A Weird Chip Inside This ZX Spectrum.
Vložit
- čas přidán 1. 11. 2023
- A fairly normal ZX Spectrum + 48k repair turned into a bit of an adventure with a video encoder chip I had never seen before!
Also, I delve into bad capacitors and why they are so terrible in ZX Spectrums.
@RoseTintedSpectrum 24 Hours of Dizzy event • You DO NOT Want to Mis...
Ko-fi page
ko-fi.com/morefunmakingit
Amazon Wishlist
www.amazon.co.uk/hz/wishlist/...
SECOND CHANNEL!
/ @morefunmakingit2
Discord Channel
/ discord
More Fun Website
www.morefunmakingit.co.uk/
My Twitterings can be found here
/ morefunmakingit
Facebook
/ morefunmakingit
Mastodon
oldbytes.space/@MoreFunMakingIt
BlueSky
bsky.app/profile/morefunmakin... - Věda a technologie
I have gone into electrolytic capacitors in great depth as a project manager for a 3rd party design house. There are all sorts of issues. For a start the Boric Acid wants to eat the aluminium oxide insulating layer on the foil plates. Store a cap too long and it will break down. You also need to "re-form" electrolytcs by charging them to about 1V less than rated voltage for 24 hours after 6 months storage. Then there is the whole water metering issue. Too much water and you still have a significant amount of qctive boric acid eating the aluminium oxide. This reduces life radically. Basically it's an equilibrium reaction where you use electric charge to drive it to one end (the aluminium oxide end).
Excellent information, when i was a camcorder engineer way back, we found that machines in regular use suffered much less from leaky surface mount caps than machines used only from annual holiday to holiday.
Yep.
The rubber seals on electrolytic capacitors have a ten year design life - if they leak they die.
That’s why they are avoided in modern industrial equipment designs.
Older designs used nicotinic acid as the electrolyte.
I used to work at BT when Spectrum was about. In the exchange we had a module called a translater ( adjusted numbers ). It was old tech with relays etc. They used to show visitors around the exchange. The guy that showed people round, used to pull the cover off and say this is a computer. For a joke we blue tacked a spectrum onto the relays . When he opened it next time he announced " This is a computer". Everyone laughed. He didnt see the funny side ( old school strowger engineer). Great you are keeping these going. My first computer was a Nascom. Kit was £225. Eventually made it into a CPM machine with Gemini extra cards. Great learning experience.
Thats an amazing story. These should be recorded somewhere!
The SN94459 and LM1889 are indeed pin and functionally compatible for the pins that exist on the SN94459.
Difference being the SN94459 does not contain a VHF RF modulator and the LM1889 does, which the Spectrum does not use anyway. PAL systems always used an external UHF modulator.
Just found your channel, love your work.
or LM2889N a later version of LM1889N/LM1886 (Version 2?)
only difference seems to be a wider voltage tolerance 10-16v (LM1889N 12-16v)
neither is officially manyfactured any longer,
seems ther a questinable manufacture in XXXX that can make the old TI version of LM2889N, min order 1000 pcs...
If that transistor mod at the beginning of the video is the one I've seen (a single transistor follower), do note that the output impedance won't be 75 ohms, it'll be very much lower than that (but if it works for you, it works). If you want a proper 75 ohm output the easiest way to do it is with an appropriate composite video driver IC, they are still available at the usual electronics suppliers.
Thanks for the video. My Issue 3 has new Panasonic capacitors in it. I also swapped out the 7805 with a switching reg no heatsink needed. Also swapped the transistors with BC equivalents. Another mod was a 2A bridge rectifier so any polarity on the 9v in.
I've seen more boards fully come to life by only swapping capacitors!
So you showed what is the best: never start with swapping 4116 chips: always start with the capacitors.
Especially the ones on the 12V rail(s): the 100uF's.
Though that does not mean the 4116 chips, ULA, or whatever part, are still reliable.
Hey Ben!
I have seen examples of socketed chips being the issue due to oxidation in the sockets.
@@ehsnils definitely, i had a 128 that'd crash if you so much as tickled it, it was dirty ula socket!
@@MoreFunMakingIt i'd say c46 is the critical one, if it goes bad, the -5v line can drop or disappear resulting in damage to the 4116s , its the pulse feed for the -5v diode pump circuit
I managed to short the -5V killing all 4116. So I now have a 48 kB Spectrum with two 64 kB banks. First delivers 16 kB and second delivers 32 kB. Strange? Definitely. But was a quick fix and now no -5V is needed.
Only strange outcome is that the ULA seems to become a bit hotter than originally.
As you can imagine back in 1982 when I was repairing these on my bench and doing these every single day of the week they where a S*d to unsolder. I remember we used to cut a pin on the memory chip to see which one was faulty and monitor the resistance. I remember we had so many broken at one stage we loaded up 2 big artic lorrys full of broken ones and they went to an airport to be storred....... we had to do something as we could not move in the factory....LOL
I would LOVE to hear more about your days fixing these back then. Get in touch via either my email or discord so I can hear more!
I used to work for Thorn EMI back in 1984/85. We were the main production line for the Sinclair Spectrum under contract from Sir Clive. Many of the machines that went out contained refurbished boards from faulty units that came back. I was one of the production engineers responsible for the repair of these faulty boards. That vertical line pattern immediately triggered "RAM fault" in my head. It must be stuck in there somewhere. LoL. One contribution it repair that I made fairly soon after starting was the diagnosis of a faulty base RAM chip that shorting the 12V supply and blowing TR7, part of the 12V supply. The standard repair approach, that I was shown and told this is how you do it, was to isolate each 12V pin by removing the solder and floating it in the centre of the through plating hole, change TR7 and move each pin to make contact until TR7 blew. The RAM chip and TR7 were then red dotted and sent to the soldering girls for replacement. This obviously took a while to isolate each pin and a certain amount of skill. Some lazy engineers used to simply side cut each leg and re-solder them once the faulty chip was found. Very messy. I thought differently. I thought "why is TR7 blowing?" "because a chip is pulling too much power" "what if you gave it a power supply with higher amperage, like straight off the power adaptor?" So with a length of solder I tapped off the PSU which handily was positive sleeve and attached it to the 12V supply to the RAM and bingo, with the help of a little freezer spray to frost up the RAM, the ones that got hot were the culprits. I showed the established engineers what I was doing and was told that's not the way to do it, my reply was well it works doesn't it? Don't think they liked the 18 year old newbie showing them how it's done. LoL
Thank you for sharing those wonderful memories! I too really don't like the snip method of isolating chips. Really like the idea of floating the pin though, that's very clever and I'll have to remember that.
Takes me back where I was in 1980s we were using freezer spray too to look for bad ram on various makes watch for which one defrosted first. it's good thing that engineers on bench were finding new ways to improve efficiency.
Takes me back where I was in 1980s we were using freezer spray too to look for bad ram on various makes watch for which one defrosted first. it's good thing that engineers on bench were finding new ways to improve efficiency.
Thankyou for that. I was confused why my issue 6 had a manky display. After looking at the parts I took off I found that IC. I replaced it with the LM IC. That is why it did not work. Cheers
It was great to see so dedicated work. I used to have ZX-80, ZX-81 AND ZX-Spectrum MANY years ago. I also made a new keyboard and even 5 -1/4 inch floppy drive. But that was a LONG time ago.
Thank you! Bet you wish you still had that ZX80! 😉
The bad video quality you are refering to may have several causes. The most likely ones are noise in the power source (not ultimately the psu, can be induced by another nearby device, - maybe try powering from 3 LiPo cells in series -> no noise, maybe add additional decoupling to the ULA, video encoder… - different values 0.1µF, 0.47µF, 1µF) or the still running modulator (but you did the composite mod outside the can and cut the power to it). Other possibilities are noisy diodes or transistors, bad contact in the ULA Socket. Check the video out of the ULA on an oscilloscope and see if you can identify the noise (looked not very high frequency - in the kHz range maybe even less, do this on an grey screen and check ist AC coupled - check the noise level and try to meassure the frequency - then search other nearby signals that may be the source of the noise. Recently someone else had a similar problem (I forgot who it was Adrian, Jan Beta, … can't remember)
The professor has spoken!! If I could only gleam a weeks worth of apprenticeship.
I prefer the ZTX213's at the video out, since other ones often blur the picture a bit.
And the interference could be caused by the small caps at the 4116 chips: I have axial 100nF ones to replace them (only if needed) nowadays.
Excellent! I will look into those caps. Although hooked up to a TV the output looks great
In all my time repairing stuff, I've replaced more 0.1μF ceramics that have DC across them - usually decoupling caps - than electrolytics. Probably hundreds more!
Good Video as always, I have come across one of those chips before in a PCB I fixed up for a friend.
Bug Byte in the system, Manic miner Lives on..
@10:45 I fell for your trick of pressing the screen to vote but soon got over it when you blu-tacked the socket to your finger :D @11:10
Gotcha! 😆
That reminds me, I still have a cap kit for my 48k that I really should use to replace it's old caps. Lovely video and nice to see the results of testing this machines old caps. I'd never heard of an alternative LM1889 chip either, so that's interested - I'd have been interested to see how this speccy performs with it popped into the sockets after replacing the video transisters.
Cheers Chris! I've kept the odd chip and will test it in another board at some point
I bought myself a spectrum a couple of weeks ago but it had problems, long story short all tests said "lower ram fault". I got some replacement chips and no luck. Then the other day I saw your video and said naah, it's not that, the caps look fine. This evening I found 2 100uf caps in my spare parts bin and in desperation I gave it a shot and she sprang to life. I'll replace the rest of the caps now. Thank you so much you turned my faulty Spekky into a really easy fix ;)
Fantastic! That's great to hear. I wonder how many of these have been ruined after someone started chasing faults in the chips when it was the caps all along?
5:06 The poor conductivity czcams.com/video/BCBcKRPGOhc/video.html can be revealed by bridging a suspect cap with a new one, since a dry original will act, very-nearly, to be "out of circuit". Often an instant "healing" like a cracked trace closing up and completing the circuit.
The transistor gain (hfe) has raised the chroma threshold in the lm1889n chip. By raising the voltage it has shifted the dark scale to lighter by my understanding
i rember being bought a 48k zx spectrum for Xmas and playing attick attack and spellbound .. i loved it soo much..
10:56 If you have to go back with wick and hot air, you’re using the desoldering tool incorrrectly. After ensuring the desoldering tip is nicely tinned, apply it to a joint. Once the solder is fully melted (as determined by being able to wiggle the component lead freely back and forth), turn on the vacuum. Now continue wiggling with the vacuum on - this ensures you get all the solder, and the continued vacuum causes cool air to flow through, cooling the component leg and preventing it from resoldering itself. Then remove the desoldering iron, and then release the trigger. This entire process should take around 3-5 seconds. The wiggling is essential - without that, you don’t know if the joint is melted, and you won’t release the leg from the side of a plated hole.
It looked like a spring-loaded solder sucker to me.
I personally pre-heat the area up with a heat gun, then just use a manual solder sucker. Learnt that at the age of 16 on 4 layer boards in the 80’s as a YTS “yop” at the Olivetti repair lab. They were impressed that I could replace chips without damaging the board (apparently i was the first such yop that could do so). Preheating the board came from my dad who would preheat jobs before tig welding.
@@stevenclark2188 Look again (and listen to the audio): it’s clearly a continuous-vacuum desoldering tool.
@@eliotmansfield Definitely a 👍 for using preheating, but how do you prevent the component leads from re-adhering themselves to the inside of the plated through-hole?
@@tookitogo you still use a soldering iron with the solder sucker. The pre-heat just means the pins warm up quicker and doesn’t suck all the heat out the iron when doing the power plane pins
Great video - subscribed! I started with a ZX81 & pre-ordered a ZX Spectrum (well, my parents did...) so had one of the original 48k machines. I've no idea where it is now though. At some point, I rehoused it in a Dk'tronics case (which was made out of some weird resin and not ABS as some sites say) so I even had a proper keyboard!
Thank you! You need to track that machine down! That would be a lovely thing to find.
I had the same memory issue with the C45 capacitors which I replaced and fixed it.
Another good one, and I pressed the correct answer on the screen. Hope you are feeling better, it’s that time of year for getting sick.
Thank you Curtis. Feeling a lot better now
i've never yet come across the SN chip, all my spectrums used the LM (or TEA in the 128 models)
Looking at that interference, I think it's still possible you've got grotty supply rails. Have you put your sillyscope on the +12V supply to the LM1889 yet? What does it look like in AC-coupled mode?
The interference may be coming from the -5v bias DC to DC converter, just change the electrolytic caps that have probably dried out.
dont necessarily HAVE to replace all, C46 is the critical one
Excellent work and beautifully presented. In my experience blurring and interference can sometimes be caused by "old man eyes" so maybe an option is to get yourself some younger ones. Green ones are a bit harder to find but I believe it will really make you pop and set off that new beard to its full potential.
As I touched the screen I realised ... this isn't a touch screen!!! How I laughed. I will watch it on my daughter's laptop and try again ;-)
I prefer to stick with my OG Vintage eyeballs that happen to be green quite often!
Keep jabbing those screens! If it doesnt work try using a sharp stick.
Nice to find a video I hadn't seen yet when I got home from the RCF 👍 Good to see you there today (I'm the guy who dumped a Falcon 😳)
Hiya! I hope you get your falcon back one day!
@@MoreFunMakingIt Unlikely as it went into the council dump about 20 years ago! Anyway, that is in the past and I'm not interested in that kind of machine any more - it's too complex to tinker with and too primitive to do useful work. Dragons, BBC's and my homebrew 6809 are much more fun 😁. Keep up the good work and keep the videos coming 👍
I’m no expert at any of this, but as a software guy for years, I doff my cap in deference to people such as yourself that make old gear like new 👍👍
Thank you!
capacitors can and do last much longer than many think , i have many old transistor radios MUCH older than spectrum still unrecapped as didnt need it ,
If you want longevity on CAPs, I'd go with Nichicon, although Ilonios Capacitors, (IC) are good. Those Gelec are crap!
probably, sinclair cut the cost as much as possible, that said i have several spectrums, unrecapped, that still work perfectly, one of those caps in the pic was a 'famous' blue philips, i find they usually last well but if left unused for many years can 'deform', but if gently reformed, can be ok again , the only issue i had with one was in a roberts r505 radio, and i think that was 'cause they used one too low voltage rating
Why "test by substitution" the capacitors (probably fine for another 50 years)? Laziness? Market forces?
Why not test each cap first before condemning the whole lot. The replacement caps how long are they going to last for? 2 or 3 years? maybe 5?
interesting - I noticed on my ZX spectrum that the video level with the capacitor only is just 200mV and I guess that's why you get a dim picture.
So, what is your amazing capture device which works better than the SIX I've tested?
Thanks for the video!
Thanks Tony! My capture device is the knock off 2x from China. It's not perfect, but better than pointing a camera at a screen
I wonder what picture would you get with SN94459N chip and good transistors (TR1 and TR2)?
I have kept the chip and one day I will experiment...
An excellent repair. Subscribed. That tester looks fantastic. You may know why my issue 1 +2A makes the sound of a key held down if with the keyboard disconnected. It was working fine, then just went like that.
The +2A has it's ULA chip suspended in a hole in the motherboard. They're prone to dry solder joints. I would check there first
@@MoreFunMakingIt I suspected that chip. It's really odd how they mounted it like that.
Your setup is nearly as messy as mine. Well done.🤓
Always fighting the mess!
Great video. I own a ZX Spectrum 16K from 1982 with rubber key board. After many years of not being used i tried to start it on but a similar failure to yours happened to me. The screen shows 4 or 5 vertical lines and makes nothing.
I am not skilled on electronics but knowing that it could be a capacitor issue i would try to repair it.
Thank you!
Good luck!
Damn SN94459N! How dare it not be the problem😆
So rude!
New Vfd drive are making electrolytics work super hard these days
I pressed the screen.
My monitor fell off the desk.
I need a new monitor.
LOL!
Oh no!
Maybe I should do a tick the box with a pen 😆
Interesting!
The electrolytic caps have a useful life of about 3000 hours. After that the capacitance goes down and eventually zero. Use tantalums to get a stable system
depends entirely on their conditions of use, they can and do last much longer , the datasheets specify life at everything maximum, reduce any and life increases rapidly , i have a 1958 transistor radio completely 'virgin' unrecapped
@@andygozzo72 Sure, true. However, if they are continuously powered, they simply die. High temperature accelerates it. I also have an Emerson portable "pocket" radio from 1956, still working.
@@henrikstenlund5385 yep, heat and high ripple are the biggest 'killers'
@@andygozzo72 You are right. Excessive current is one killer of caps. Power supplies need special caps for that.
@@henrikstenlund5385 these days, yes, but they still have a habit of putting them near hot parts like heatsinks 😉
Im no "computer expert, however im very into radios and the relevant part is interference, I wonder if your bars that show up are caused by local rf interference, in my case the best way to check was to go around the house and turn everything off one by one then eventually the switchboard, ive found things like wifi routers to fridges even insulators on powerlines across the road or wireless powere box meters have cause similar issues with my Software defined radios and, shielding and filtering help.
I have a toastrack spectrum but the large heatsink has all the paint flaking off, any idea what I could respray it with?
Should be able to spray that with just about anything really. It wont be getting excessively hot, so I would just pick some automotive paint. Clean it thoroughly, maybe even prime it, and then multiple thin coats.
A 5vdc buck converter,
caps and shielding should do it
I wonder what kind of capacitors the voyager probes have in them.
I did try looking this up, and I couldn't find anything definitive. The whole thing is powered by massive capacitors, but they have to be of a solid composition. Electrolytic fluid would just freeze in space.
But interesting question!
I'd really love to get my hands on some kind of table that lists "good" and "bad" ESRs for various different caps... for me, at the moment, ESRs are just "some arbitrary number".
It depends on the circuit tolerance. It's generally accepted that most caps should read below 1 ohm unless they are small values such as 1uf, which may read a few ohms, say 5 ohms. If any cap reads over 5, I'd toss it. If you check the data sheets, you can find max ESR when new. They don't change much until end of life
Nice! Thanks!
I had a 128K toast wrack. An fantastic thing
Still on my wishlist!
@@MoreFunMakingIt they use very different video circuitry which has its own 'peculiarities' 😉
Made me laugh - Subscribed
Thank you!
I suspect we could make a replacement for the LM1889 chip that would be good enough.
The multiply action wouldn't need to be anywhere near as linear as what is needed for real TV.
The channel-3 vs channel-4 oscillators is fairly easy
The color subcarrier is not too hard to generate.
The only issue I see is making it fit mechanically even with surface mount parts.
Sounds complicated!
@@MoreFunMakingIt Yes it is a bit complicated but not too bad if you want only the composite output and not the channel 3,4 modulator. Take a look at the block diagram of the chip that is in its data sheet.
@@kensmith5694 the channel 3/4 thing is irrelevant, this is a UK channel 36 ish UHF machine , the modulator part of the LM isnt used, the modulating is done in the UM1233 modulator
@@andygozzo72 If you are bringing out the composite video instead, you don't need either modulator. You only need the 3.(whatever)MHz crystal oscillators and the multiplier action. As it isn't a movie or anything, the multiplies can be fair crude and still good enough. No linearization would be needed on a gilbert cell. Matched transistors can be had in SOT23-6 packages so temperature gradients won't be an issue.
I would've painted every circuit a different colour to get it working
That's because you're a pro
Can you piggyback capacitors or reinforce them?
Depends on how the capacitor has gone faulty. You can certainly connect them in parallel
I was tempted to press the screen but....
Press the screen!
I pressed the screen (and got it wrong 😁)
Does the screen even _have_ a but? And wouldn't that be electronic harassment?
So what exactly was the problem? The transistors or the modulator? You said you replaced both.
A mix of both. Plus some added noise from capture.
Since the culprits were TR1 and TR2, hoe about putting that SN94459N back in? It's socketed so give it a quick try!
It's already gone. But I've kept the chip and will test it on another machine
Isn't the reason the LM1889N replacement wasn't changing anything that this chip is a TV modulator? You are bypassing modulation altogether I think.
The LM1889N has a built in modulator section thats completely bypassed on the Spectrum, they chose to use a separate modulator.
Interesting. So what does the chip do? Mix in audio?@@MoreFunMakingIt
Its always the capacitors!
Conspiracy theory time:
Electrolytic capacitors are only used because the factory wants your device to stop working soon, so they can sell you another.
As I sit here in my tinfoil hat I cant help but wonder if you are onto something there!
And joking aside - probably is some truth in it!
@@MoreFunMakingIt its NOT 'always' capacitors, they can and do last much longer, i have lots of old equipment unrecapped or mostly unrecapped that can prove it, you should not assume blanket recapping is some magic cure all
11:12 what kind of clay do you use here to hold the socket in place?
Clay? You have to use chewing gum from the 80s. It's the only fully compatible option.
Blue tack 😁
it can just be the cable your using between the spectrum and monitor doesn't have enough shielding.
I think it was actually the capture device
@@MoreFunMakingIt spectrums never have a 'perfect' display,
Most Sinclair faults actually originate from users playing Kokotoni Wilf or Subsunk on them in the past. Not a lot of people know that.
Could you elaborate more on the topic for people like me who are not very familiar with this platform experience?
Its one of the hidden dangers
Manic Miner FTW!
THAT VIDEO MOD IS INCORRECT SIR 😊😊😊
All RAM is bad = too high voltage on the 5 V line (even for milliseconds).
Probably down to the bad capacitors in this case
@@MoreFunMakingIt that wont and cant cause the +5v to be high, that may be down to the 7805
TRY BUYING DESOLDERING TOOL MAKES LIFE SO MUCH BETTER 😊😊😊
Whahaha, no I didn't click. And have a bunch (like 30?) pulled LM1889 chips, to replace (always) those garbage ones.
I bet you did click!
So those other chips are really that bad?
At least half of the ones I encounter during refurbishment, have issues.
They're fortunately not that common, but I always replace them to be sure.
Especially since I include warranty.
@@ByteDelight good to know!
Ref: Attention Span Lowering Agenda : All Micro Edits & Clothing changes are noted.
Have seen
Thank you!
I just sigh that guys like you have the knowledge and skills to fix stuff like this. Are you all self taught or degree level?
Totally self taught. Always been a tinkerer, but everything I've learnt over the last couple of years has been from the internet
@@MoreFunMakingIt Always interested in electronics from a kid when i went to the village fete and bought valve radio for about 2 bob. Then TV's. Nearly killed my self when i stepped off a chair and put my foot on the floor whilst touching the chassis. Did an apprenticeship with electronics. Now at 73 still tinkering. Just repaired HP 2009v display. Tried to do it long hand but my hands arent steady enough to handle SM components. Whats interesting is that I spent £20 quid on bits but then got a used board from Ali Express for £7 all up and it works. Happy days
@@CASHSEC sounds like you're having great fun!
0:31 aaugh stop touching the chip I'm sure you're grounded but I can't watch 😅😅😅
Old cap or old crap? 😜
Why not both?!
i wrote my BSc. thesis program with my soectrum. it was an a.i. expert system for economical decision making. it was 1986. my today machine is doing 45 Tera Flop with 6 Nvidia GPU and 5KW, refrigrant cooling system. do not know why these days anyone wants to fix this or use it. use it for what? anyway, as long as crazy people are out there you fix these machines. have a happy time. moses
'worser' ? lol
Worser??? WORSER???? Worse!!!! No such word as "worser".
Are you sure about that? Would you like a minute to go check. Here. I will save you the trouble :) www.oed.com/search/dictionary/?scope=Entries&q=worser
😜
Worse -> Worser -> Worsest -> Worsester -> Worsestershire (sauce)
Yes, i'm sure. We ALL know the OED has gone "woke" and now just includes made up words that didn't exist 30 years ago not for the benefit of language but so the thickos can play Scrabble. My English teacher would rap our knuckles for saying that (this was the late 70's) and DON'T get me started on this "would of" garbage. No, despite what the oh-s0-woke OED says "worser" is not a word. Funny how my spell checker flags it up however!!! So, in conclusion, the OED is woke garbage and worser is STILL not a word!@@MoreFunMakingIt
You also look about the same age as me, late 40 to mid 50 and i bet you a months salary you were NEVER taught worser was a word when at school. You may have used it but that doesn't make it valid.@@MoreFunMakingIt
@@RFC3514Worcester. Pronounced "Wuste" - not to be confused with "Würste" for the German Readers.
New equipment for me but no TV. Still have internet but overall, Spectrum sucks............................
Huh?
Could it be a power supply issue? Maybe input DC isn’t being properly filtered.
I think it might have actually been introduced by my capture. Or at least enhanced by it. Worked fine straight to TV.
Oh!! Let’s see the A1200!
Sooooooon!