The Spectranet Ethernet interface created by Dylan Smith is in theory compatible with the SPECTRA interface but in practice Spectranet PCB versions up to at least issue 2 are not directly compatible.
The reason for the incompatibility appears to be coupling of the Spectranet's network frequencies on to the /CLK line. A work-around exists for SPECTRA but not SPECTRA+128.
Spectranet can be made compatible with SPECTRA as follows:
- Solder a wire on Spectranet to connect the /Y video signal from its edge connector to its rear expansion bus.
- Solder a wire on Spectranet to connect the +12V line from its edge connector to its rear expansion bus.
- Cut the /CLK signal at the edge connector of the SPECTRA interface.
- Solder the /CLK line on SPECTRA to the 0V line.
- Connect Spectranet in front of SPECTRA.
With the SPECTRA interface connected behind Spectranet, it is not possible to fully use the onboard ROM facilities of the SPECTRA interface (the Spectranet interface's ROM will always take priority). Attempting to connect Spectranet behind SPECTRA introduces a completely different set of problems, and a work-around for this configuration does not yet exist.
The modifications above do not work with SPECTRA+128 since it uses the /CLK line. When SPECTRA+128 is connected behind Spectranet, a stable television picture is produced but network access fails. Introducing a ferrite bead on the /CLK line between Spectranet and SPECTRA can result in the network working but introduces instability of the television picture during network access.
The divIDE hard disk / SD card interface created by Pavel Cimbal is in theory compatible with the SPECTRA interface. It should be connected behind the SPECTRA interface. However, intermittent (and sometimes continual) reboots of the divIDE has been known to occur on some Spectrums. Investigation revealed two causes:
- The divIDE Plus pulls the /RST line down to 3V and introduces electrical noise on the line.
- The Spectrum ROM did not page back in quickly enough after the divIDE Plus relinquishes controlled via the /ROMCS line.
The /RST line issue can be overcome by replacing a 1k resistor with 4k7 and by fitting a 0.1µF capacitor, as shown below:
The /ROMCS issue can be overcome by replacing resistor R14 on SPECTRA to 2k2. Resistor R12 performs a similar function but for the ROM cartridge socket and so it is recommended to change this to 2k2 also.
Note that the divIDE interface cannot be connected in front of the SPECTRA interface since the divIDE contains onboard ROM and ensures it has exclusive access to the ROM address region (0-16K) by not passing the /MREQ line through to its rear expansion bus. The SPECTRA interface needs the /MREQ line so that it can monitor writes to the Spectrum's video RAM. Without it the SPECTRA interface is unable to reproduce the Spectrum screen, and will typically generate a display of randomly coloured attributes. A workaround solution would be to solder a wire from the /MREQ line at the edge connector of the divIDE interface to the corresponding contact on its rear expansion bus. It is still recommended to make the /RST modification described above.
The divIDE interface should really monitor the /ROMCS line on its rear expansion bus and should another device raise it then the divIDE interface should disable its onboard ROM. This priority mechanism was introduced on the ZX Interface 1 so that it would coexist with the ZX Interface 2 ROM cartridge mechanism. A consequence of the divIDE interface not monitoring the /ROMCS line on its rear expansion bus is that it is not possible to use the onboard ROM facilities of the SPECTRA interface.
The RAM Turbo joystick interface produced by RAM Electronics Ltd supports the Kempston joystick format but does not decode using I/O port base address $1F as would be expected, i.e. address line A5-A7 all at logic '0', but only checks for address line A5 to be at logic '0'. This causes a conflict with display mode I/O port $7FDF used by the SPECTRA interface since this happens to have address line A5 at logic '0'.
The SPECTRA interface accommodates this type of incompatibility by not passing I/O requests for port $7FDF through to its rear expansion bus. Therefore, connecting the RAM Turbo interface behind the SPECTRA interface will allow it to be read using I/O port $1F.
The ROM Turbo interface has been known to cause picture break-up with used with a SPECTRA+128, which appears to be due to the additional loading on the /CLK line. It is therefore deemed incompatible with SPECTRA+128.
The Kempston Mouse interface produced by Kempston Micro Electronics Ltd operates using I/O port base address $DF, i.e. it ignores the state of the upper address lines. This conflicts with the I/O port address of $7FDF used to operate the new display modes provided by the SPECTRA interface.
Connecting the Kempston Mouse interface behind the SPECTRA interface provides partial compatility due to the filtering of I/O port $7FDF accesses from the rear expansion bus of the SPECTRA interface. A conflict would not occur so long as the driver for the Kempston Mouse does not place $7F on the high address lines when it accesses I/O port $DF. There is no guarantee that existing Kempston Mouse driver software will avoid placing this value on the high address lines, but all newly written software could be deliberately prevent it.
Should a conflict occur when using existing Kempston Mouse software, then a workaround solution is to simply disable the new display modes using configuration switch 6 on the SPECTRA interface. Existing software will not know how to control the new display modes and so having them disabled will not present a problem.
The Cheetah SpecDrum produced by Cheetah Marketing Ltd operates using I/O port base address $DF, i.e. it ignores the state of the upper address lines. This conflicts with the I/O port address of $7FDF used to operate the new display modes provided by the SPECTRA interface.
For the same reason as described for the Kempston Mouse, connecting the Cheetah SpecDrum interface behind the SPECTRA interface provides partial compatility. A conflict will only occur if the SpecDrum driver happens to place $7F on the high address lines when it accesses I/O port $DF, and so new driver software can be written to explicitly prevent such an occurrence.
As for the Kempston Mouse, a workaround solution to prevent conflicts occurring when using existing software is to simply disable the new display modes using configuration switch 6 on the SPECTRA interface.
Several models of TVs by Panasonic do not appear to implement the SCART specification correctly and will either shift the picture to the left or cause the picture to continually jump left/right. It appears to be due to the TV expecting the video synchronisation signal to contain a colour burst signal, but this should only be required for a RF modulated signal.