Some 48K Spectrums suffer from vertical ‘banding’ on plain, light colours that persists even after the capacitors have been changed. These are colloquially known as ‘jailbars’ and can be a quite annoying visual artifact.
The exact reason that some machines suffer more than others from this effect is not currently known, but it is suspected that it is caused by issues with the ground plane on the Spectrum PCB (see here for a slightly expanded explanation).
This effect can be somewhat alleviated by replacing capacitors C5 to C8 (4 ceramic caps on the south edge of the PCB, below the lower RAM) with ones with a value between 150-220nF. These capacitors filter the +12v voltage supply rail to the lower RAM IC’s, and help reduce noise on this rail.
Shown below is the screen display of the same machine after this modification. The jailbar effect is not completely eliminated but is significantly reduced.
A similar effect can be seen on 128K ‘Toastrack’ models when using RGB video output, that doesn’t otherwise manifest itself over RF or composite output:
The fix in this case is as follows:
- Replace C28 with a 47uF axial capacitor (original is 22uF, right of the 5 way keyboard connector)
- Replace caps C7 and C8 with 1uF ceramic capacitors (next to IC12 and IC13 RAM chips, bottom left of the board)
Again, shown below is the screen display of the same machine after modification, with significant reduction of the jailbar effect.
On a Grey +2 (Z70500 Issue 3 PCB), the capacitor to change is C52 (just to the right of the ULA), and a radial capacitor with a value of between 47uF – 100uF can be used for replacement (thanks to Stuart Brand for proving this modification).
Stuart has also been able to improve the image quality further by adding 1nF ceramic capacitors across the footprints to C72-C79 underneath the PCB (the originals could also be removed and the 1nF parts used as replacements). His video detailing the modifications in full to a Grey +2 can be found here.
- The original 47nF capacitors had negligible ESR and a vloss of 0.5%. The new 220nF Vishays had an ESR of 2.6Ω and vloss of 0.2% (at least, according to my cheapy component tester).
- As before, changing to the 220nFs reduced average peak-to-peak noise by around a half.
- The picture output from the Speccy was visibly improved.
- See pics for spectrum analysis before and after (average). Shows a significant reduction in magnitude plus, interestingly, apparently most of the higher frequency harmonics have completely gone.”