick), and temperature sensor (but also works with BLE sensors), and an IR receiver, but that is all.

The idea is it is cheaper and just works as a Faikout Remote. The IR is because you can use it with a Daikin IR remote (cover the actual IR receiver in the Daikin aircon).

This is now sold on shop.revk.uk with resin case and display. 

Env monitor board

Having removed all the environmental stuff from the Faikout Remote I would make a new Environment board. It would handle a display, but obviously this is optional as many cases do not need a display. It would include the sensors that were in the Faikout Remote...

• CO₂
• Humidity
• Temperature
• Pressure
• Noise
• Light (including colour)

The temperature uses a built in SHT40 temperature sensor (on thin long track with no ground plane, for some thermal isolation), temperature from SCD41, and from GZP6816D, but also WAGO for DS18B20, and, of course BLE sensors. The DS18B20 are commonly available as a probe on a lead, and one or more can be connected. This is ideal for things like data centre use, for front and back of rack, or home usage for inside and outside temperatures.

I have also made it so the display can be fitted to board, or use the supplied lead and plugged in to board.

It would not have the joystick nor IR receiver.

Connectivity and power

This is where it gets more fun - the plan is Ethernet, so it can work on WiFi or Ethernet, and PoE so it can work on PoE or USB-C or DC 5V-28V. The PoE would be a module (solder in), so optional.

• Home use would typically be USB-C and WiFi
• Office use (and well cabled home use) would typically be PoE
• Data centre use would typically be Ethernet and USB-C, as data centres usually do not have WiFi nor PoE.

This should make it a very flexible environmental monitor, which is increasingly important for things like workplace monitoring for health and safety, as well as home monitoring.

This would be sold on shop.revk.uk with options for PoE, Display, and DS18B20 sensors.

I expect to get started on these soon, but obviously I really need to wait for my proof of concept first - that does not usually stop me :-)

Latest render...

Rendering

I have a simple task, it is auto generating some images for PCB designs.

This all worked, nicely. There are two parts

• Get KiCAD to render the 3D model of board, and make an image.
• Get OpenSCAD to render a view of my case design.

In both cases I want a transparent background, anti-aliasing, and cropped around the edges.

KiCAD

KiCAD has a command line that lets you make a nice image render, on a transparent background. However I was not easily able to control which layers show, and it showed stuff I did not want shown. This may have improved since. But also I wanted to change a user layer to an Edge Cut and remove the original Edge Cut to allow me to render a broken out version of the PCB. So I wrote a simple function to edit the KiCAD file before rendering.

I rendered larger and then auto-cropped in gimp.

In panel view

Cut out of panel

OpenSCAD

For the case designs, I have OpenSCAD make STL files ready to print, but I also wanted a render as a PNG. So I asked OpenSCAD to render the STL. STL has no colour so I picked white. But OpenSCAD does not render on to a transparent background.

I got around this by setting white/shades for the object and a very specific light blue background which is not in the object. OpenSCAD did not do anti-aliasing which makes if very easy - render larger - change the blue to transparent, and scale down for anti-aliasing.

I did all that in one imagemagick command. Worked well.

Upgrades

I have since upgraded KiCAD, and OpenSCAD, and gimp.

GIMP

GIMP is really good, and I do not want to dis it, honest. But gimp scripting is nightmare.

You would expect something like a simple auto-crop would be a simple command line. It is not. I had to make a script-fu script and put in the right directory (which was version specific). On upgrade I have to move it, but did not work. I have to tell the command line which interpreter to use now.

Even then, the script is a nightmare. I just want to auto-crop and images, so maybe three steps: load image, auto crop image, save image.

If you search you find examples (not for latest gimp) like this!

I managed to make a much smaller script, but still. One annoyance seems to be that there is no one simple place to find documentation of what commands exist and what arguments for scripts. Also, confusingly, some are plugin- and some are gimp-. I got it working but mostly by trusting examples I found not finding clear documentation.

When I upgraded, it did not work. This was difficult. Most commands had changed name, as had some arguments. I eventually got loading the file to work, and setting background colour, and even autocrop. But save file would not work. It seems a path and a string are not the same somehow, but my filename string worked for load, not for save, so inconsistent.

I gave up. I realised that I was doing this for the KiCAD PCB images as well, and in that case I used imagemagick.

In imagemagick it is magick filename -trim filename, so really simple, and worked.

KiCAD

The KiCAD update created something odd where it was showing the User.1 layer I used for V-Cuts, which it did not before. I had to change my tool to strip that layer. And now the images cropped correctly.

OpenSCAD and Background colour

But the OpenSCAD was bugging me - this was not a new issue as such, just I ignored it before. The PNG images seemed to have a black background. Noticeably when I look at a thumbnail on MacOS. This was not happening on the KiCAD based images.

I assumed it was the background setting in the PNG, and that was somewhat confirmed when I used pngcheck which reported the image was a 16-bit greyscale image with a background of 0x00FF. That is nearly black. I found no way to change this. imagemagick seemed to be setting only an 8 bit value for background on a 16 bit greyscale image.

In desperation I coded some C to update a PNG to set the background to a white level, got it all working, but nothing changed! My code was setting to 0x00FF. I tracked down, the image is 8 bit greyscale not 16, so 0x00FF is correct! pngcheck was misreporting as 16 bit (or was counting the 8 bits alpha as part of it, confusingly).

The fix, eventually, was to avoid greyscale as it clearly upset the thumbnails. I forced the original format which was 8 bit RGBA but simply prefixing the output file in imagemagick with PNG00:

This was a lot of yak shaving this morning.