EMC who cares?
Really, the BT power line modems are also an abject EMC failure but due to the money behind them ofcom, etc, don't care. The solution? They re-draft regulations to allow more noise...
A pal of mine suggested a short while back that it might be fun to obtain the blueprints for Sinclair Research’s ZX Spectrum and have a batch built up to sell to fans of retro computing. It’s a good job this plan never made it out of the pub: the dear old Speccy would have immediately fallen foul of modern electromagnetic …
You should try the Home Hub 3's standard PSU.
When I first got Infinity I couldn't work out why my previously reliable network of power line kept failing to link.
A few tests later, and I found it's the cheap PSU BT supply chucking too much noise back down the mains!
Surely a simple circuit diagram ?
My memory may be playing tricks on me, but I'm pretty certain there was a very short-term release of DIY spectrum kits* so there might be something already out there. Trying to build a modern compliant version strikes me as the IT equivalent of fitting wheels to a tomato - time consuming and utterly pointless.
*The genius that was Sir Clive realised that you could get your customers to build the unit, thus avoiding unrealistic delivery schedules *and* sell it as a "feature". Certainly the ZX80, and early ZX81s ....
I think (but don't have clear memory or facts) that the ZX series were cheaply made and used a double-sided PCB and not multi-layer boards with power & ground planes. That, if true, is probably the #1 reason for the poor EMC performance.
Also note they tested it without cables/peripherals, so real-world use would be significantly worse that observed in El Reg's article.
With modern components behind it, you could likely put the entire thing in a single FPGA, with the only signal lines coming out being the cassette interface line, the video, and the keyboard scan - and in today's designs, those would all be 3.3V rather than 5V, reducing the RF even more.
Then I guess the solution is to buy 'The ZX Spectrum ULA: How to Design a Microcomputer' (ISBN-10 0956507107; published in 2010 so still widely available) as the ULA is fully documented and imaged within. You could definitely build an entire new ZX Spectrum with that and even have the correct horizon on Aquaplane, the correct multicolour text on Uridium, etc.
I can't remember making music on the Spectrum this way (it had a speaker after all), but I do remember a piano program for the ZX80 that worked that way.
I did use a radio to monitor long-running programs - in fact I can still remember the "melody" for one of them. The trig functions had very distinctive sounds.
The oddest case of this I came across occurred with an Amstrad PC 1512 running Digger. The actual game music could be picked up by a radio tunned to just below Radio 1's FM frequency.
I'm not sure if this was a deliberate feature of the game that worked with any PC or just a peculiar interaction with the Amstrad's hardware. It didn't seem to be a general occurrence with the 1512, I never managed to do the same with other games.
I REM this. My neighbour used to complain loudly when I ran my Speccy because he said it used to POKE his TV reception. Said it also upset his DIM CAT. The CONT.
I got the POINT and so I went OUT to Tandy, had a PEEK and bought some metallic spray. Upon my RETURN I coated the inside of the case and then insulated with electrical tape. Sorted the problem FOR sure. CLEAR result. Try it out to VERIFY.
Its a marginal fail, its only 4 or 5dB above the limit, and at a couple of spot frequencies. A few minor modifications would fix it - either an "R" or a ferrite bead in the clock line, or a little work on the grounding scheme. I've seen much worse initial plots than this. Oh, and the emissions above 100MHz should be fine, look a the plot, there is no energy in the higher frequencies, it is just the test equipment noise floor. OK there might be problems with spurii and harmonics of the UHF modulator, once it gets enabled again. Not bad for the time and rather exceptional for our Clive - or Jim Westwood (see Reg's passim) to give him a proper name.
It's still available, and is still used in a couple of our 20+ year old designs! (Our industrial customers are a conservative lot and still want "what they bought last time" - who am I to argue if they're willing to pay for it?)
Mind you, the updated versions use all the techniques mentioned in the original article and end up costing far less overall, not more. In short, stop the product from radiating in the first place and you don't need to stick expensive filters everywhere.
--stop the product from radiating in the first place --
Got THAT T shirt. Odd how spending a few pennies on prevention is begrudged, but a few hundred thousand dollars may be spent to find and fix RFI later.
I'm retired now and get the occasional job via a contracting firm when someone realizes he's ignored High School (6th Form for you lot) Physics.
Or can't figure out his own test equipment.
Take two ferrites and call me in the morning.
Reminds me of my first computing job in the mid 80's.
We had an old 'British Micro' running CP/M with an orange monochrome display. If you placed the display on top of the case of the main unit, then the display would distort as the electro-magnetic fluctuations from the unit would effect the cathode tube.
Add some floppy access to it and the screen would become unreadable.
To solve, we had to use a 1/8" steel sheet between the box and the display.
Tight loops produced a different noise (more whine than ticking hum) on a nearby transistor radio than main control loop of a program (presumably DRAM refresh cycle rate changed), so I was able to find errant code by listening to the radio tuned to some interfering harmonic frequency on similarly Z80 equipped Jupiter ACE.
I always assumed the cause was the plastic case not providing any shielding.
If it makes your radio buzz, so you have to move the radio to the other side of the room, maybe no big deal, but it's "real".
Then put 10 Spectrums (Spectra?) in a classroom, and discover that it causes patterning and noise on the TV upstairs where they're watching a schools broadcast. Is that "more real".
Then you get the really awkward ones, like when it upsets the hearing aid of the deaf kid so (s)he can't use the computer lab.
You ned to draw a line somewhere, and enforce it.
I would have expected far worse to be honest. I recall my Spectrum+ would knock out Radio 1 on FM!
Not sure about “rather fishy power supply and cable”. The brick was a simple transformer/bridge rectifier/capacitor jobbie and onboard the regulator was a basic linear 7805. Nothing noisy there.
Back in the days of the ZX81 here in the US, the requirements for computers were the less rigorous Part 15A, intended for office work (since nobody would have a computer in the home where Part 15B ruled).
There was one computer of which I am aware that would stand any chance in the modern regulatory environment, and that would be the old Atari 400 and Atari 800 - all the guts were in a potmetal shell weighing more than the entire ZX81. It would be fun to try one of them out in an RF chamber.
Then there was the good old TRS-80 Model 1 - a machine so RF noisy that several games did sound effects by instructing you to tune an AM radio between stations and place near (within 2m) of the machine. A machine so noisy it "passed" TEMPEST because it put out SO MUCH hash you couldn't recover what was on the screen.
It had a metal shield on most models. They were more compliant than Sinclair who were way too obsessed with cost and this cheapness is the reason why I think we have very little in the way of a computer industry.
People grew out of the speccy and its limitations and the lack of profit meant no investment in a decent 16-Bit machine, Sinclair had wasted all the money on a QL.
I doubt your reasoning. We have perfectly cromulent hardware success stories coming from Cambridge including the ARM and we punch way above our weight in software terms. It's hard to argue the cheapness of the Spectrum isn't part of the reason why so many were able to get into software. It helped in my case. (Age 7).
My guess for a source of the radio noise would be the power supply section on the main board (not the adapter). At the time it was a common source of breakdown and if I remember correctly it was a cheap self-oscillating circuit generating -5 and +12 in addition to the +5 from a common regulator using the 9V main. It's hard to tell how much EMF could be generated this way or which harmonics but it was not shielded at all.
You're right about the circuit. Strange thing is that if they had run 9VAC into the machine, they could have had their 12V and -5V without faffing with that oscillator, just simple diodes, caps and regulators. There would have been a larger smoothing cap inside the Speccy (maybe a few smaller ones in parallel according to available space), but apparently the way it was done was the absolute cheapest.
While I understand the need for some regulation of the EM radio spectrum, this kind of thing is disappointing. Once upon a time, anybody with the knowledge necessary to build a circuit and shove it in a box could then go and put it on sale. Nowadays it looks more like you'd need a legal qualification more than the electronics knowledge to put something on sale. You have EM Compliance, you have the god damned WEEEEEEEEEEEEEEEEE directive and no doubt another few regulations I am not aware of!
I wonder how many innovative products are left rotting in peoples sheds because they've looked at the cost and effort of taking them to market and have turned round and said sod that for a laugh...
I wonder how big the pile of non-rotting obsolete products would be if we didn't have said regulations, and personally I'm quite glad that I can listen to the radio/get wireless reception without my next door neighbour spewing EM noise out with some bit of kit knocked up in someone's shed.
The regulations aren't just there to stop people having fun you know...
Then look at it another way, we have all these new regulations and now have electrical appliances and TVs that generally only last the warranty period of a year and then fail.
Whats the average life of a modern LCD TV? 2-3 years? I remember a time when a TV would last 15+ years.
RoHS has just meant brittle inferior solder that cracks at the first sign of heat/stress and then it's landfill bound. Now that's eco friendly.
They giveth with one hand...
Unfortunately I cannot remember where I read it, but I remember once reading that in modern electronic hardware, the quantity of valuable metals is actually being reduced. Where once a circuit board would have significant amounts of gold on connectors, and lead in the solder as you pointed out, now there is very little to make recycling economical, and I have to wonder if ironically that has actually countered the increased concern in recycling old equipment?
As for my earlier point, I am not disputing that some regulation is both necessary and good, I too expect to be able to use my laptop and phone to get reliable connections to all manner of wireless signals. However I also have to wonder how bad things would be in an unregulated environment. I mean, when these devices were being produced (literally in sheds and garages to begin with) and there wasn't huge amounts of regulation, I don't hear stories of how they destroyed TV and radio transmissions. Nor do I necessarily see the need for CE marks, since electrocuting your customers or burning their houses down tends to be bad for business!
What I am really suggesting is perhaps a lighter set of regulatory requirements that allow inventors to come up with new products for minimal overheads, although perhaps only where these do not involve the use of Wireless RF components.
This post has been deleted by its author
Caution: long. Contains facts, logic, and small amounts of technology.
" I mean, when these devices were being produced (literally in sheds and garages to begin with) and there wasn't huge amounts of regulation, I don't hear stories of how they destroyed TV and radio transmissions. "
You're utterly clueless on this subject aren't you (or you're trolling). It's OK to be clueless, it's not OK to make ignorant ill-informed comments based on that cluelessness.
When devices were being produced in sheds, there wasn't much RF spectrum in intensive use, there weren't that many electronic devices being made or in use (how many radios in a typical house back then?), and pretty much none of the electronic devices in widespread use were devices capable of wideband RF emissions (such as the switched mode power supply as found in every modern piece of consumer electronics and loads of other places too).
When devices were being produced in sheds, there were also no miraculous applications of modern technology such as the various flavours of DSL broadband, which is basically long, medium and short wave radio down phone lines rather than through the Ether. Phone lines never intended for that use, but which conveniently (a) happen to work well enoug for most folk to use (b) happen to work well enough without causing significant disruption to other simultaneous users of the same RF spectrum, in the same place or elsewhere.
By the miracle of DSL, every available station frequency is in use at the same time, not for voice or music over the Ether but down your telephone line to encode a few kbits per second (per station) of data, upstream or downstream. Telphone lines used this way can pick up RF noise from outside (though mostly they do OK at rejecting most of it). The cleaner the overall RF environment, the more station frequencies are usable and the better the data rate is for each station, so the faster the overall connection is. The dirtier the RF environment, the worse the DSL performance. [Apologies if anyone is offended by my gross oversimplification]
I have no idea how old you are, but I'm sure some readers here will remember how the arrival of increasing numbers of TVs in the UK rendered long wave broadcast reception useless across much of the UK. That would be an example of how unregulated devices destroyed the usability of radio transmissions, something which very definitely happened even if you didn't notice it.
The CE regulations, and EMC best practice in general, are supposed to (amongst other things) help avoid repeating such entirely avoidable disasters, so long as the 'light touch regulation' folk get a clue (or just shut up).
"a lighter set of regulatory requirements that allow inventors to come up with new products for minimal overheads, although perhaps only where these do not involve the use of Wireless RF components."
Show me a device that demonstrably cannot involve the production of significant RF emissions and I'll show you a device whis is ALREADY exempt from the RF parts of the CE regulations.
"electrocuting your customers or burning their houses down tends to be bad for business!"
And everything less serious than that is OK? Even if it stops other legitimate uses (TVs and long wave broadcast, for example?)
Thanks for reading, have a low-emission weekend
“rather fishy power supply and cable”
I may be mistaken, but I seem to recall that it was a bog standard iron core transformer and rectifier, with a linear reg inside the machine itself. Might be a bit crap, but not as noisy (rf-wise) as modern swicthers. (or was that the zx81?
probably both were like that.
Fit a metal shield and it might be okay.
The Apple 2 was a bastard for giving off RF too, so much so that they sold it without the modulator. You don't tend to find much knowledge of this on the web as it doesn't fit in with the beautiful, magical history of Apple and the legendary genius of Woz (who was a poor engineer at first).
Haven't looked in any detail, but there seem to be MAME-based emulators for the ZX Spectrum.
MAME: Multiple Aracde Machine Emulator
Runs on multiple platforms supporting gcc/qt/python.
Apparently available on Android (and therefore by implication on ARM/Linux?)
Far from an abject failure, OK it does not have cables connected and that could make a BIG difference but a failure at two frequencies by not too much should be easily fixed and would not be too unusual for a test run on a first prototype for a modern product. I have seen worse test results than this on CE marked equipment!
I get the impression the writer knew what story he wanted to writer and was going to write it whatever the test result unless it passed on everything.
My Amiga a500 ocs used to almost blanket the entire radio spectrum for about a foot around it with its interference . The ecs one I got later wasnt nearly as bad. I think the later Amiga had an rf shield encasing the top of the motherboard that the old one was missing.
Dont recall the speccy 48k chucking out much interference though.
Conformance to EN55022 is non-mandatory. It just shows due diligence to the EMC regulations. In fact, conformance to the law seems to be non-mandatory if you have enough cash and can lobby hard enough.
A lot of LED lights that are on the market do not conform to EN55022 and have been seen to cause interference to DAB receivers...
And VDE 0871!
Thia sort of stuff's been my living since 1983, though less now, semiretired. In the "good old days" clocks* ran in the SW spectrum, where we in the US could listen to the BBC,**. Or the neighbor's computer.
* Except cuckoo.
** No US broadcasts now.
Biting the hand that feeds IT © 1998–2020