Stuck with repair of Spectrum+

[speccyplus]

Which Z80 did you buy?

Any rated at 4MHz or a faster speed should work. The original fitted chip would have been a Z80A or equivalent. CMOS Z80s do work as well and are available brand new from the larger electronic companies (e.g. RS components).

Mark

Got this one, slightly faster at 6mhz than the 4mhz one in there now.

[speccyplus]

So I pulled IC3 and IC4, they are 74HC257s. Not 157s like the diagram you posted, I hate that all the pins are not clipped, but folded over, pain to solder suck, have to bend them and risk damage to the pads on the bottom. Second one took forever to pull so I would not damage the board.

I don't think the LS257 I had would work, since HC I think float high and and the LS use more power.

Anyways pulled them both, they tested OK. However pins are so mangled and would not fit in the sockets clean, ordered some exact replacements.

Gonna pull the other ones and test as well when I have time again.

Anyways same results putting back in so means my socket work is good!

Spent some more time scouring the board, I see nothing bridged on the bottom, except 3 pins that look like they are soldered together on the IC 14 chip, pins 13,14 and 15 I believe, on the bottom. This was the replaced chip by previous owner.

I don't want to pull the CPU socket, as I am worried the crappy previous soldering job may cause damage.

For giggles I ran without each of the 74HC257s to see the screen change, and yes, it looks similar in behavior but different.

[Firefox]

Spent some more time scouring the board, I see nothing bridged on the bottom, except 3 pins that look like they are soldered together on the IC 14 chip, pins 13,14 and 15 I believe, on the bottom. This was the replaced chip by previous owner.

Pins 14, 15 and 16 should be shorted together...

(edit) That looks correct in your photo of the PCB underside.

[speccyplus]

Spent some more time scouring the board, I see nothing bridged on the bottom, except 3 pins that look like they are soldered together on the IC 14 chip, pins 13,14 and 15 I believe, on the bottom. This was the replaced chip by previous owner.

Pins 14, 15 and 16 should be shorted together...

(edit) That looks correct in your photo of the PCB underside.

Thanks!

[speccyplus]

So I am taking a closer look at all the previous replaced parts. This makes me take a closer look at the transistors.

TR4 is the correct part, a 650
TR5 is a 213, on the 4b schematics, says it should be a 315, one is NPN and the other is PNP so the collector and emitter is swapped? Not sure on these parts. However my voltages look OK when I check on the ram. This was the one that was confusing me before doing testing and seemed backwards compared to others.

For TR2 I have a 548, not a 313. I can't see TR1 but looks physically the same and they look different than all the other transistors, packaging is larger and round. So I assume replaced at the same time.



I used this video to do my voltage tests: https://www.youtube.com/watch?v=IzgCmldm2H4

[1024MAK]

A 74LS257 can replace a 74LS157. See the datasheet. And in a circuit designed for a 74LS157, you can use a 74LS157 to replace a 74LS257.

74HCTxxx can replace 74LSxxx in most designs.

74HCxx should not be used in systems using TTL logic levels, as they have different input switching thresholds.

The original series were 7400 etc.
Then various types were produced, but 74LSxxx became the most common, as they were lower power than the 7400 series, but just as fast.
These have, for the most part, been superseded by low power CMOS versions. Hence your board having the 74HCTxxx which are CMOS versions that are compatible with TTL switching levels.

Having said that, my bad for not noticing that your board has 74HCT257 chips

Mark

[1024MAK]

I hate that all the pins are not clipped, but folded over, pain to solder suck, have to bend them and risk damage to the pads on the bottom. Second one took forever to pull so I would not damage the board.

I normally recommend that unless the chip being removed needs to be saved, that all the pins are cut as near to the top of the plastic package as possible. Then the pin can be moved with long nose pliers while keeping the solder molten with the iron. If the bottom is not bent over, you could gently pull it out. Obviously if the bottom is bent over, it’s not that simple

I don't think the LS257 I had would work, since HC I think float high and and the LS use more power.

7400 series and 74LSxxx series have inputs that tend to float high if left unconnected (but this is definitely not recommended practice).
74HCxxx and 74HCTxxx have to have their inputs tied to a defined logic level, otherwise these CMOS chips will react to any static charges, or nearby electrical potentials.

The ZX Spectrum was designed for 74LSxxx logic, so using a 74LS157 or 74LS257 is fine. Yes 74HCTxxx are lower power devices.

Spent some more time scouring the board, I see nothing bridged on the bottom, except 3 pins that look like they are soldered together on the IC 14 chip, pins 13,14 and 15 I believe, on the bottom.[/url]
That’s as per the design. IC14 is the video colour encoder.

Mark

[speccyplus]

Interesting, so I replaced all the logic chips with the LS257 and noticed the patterns on screen are different, don't know if that means anything.

Trying to save the chips, but this last 257 was a bugger, and almost cut it out. Keeping them for testing only.

I have removed all the 74HS257 chips, all tested OK. If I want to disable upper ram completely, do I need to remove the 00 and the one next to it as well?

This is the state of the board now:



Also any thoughts on the TR5 being NPN vs PNP? I know that video is an earlier issue, but my tests were the opposite for TR5 when measuring the collector and emitter.

TR5 is a 213, on the 4b schematics, says it should be a 315, one is NPN and the other is PNP so the collector and emitter is swapped?

[1024MAK]

So I am taking a closer look at all the previous replaced parts. This makes me take a closer look at the transistors.

TR4 is the correct part, a 650
TR5 is a 213, on the 4b schematics, says it should be a 315, one is NPN and the other is PNP so the collector and emitter is swapped? Not sure on these parts. However my voltages look OK when I check on the ram. This was the one that was confusing me before doing testing and seemed backwards compared to others.

A ZTX213 is the correct specified transistor for TR5, the ‘315’ is yet another error in the schematic diagrams... (yes, there are lots of errors).
TR4 can be a ZXT650, ZTX651 or a ZXT653.

For TR2 I have a 548, not a 313. I can't see TR1 but looks physically the same and they look different than all the other transistors, packaging is larger and round. So I assume replaced at the same time.

Most boards use ZTX313 transistors, but some use other types. TR1 and TR2 are the final stages in the analogue (composite) video stages. If your video signal is acceptable (disregarding the multicoloured mess caused by the fault on the processor side of the board), then they are fine. BC548C or BC549C are often used for replacements, as ZTX313 transistors are obsolete.

Sinclair used multiple manufacturers and they used multiple compatible parts in order to keep the price low. So it’s not unusual to find suitable alternative parts used, either when originally made, or as a result of having been serviced or repaired.

Mark

[1024MAK]

Ahh, a Mr Zigg video. They are good. Although I can’t get used to his voice.

Mark

[1024MAK]

If I want to disable upper ram completely, do I need to remove the 00 and the one next to it as well?

The 74 series logic chips used in positions IC23 and IC24 are normally very reliable (they are the simplest digital chips on the board). And IC24 is needed to control IC3 and IC4 for the lower RAM.

To fully eliminate the upper DRAM (the TMS4532) chips, the only sure fire way, is to remove them. If you remove only IC23 you will leave the DRAM control inputs (RAS and CAS floating, which would not be good. And if one of the TMS4532 chips is affecting the data bus, you are no better off than the method I outlined a few days ago in an earlier post.

Indeed, with IC25 and IC26 missing, the address inputs to the upper DRAM chips are already floating. If they are disabled as per my earlier post, this does not matter.

Mark

[speccyplus]

Ok, so other than for the fun of removing and testing chips, I will wait for my CPU to show up.

Will read up on doing some tests with the oscilloscope.

[speccyplus]

Ok, got some parts in over the weekend.

• 2 extra ram chips for lower ram.
  CPU
  Cap Kit

I recapped the whole board, installed the correct ram vs my hacked ram.

Same results.

Swapped CPU. Same results.

This at this point I assume it is upper RAM or the ULA maybe?

The one ram chip that started the whole investigation was pretty darn hot, it was the second from the bottom left. I know the RAM is all tied into the ULA.

The other option is to start cutting out the upper ram.

Or ROM issues?

Anyone have any more thoughts?

Here is the board now:

[speccyplus]

Ok, removed all the upper ram chips, last person to work on this used too hot an iron, lots of burned through holes and I have to deal with lifted traces Started some bodge wire work on the bottom row, top needs it too.

Scoured the area for any shorts I could find.

Anyways, after the board was liberated of all its upper ram, it looked like this:



And did the same thing as before.

So I removed the lower ram, this looks a bit better as I saw on other videos the way it should behave with no ram.



I will post a video shortly.

[speccyplus]

Here is what it should look like with no ram:

https://www.youtube.com/watch?v=izcGFap8HGI

And here is what I get:
[media]https://streamable.com/g7frud[/media]

It looks like something kinda works, but still hangs... Similar but not the same...

[speccyplus]

I just went and socket the rom, hoping my reader can read it. Does anyone know the specs for that chip?

I also suspect maybe a through hole was damage on the mobo when the CPU was replaced, may have to go through that and check things...

[1024MAK]

I just went and socket the rom, hoping my reader can read it. Does anyone know the specs for that chip?

The ROM is type 23128, which is similar to a 27128 EPROM but the ROM has two CS chip select / chip enable control signal input pins, unlike the EPROM which has only one. Both have an OE output enable. Or the ROM could have three CS chip select / chip enable control signal input pins. It depends on which ROM from which manufacturer was fitted.

On the ROM used in the ZX Spectrum all these control inputs are active low, but note that an option available at time of ordering was to have one or two active high control inputs.

The control inputs are on pins 20, 22 and 27.

Mark

[speccyplus]

Thanks, I ordered some blank chips.

[speccyplus]

So, a funny thing happened on the way to the forum.

I want to slap myself if this is true.

Playing with the rom chip a 23128 GI, I was able to apply power to pin 27 or 20 or 22 and always pin 1 and was only able to do a read when I selected a NEC 27128 chip on my reader, not Hitachi.

My spectrum board is set for Hitachi.

The chip read as blank (All FFs) If this was working correct, then I have a blank ROM chip!

Looking at my board again, I can say yeah, it looks like the jumpers have been changed, and as jumpers were not flush and solder blobs were big as in joint was not heated.

I googled pics on the internet, and every GI ROM Chip had jumpers set to NEC and nice clean installs.

Doh. So will see when I get my blank 27128s and program those if this works correct.

are the GI chips PROMs and not ROMs?

Thinking about the desoldering, all pins were straight and not cut, so this seems to me that a BLANK PROM was used as a replacement, didnt work and they changed the jumpers.

Doh!

Not sure if true, but if it is, this is why I can hate other peoples failed projects!

[1024MAK]

The ROM chips used by Sinclair (type 23128) are mask ROM chips. These are not programmed as such. Instead they are manufactured with the required data embedded in them. Hence the reason they are normally considered to be fairly reliable. If you are making a lot of machines in a production run, this is by far the cheapest type. Hence nearly all 16K/48K ZX Spectrums made used these.

GI ROMs are mask ROMs.

EPROM (and EEPROM/flash) chips internally are analogue, the stored value is a charge in the capacitance of the gate of a transistor. Hence over a long period of time, they may suffer from “bit rot” as some cells forget their stored value. A blank device reads each cell bit as a logic “1”. Note that windowless EPROM chips are also available, these are described as one time programmable (OTP), which means they cannot be erased.

A PROM is different again, generally these are programmed by destroying a “fuse” for each cell that you wish to change. These were never very common, as EPROMs were more widely available.

To use a EPROM, follow the instructions here.

My experience is that if a board like a ZX Spectrum board has been worked on (something removed by (de)soldering) by a non-professional, there is a good chance that there is damage to some PCB tracks or the through hole plating. You only need one fractured track or one plated through hole to be open circuit to wreck the entire operation of the machine.

Mark

[speccyplus]

Thanks again Mark, that is what I thought, and entertained the possibility I am reading it wrong.

I am not a professional, but I know enough if I screwed up a through hole, lifted a trace, etc. I check everything after I pull a chip. But yeah will do more passes again looking around where I suspect work has been done.

I have already checked continuity for all through holes with traces on both sides on the CPU sockets. The Ram before I socket, same deal. Same with the ROM I just pulled.

I scoured the net a little last night to find a pinout for the following chips since they are all the same:

NEC D231128C
GI 9128C
Skk SBB23128

I could only find the GI.




With pins 20, 22 and 27 do you think it is worth tying them all to +5v at once, to see if I can read anything other than FF FF ...? I have only tried one at a time, and no other combinations.

I want to find out if this rom is bad.

[1024MAK]

The control input pins 20, 22 and 27 on the ROMs used in ZX Spectrums are all active low. This means that they all have to be at a logic low level of between 0.8V and 0V in order for the ROM to output data.

This Commodore ROM is similar.
I don’t have links for NEC, Hitachi or GI ROMs.

Mark

[Firefox]

I could only find the GI.

If that pinout is accurate then the ROM links are correct in the "H" configuration, as they are in your latest photo.

[speccyplus]

Was finally able to dump my rom. Mine is the zx16k.rom file dumped as a 256bit chip, I then had to split the file to get the upper and lower halves. I guess a trick to get a burner to toggle pin 27 hi and low as a memory select setting and thinking it is reading 2 16k banks...

me@home:~/tmp$ minipro -p AM27256@DIP28 -r zx.rom -y

me@home:~/tmp$ spilt -b 16k zx.rom

me@home:~/tmp$ mv xa zx16k.rom


me@home:~/tmp$ md5 48.rom
MD5 (48.rom) = 4c42a2f075212361c3117015b107ff68

me@home:~/tmp$ md5 zx16k.rom
MD5 (zx16k.rom) = 4c42a2f075212361c3117015b107ff68
me@home:~/tmp$

I have a TL866+ II EEPROM burner BTW, and the GUI is nice, but the open source CLI too lets you bypass all the silly error checks, enabling me to do the dump above!

[speccyplus]

I went and pulled the last 2 74ls chips, tested OK as suspected. I now have a 16k spectrum in theory

I scoured the board, and found 1 broken trace from the previous work someone did on the board, it was pin 3, a chroma signal pin with a through hole on the video signal chip. I bodged it below to the resistor it should connect too, didn't expect and change any and there was none.

At this point only chips left are the ULA...

I have some EEPROMS coming still, maybe burn a zero memory usage test rom? Any suggestions on the best one?