The Sinclair QL Forum

One of my QLs has a mod on ULA - does anyone knows what mod is this ?
vDriveQL works on this QL (kinda works - works on QL+Tetroid GC but not on QL+original SGC), but it does not work on my other 3 QLs (one of these has the diode mods on microdrives, 2 do not have it)

IIRC this was one of Sinclair's mandatory modifications to bring the QLs up to issue 6 (?) standard but that's about all I can remember - Dave will probably know more

There's some information on modifications in the QL Service Manual - http://www.sinclairql.net/srv/index.html

Search for the word "modification", or look for resistors R102/103 (33kilohm).

The two 33K resistors appear to work as a mild pull-down on the two data lines for the Microdrive bus. The presence or absence of this mod does not appear to affect vDriveQl performance and what the mod is trying to fix is a mystery.
I've been working with Andrew on a fix for the 'vDrive not responding issue' for a few months but due to not being able to repeat the symptoms (I have most of the same kit - GC/Tetroid GC/QubIDE etc. - but all combinations work fine) makes it difficult to come up with a solution. There was an issue with compatibility with Tetroid's SGC (I don't have one) but this has been fixed. There was a report of the PRINT command not working after loading the toolkit with a GC attached - I have not been able to reproduce this error (can someone confirm this?). There are some [undocumented] mods on the Microdrives which affect the rise time of the Microdrive R/W signal from startup which definitely affect vDriveQL performance but I think we've ruled this out in this case.

I continue to work with Andrew on this to get it sorted (and I'll no doubt learn some more QL quirks - my list is quite long already

As an aside: I must admit that the popularity of the vDriveQL has pleasantly surprised me. It was never designed to be a 'modern' storage solution for the QL (and comparisons with other solutions I think miss the point) but simply a way of providing a real-time emulation of what using Microdrives would have been like, as well as helping to preserve the physical cartridges and drives. It was also intended to be used to provide access to all the archived Microdrive content that may (or may not) become available.

The 2 resistors connect to -12v supply and each resistor is in turn connected to the raw data inputs to the ULA from the Microdrive.

Some kind of negative bias to better address the input threshold switching maybe ?

[quote="Bloodnok"]
I've been working with Andrew on a fix for the 'vDrive not responding issue' for a few months but due to not being able to repeat the symptoms (I have most of the same kit - GC/Tetroid GC/QubIDE etc. - but all combinations work fine) makes it difficult to come up with a solution. There was an issue with compatibility with Tetroid's SGC (I don't have one) but this has been fixed. There was a report of the PRINT command not working after loading the toolkit with a GC attached - I have not been able to reproduce this error (can someone confirm this?). There are some [undocumented] mods on the Microdrives which affect the rise time of the Microdrive R/W signal from startup which definitely affect vDriveQL performance but I think we've ruled this out in this case.

As soon as I get my Drive we will see if I have the same problem. My QL has the same 2 resistor mod on top of the ULA and I have a tertroid SGC coming. I will let you know the results once I get my hands on the goodies

Pr0f wrote:The 2 resistors connect to -12v supply and each resistor is in turn connected to the raw data inputs to the ULA from the Microdrive.

Some kind of negative bias to better address the input threshold switching maybe ?

Some time ago I've been told the RAW lines mean Read And Write. When the MDVs are not in use, the lines are 'floating' which should not be the case. However, there also seems to be the issue of some current leakage created by the MDV ULAs, that can make these lines 'float' to an undefined level, which, in turn, can cause the inputs on the 8302 ULA to 'latch up' and stop functioning normally. To prevent this, a small negative bias is applied to cancel out the small positive leakage currents, using resistors from the -12V power supply, which is avalable on QL motherboards.
Originally one place where this fix was applied in a slightly different manner is the QIMI mouse interface. On it, there is a -5V generator, which is also used to apply a small negative bias to the RAW pins, using smaller value resistors (due to the smaller negative bias voltage). The QIMI PCB plugs into the 8302 socket and the 8302 goes 'piggyback' onto it, which was a nice opportunity to apply the fix without having to solder onto the ULA or introduce 'flying' components.
The fix was also incorporated into the Aurora circuits, as it also uses the original 8302 ULA, even though no microdrives are supported - just in case.

Thanks Nasta, mystery solved (I've always wondered why RAW was used on the QL and not just DATA as in the Spectrum). It is a mandatory mod then - quite a few QL's I come across in supporting the vDriveQL don't have the mod so I'll be recommending it be added.

Bit of a coincicdence

I recently resurrected an old QL motherboard which I had removed the 33k resistors from the ULA (not required when I used QIMI). With no microdrives attached (and no QIMI) I sometimes had trouble getting the cursor to appear after pressing F1/F2. But then I noticed if I waved my hand over the QL the cursor would then appear.

Problem was without the 33k resistors pulling down RAW1/2 the floating pins read it as a phantom microdrive.

Just soldered some resistors on the underside of PCB, but it was a bit of fun for a while waving my hand over the QL "These aren't the microdrives [droids] you're looking for".

Silvester wrote:"These aren't the microdrives [droids] you're looking for".

These are not the signals you are looking for
Quite literally, I think the problem is the MDV circuitry generates spurious interrupts.
To be honest, I'd be tempted to keep the machine just to show other people it had gesture recognition 35+ years ago )