## Friday, 9 March 2018

### Limits and tolerances

Frustrating how this is dumbed down for the public and, IMHO, for no good reason.

We all know that the ASA have a big issue with someone saying that broadband can be "up to 80Mb/s". They need it dumbed down to a 90th percentile or some such.

Indeed, even OFCOM want a "range" of speeds on an individual line estimate, but rather than a range of the "lowest speed we would ever expect without a line being faulty" to "the highest speed physically possible on that line length", they actually want 20th to 80th percentiles, to dumb it down, and create a case where 40% of people see line speeds that are not in the "range" that was quoted.

So, some simple bits of logic here, which any scientist or engineer will understand.

Tolerances

If I quote a simple single figure for something, such as 100mm, there has to be some tolerance for that. E.g. let's say I am telling you the length of a metal rod that is to be used as a part in some machine.

You cannot make a metal rod that is 100mm long. Basically, at the edge of that rod will be atoms, and at the very best you can make it one atom more or one atom less in length, and even the "edge of the atom" is questionable. If 100mm is not an exact number of atoms long, you have not made a rod 100mm long. It is just close, even if very very very close...

In practice I may say 100mm ±1mm. This creates a range of 99mm to 101mm. You can aim for that, and will meet it or not.

Whatever analogue metric you are quoting, length, temperature, voltage, as a fixed value, ultimately it has to have a tolerance. It should all be explicitly stated, though some specification may say "all lengths are ±1mm" or some such. In any case you need to know the tolerance to understand the fixed value you are given.

Errors

Very similar to the above, any measured analogue quantity will have an error margin, and it is just like a tolerance. If you say it is 20 in here, then that will have a margin, maybe ±0.5

Limits

If you specify a limit, e.g. a minimum, or a maximum, or a range, you do not need a tolerance.

A limit can be absolute. E.g. that metal rod must be <100mm. A rod that is one atom longer is not within limit.

Stupid regulations

Some of you may have seen my example label for the new FireBricks. Based on the specs of the parts we quoted 85V-264V AC. That is a range.

These limits are absolute, 84.9999V is too low, 85.00001V is fine (as it 85V).

But guess what? The figures we quote have to allow for 10% tolerance. WTF? Limits do not need tolerances, they are absolutes.

End result, we have quoted 100V-240V. The 240V with 10% is 264V. The 100V with 10% off is 90V not 85V, but we can handle down to 85V. We decided to give in to the madness as say 100V. We could have gone for 110V maybe. The same issues with frequency of supply, a range but with a tolerance on the values?!?!

But really, what is wrong with actual limits - why is this remotely hard - limits do not need tolerances.

As an engineer I would read that as 241V is too high, but apparently that is not the way the regulations work!

P.S. I saw a strange use of ± on train times, I think in Holland, but not sure. Basically they would say train due in ±5 minutes, when they definitely did not mean due -5mins. They actually means 5mins ±something, but was a strange way to quote it. It was almost as if they used ± as "about", when it means "plus or minus". Using ~, e.g. train in ~5mins, would make a lot more sense as "about 5mins".

1. The problem with “up to”in advertising is that a lot of companies can’t be trusted not to mislead consumers (such as those ones which sell the most powerful WiFi in the world). If I’m paying for “up to” 24 Mbps, then I hope they won’t mind if I pay them “up to” £29.99 per month.

1. Good grief! The "up to 24M" is "your line, has a length and loss and that is a matter of fact, and depending on that you may get any speed from 0 to 24Mb/s", i.e. "up to 24M" and for that you must pay "£29.99" (note, not "up to £29.99"). And you can buy a service on that basis. Why would you think that getting under 24M is anything other than exactly what you were sold for the FIXED price of £29.99, not "up to £29.99" (a term you made up) and not as advertised. Arrrg, Troll (I assume)

2. Sorry, didn’t mean to cause controversy or frustrate anyone, honestly... I think the issue is that from the consumer’s perspective, people want certainty about what they’ll be getting for their money. Most people have no idea how long their line is but can see that some houses get much faster broadband for the same price and it seems unfair to them. “Up to 24Mbps” might as well be “Up to 10Gbps” for all the difference it makes to the poor consumer who is on a 5 mile line and gets a 256Kbps connection.

3. Fair enough, but that also gets me annoyed that even with specific line details we are expected to quote 20th to 80th percentile, not any sort of certainty of the range!

4. Yes, that’s crazy... if the provider has the line details then it would be far better to say (for example) “12.5Mbps +/- 2.5Mbps for your particular line”. I would have thought most consumers would prefer that, actually! :)

2. The stupid thing about all this ASA crap with internet speeds is you have "UPTO" speeds that the service can actually exceed.

3. The mains in my house is often 250V because I am next to the substation. I have deliberately avoided buying things that say their limit is 240V because I worried my house mains might be a bit much for them, and looked instead for stuff that says 250V or better still 264V which seems quite common. So despite me having applied proper engineering principles, not to mention obvious common sense, I now find stupid rules undermine that.

50-60Hz doesn't fill me with confidence either, the UK mains is allowed to be below 50Hz.

1. Exactly my point, and why we originally said 264V, but the testers picked up on it so it had to change to say 240V

2. Testers are only doing their job. The people that wrote the rules on the other hand are clearly dickheads. What if the tolerance is changed to 5% and my 250V mains exceeds that? How would I know?

3. Not just allowed. The UK mains is *required* to be below 50Hz as much as necessary to bring the average frequency down to exactly 50Hz. :)

4. Fascinating... could you tell more about this please? Do they somehow mix frequencies? Wouldn’t that create a beat pattern? Also does this mean there is a natural tendency for the supply to be above 50Hz for some reason?

5. The Swissgrid explanation - https://www.swissgrid.ch/swissgrid/en/home/experts/topics/frequency.html - is pretty good.

Basically, though, the frequency of the grid is set by the relationship between power supplied and power consumed.

A big power plant is "just" some source of mechanical energy connected to a spinning generator. If electrical energy consumed is lower than mechanical energy supplied, the spinning generator speeds up, and the frequency climbs; conversely, if consumption is higher than production, the spinning generator slows down, and the frequency falls.

The grid operator tries to hold a flat 50 Hz, but can't - they can't adjust either supply or consumption fast enough, so they end up with frequency shifts across the course of the day.

Voltage is easier to hold steady - variacs and similar can be designed to adjust the voltage automatically to match the grid's chosen operating point.

4. The 10% tolerance does make sense if you think about it from the PoV of what power supplies can be connected.

Your 100V means that I can connect a power supply that is nominally 100V rather than having to look at the limits for each power supply (which often aren't given if you buy online)

Sometimes the spec does set *one* limit. The ancient USB spec had clients limited to drawing a *maximum* of 500mA and the server required to be able to supply *at least* 500mA

Which was why modems like the speedtouch frog would often work with lower quality computers, that didn't current limit to exactly 500mA but would have problems on higher spec machines.

But for normal use cases I think the 10% rule makes sense. Your equipment is 100V-240V. His equipment is 50V to 100V. My 100V supply works with both of them.

1. Well, yes, I sort of see what you are saying, but two issues:

1. That only works if the tolerance (10%) is the same for both things. It is in this case, but if you were in a country with 100V supply and 15% tolerance, the 100V-240V device may not work. To actually check you have to know the tolerances, and if different you then have to work out the actual range to tell!

2. If the device said 90V to 264V, you would also assume your 100V supply would be fine, and similarly equipment 45V to 110V would be fine. So forcing the labels to be basically lies does not actually help matters does it?

5. Can't see why anyone should want to refer to 264V these days (i.e. 240V + 10%). The EU ruled that the European systems should be be standardised at a nominal 230V but in reality nothing much happened, the existing 220V and 240 systems just carried on regardless. To reflect this reality, the UK limits were bodged to become +6% and -10% so that the upper limits (253V and 254.4V) are now virtually the same for both systems.

So if you really wanted to set absolute limits for a product that would work anywhere, surely they should be 90V (Japan's 100V less 10%) and 254.4V (UK's 230V plus 6%)?

1. Or original intent was simply to report the actual limits based on the specifications of the power supply module, i.e. 85-264

2. I see a suprising amount of equipment that does say 85-264 so clearly this is a common standard a lot of stuff is built to. But why did the 10% idiocy not apply to this equipment? 47 - 63Hz seems quite common too, which is much clearer than 50 - 60.

3. Well, these safety standards are new, we did not have to do them last time, so maybe that, or maybe just people not quite complying correctly. After all, there is not really a lot of enforcement of this shit.

6. Just state that the supply voltage should be between 85-264V including tolerances.

1. We would, if not for the requirement for a CE mark on the product before we sell it, and that the testing and compliance to be able to do that mean we have to state the range of nominal supply voltages, i.e. in our case 100V to 240V.

2. Apologies for bringing everything back to a popular/pet topic but...

Tesla used to quote 85-265V as the input voltage range of their cars' onboard chargers. But they've recently removed that claim (and are now silent on the subject).

For them I doubt the change in declaration was a regulatory thing - there's an odd loophole that electric cars aren't subject to CE testing because they're not considered to be 'appliances'.

Instead it's the charging points that have to be tested and approved, which is a bit silly since really they're just mains extension leads with safety interlocks.

3. Does the Tesla charge faster at higher voltages?

4. In short, yes.

Same is true of most electric cars - the AC chargers (and the AC charging specs) are all based around defined current rather than defined power so if the supply voltage is high then the car charges faster.

7. I can confirm that in Dutch, the symbol ± does, in fact, mean 'about'.

1. Thanks, but how does anyone Dutch ever real any technical drawing if that is the case? Wow.

2. This raises an interesting question: I wonder what they do in Germany. I bet over there they state the figures just like an engineer would. Does anybody on here know?

8. What's the tolerance on the 15W?

1. Interesting point. I wonder if it’s just voltage, frequency and current figures that are regulated - but not power consumption. It’s a crazy old world, not least because people are only ever going to run AC things at 240v or 110v anyway. I wonder if these crazy regulations also apply to DC devices.

2. Quite. We are working in the DC labelling next.

3. I have no mains feed to my garage. I use a UPS instead to run the lights, I have a pair of the UPS and alternate which is in the house on charge and which in use. The UPS has switchable output voltage, the higest it will do is 196V. I have to be careful what lights I use, I currently have a 32W electronic ballast fluerescent in there.

So anyway, there is a use for things other than 110V or 240V. For a start there is 100V and 220V nominal in some places, and then there are UPSs.

4. Lived in Japan for many years. 100V on the nose day in day out. Even after 3/11 caused significant generation outages (Fukushima etc.). The Japanese don't do tolerances, that requires someone to make a judgement call. All domestic kit is labelled simply 100V.

Here in the UK, our mains regularly gets into the high 250's causing the UPS' trim function to kick in.

9. Hi Everyone, apologies for the slightly off topic post. However I wanted to say congratulations to RevK on getting the nickname trademarked (I’ve only just noticed the little R in the circle in the blog title; don’t think it was there yesterday and the patent office website is down for maintenance at the moment so I can’t check when it was registered).

Also why not look at this photo of an underwater sea cable. Isn’t it interesting, all that protection for those tiny little strands! https://i.redd.it/bbw2h6qf70l01.jpg

Have a nice day everyone

10. Is it a risk having manufactured squillions of these things, having them sat on a shelf and potentially now finding that they will never receive approval and every single one will have to be scrapped? Is it safer to wait for approval and then commence manufacture?

11. For the mains in the UK its 230Vac +10%/-6% (might be -10% now) so upper limit is 253Vac.

You get paid compensation for every day its above 253V - I know that from personal experience. Was hard to readjust to having to pay them once the substation got sorted :D