ZX Spectrum (48K & 16K) Initial Tests 

I have got the same problem with one of mine same issue also. How did it work out?

It sounds as if there is a short circuit between the 5 V and 0 V supply rails. That could be tricky to track down because so much is connected to both. Examine the circuit board very carefully to make sure it's not due to debris/a damaged track/out of position component lead/etc. Sometimes using a high resolution resistance meter to check the resistance between these rails at each component can find such a fault; the resistance will be very slightly lower near to the short. However, often the only way is to remove components until the guilty one is found.

Have you tested the ohms on the speaker? If it is good, try testing the diode at D9 if you have changed TR7 it was the advice i was given that fixed my deaf speaker

If none of the other measurements indicate a fault, then you shouldn't worry about the result of your in-circuit test of TR5. The reverse bias voltage reading during this test depends very much on the particular multimeter used. Only if a very low reading is obtained would there be cause for concern.

First of all thanks for your clarification, now my Speccy is in full working order, new capacitors improved a better video signal, composite video, spot on !! Besides, I have an Issue 4B to repair, without switch it on I made all the initial tests and I've found TR4 shorted: do you suggest other test before change TR4 and switch the power on? If there are some lower or upper ram that failed, replacing TR4 will fail another time itself? Thanks a lot for the answers !

Remove TR4 and repeat all the power-off tests. If they pass, replace TR4 with a new ZTX651/ZTX652/ZTX653 (very few other physically small transistors will work reliably here) and do the tests again. A vicious circle exists whereby sometimes faulty lower RAM overloads the voltage converter circuit and causes it to fail (possibly damaging more RAM ICs) or something else causes the voltage converter circuit to fail (e.g. misuse of the edge connector) and that causes RAM to fail, sometimes in a manner that'll overload and damage a repaired voltage converter circuit. Therefore all faults in the voltage converter circuit and in the RAM (where the fault is of a type that might overload the voltage converter circuit) must be corrected at the same time. If you have a bench power supply with current limiting, set it to 9 V DC limited to 750 mA and use it to power the Spectrum (remember the input socket is centre negative). If the current limit operates for more than a couple of seconds, turn the power off immediately and you might save TR4 from being destroyed. If you have to use the normal ZX PSU and there is still a fault in the voltage converter circuit or the lower RAM overloads its output, TR4 will probably be destroyed instantly. If this doesn't happen, immediately test for the presence of the three power supply voltages at the lower RAM. There are a few diodes in the voltage converter circuit. It is not all that common for these to fail, but it is worth testing them if you don't find anything else wrong. Consult the circuit diagram to see which ones they are and to determine what test results to expect. If one of the resistance tests gives an unreasonably low reading or the voltage converter circuit still fails to work after having tested all of its key components, you'll probably have to remove all of the lower RAM ICs. With the eight lower RAM ICs removed, you should get the voltage converter working before fitting any new RAM. Once it is working, I recommend you fit standard (dual wipe) IC sockets and a new set of 4116 ICs or one of the lower RAM replacement modules available at zx.zigg.net .

Thanks a lot, I've replace TR4 with another ZTX650, replaced all electrolytic capacitors, all voltages correct, -5,11V, 4,96V and 12,11V should be ok, ROM and ULA swapped on my working 48K and they are good, tested, but I have surely a fault in lower or upper ram or both, the system hangs with magenta border, flashing colored blocks and garbage on the screen...weird measure of resistance on pin 2 of IC8, 505 ohm and IC17, 47,4 ohm, other pin 2 of each ram IC's are around 7,1k - 7,6k: can we say IC8 and IC17 are at fault?

I assume that above you are saying you measured 505 ohms between the 0 V power supply rail and pin 2 of IC8 and 47.4 ohms between the 0 V power supply rail and pin 2 of IC17. If that is correct, then I think it very likely IC17 is faulty. IC8 probably is not as pin 2/14 of IC8 is connected to pin 2/14 of IC17 via a 470 ohm resistor. Remove IC17 or cut through pins 2 and 14 and bend them so they are no longer connected. With a bit of luck, the computer might then initialise as a 16K model. If it does, run the Memtest program from zx.zigg.net and tell it to test the upper RAM. It is possible there are other faulty upper RAM ICs; they just haven't failed in a way that prevents the computer from initialising. Faulty upper RAM ICs can be replaced with 4164 DRAM ICs (150 ns access time or faster) or all eight with one of the upper RAM modules.

Are you sure it's an issue 3? Those were generally only ever sold as the 48K model. There are (plausible) rumours some faulty issue 3 systems were sold as 16K models. What leads you to conclude it has been upgraded? Can you post a photo of the circuit board? I'm not quite sure how to interpret the numbers you've posted. The 5 V output from the regulator is a good start. What voltages do you measure at the lower RAM with the power on? See towards the end of the video where they are labelled. From your photograph it looks as if the TV is not well adjusted or the video modulator in the Spectrum isn't working well - you could tap off the composite signal before the modulator. However, the display does look rather like what often happens when the lower RAM is faulty, but it's difficult to diagnose from just a photo. I strongly recommend you get some hands-on help from someone with electronics experience or make use of a repair service. It's very easy to make things worse if you're not sure of what you're doing.

JoulesperCoulomb Thanks for your reply. Back in the day my parents got me a ZX Spectrum which was 16k. A year or so later I was with my father in Stockholm city in a toy store called Stor & Liten to have them upgrade it to 48k. This being said, I only assume that this is my original one... they might have swapped it with a "real" 48k, who knows? As you will see in the images below, this Spectrum is indeed an Issue 3. My multimeter doesn't have a 400k Ohm setting but only a 200k one so that is why i thought the numbers i provided my have been different. Not sure how Multimeters, Ohm, Voltage and all that works but i want to learn. Anyway... so i tested the lower ram with Voltage DC setting at 20 and the values are the same for all lower memory chips: 5 V pin shows 4,96 V (seems ok) -5 V pin shows only -1,76 (not even close) 12 V pin shows only 8,03 (not even close) BTW, do you have a video on how to connect a composite cable and which alterations you need to do to the Spectrum? I tried plugging it in as is but got no video signal. This is the image of my ZX: imgur.com/a/DkOxy

It does look rather like an issue 3 that was sold without the upper RAM. Unusual! I think it very likely you'll find transistor TR4 (and possibly TR5) is faulty. The question is did it fail because of something shorting one of the supplies at the edge connector or because of a failed 4116 IC doing this instead? Your multimeter is very unusual if it hasn't got a diode test mode, so you should be able to try the tests on TR4 & TR5 in circuit. What results do these give? If necessary, TR4 can be replaced with a ZTX651/652/653. You might get through several if the lower RAM overloads the voltage converter circuit and you don't use a bench power supply with a current limit. Faulty lower RAM ICs can be replaced with 4116s or all 8 with the ZX Spectrum lower RAM replacement module (zx.zigg.net/LRR). The composite video signal is carried into the modulator by the wire that passes through the plastic insert on the left hand side of the can. The other connection is the 5 V supply. Both can be de-soldered from the circuit board and bent up out of the way or cut off (make sure they can't break off and become loose inside). Take the top off the modulator can and disconnect the resistor that couples the RF signal to the output connector. Install a new wire to carry the composite signal from the PCB to the output connector. I usually use a 100uF capacitor (electrolytic, positive terminal connected to the Spectrum PCB) instead to do this job as it removes the DC offset that some displays don't like, but most will work without this. Again, I strongly recommend you get some assistance if you have no experience of the repair/assembly of electronic circuits.

Superb... i'll see if I can replace the TR4 & 5, just need to find the correct parts. It seems to me too that they might be faulty. I'll get back to you in a few weeks i guess since i need to order the parts. My multimeter do have a diode test mode but i never got anything close to your results. Thanks again ^^

+cb meeks If the 0.4 ohms you measured was across the power input, then there certainly is something wrong, very likely the 7805 has failed. Capacitor C50 is another candidate for causing this problem. If the Spectrum still has its original electrolytic capacitors, I recommend replacing them anyway. If you want guidance on doing this: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-TeCVt2T-YSo.html 1.8 k ohms between the 0 V and 12 V rails doesn't suggest a fault. All the Spectrum models that use a DC barrel connector for their power supply input require negative on the centre pin, as do quite a number of other items of electronic equipment the Sinclair brand was in some way associated with. I suggest putting a label of the rear of the case to remind you of this.

JoulesperCoulomb Thanks for the reply! Do you think it's worth the effort to replace that barrel jack with one that's reverse?

+cb meeks I think you'll struggle to find a new socket that does that, but I recommend you don't anyway. You'd then have a Spectrum that is different from any others and one day someone will plug a standard ZX supply into it. If you're not using an original ZX PSU, I suggest, if necessary, replacing its output plug with one wired correctly for the Spectrum or permanently glueing on an adaptor to achieve the same thing, then put a big label on the supply saying it is for ZX Spectrums only.

JoulesperCoulomb That's a good point. My problem is that I don't have a PSU for a ZX Spectrum. And, I'd rather use something modern for that anyway. So, it sounds like what I need to do is get a modern PSU and make sure the polarity is correct (with an adapter) and label that PSU as ZX Spectrum. So, what ratings would you suggest for a modern replacement? 9V, 750mA? I do plan on plugging in expansion cards like something with an SD card and perhaps an Arduino/AY-3-8910 combo. So perhaps a 9V, 1.5mA? Thanks again!

+cb meeks Look for a regulated 9 V supply capable of outputting at least 1.5A. If you wish, I can sell you such a device with a rating suitable for powering all of the ZX Spectrum models that have a two way power input socket. Contact me at zx.zigg.net if you are interested.

Your message doesn't make much sense, but I think you might be enquiring about the suitability of the power supply unit you have obtained. The one you describe should work, so long as it has a DC output, HOWEVER, note that the centre terminal of the ZX Spectrum's power connector must be connected to the negative side of the supply. The computer will probably be damaged if you get this wrong.

+TheGoldieOldies It doesn't matter that your multimeter has different ranges. This method of testing is a bit inexact and the results will vary from meter to meter and depend on the range any particular meter uses. To some extent, it's a matter of getting to know what results should be expected when a particular meter is used. The results vary a bit between the different versions of the ZX Spectrum too. However, the types of faults being tested for here will generally result in very low resistance readings (often less 150 ohms or so). I don't fully understand your description of the outcome of your tests on TR4 as you don't explain exactly what you were doing. What is 1415? What units does it have? Most meters don't beep in any circumstances when set to the diode test mode and if I encountered one that did, I think it likely it'd be to indicate a short circuit and not an open one. However, I know nothing about the meter you used. Have you checked all three power supply voltages at the lower RAM ICs? Given the picture produced, I rather suspect they will be present and correct, but they should be checked. Can you hear a faint buzzing or similar noise coming from the inductor? The display in the linked photograph is a rather unusual one for a faulty Spectrum. It makes me have a slight suspicion the ULA may be faulty and I recommend you test it in another Spectrum if you are able to. Don't try another ULA in the faulty Spectrum at first, just in case it has a fault that causes it to damage the ULA, though this is unusual. If the ULA is not faulty, I'd next suspect the lower RAM (or its power supply), followed by the upper RAM.

+JoulesperCoulomb Hi, i'm using this multimeter: s21.postimg.org/onktybvp3/196402037_hires.jpg I put an arrow there to show you which feature i used to test TR4. 1415 is a value i got when i placed the red probe on TR4's base and black probe on its emitter. When I placed the black probe on its collector the display on multimeter just showed OL and it was beeping. I've just measured all the voltages as quickly as I could and here are the results: Lower RAM: Bottom right: 4.98, Bottom left: 8.24, Top left: 0, Top right: 0 TR4: Emitter: 0V, Base and Collector: ~8.95V TR5: Emitter, Base and Collector: ~5V There is no buzzing from the inductor coil.

+TheGoldieOldies The incorrect power supply voltages and lack of noise from the inductor indicates the voltage converter circuit isn't running. I'm very surprised the lower RAM is able to accept and hold enough data to produce the display in your first photograph. The diode test mode of a multimeter normally shows the voltage drop across a diode when it is forward biased or open-circuit (it usually displays OL, but doesn't beep) when there is no conduction. Perhaps it means 1415mV for TR4's base-emitter junction, though that's higher than it should be. I'm fairly sure the beeping when you test TR4's base-collector junction means there is a short circuit; this is how TR4 often fails. Try measuring the base-collector junction with the meter set to the 200 ohm resistance range; it'll probably give a very low reading. You ought to have noticed this if you had measured the resistance between the two windings of the inductor (this is the same as measuring TR4's base-collector junction). It seems TR4 has failed and it should be replaced with a ZTX650/ZTX651/ZTX652/ZTX653. Very few other small transistors will work reliably for TR4. With TR4 replaced, try powering the Spectrum again. If you have a bench power supply, set it to output 9 V with a limit of 750 mA. Remember the centre pin of the Spectrum's power connector is the negative terminal. If the current limit activates for more than a couple of seconds, turn the power off immediately; there is still a fault and the new TR4 will be destroyed if you don't. If you have to use the standard ZX PSU and there is still a fault affecting the output of the voltage converter circuit, then TR4 will probably be instantly damaged. If you're lucky you'll now have a working Spectrum. If not and TR4 fails again or the computer activates the supply's current limit, it's likely one of the lower RAM ICs is overloading one of the power supply rails. It's difficult to determine which IC and so you'll probably have to remove all of them. See another video I've made about this process should you want guidance. Unless you are very confident you can remove them without damaging the PCB, I recommend you cut through their legs. If, however, after replacing TR4 the replacement does not fail and all power supply voltages are correct and present, but the computer still doesn't work, it's still likely one or more of the lower RAM ICs is faulty. This is often the case if they have been incorrectly powered, as the ones in your Spectrum have. If the Spectrum still has its original electrolytic capacitors and the voltage converter circuit isn't producing 12 V and -5 V, then I recommend you replace these before removing any ICs. Failed capacitors are a very common cause of faults. Faults in C46, C44, C45 & C47 in particular are likely to affect the voltage converter circuit. Even if you get the computer working without doing this, I recommend the electrolytic capacitors be replaced anyway. There is a fairly high chance they will cause a failure soon.

+JoulesperCoulomb Thank you very much Sir. I've just ordered couple of ZTX653s. They should arrive next week. As for the ULA testing you suggested..i have Spectrum+, issue 6A with 6C001E-7 ULA. Not sure if the ULA from the faulty speccy (mod. 5C112E-3) would be compatible with issue 6 motherboard.

+TheGoldieOldies Strictly not, however the 5C ULA might well work in an issue 6. Anyway, I suggest suspending testing the ULA until you've got the voltage converter circuit working.

How wide are the vertical bars? If a character width or similar, I'd first suspect the contended RAM; IC6 - IC13. Note that the RAM in the 128K Spectrums is implemented rather differently from that in the earlier models. Much of the information in repair video 7, particularly part 4, is relevant to your Spectrum, though the component numbering isn't the same.

They are 1 character wide, thank you again, I'll spend some time on it tonight.

JoulesperCoulomb I've tested for shorts etc, finger probing causes some graphical glitches but I can not pinpoint it? It looks like it maybe ram issues, do you recommend that I get it repaired or should I sell it as spares/repair? & is there anyone who can repairs these?

Obviously, it's up to you to decide what to do with it. It is one of the most desirable Sinclair computers, so I'd think it'd be worth getting it repaired. I can offer a repair service; send me a message at zx.zigg.net if you want to know more. Also, a few regulars in the Hardware forum at www.worldofspectrum.org offer repair services too.

JoulesperCoulomb Thank you such much, I will contact you soon. I would love to see it work again as my late father taught me how to operate the spectrum. We had the 128k version back in the day & I'm so happy to have one again... This one will stay in my collection forever.

I haven't got full details immediately to hand. However, I suggest you do the following... Look up the pin-out of the power input DIN connector. Pay attention to whether it shows the socket on the computer or the plug on the PSU, but you'll be able to work out what each input is so long as you know that. Then measure the resistance between the 0 V pin and each of the other voltage inputs. It'd also be worth checking the resistance between the supply rails (e.g. the 5 V and 12 V input), but you're less likely to find a problem there. This is a rather inexact process as the circuit being measured is far from a pure resistance, but just as shown in this video, an unreasonably low reading suggests a fault. I can't remember quite what the results should be for a non-faulty +3, but I would be worried about readings much lower than approx. 150 ohms and I suspect they ought to be at least a few hundred ohms.

+eLJaybud What is it missing? Almost every basic multimeter should be adequate for the tests demonstrated in this video; even the most cheap and nasty. If it has manual range selection, then its range divisions don't have to be the same as the ones on the meter I used. Simply select a range that covers the expected likely reading. As noted, the results for the resistance tests vary somewhat depending on the multimeter used and the version of ZX Spectrum. The aim is to make sure none of the circuit paths measured have an unreasonably low DC resistance.

I'd first check the zener diode in the voltage converter circuit. There's a possibility this will have killed the lower RAM ICs.

There shouldn't normally be any text near the top of the screen immediately after power on, so I'm not sure what you're referring to there. There are not many small transistors that will work reliably for TR4. I suggest fitting a ZTX651/ZTX652/ZTX653 for TR4. Unless other modifications have been made, the voltage converter circuit cannot work without TR5. Many small PNP transistors can be used for TR5; I usually use a BC557. I recommend fitting the correct transistors as described above and then briefly power the computer on and check for all three power supply voltages at the lower RAM. If it has been powered on without all of these present and correct, it is likely one or more of the lower RAM ICs will have failed.

The +3 is a rather different design and most of the tests described don't apply to it. Also, the +3 does not contain 4116 RAM ICs that are likely to be damaged if the power supply system isn't working properly. However, before switching on a suspect +3 for the first time, I would take resistance measurements across the power supply rails to check there isn't a short or near short between any of them. The +3 doesn't have internal voltage conversion circuits like many of the earlier models; the external PSU provides the various voltages needed. So, this test can be done at the external DIN power connector without needing to open the case. If possible, power the +3 for the first time from a bench power supply with current limits set and this might avoid damage in the case of some faults. Most functions of a +3 should work with the +5 V and + 12 V supplies present. If you have to use an original +3 PSU, check it is producing the correct voltages before connecting it.

If the Spectrum works perfectly apart from the display problem you describe, we can conclude that the fault almost certainly lies in the video circuitry. An oscilloscope would be very useful here; I'd want to use one to view the final video signal and the signals at some of the encoder IC's pins - the Spectrum service manual has some diagrams showing what should be found here. You might have to consider the possibility of the failure of one of the components that almost never fail; resistors and the small capacitors. I have encountered a few very strange video faults that were caused by these components. I recall one case where a ceramic capacitor had gone short-circuit and another where a resistor had broken in two, but the break was only apparent with very close inspection under a magnifier.

If the meter is in proper working order and you know how to use it to measure resistance and voltage, then you should be able to do the job. I don't know what resistance ranges your meter has, but use of one of the lower ones (say 1 kilohm or maybe even 10 kilohm) should be appropriate for most of the tests. The readings may well be rather different from those obtained using most digital multimeters, but it ought to be possible to use it to detect many of the most common faults.

Did the Spectrum work? I suspect you simply had the polarity of your test leads reversed.What voltage did you measure on pin 9 of one of the 4116s?

JoulesperCoulomb hi yes sorry you're right had my leads reversed! just panicked and turned it off. but still nothing on the screen!

Steven Swaffer will be looking a TR1, TR2 as per your repair No 1 video cheers!!!!!

It's rare for those to fail, but faults with them can cause a lack of video output. How are you connecting your display to the computer?

Is there a click from the sounder when you press a key? If there is, the Spectrum is probably working properly apart from the lack of picture.

+ISFiYIywAFIBc6qAIIIIIIIIIIIIIIIIQrXTJiCtY3Asd4WF Since so many components are connected to the 5 V supply, there are many candidates for short circuiting it to the 0 V rail. I'd start off, though, by suspecting a failed 7805. Any power supply related fault is liable to affect transistors TR4 and/or TR5. I recommend removing them and testing them out of circuit. What result do you find if you measure the resistance between the two windings of the inductor? It should be very high/open circuit. If not, most likely TR4 has failed with a a base-collector short.

+JoulesperCoulomb Measuring the resistance between the two windings of the inductor gives me a reading in the Mega Ohms range. I have ordered some flux to use with solder wick, as this is my preferred method of desoldering and I seem to have misplaced my previous bottle of flux. (Solder wick + lots of flux = great results.) So until this arrives, I will not be desoldering any components. I do not want to ruin the pads on the board. Incidentally, in case I find that TR4 or TR5 are faulty, what readily available replacements would you suggest I use? It seems that ZTX transistors seem to go at ~£1 per piece which seems rather expensive. When I have that flux, I'll de-solder the transistors (and the regulator) and test them out of circuit. Thanks for your help.

+ISFiYIywAFIBc6qAIIIIIIIIIIIIIIIIQrXTJiCtY3Asd4WF There are very few physically small transistors that'll work properly and reliably in place of TR4. I recommend you use a ZTX651/ZTX652/ZTX653. These are not especially difficult to obtain. I think ZTX653s were about 60p each in small quantities when I last bought some. Most small signal PNP transistors ought to work for TR5. I usually use a BC557. Check the pin-out of any replacement carefully; it may not be the same as the ZTX213. If I remember rightly, the BC557 has its emitter and collector swapped when compared with the ZTX213.

+JoulesperCoulomb You are indeed right, I can find ZTX653s much cheaper than ZTX650s (about 30p a piece). Thanks for sharing your knowledge.

+JoulesperCoulomb while I am fixing this ZX spectrum, I would like to apply a composite video mod, there seem to be two popular mods, one using a capacitor and one using a transistor (apparently the transistor mod is brighter) I happen to have a few transistors which would work for the transistor mod (and capacitors which would work for the capacitor mod) do you have any preference? Does it really matter that much?

An issue 4 Spectrum should have -7 type ULA. The -6 will often work, but sometimes leads to unreliable lower RAM accesses (with all sort of strange symptoms). You say the -5 V supply is missing, so I assume you have measured the power supply voltages at the lower RAM with the power on (though you don't directly say you have done this). What voltage did you find at the 12 V rail? I'd expect the computer to not be able to initialise at all with -5 V missing and this could well lead to the lower RAM ICs being damaged. If this is correct, I'd suggest one of the following has happened: 1. The voltage converter circuit is broken. 2. The voltage converter's -5 V output is being overloaded. 3. A broken PCB track is preventing the -5 V supply from reaching the lower RAM. TR4 burning out is the most common failure mode of the voltage converter circuit. I assume your tests of TR4 and TR5 were made in-circuit using your multimeter's diode test mode. They suggest that both of these transistors are in working order. Barring 3, problem 2 is quite likely and the only easy way to find out is probably to remove all of the lower RAM ICs; though you could take them out one at a time and see whether the -5 V supply returns after removing each one. With the lower RAM removed you should make sure that all power supply voltages are present before fitting any new RAM. Lower RAM can be replaced with new old stock/used 4116 ICs (150 ns access time or faster) or with one of the modules available at zx.zigg.net . If you aren't absolutely confident you have the skills and equipment needed to de-solder RAM ICs without damaging the PCB, I recommend you remove them by cutting through their leads and then remove these one by one. It is easier to replace the RAM than it is to replace the PCB!

Hey Joules, I didnt power up this board since i experience these artifacts on screen after boot. I still need to take care of power supply higher voltage issue. But i follow up your initial tests with resistance measuring. I didnt said i dont have -5 volt rail. What i said is that i have cca 99 kohm of resistance in multimeter auto range but in manual ohm range i have OL on screen so that is odd. If i recon you have few hundreds of ohm at -5 volt rail!? Lower RAM on my board is socketed luckily so i can use extractor to remove it from board. I would like to replace all eight chips with new guys in 4116 packets. I would like this board to be closest to original. Do you know where to get these in NOS or brand new? What do you think about these?: www.jameco.com/z/4116-15-Major-Brands-Dynamic-RAM-16kx1-150ns-DIP-16_41339.html I still need to do recap on that board but i am absent from my home currently. It would be nice to remove tr4 and tr5 and do measures out of board and change it in case of failure. It is odd to me to see that ULA on 4A board. Is there any way i can find proper ULA chip for my board? So before i do recap and wait for new chips to arrive i cant do more. And still need to take care of power supply :)

You said 'Yeah thats it, so it seems i dont have -5V rail', which I took to mean you had measured the voltage somewhere on the -5 V supply rail and obtained a result other than -5 V. Anyway, if the lower RAM is fitted in IC sockets, you should remove it all and make sure all the power supplies are working before re-fitting it or fitting new ICs. You definitely should replace all of the electrolytic capacitors as failed capacitors are a very common cause of power supply problems. The resistance readings you have taken do not obviously indicate the presence of a fault. Because of what is being measured, the readings may vary considerably depending on the multimeter used and the range it is set to. However, in general, very low resistance readings are indicative of a fault. The 4116 ICs described at the URL you give should be suitable for use in a ZX Spectrum. ZX Spectrum ULAs aren't all that easy to obtain, other than by removing them from other Spectrums. I believe that one or two individuals have finally completed projects to create new (CPLD based) replacements. I haven't any details on hand, but I have seen some advertised for sale somewhere.

Yes sorry for misunderstanding, English is not my native language. I would like to find such ULA chip. I will definitely recap this board and change RAM chips before fitting new power supply.

Whatever you do, make sure all the power supply voltages are present and correct before fitting new RAM, otherwise the new RAM will likely be damaged.

+derekjc777 I've never tested an issue 1 Spectrum. The readings you report do not indicate the presence of any serious fault.

Either the voltage converter circuit isn't working or one or more faulty lower RAM ICs are overloading its outputs. There is a potential circle of failure and all parts must be corrected in one go. The voltage converter circuit most commonly fails for one of the following reasons: - Aged electrolytic capacitors. If your Spectrum has its originals, I recommend you replace them (see another video I've made about this process if you need advice on this). - Incorrect use of the expansion connector has shorted one of its outputs. - One or more faulty lower RAM ICs has overloaded its output. In all of the above cases, TR4 usually burns out and sometimes TR5 as well. It's rare for the inductor to fail, but it can happen. Lack of the supplies generated by the voltage converter circuit often causes one or more of the lower RAM ICs to fail, so ideally, you must make sure the converter circuit is fully working and none of the lower RAM is faulty before applying power again. If you use a bench power supply to feed the Spectrum 9 V DC with the current limited to 750 - 800 mA, then usually if the current limit activates for more than a few seconds, you'll probably have a few seconds further before TR4 is destroyed should one or more of the above faults still be present.

Thanks for the speedy reply. I'll change the caps and possibly Tr4 & Tr5. I hope the lower ram is ok as I'm able to enter "PRINT PEEK 23732 + PEEK 23733 * 256" and get the 65535 (48k answer). It wont load tapes either and I hear no loading sounds although beep command and key click is ok. hopefully not a faulty ULA, even though it gets quite hot... I have already purchased a new keyboard membrane and upper ram module. I'm losing faith with it, it's starting to cost more than it's worth.

I'm a bit confused. If the lower RAM 4116 ICs haven't got all three of their power supply voltages, then there's no way the computer will initialise and give you the Sinclair Copyright message. I think you must have made a mistake in measuring the voltages. If the computer produces the correct initialisation display, then the voltage converter circuit must be working and the lower RAM is probably not faulty (though it might have a partial/intermittent fault). I still recommend replacing the electrolytic capacitors, however. If audio output is working, then the ULA's audio I/O is probably (but not certainly) not damaged. What are you using as the audio source? Many modern devices (e.g. phones/laptops/MP3 players/etc.) cannot produce a signal of sufficient amplitude for a Spectrum to be able to load from.

I can assure you, those are the voltages at the lower ram chips tested with my Fluke 77. the Spectrum works set on at 8.42v. Slightly less then I boot to black screen with white border and If i up the power to 8.75v or more I get black squares starting to appear on the copyright screen. As for audio, I have tried phones, pc Soundcard out and a cassette player ( only radio noise as I have no tapes) I hear nothing through the Spectrum's speaker at all. I did trace the audio back to pin 28 (i think it was) of the ULA socket and sound "in" is present but very quiet when the ULA chip is installed and clear as day when the ULA is not in there. (hope that makes sense) . I plugged headphones into the mic socket and wrote a quick hello world program then saved it. sound was spot on and loud through the headphone. Think I'm getting far beyond my capability's here but I do appreciate your help :)

I'm still a bit doubtful. If your information is correct, then your Spectrum has miracle 4116 ICs that are unlike any other! Perhaps they're just about managing to work in some way with the voltages you've measured, but I'd certainly expect the whole system to be unreliable. Are the voltages you refer to in your message immediately above what you are feeding into the Spectrum's main power input socket? A Spectrum shouldn't be very sensitive at all to variations in its supply voltage. If so and the regulated 5 V rail is remaining steady, then I think your Spectrum has at least one of the faults I described in my first reply. Can you hear a soft buzzing sound from the inductor when the power is on? Have you tested TR4 and TR5 (in-circuit measurements are usually enough to tell whether these have failed)?

I could be more certain if I were able to see the thing you describe. However, I suspect it is most likely a necessary modification because the computer has or had a 5C102 ULA fitted. This has a bug in its handling of I/O contention, which was (mostly) corrected by fitting an extra IC in the manner you describe. I assume this is an issue 2 or (if you're very lucky) issue 1 Spectrum. If it hasn't been upgraded to a 48K model, is the software you are trying to load for the 16K Spectrum? Not too surprisingly, software expecting to find 48K of RAM won't work in this case. Otherwise, faulty RAM is one of the most common causes of this type of problem. Try loading and running the Memtest program available at zx.zigg.net . Tell it to test the lower RAM. If your Spectrum has been upgraded to 48K, also have it test the upper RAM on a separate run. If the program should stop with an error report, post a copy of its output.

JoulesperCoulomb Thanks for the reply I tried loading the memtest and a 16k game the program is found but then I just get the "M RAMTOP no good, 0:1" error. Would a new lower ram fix module this do you think?

It might, but I'm a bit doubtful. It's unusual, though not unknown, for a Spectrum to have a lower RAM fault and there not to be corruption of the display, which you haven't mentioned. Are you sure it hasn't got the upper RAM fitted? The symptoms you describe are more often caused by a faulty upper RAM IC interfering on the data bus. What result does the computer display if you execute the BASIC command PRINT PEEK 23732 + 256 * PEEK 23733 ?

Upper ram sockets are empty and has 16k sticker on the back. I get 32767 result from commands, there is no corruption the spectrum starts up ok and simple basic listings work ok. But when loading from mp3 player via ear port it gets past the found "program name" part but it either resets, or gives the ramtop error depending on the game or rom i try. I have also replaced the capacitors.

This is certainly a rather unusual fault. If I wanted to be fairly sure of uncovering the cause, I'd connect a many channelled logic analyser and capture what the CPU is doing at the time of the crash. I could then work my way through that, with a listing of the ROM disassembly on hand and I'd very likely be able to identify what went wrong. However, this process can be extremely time consuming and tedious. If I wanted a quicker option, I'd probably try replacing the lower RAM as that has to be a prime suspect, but we can't be at all sure it is the cause of the fault. WARNING: If you're going to remove the 4116 ICs and you haven't the equipment, experience and confidence to remove them intact without damaging the circuit board, I strongly recommend you use the method of cutting their legs off, before removing each leg from the board individually. It's much easier to obtain replacement RAM than it is to repair a circuit board that has had copper pads and tracks torn off. For further guidance on this, see: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-u_NnU48bmv4.html

leshurex I would think that bypassing the modulator in the same way as in earlier versions of the Spectrum would work. However, in the interests of keeping the computer in close to original condition, I wouldn't bother; the composite video signal is available on the DIN video socket. To use this output, one would wire the screen of a 75 ohm coax lead to pin 2 and the core to pin 1 of a suitable eight way DIN plug and the other to whatever type of connector is required for the display.

JoulesperCoulomb Thanks for your quick reply! I bought that model some months ago, along with a faulty 48k model (displays flashing coloured blocks). At that time I performed the composite mod for both of them. Sadly, I never got to test the 128k given that it won't turn on (probably because I'm using a generic multi-voltage psu). Anyway, since then I was worried that the mod could've damaged the Speccy.

leshurex A basic power supply that outputs approx. 9 V DC and is able to supply around 1 A will do for either computer. Don't forget that the centre terminal of a Spectrum's power connector is the negative one. Getting the polarity wrong is likely to cause damage. There is a video in this channel about ZX Spectrum power supplies, should you be curious to know more. In the case of the 128K Spectrum, if there is no signal on the composite output, it's a good idea to try the RGB output. If you can obtain a picture from the latter, it's likely the composite encoder IC or the voltage converter circuit that generates its power supply has failed. As always, there could be many causes, but a display of flashing coloured blocks in the 48K Spectrum most likely points to a lower RAM fault or failure of the voltage converter circuit.

JoulesperCoulomb I remember testing the 48k's RAMs following this video and found any wrong values (I don't know if that rules out a RAM fault or not). How can I test the voltage converter circuit you mentioned? If it's worth mentioning I found that the X1 crystal was (pretty lousy) replaced with another, but not with the same component. As for the 128k, I believe there was some problem with the power feed, but I would have to verify that later.

leshurex Good results from the tests described in this video don't guarantee the Spectrum is free from faults. Have you measured the three power supply voltages (with power on) at the lower RAM? X1 should be a 14 MHz quartz crystal. If the Spectrum is producing a picture, then there is almost certainly not anything wrong with X1. If you get completely stuck and want to have your Spectrum(s) working again, I can offer a repair service (contact me via zx.zigg.net).

Carry out the tests described in this video and report back on the results.

JoulesperCoulomb I will do, thanks

JoulesperCoulomb Hi, I have sent you a private message with a link to video that will hopefully explain what is happening. Thanks