Nemesis accelerator wiring/diagram

[dml]

This might seem ironic but does anyone have a drawing of the F030 Nemesis accelerator board with input/output, signal labels and wire colours assigned to each of the pads? i.e. something that should really have been in the manual.

I've been slowly reworking some Falcon030 boards and bringing them back to life - one had a Nemesis board fitted (by me, long ago and with the wiring not matching the manual directions whatsoever!) and a 2nd Nemesis board is just loose, ready to be fitted. If there is a resource handy which describes the pad functions I'll could use that for reference. If not, I'll make it up as I go along and post the result here later.

IIRC the Nemesis is a glorified clock buffer with with 3 selectable clock sources separated roughly into two separate channels ('main' and 'videl') each with it's own switching mode, and with multiple (identical) buffered outputs on each channel. This is probably why I got away with the wiring as it is - i.e. not strictly following the instructions. And the reason for all of the different wiring permutations offered in the manual is more to do with minimizing intrusive changes vs getting a better result - e.g. removing resistors or cutting tracks to isolate bits of the board and applying signals closer to the intended target

[alexh]

It's not what you wanted but there are several pictures in this thread over at Atari Age of a Nemesis in place which allow you to work out which pads have which colour wire.

http://www.atariage.com/forums/topic/12 ... ir-falcon/

http://www.yardley.cc/atari/Nemesis%20i ... manual.pdf

Nemesis_2___.jpg

Falcon_Repair_013_sml.jpg

[dml]

It's not what you wanted but there are several pictures in this thread over at Atari Age of a Nemesis in place which allow you to work out which pads have which colour wire.

Thanks - I haven't seen a pic yet where the wires actually match the original colours but between the two boards I have and the pics posted here things are making more sense

[Guest]

http://www.atari-forum.com/viewtopic.ph ... celorators
i have the manual its also online ill photocopy it and send it over,., let me know {there is a nice passage at the end that gives info on how to use 55mhz } ill also send over a TTL 55Mhz TTL usual square guy...

nemesis install look on google someone scanned it a while back dont know the person but i have this archive somewhere on the falcon cf ill look it out
ill help with some pics if needed
most info is in the instal etc guide
ill help as i can doug...
its so RF non friendly and nemesis never matured
to use just a single core wire to carry VHF is just daft..
as you get 'chipset bleed over' as a result of no screening

same goes for other falcon mods 'sealed' in wax or potting compound


just daft no wonder you want to rewire it
since i reworked it my chipset is very stable even running at 55mhz clock and 55/2 bus
i derive the 32mhz clock from the dsp TTL clock to reuse it
the dot clock is not much use if you use RGB svga monitor etc

dont see the point in having a dot clock {as per American falcons they just use a single 32mhz clock from a TTL xtal}


ps this is not the correct place to post falcon or TT stuff
drives me mad

i used to have to move about 5 topics a week over to the right other place....

[dml]

http://www.atari-forum.com/viewtopic.ph ... celorators
i have the manual its also online ill photocopy it and send it over,., let me know {there is a nice passage at the end that gives info on how to use 55mhz } ill also send over a TTL 55Mhz TTL usual square guy...
nemesis install look on google someone scanned it a while back dont know the person but i have this archive somewhere on the

Thanks - I found the scanned nemesis manual at Simon Yardley's site (I have an old hardcopy but its full of my illegible scribbles). I also have a 55MHz oscillator can which I'll try on the other (non-Afterburner) board later.

its so RF non friendly and nemesis never matured
to use just a single core wire to carry VHF is just daft..
as you get 'chipset bleed over' as a result of no screening
same goes for other falcon mods 'sealed' in wax or potting compound
just daft no wonder you want to rewire it

I ripped out the old wiring and used your RF coax as instructed and it is more stable now . Especially after reworking the PSU (that was another problem!)

ps this is not the correct place to post falcon or TT stuff
drives me mad
i used to have to move about 5 topics a week over to the right other place....

Ok, noted. It looked like a general hardware discussion area. :-z

[Dal]

Topic moved to Falcon area

[viking272]

Thanks - I found the scanned nemesis manual at Simon Yardley's site (I have an old hardcopy but its full of my illegible scribbles). I also have a 55MHz oscillator can which I'll try on the other (non-Afterburner) board later.

Which reminds me that I still haven't attempted to remove my faulty Nemesis and install my CT63. It's been in bubble wrap for 3 years, ahem.
However I have spent 6 hours tidying the office over Xmas and located all my Atari kit now so if I can find a decent case for the Falcy I'll re-house and rebuild it all.

Cheers

[exxos]

probably a bit late, though I scanned the manual in 2006, drivers are there too.

exxos.www.idnet.com/IMPULSE/atari/temp/nemesis.zip

I did scan the book again a few months back, though not sure where I put all the book scans I did right now

[SoundDoctor]

Which reminds me that I still haven't attempted to remove my faulty Nemesis and install my CT63. It's been in bubble wrap for 3 years, ahem.
However I have spent 6 hours tidying the office over Xmas and located all my Atari kit now so if I can find a decent case for the Falcy I'll re-house and rebuild it all.

Cheers

Faulty Nemesis? Any idea what's wrong with it? This is a bit of a cheek but...If you do pull that faulty Nemesis out of your Falcon any chance you could donate/sell it to a madly optimistic field service engineer (namely me) to play with?

I've been looking for one for a while to speed the CPU / bus and FPU only. Not so interested in DSP speeding. The fault may be something simple that I could chase down and fix. Maybe.

If you're going to fit the CT63 you probably want to get rid on the Nemesis buffer mod, re-instate the 33 Ohm resistors, swap out the cheap low output factory transistors with BC547C or similar, and put the term res between the Ajax and SDMA clock line. I've just done all that hoping for a Nemesis to turn up so I can test it out!

Let me know if your Nemesis is worth a punt. Thanks!

[DarkLord]

If you're going to fit the CT63 you probably want to get rid on the Nemesis buffer mod, re-instate the 33 Ohm resistors, swap out the cheap low output factory transistors with BC547C or similar, and put the term res between the Ajax and SDMA clock line.

Hey SounDoctor...just curious - why wouldn't you want those mods (Nemesis) with the CT series? I've got a CT60 and the motherboard mods that speed the bus, DSP, etc, etc,... Seems to work fine here.

Or is the Nemesis not compatible with the CT series?

Thanks!

[SoundDoctor]

If you're going to fit the CT63 you probably want to get rid on the Nemesis buffer mod, re-instate the 33 Ohm resistors, swap out the cheap low output factory transistors with BC547C or similar, and put the term res between the Ajax and SDMA clock line.

Hey SounDoctor...just curious - why wouldn't you want those mods (Nemesis) with the CT series? I've got a CT60 and the motherboard mods that speed the bus, DSP, etc, etc,... Seems to work fine here.

Or is the Nemesis not compatible with the CT series?

Thanks!

No, not the Nemesis speeder itself but the early buffer mods that proliferate in European Falcons. Our machines have slightly different clock issues due to the PAL pixel clock. The Atari design to incorporate this was not their finest hour IMO. If we're being honest the whole Falcon thing was a bit of bodge job really to give ST compatibility that never quite delivered...but that is probably the real appeal of these machines - they CAN be improved. Speeding the bus (to a point) in a clean and stable way that is swichable is generally a good idea

You'll see a lot of threads on the forum about the various buffer mods that were employed to resolve issues with weak clock signals. Most of them DIY and badly done. The earliest and probably most common of which involved removing SMD resistors and adding wires from a 74F04 Nand Gate or 74F08 hex inverter IC piggy backed on a GAL for +5v and Gnd. One of the better ones came with the Nemesis but there are much better ways of resolving the wobbly clock today. Rodolphe's CT6x fitting instructions has a procedure to remove the old buffer clocks for European machines.

I'm guessing that you have an NTSC machine so didn't need to fit any sort of buffer mod or, if you did, it didn't cause too much trouble!

Viking mentioned removing a faulty Nemesis so it would be best to remove any old buffer mods and put the Falcon back to stock then put the more modern better thought through one suggested by Rodolphe in place. Simbo has improved on that and come up with a more complete solution. As far as I'm aware he has had the best success with a Nemesis I've read of here.

I'm sure most techs would agree that the bulk of stability problems seen on any Falcon when adding in new speeders or upgrade boards are due to previous incomplete, incorrect, or half removed buffer mods. I would always urge anyone to put the Falcon back as close to stock as possible before working through the fitting of a new and different board. It's a real problem when buying a 20 year old enthusiasts machine. It's pot luck how hacked about these old systems are. I bought another Falcon to accelerate from eBay...it wasn't cheap but I was really lucky, it was completely untouched by any enthusiastic but untalented modder. Hurrah! There is a downside tough.
I toyed with the idea of adding in my own clocks to speed the Falcon but it's so close to factory condition but I just couldn't bring myself to do it. Feels like vadalism
My only problem now is finding a Nemesis and /or waiting for Rodolphe's new CTX60 to do it properly. My Falcon is all dressed up with nowhere to go.

Sheesh I've written a book....shut up now Mark.

[DarkLord]

It's all good - I appreciate detailed replies/explanations. Helps people like me that
need a lot of hand-holding.

Mine is an NTSC machine.

I'm not exactly sure about how my mods were done. On the original Falcon I had
with a Nemesis, Alex Yu at ATY Computers installed it. Never had any problems
with it, but it wasn't maxed out - I think it was running at 20mhz instead of 25,
because it became unstable at the higher nemesis speeds.

On my current Falcon with the CT60, Mark Ducksworth did the mods. Hmm, I
think I've got a picture, very low quality though, need to redo it:

SDMA2.jpg

Hope you get something soon for yours - the Falcons deserve the best.

[viking272]

Which reminds me that I still haven't attempted to remove my faulty Nemesis and install my CT63. It's been in bubble wrap for 3 years, ahem.
However I have spent 6 hours tidying the office over Xmas and located all my Atari kit now so if I can find a decent case for the Falcy I'll re-house and rebuild it all.

Cheers

Faulty Nemesis? Any idea what's wrong with it? This is a bit of a cheek but...If you do pull that faulty Nemesis out of your Falcon any chance you could donate/sell it to a madly optimistic field service engineer (namely me) to play with?

I've been looking for one for a while to speed the CPU / bus and FPU only. Not so interested in DSP speeding. The fault may be something simple that I could chase down and fix. Maybe.

If you're going to fit the CT63 you probably want to get rid on the Nemesis buffer mod, re-instate the 33 Ohm resistors, swap out the cheap low output factory transistors with BC547C or similar, and put the term res between the Ajax and SDMA clock line. I've just done all that hoping for a Nemesis to turn up so I can test it out!

Let me know if your Nemesis is worth a punt. Thanks!

I am semi-interested as I thought I wouldn't really need the Nemesis if I have the CT63 to fit.
I don't think it's faulty per se as it was working for many years, but more likely something has come loose or needs refitting. But in operation it's not too stable when I last booted up my Falcon.
I'll send you a message.
Cheers, Simon

[SoundDoctor]

It's all good - I appreciate detailed replies/explanations. Helps people like me that
need a lot of hand-holding.

Mine is an NTSC machine.

I'm not exactly sure about how my mods were done. On the original Falcon I had
with a Nemesis, Alex Yu at ATY Computers installed it. Never had any problems
with it, but it wasn't maxed out - I think it was running at 20mhz instead of 25,
because it became unstable at the higher nemesis speeds.

On my current Falcon with the CT60, Mark Ducksworth did the mods. Hmm, I
think I've got a picture, very low quality though, need to redo it:

SDMA2.jpg

Hope you get something soon for yours - the Falcons deserve the best.

Thanks!

Yes, Simbo nailed it with his work using high quality mini coax and copper foil wrapping to keep the RF in check and allow the HF clock signals half a chance by adding additional grounding to constrain the ground bounce. I think solving the poor grounding / ground bounce and +V bounce and changing transistors to low noise high output is probably the bigger factor. The ground rail and +5v can be all over the place in a Falcon making the motherboard noisier than my wife when I come home with yet another computer
That's also why the 1st release machines had abysmal audio quality, so Atari changed the inductors on audio paths to the 3.5mm I/O to wires coiled on toroidal cokes which helped a bit. A bit of Atari trivia is one of the main engineering changes for the Falcon included the addition of tiny bead chokes/inductors put on the legs of the 8 pin 25MHz and 32MHz XTAL cans probably in an attempt to constrain lead inductance. I seriously doubt it did anything to help because the clock signals would still be jittering about all over the place regardless, add to that a half assed slooowww hex inverter buffer mod inverting the clock and you have asynchronous inverted jittering clock signals at various levels crashing into each other at various points all over the MB. OK, it's not quite that bad (at 16/32MHZ) but it can be if you increase the clocks to 20/40 or 25/50 or even 27.5/55 and then you end up with 1s and 0s rattling about like raw rice on a trampoline. The original Nemesis was struggling away with unshielded thin wire carrying the clock signals and control lines to various points on a MB that in all honesty is a dogs breakfast on the clock path front. I think the problem is probably worse now all the caps on the boards are 20 years old but that's what makes Falconry much fun.

My 100% stock Falcon is hysterical on the RF front - rock solid-ish but I sometimes try to listen to a DAB digital radio nearby when running the Falcon and even with the shielding in place you really get an idea how much high frequency RF shash is spewing out of it when you try to shzzsht f@lllløw a BBbbrøad.c.a.sss.t on,on,on,shzzzt,diggggital rrrraradio *shzzshd*//bzztktzt....If you know what I mean Imagine what it's like INSIDE the case. I used a Falcon 18 years ago to record audio in Cubase, and knowing what I know now, it's a miracle we managed to produce anything at all back then. But the Falcon was still an ambitious project. If they had the time and could have increased the COG / Price point it would have been a killer pro audio machine.

Looking at your pic you seem to have all the Rev J work (that can be seen anyway). The Ajax / SDMA term is there as needed and you have a permanent dedicated DSP clock feeding the 56k DSP chip wired to the clock i/p pin on the chip. The other way of doing that involves cutting the clock trace on the other side of the board under the clock can between the 8 pin and 14 pin can solder pads, bending the new clock pins to 45 degrees and soldering the can on its 'hind legs' into the unused pads on the factory clock. Yours looks less invasive / kinder - looks like a great job by Mark Ducksworth.

There I go again yadda yadda yadda...

[dml]

Viking mentioned removing a faulty Nemesis so it would be best to remove any old buffer mods and put the Falcon back to stock then put the more modern better thought through one suggested by Rodolphe in place. Simbo has improved on that and come up with a more complete solution. As far as I'm aware he has had the best success with a Nemesis I've read of here.

I can confirm that having repeated Simbo's approach (using high quality microwave-class coax with correct grounding and termination) coupled with replacing capacitors and being careful about shielding and wire length I've had good success with silly clock rates. I had no trouble at 55MHz and I'm going to have a go at 60MHz+ when I have finished the cleanup.

I'm still working on the Nemesis wiring/function diagram - it will be ready soon.

Happy new year all.

[dml]

It's all good - I appreciate detailed replies/explanations. Helps people like me that
need a lot of hand-holding.

That right there is a DSP overclock - a hand-fitted oscillator taking power from nearby and feeding the DSP's clock pin directly. The track between the DSP and the other oscillator will have been cut. The original 32MHz oscillator will still be feeding other bits of the board.

I'd say it probably isn't ideal to have it stuck on top of the SDMA chip like that in case it comes loose but the DSP clock isn't the most troublesome mod - it's probably best to leave it as it is.

There's a termination resistor mod in the upper left of the SDMA chip. This is for SCSI and audio problems.

[dml]

Here's a first go at the Nemesis board pinout diagram, using (what I believe are) the wire colours it was shipped with.

This diagram is based on A) my own working installation and B) the Falcon030 schematic. It's not based on the supplied fitting instructions in the manual. The manual is a step by step howto with more than one fitting permutation described - it explains very little about the function of the pads/wires. It is not a complete reference. The point of the diagram was to help disambiguate some strange looking installs and help 'refits' where needed.

WARNING: This is not intended to replace existing fitting instructions! Fitting the board requires tracks to be cut and/or components removed and there are different ways to go about it. Use it for reference only, to help rewire/restore an existing board or figure out what's going on with an existing installation...


Some of the wires are interchangeable (e.g. the four main clock outputs J1-J4) and might not match the wire colours from your installation so don't be put off by slight differences.

The FPU and DSP outputs are optional. In fact the DSP and FPU are better driven by separate clocks unless mini coax is used, otherwise it may not be reliable due to the extra wire length needed.

Nemesis used no proprietary parts or firmware - i.e. no PALs/GALs or controllers. It's all TTL logic so dead boards can be repaired easily with off-the-shelf bits.

The 'LO/HI' switch lines (J8/J9) are supposed to connect to the MC6850 controllers. I placed a 220R resistor in series with the J8/J9 switch lines.

The RGB dot clock switch line is supposed to be a physical switch (+5V=on / GND=off). Again a series resistor is recommended.

The RGB dot clock mod (J5/J10/J11) is optional - it provides an RGB-monitor (32MHz) compatible dot clock when the main 32MHz clock has been modified to 40+. I believe it does this by hijacking the VGA 25MHz clock and forcing it to 32MHz. It requires an auto folder patch as well to force the VIDEL to use the VGA dot clock in RGB mode. If you're only using VGA output this mod can be ignored (or undone).

The three electrolytic capacitors are not power filter/bypass capacitors - they are switching/debounce filters for the 3 switching lines. If your machine runs nice and stable with Nemesis in any mode, but switching *between* modes sometimes causes a crash, these caps may need changed and/or series resistors on J8/J9 need attention and/or shielded wire (mini coax) is needed for the switching wires.

Ground as many of the 'G' pads as possible, using the shortest links you can get away with and preferably using thick wire, copper braid or foil (I fitted mine at the outer ground rail since it's in a C-LAB/X case).

All signal lines (J1/2/3/4/5/7/10/11) benefit from using mini coax with the shield properly connected to GND.

Nemesis still requires the buffer patch. The buffer patch is needed to 'fork' the COMBEL output clock to various parts of the board (CPU/FPU etc.) using buffer gates, instead of a resistor 'splitter' (The resistor splitter was a cost saving adventure by Atari, and a bad one - the signals are not properly isolated from each other, causing stability issues). Also the buffer patch sits on the far side of the COMBEL clock divider (32MHz -> 16MHz) and forks the pre-divided 16MHz signal. Nemesis deals with signals 2x that frequency, on the input side of the COMBEL divider. You can bypass this arrangement in some cases (e.g. wire Nemesis directly to FPU for double clocking) but it's unlikely to work with the CPU and will mess up the SDMA too. In summary - Nemesis is on the input side of the COMBEL and VIDEL, and most of the other devices are on the output side of the COMBEL, working with master clock divided by 2. Therefore, buffer patch is still needed to fork the divided signal.

Nemesis might benefit from adding a local bypass capacitor (tantalum) although I haven't found it necessary so far with the mounting site I used.

The Nemesis board was designed by Peter Green IIRC. I didn't have anything to do with the hardware - although I did make some software bits and pieces for the install package...


If I any mistakes surface I'll fix and repost an update. Most likely error is NEM_LO/HI switch lines are reversed e.g. HI/LO instead of LO/HI. I'll check this later with a probe.

[UPDATE] errata #1 - the purple wires (from J6) are not in fact return lines for the HI/LO switches - they are a replacement 500kHz signal for the mouse/keyboard serial chips - otherwise these don't function properly when the master oscillator is changed from 32MHz. Since this is another clock feed it's probably a good idea to use mini coax again, to stop it interfering with other stuff.

nemdg100.png

[SoundDoctor]

It's all good - I appreciate detailed replies/explanations. Helps people like me that
need a lot of hand-holding.

That right there is a DSP overclock - a hand-fitted oscillator taking power from nearby and feeding the DSP's clock pin directly. The track between the DSP and the other oscillator will have been cut. The original 32MHz oscillator will still be feeding other bits of the board.

I'd say it probably isn't ideal to have it stuck on top of the SDMA chip like that in case it comes loose but the DSP clock isn't the most troublesome mod - it's probably best to leave it as it is.

I assumed that the clock pin of the DSP has been lifted...breaking my own rule...never assume!

[dml]

I assumed that the clock pin of the DSP has been lifted...breaking my own rule...never assume!

You could be right - I assumed the track was cut because that was the most common way. It could just as well have been lifted.

[dml]

An old webpage by the same Peter Green seems to describe a precursor to the Nemesis board - the latter incorporates most of these mods into one package and improves on them a bit. This makes it quite a good addendum to this thread

http://atari.nvg.org/cpu_accel/

[DarkLord]

That right there is a DSP overclock - a hand-fitted oscillator taking power from nearby and feeding the DSP's clock pin directly. The track between the DSP and the other oscillator will have been cut. The original 32MHz oscillator will still be feeding other bits of the board.

I'd say it probably isn't ideal to have it stuck on top of the SDMA chip like that in case it comes loose but the DSP clock isn't the most troublesome mod - it's probably best to leave it as it is.

There's a termination resistor mod in the upper left of the SDMA chip. This is for SCSI and audio problems.

Okay, thanks for the info Doug.

[SoundDoctor]

An old webpage by the same Peter Green seems to describe a precursor to the Nemesis board - the latter incorporates most of these mods into one package and improves on them a bit. This makes it quite a good addendum to this thread

http://atari.nvg.org/cpu_accel/

Yes it's nice to look at the first seed of an idea retrospectively. I often wondered how many people actually sent that registration fee!

I'm not sure I agree with PGs apraisal of wind turbines though!

[Guest]

i reused the 32.000,000 hz dsp TTL clock
and connected it as the dot clock in to the nemesis {main 32mhz clock instead of L102 }
this is the same as the USA ntsc machines setup with just 32mhz direct clock
i don't need a TV so WTF ???
do i need a RGB pollutant and all 32mhz derived clocks modulated dot clock for??? + the CCLK i remove ...
adds to the stability of the audio sdma etc...
you never see the Americans moan about a bad sdma audio stuff
in there world it doesn't exist
i wonder why
it just pisses me off the nemesis from an RF point of view its fitting is like a dogs breakfast
hence the coax and RF clean up
i get 250% stable chipset @ 55MHZ main and i use the 40mhz
and 32 standard .. and thats it ...
55mhz when compile and 32 any other time

heatsinks on all ics now using bonded heatsinks {ebay bonding pads}

what you need to ask is what is a compatible resolution for your STE games in RGB {next harmonic???}

[DarkLord]

you never see the americans moan about a bad sdma audio stuff
in there world it doesnt exist
i wonder why

Hmm, well in my case, according to Doug and that picture I posted, mine has
the mods to add stability done.

[Guest]

not sure

did you use the 32mhz DSP TTL clock to supply 32mhz to the chip set @ L102 ?

its AOK to overclock the DSP to 50mhz or so and leave the chipset @ 32mhz


my mod would of choice for audio and midi {i dont use it only pcb making and simulations in pure c}
be to use the ttl DSP clock @ L102 to inject 32mhz to the main chipset
and to overclock the dsp to 50mhz as youve done {i think}
so i would also extend the 50mhz to the coprocessor
and the clkin to the videl {from the video socket external clock like the blowup}
NO CLOCK BUFFER MOD
all resistors are 33R otoh 216 217 221 220?? 222 not sure the four
most falcons are fitted with a zero ohm at 221? needs changed as per ct63 install
and at the sdma a 50R carbon resistor or 4X 75R from old ntl terminator plugs
to make 50 with no inductive accepts if you know how termination is made non inductive using 4 resistors metal film
or just one carbon you can use a 2.2R and a 47R
in series
basically it should be 50R as close to the clock in pin as possible on the sdma
using the pcb to carry the clock to the sdma is a bad idea

the actual problem is the pcb as there is no null loop inductors and close coupled capacitance added {before there time i am afraid }
forinstance vero board has a capacitance between its tracks of 15 pF
modern pcb designs are far more careful to add pcb tracks inductors as chokes for decoupling rf and stray rf
and avoid resonant size ...and width tracking sadly missing from the falcon and the TT as i say before the time of pcb simulations etc...

rf above 30mhz is considered as VHF and conforms to a different set of conductive rules
to frequencies below 30mhz like citizens band {26-28MHZ} or 10M ham bad {28-30MHZ}

it tends to 'cling' to tracks rather than conducted
so a clean path is essential and bare copper is often replaced with silver and even gold plate
hence my use when connecting the nemesis to use silver coax .... 50 OHM

to allow the RF to flow without having to climb across the 'bolders' of copper clumps etc
where fine pcb are smooth and silky {stbook} its the best built atari by far

made in japan

im highly impressed by its pcb... other st pale into dust beside it

GM1KMK is my call.... if your interested and i repaired most radios that needed it in glasgow for about since i was 15