Sinclair ZX80 / ZX81 / Z88 Forums

Some time ago I bought a non-working issue 4B Spectrum that looked in good condition because I didn't have an issue 4B motherboard and it was low priced because of a lot of desoldering work on it and missing keyboard connector and speaker amongst other things.
I recently got round to doing some work on it and I've noticed some cut tracks on the PCB connecting pin 8 of the upper ram chips to (I presume) the -5V rail and I'm wondering if this is for real.

The upper ram chips are TI TMS4532-20NL3 and I've desoldered and tested them and they all appear to be ok, three of them had the full 64K working.

The upper and lower ram multiplexer chips aren't the usual SN74LS157N, they're PC74HCT257P and they appear to be original.

I've looked at the circuit board schematic url;

https://spectrumforeveryone.com/wp-cont ... matics.gif

and there's no mention of any track cutting so I was wondering if any forum members know of or have seen a mod like this.
I don't want to fry any chips so I thought I'd ask.
Only IC22 appears to have a connection from pin 8 to the (I believe) -5V rail.

First of all, there is no -5V on the UPPER RAMS. Pin 8 is connected to +5V (some schematics available in the WWW are really bad to read or faulty)! I think someone has cut the lines to isolate a chip one by one to find a faulty one or added an expansion board with extension RAM and "switched off" the upper RAMs therefore ...
IMHO the 74HCT chips where produced later then 1983 and so I think they are replacement parts and not original on the 4B board.

Tiger wrote: ↑Tue Jul 12, 2022 10:01 pm First of all, there is no -5V on the UPPER RAMS. Pin 8 is connected to +5V (some schematics available in the WWW are really bad to read or faulty)! I think someone has cut the lines to isolate a chip one by one to find a faulty one or added an expansion board with extension RAM and "switched off" the upper RAMs therefore ...
IMHO the 74HCT chips where produced later then 1983 and so I think they are replacement parts and not original on the 4B board.

Hi Tiger

Thanks for your explanation, that seems right.

I searched 'Sinclair Spectrum issue 4B motherboard' and looked at the images, several of the motherboards had the same combination of chips so somehow they may be originals.

I think I'll solder up the tracks and continue with the repair!

Yes, pin 8 is VDD, which on these “+5V only” DRAM chips is the +5V supply. They don’t need a -5V supply.

As Tiger says, the only possible reason to cut the track to pin 8 is to power them down, either to isolate a faulty chip or to use external memory. It’s a bizarre way of doing though. Cutting the leg of the chip(s) is far, far easier.

74HCT257 are more modern equivalents to the discontinued 74LS157.

Mark

1024MAK wrote: ↑Tue Jul 12, 2022 10:54 pm
As Tiger says, the only possible reason to cut the track to pin 8 is to power them down, either to isolate a faulty chip or to use external memory. It’s a bizarre way of doing though. Cutting the leg of the chip(s) is far, far easier.

You can't re-use chips with no legs, is the only problem with that.
Using a solder sucker and then working the legs loose first with an iron and then a small screwdriver or similar means you can do this with the chips!

sP1d3r wrote: ↑Thu Jul 14, 2022 9:26 am

1024MAK wrote: ↑Tue Jul 12, 2022 10:54 pm Cutting the leg of the chip(s) is far, far easier.

You can't re-use chips with no legs, is the only problem with that.

I meant cutting just pin 8 on each DRAM chip (halfway between where it comes out the chip and the board) for fault finding purposes is less destructive compared to cutting the PCB tracks. If only pin 8 on each DRAM chip was cut, if the relevant chip(s) were not found to be faulty, you can then bridge the cut in the leg with solder. Hence restoring it to operation.

Yes, it’s not going to be pretty, but then, patching up a PCB is not pretty either.

Obviously, if you are good at desoldering chips from a PCB, then that’s even better. But not everyone is good at desoldering chips from plated through double sided PCBs without causing damage.

Mark

1024MAK wrote: ↑Thu Jul 14, 2022 3:05 pm Obviously, if you are good at desoldering chips from a PCB, then that’s even better. But not everyone is good at desoldering chips from plated through double sided PCBs without causing damage.

The ideal solution would be to desolder and socket the chip, then lifting the legs on the chip instead of cutting them.

I've tried a lot of desoldering with varying degrees of success, it's reliant upon the correct iron temperature so that you don't lift tracks and hole plating, a good solder sucker and correct manipulation to get a good draw when the plunger springs and then working all the chip legs loose prior to carefully lifting the chip, testing to see if it's stuck.

It's worthwhile because you can save chips and check the pcb tracks but it does take a lot of practice.

sP1d3r wrote: ↑Thu Jul 14, 2022 3:47 pm I've tried a lot of desoldering with varying degrees of success, it's reliant upon the correct iron temperature so that you don't lift tracks and hole plating, a good solder sucker and correct manipulation to get a good draw when the plunger springs and then working all the chip legs loose prior to carefully lifting the chip, testing to see if it's stuck.

It's worthwhile because you can save chips and check the pcb tracks but it does take a lot of practice.

Yes, I’ve done the same.

Mark

I've now repaired this motherboard and it boots up and works but the speaker doesn't make any sound at all, neither with key clicks, Diagrom or games.
I've tried a known good speaker which also doesn't work and I also tested TR7, a ZTX450, which tested good on my LCR tester which has one scrambled digit from testing an undischarged capacitor after assembling the kit, it's always worked perfectly despite this.

I've recently found a website contain a lot of useful info and files about the firmware for my LCR tester, a Hyland M12864.
It currently has the original firmware, version 1.12k which I'm going to try to reinstall although there are newer firmwares for this type of tester, it's also for GM328 and T4 testers, which have similar hardware.
The firmware needs to be configured for the particular tester, I'm also getting a newer GM328 tester which like my M12864 has an Atmega 328P, which with min. 32K can run the ST7565 screen and also the 160 x 128 display on the GM328.
This website has links to the source firmware and also more recent firmwares and details of how to mod standard testers with 0.1% tolerance measurement grade resistors and a high tolerance regulator and 16MHz crystal to significantly increase the capability of the tester, which can test many different items though not in-circuit and also provide a PWM signal.
The resistors can be obtained in both through-hole and SMD, the SMD type seem to be more common and cheaper but only 3 x 680R and 3 x 470K are required.
I may get a through-hole and an SMD version and compare.
Care must be taken to fully discharge capacitors before testing as there is little protection for the Atmega 328, hopefully the one in my M12864 will be OK once new firmware has been burned.
It now appears that using the known good speaker blew it because it doesn't work back in the motherboard it came from

I hope that it wasn't my soldering that blew the speakers, it might have been because if the speaker terminal gets too hot, the coil wire might melt the plastic parts in the speaker!
I'll check the output to see if it's faulty.

Issues with no sound are often TR7 failed, D9 or D13 failed, or the speaker.
The DC resistance of the speaker (as measured using a multimeter) should be less than 45 ohms.

On the low resistance range (200) or the continuity/diode test range, the speaker should faintly click as you connect and disconnect one of the test leads. Obviously do this out of circuit or with the board disconnected from power.

Does/did the machine produce “loading” sounds when (if) you tried loading from tape?

If you get “loading” sounds, but no beeper sound and D9, D13, TR7 and the speaker are good, then it could be the ULA, but the ULA is rarely the cause.

By the way, the collector of TR7 connects to the +9V (nominal) rail, NOT the +5V as shown on the schematic.

Mark