An Enhanced MK14 (MK14E)

[circuitryboy]

I believe one problem with the MK14 has always been the PCB size/shape.
Is this not an opportunity to adopt a 'standard' board size, with separate keypad?
And for 64K/multiboard I recommend buffering address/data buses. Latching the SC/MP status flags can be useful here.

[ChrisOddy]

There are actually only 2 pullups on A10 (one is on NA10) which is why I increased the pullups to 10K though I take your point about all the LS inputs loading it as well.

I'm beginning to think there are actually two different requirements here - for a slightly improved MK14 and then an all singing all dancing SC/MP computer ?
I could leave the MK14E pretty much as is but then develop a second board intended for NIBL rather than SCIOS ? (no display/keyboard but with the same backplane for some compatibility with e.g. VDU)

A question for those more knowledgeable on SCIOS - I thought it wouldn't work with the full 16-bit address bus decoded i.e. with LK5 in the '0' position and that it would have to be disconnected i.e. 'X', but it seems to work fine ? Is that link position therefore unnecessary ? (probably don't need LK5 at all if this remains a MK14)

[Phil__G]

Originally Posted by ChrisOddy

I thought it wouldn't work with the full 16-bit address bus decoded

I was about to say, dont forget it relies on reflections for storage from page zero, eg a negative offset from a low ROM address puts you in RAM, this is extensively used for the stack & variables...

Re the 2nd option, its sounds like a great opportunity to run Karen's 'PAGE2' system!

[SiriusHardware]

I have another suggestion, slightly more mundane than the others - the MK14 PCB has never had any provision for the mainboard to be mounted on a backplate or in an enclosure, the only holes ever drilled through the PCB were those provided for the mounting of the keypad.

While those existing holes can also be used with standoffs below the PCB, the PCB could do with another two mounting holes further back. I realise this doesn't sit well with an already closely packed design, but if it can be done...

[SiriusHardware]

It would mean more work for Chris but I would prefer to see the proposed revision remain recognisably an MK14 with the sensible changes that have been made, namely the move to more affordable EPROM and RAM, the original PROM images removed and filled in with RAM, the awkward dangly single-step hardware moved onto the PCB where SOC should have put it in the first place, the system buses and control lines tracked to the rear edge. Those improvements, retaining the original form factor, are enough for me.

It seems to me that if you want anything more advanced it would be better to design a new SC/MP computer from the ground up to cater for as many of people's special needs as may be possible.

[ChrisOddy]

Phil__G - so I presume the LK5 'X' and '0' positions are both useful ?

SiriusHardware - I'm sure I can find a space for some mounting holes.

Can anybody advise on the polarity of the Rx and Tx lines when used with NIBL, do I need to invert them for connection to e.g. an TTL FTDI serial cable ? I've looked at the Elektor BASIC computer and original Introkit but its not easy to work out as they are RS232 and 20mA ?

[Mark1960]

Originally Posted by ChrisOddy

There are actually only 2 pullups on A10 (one is on NA10) which is why I increased the pullups to 10K though I take your point about all the LS inputs loading it as well.

I missed that one is on NA10.

While checking that I realised you are clocking the high address using 74ls175 with NADS inverted, which would be the leading edge of NADS. With an edge triggered register its better to clock this on the trailing edge.

I think the inverter on NADS is not needed.

[circuitryboy]

Originally Posted by ChrisOddy

A question for those more knowledgeable on SCIOS - I thought it wouldn't work with the full 16-bit address bus decoded

Internally the SC/MP doesn't access those latched top Address lines. The p-counter goes up/down a single 4K 'page' like a hamster in a wheel.

[Realtime]

Originally Posted by Phil__G

Question for Realtime, DavidMS and Coolsnaz, do you think an RP2040 emulation of the 8154 is feasible?

From a purely HW pov you would need 38 I/O pins on an RP2040 module.

The RPi PICO module is a 40 pin dip but only 30 gpio are available (which is the max gpio from the RP2040 chip). Even if there was enough I/O the other problem is that the RP2040 is 3v3 part, so level translators would be needed to interface to the MK14 5V TTL bus.

Even so, it would be an interesting project to try a Port-A only emulation using a PICO or similar. I suppose you could include a port expander chip to be able to provide A and B ports.

[ChrisOddy]

Originally Posted by Mark1960

While checking that I realised you are clocking the high address using 74ls175 with NADS inverted, which would be the leading edge of NADS. With an edge triggered register its better to clock this on the trailing edge.

I think the inverter on NADS is not needed.

Ah yes I've just checked this myself and you're correct its not needed for the 175 however I still need an inverted NADS for the Single Step circuit (taken from the original MK14 manual).

[Mark1960]

Originally Posted by ChrisOddy

Ah yes I've just checked this myself and you're correct its not needed for the 175 however I still need an inverted NADS for the Single Step circuit (taken from the original MK14 manual).

I’m not so sure about the need for the inverter in the single step circuit, does it really matter whether it counts leading or trailing edges, just that it counts the number of NADS.

Edit: OK I see it now, uses inverted NADS and D7 to reset the counter.

[Mark1960]

Its probably still better not to invert NADS for clocking the upper four bits of the address. If you use the trailing edge it allows overclocking the 8060 to 8MHz, but using the leading edge doesn’t.

[Mark1960]

I think the flag output needs to be inverted for serial output but not the sense input. I seem to remember this was the same for the 8072.
https://www.dos4ever.com/SCMP/SCMP.html#introduction

[circuitryboy]

Originally Posted by Mark1960

If you use the trailing edge it allows overclocking the 8060 to 8MHz, but using the leading edge doesn’t.

In my experience NADS at the '8060 is a seriously degraded pulse and needs a buffer.

[Mark1960]

I think NADS can be improved with an external pull up. 74ls ttl inputs seem to pull up enough to make a difference, but the rise time slows down after it rises above the logic high threshold. I have noticed with 74HCT inputs that the rising edge timing on the output can vary between different 8060s.

I was wondering if maybe the 8060 doesn’t have depletion mode pull ups on the outputs, but maybe there is just something different about the NADS output driver.

[Realtime]

When developing RealView it took me a while to spot that NADS has a very slow rising edge, but interestingly it's even worse just after NENIN has been asserted. From what I recall NADS would reach about 50%, stall and then continue rising at a slower rate. I ended up using a 1K pull up to ensure the edge was good enough for direct interface to a XILINX CPLD (5V tolerant but not slow-edge tolerant, it turned out).

[ChrisOddy]

Originally Posted by Mark1960

I think the flag output needs to be inverted for serial output but not the sense input. I seem to remember this was the same for the 8072.
https://www.dos4ever.com/SCMP/SCMP.html#introduction

I've had another look at the Elektor BASIC Microcomputer circuit and that ties up with what you suggest, it uses RS232 and the Tx output is inverted. I still struggle with the Introkit as it uses 20mA which I thought was no current for a high which would be the wrong way around ?

I have more confidence in the Elektor BASIC and dos4ever computer so will go with Tx inverted !

Chris

PS revised schematic and layout almost ready for final (?) inspection.

[TonyDuell]

The normal current loop interface has the current flowing in the 'mark' state. This corresponds to a -ve voltage on the RS232 TxD line and a high level -- logic 1 -- at the TTL level output of most, if not all, UART etc chips.

[ChrisOddy]

Thanks for that Tony, in that case it ties up with the Introkit circuit as well. Its been over 40 years since I last went near a 20mA current loop !

[Silvester]

On my Elektor SC/MPuter clone I just used 74LS14 inverter from FLAG0 to MC1488 RS232 driver and MC1489 RS232 receiver directly connected to SENSE-B input.