Cant load anything

[Adrian45]

Hello all - Newbe to the forum

Got my hands on what I believe is a 16k zx , issue 4A

It powers up ok

Keys didnt work , found breaks in the ribbon , cut them back with a sissors , got all the keys working.

But the machine wont load any software.

I had an Oric Atmos back in 1984 so im well used to levels , treble etc etc ( the oric was a dam nightmare for loading /saving , remember being so jealous of speccy owners)

I get the usual blue to red background - when the signal starts i get an intermittent horizontal bar pattern , as if the sound jack was being plugged in and out !!!

Tried mp3 , wave files ....... volume up , down , mono , stereo ....... no change. Have not got one register on the machine of a program name.

I think this machine has been recapped as some of the caps are radials that have had their legs wrapped around the cap to serve as an axial

Looking at the back of the board i see it was an ok soldering job but loads of flux everywhere.

cleaned it all up with iso but no change. I think some of the caps are the originals. Not too sure

It has had the composite mod done to it but the display is not great at all

There is also some kind of mod around the speaker , diodes and rerouted cap - guy i got it from says it was a mod to cut the sound to the speaker , but doing a bit of searching it looks like a standard mod to improve the power circuitry ....... again not sure

I want to fully restore this machine , switching power 5 volt , new membrane , new caps and a 48k upgrade.

all the sites i have looked at re the 48k upgrade give instructions of just simply installing the new chips in sockets , soldering in jumpers........ this board has no sockets , so i will have to remove old chips , install sockets , install new chips ......... so im a bit confused ... why no sockets. I typed in a few peek commands that confirmed it is indeed a 16k machine ...... but again i could be wrong

Im wondering if old failing caps would degrade the tape loading - i used speccys back in the day and they were bullet proof for loading ( compared to my crappy oric) - you could probably load a game over the phone the speccy was so good !!!

as for me - ive restored many old machines , amiga , macintosh , atari - im good with the soldering iron , cant read schematics , no scope and know basically nothing about spectrum's , other than i spent hours and hours playing games on them 30+ years ago !!!!

any advice would be greatly appreciated folks


heres a pic of the board with heat sing removed and the display im getting

cheers

[dfzx]

Your machine is a 48K, the DRAM slots in the lower right are all populated. It sounds like it was sold to you as a 16K because the upper RAM isn't working?

You might still be able to load a 16K game, though. What device are you using to play the tape? Spectrums require significant volume from the tape player, many modern devices like phones can't go that loud and you see the effect you're describing.

[Adrian45]

Thank you for that reply dfzx

That makes allot of sense. I was wondering how it could be a 16k with all the slots populated , and didnt know the machine could power up at all if there was a memory problem.

I typed in 'print peek 23732 +peek 23733 * 256 " and got a result of 32767

And you guessed right , im using the hedaphones socket from a desktop , did the same with a laptop then used an mp3 player , all set to max volume ( among other settings ) - even tried VLC player as you can 'boost' the volume to 125% - no good - the lead im using is a stereo lead , so I used a small bit of silver tape to force one end into being mono ( not good for laptop i guess , but had to try) made no difference

I have an mp3 of jet pack and death chase as they are both 16k games , but as I said speccy cant even detect the header name.

Ive gone over the board with isopropanol , had a good look with a mag glass , both sides , cannot see anything obvious - But knowing now it is 48k there is a problem somewhere. I have a mono lead somewhere - ill have to dig it out and try with that to be sure im getting the full signal into the machine

What do people use today to load files into their speccys , headphone socket + small amp ?

cheers

[dfzx]

There's a bit of information in this thread about volume and amplification and the options open to you. Probably other threads as well if you search the archives, it's quite a common subject.

If you want to investigate your apparent upper memory issue I suggest you start a new thread with a suitable subject line. This one's gone a bit under the radar. Start by asking if it's possible to test upper RAM with basic tools - I'd quite like to know the answer to that one myself.

[Adrian45]

Are you supposed to hear the noise of software loading from the spectrums internal speaker during loading ?

Im hearing nothing from it - zero

[Adrian45]

just saw your post dfzx - thanks ill have a look

Ive been loading software from pc's to many old 8bit machines - i dont think its volume - i plugged the lead into my monitors headphones socket , with sound routed through the hdmi - that allowed me to use the monitors volume to amplify the signal - still nothing

im thinking it is a hardware issue - would old failing caps possibly effect the signal going into the machine

[dfzx]

Are you supposed to hear the noise of software loading from the spectrums internal speaker during loading ?

Yes! The tape loader screeching noise is absolutely characteristic of the machine and you should be able to hear it as the tape loads. If you're not hearing that, you're missing out. So is the Spectrum, it can't hear the signal. You need an amplifier or a 1980's cassette player. The latter, for choice, obviously.

[Adrian45]

Thanks a millions dfzx - that link explained the problem very well - I have one of those cassete's with the headphone lead coming from them ( used to connect cd players to car cassete players ) - ill use that , with an old tape deck and try again

Thanks again - great help man

[dfzx]

i dont think its volume ... im thinking it is a hardware issue - would old failing caps possibly effect the signal going into the machine

Here's another thread on the topic on another site. Here's another.

It isn't out of the question that a capacitor is causing a problem, there's a bit of filtering on the input line. But this is one of those "thundering hoofbeats probably means horses, not zebras" things. The easiest test is to get an amplifier. Do that before you start desoldering components.

[Adrian45]

SUCCESS !!!

Pulled out an dusty old beringer mini mixer from the attic - sound from pc headphones to line input - beringer headphones out to zx - gave me treb, bass , mid ,pan , gain and volume

first try and i got a steady signal into the zx

first try loading game and it worked

but no sound from spectrums speaker when game playing - so i reversed the signal - spectrum out to beringer in and sure enough i can hear the games sounds on my headphones ........ perhaps the little speaker is blown !!! - its not physically disconnected as far as i can see

would have never thought that would be the problem - thank god for forums is all i can say

[dfzx]

SUCCESS !!!

Hurrah!

but no sound from spectrums speaker when game playing

OK, power up into BASIC and try

Code: Select all

BEEP 1,0

If you hear the tone from the internal speaker then it's clearly working. If not, that needs looking at as well.

[Adrian45]

did that - no sound

even typed it in a prog with a goto 10 loop ........ nada

But im happy - machine is 95% ish ! working

will try loading up a 48K game and see what happens - If there is a memory problem i guess it will crash when it runs out of memory or just freeze up

[Adrian45]

sure enough it didnt load

tried to load manic miner - failed to load

problem with upper memory and no sound

but hey , the chase it what makes it fun

will have a go tomorrow

thanks people for the great advice

[1024MAK]

Got my hands on what I believe is a 16k zx , issue 4A

As others have said, it's actually a 48K board (as manufactured)

But the machine wont load any software.

I get the usual blue to red background - when the signal starts i get an intermittent horizontal bar pattern , as if the sound jack was being plugged in and out !!!

The Ear socket looks to have been butchered! I'm assuming you did not do this? If you did not, that means the previous owner did the dirty deed. So they knew there were problems.

It could well be a fault elsewhere. More later.

I think this machine has been recapped as some of the caps are radials that have had their legs wrapped around the cap to serve as an axial

Apart from C44 and C45, all the rest are original electrolytic capacitors. C44 and C45 may also be originals, as when the factory ran out of axial electrolytic capacitors, they used radial and bent the leads just like on your board. Or they may have been replaced at a later date.

Looking at the back of the board i see it was an ok soldering job but loads of flux everywhere.

Sounds like someone attempted some repairs at some point. But keep in mind the DC/DC converter mod would have required some hand soldering.

It has had the composite mod done to it but the display is not great at all

Which version? just using a wire, using a electrolytic capacitor, using a NPN transistor or using a add in board?

There is also some kind of mod around the speaker , diodes and rerouted cap - guy i got it from says it was a mod to cut the sound to the speaker , but doing a bit of searching it looks like a standard mod to improve the power circuitry

That's an official Sinclair modification of the on board DC/DC converter circuitry. I have a blog page with details if you are interested.

all the sites i have looked at re the 48k upgrade give instructions of just simply installing the new chips in sockets , soldering in jumpers........ this board has no sockets , so i will have to remove old chips , install sockets , install new chips ......... so im a bit confused ... why no sockets. I typed in a few peek commands that confirmed it is indeed a 16k machine ...... but again i could be wrong

That's a 48K board. Don't remove anything yet.

Im wondering if old failing caps would degrade the tape loading

Not if the machine will run a BASIC program, no.

That photo of the screen looks normal for an old unserviced machine on a modern LCD TV. What exactly are you not happy about?

Back to the LOADing problem. First off, the Ear socket (which is the tape input socket, don't plug anything into the Mic socket) will only work with 3.5mm mono plugs. I just want to get that in first, as two common mistakes are trying to LOAD using a stereo plug or using the Mic socket.

Does this Spectrum make sounds, say if you use BEEP 1, 12 ? Is it loud or only just audible?
When LOADing, can you hear sound from the Spectrums speaker? If yes, is the sound loud or only just audible?

Do you have a digital multimeter? If yes, is it a cheap model? Can you post a photo of it please.

As I have said, this is a malfunctioning 48K board.

Can you run this test program:

Code: Select all

48K ZX Spectrum quick ‘n simple RAM test
1 CLEAR 24999
2 FOR a=25000 TO 65535 STEP 250
3 PRINT AT 0,0;a: POKE a,0: LET d=PEEK a: IF d<>0 THEN GO SUB 8
4 POKE a,255: LET d=PEEK a: IF d<>255 THEN GO SUB 8
5 NEXT a
6 PRINT “done”: STOP
8 PRINT “error at “;a;” “;d
9 RETURN

Then at any of the addresses where an error is reported, enter into this program:

Code: Select all

RAM test, test one address only looking for a bit error
1 CLEAR 25999
2 PRINT “Enter address to test”
3 INPUT a
4 PRINT “Address ”;a
5 FOR c=1 TO 4
6 READ t
7 POKE a,t: LET d=PEEK a
8 PRINT “Value written ”;t;“ value read ”;d
9 NEXT c
10 DATA 0,85,170,255

Report back the results.

Mark

[1024MAK]

So, reading through the topic, the remaining issues are the RAM fault and no sound from the speaker, but you do get sound from the jack socket, is this from the Ear, the Mic or both?

Mark

[1024MAK]

If an admin or a mod is reading this, maybe update the title of the first post to better reflect the topic content?

Mark

[Adrian45]

Hi 1024MAK

Thanks for the help

Im getting sound from the mic port - i didnt even try the ear port - i could be a blown speaker , ill test it with the meter

I typed in my first program on a spectrum - omg it took 20 minutes !!!!!....... how in the name of god did kids type in listings from magazines on these machines back in the day ...all i remember was " J , symbol shift P,P " ....anyway results from the program

program starts at 25000 and is checking memory

when it gets to 32000 it reports first error , and every location after that is also reported as a error ,all the way to the end at 65500 . displaying a value of 16 next to every fail........ as if the upper memory is not present ......... anyway typing in second program

Getting faster typing in on this crazy little keyboard ........ results

25000
value written 0 , value read 0
value written 85 , value read 85
170 , 170
255 , 255

all good

32000
same as above
32001
same result

32010 - same good result
32500 - all good

33000 = errors
value written = 0 , value read = 16
value written =85 , value read = 85
value written = 170 , read = 186
written = 255 , read = 255

34000 = same error as 33000
44000 = same errors
54000 = same errors
65500 = same errors

problem starts between 32500 and 33000 - ill try home in on exactly where

32750 = good
32900 = same fail
32800 = fail
32775 = fail
32760 = good
32770 = fail
32765- good
32768 = fail
32767 = good........... found ya !!!!!

it seems every memory address after 32767 is not storing its data correctly
I see the exact same write / read error on every address

I said above when running first program that error starts at 32000.......it was running very fast and could easily have been 32767 ....... the numbers were just wizzing by

Looking at that neat little program we are writing to an address , reading it back , 4 times on the same address

so it is saying the first and third writes are not coming back correctly !!!! or only every second write / read is good - and the bad write/reads are decimal 16 higher than they should be .................... HELP !!!!!!!!!





sorry for the long post , i was typing in the results as i went

this is fun !!!

[1024MAK]

So, the reason for those test values is because we really want four different binary test patterns.

With the RAM in a Spectrum, it's divided into the 'lower' RAM and the 'extension' or 'upper' RAM.

The 'lower' RAM runs from 16384 (0x4000) to 32767 (0x7FFF).

The 'upper' RAM runs from 32768 (0x8000) to 65535 (0xFFFF).

There are two physical groups of RAM chips. Each group has eight chips. So for a 48K machine there are 16 DRAM chips.

The type of RAM chips are called DRAM, for dynamic RAM.

Each DRAM chip can only store ONE binary digit per address. Hence one chip per Z80 data line. As the Z80 has eight data lines, there are eight DRAM chips per group.

Value written = 0 , value read = 16
In binary this is

Code: Select all

00000000  (0 dec)
00010000  (16 dec)

so bit 4 is returning 1 instead of 0.

Value written = 170 , read = 186
In binary this is

Code: Select all

10101010  (170 dec)
10111010  (186 dec)

so again bit 4 is returning 1 instead of 0.

Bit 4 corresponds to data line D4.

Conclusion, one of the eight 'upper' DRAM chips is faulty. Assuming Sinclair have not swapped anything around and the schematic is correct, that would be DRAM chip IC19.

You can prove that IC19 is the chip for data line D4. With the power disconnected, and your multimeter switched to the 2kΩ resistance range (or equivalent), test between IC19 pin 2 and IC2 (Z80A or NEC µPD780C-1) pin 7. You should get a low (zero) resistance. Test again using the 200Ω range and you should get a similar result.

Mark

[Adrian45]

Great explanation Mark

After reading it a few times I understand and it makes sense.

Will have a look later , take some readings with the meter and report back.

If one of the chips is damaged I have sockets , same memory chips ( i think) and the soldering skills to remove offending chip , install socket and install new memory chip.

Thank you all so much for your help.

will report back later.

[1024MAK]

the replacement must either be the exact correct 32K DRAM chip. Or a 64K DRAM chip that has the same row/address layout.

Mark

[Adrian45]

I took readings those readings from ic19 and z80 , at 2kΩ and 200Ω - got zero in both reads

had a look with magni glass at ic19 , pins , traces - could not see anything obvious = all looks good

powered on machine for 10 min , took temp readings of chips - ic19 was 2 degrees warmer than surrounding chips

had a look at the 4116 memory I have and noticed the upper ram is not the same chip as the lower ram

upper ram = tms4532 - 15nl3

did some reading and I cannot use 4116 chip for upper ram

I do have a 128K +2 and a commodore 64 and it says one of those might have the chip

quite happy to scavenge the chip from another machine for now

should I remove ic19 and socket it or is there anything else i should try before i take out the iron

[1024MAK]

Yes, different DRAM chips are used for 'lower' and 'upper' RAM. 4116 are not suitable to replace a upper DRAM chip.

Be back later, as the site is about to go off-line...

Mark

[Adrian45]

well i went ahead and removed IC19

was allot more difficult that expected - couldnt get all solder out from both sides so had to use air gun for final removal

dont think i damaged any pads or traces - will have to clean it up for a better look.

Now I have to hunt down a suitable replacement chip

I have a commodore 64 and a 128K +2 spectrum - would either have a suitable ram chip ?

[1024MAK]

So, I'm back

For a replacement chip, you either need a TMS4532 - 15NL3 (which are long out of production and nearly as rare as a live Dodo).

Or a 4164 DRAM chip, preferable a Texas Instruments (TMS) one. Although various other makes will work as well. If you find any or intend to buy some, do find out what 'speed' they are.

Before trying to raid another machine, keep in mind that nearly all manufactures soldered DRAM chips. Which 128K +2 spectrum. The grey or a +2A/+2B? The latter use a different DRAM chip again.

If you want to buy some, I can recommend a seller (not me!).

Mark

[1024MAK]

Post above edited.