[News] Announcement: Technical data of the new Catweasel MK4 | ANN.lu |
Posted on 18-Jul-2004 12:20 GMT by Jens Schönfeld | 469 comments View flat View list |
During the first months of this year, we were totally surprised by the overwhelming demand for Catweasel MK3. All stock has been sold, and even our retail partners do not have anything left in stock. The demand was so high that one controller even went for more than 150,- EUR on eBay!
Unfortunately, it turned out that a new production run of the existing MK3-design cannot be done for reasonable prices, so a re-design became necessary. The new controller is now in the first stage before mass production, so we're confident to be able to show the first controllers at the Amiwest show on july 24th and 25th in Sacramento, Califoria.
Many improvements have been made compared to the Catweasel MK3 that can be summarized under the headline "bigger, better, faster".
The changes in detail
The most obvious change is the size of the card: With only 2.5 inch (63.5mm) height, it complies with the low-profile PCI standard that not only fits into any normal computer case, but also in flat models that are so famous among so-called "case-modders". The Flipper-interface will not be continued. Those who need a Catweasel for their classic Amiga can use the Catweasel Z-II S-Class, which is still availble.
Improvements on the floppy controller
Kylwalda built in
While the old Catweasel models always had their own floppy drives that were installed in addition to the existing controller and drives, the MK4 has the option of using the existing diskdrives. After the machine has started, th drives can be used just like before, and after the drivers have been loaded, the Catweasel can take control of these drives when necessary. This is especially useful for smaller cases with fewer drive bays.
We already addressed this problem earlier with an additional product called 'Kylwalda'.
Suppot for auto-eject drives
These drives without eject-button are well-known from Macintosh computers, and they're now fully supported by the Catweasel. You can also mix floppy types, one with an eject-button, and another from the Macintosh world on the same cable.
Hard-sectored disks supported
This kind of 5.25 and 8 inch disks were already readable with the previous Catweasel models, but writing was only possible with a high software effort, and it required a realtime operating system. This effort is not necessary any more with the new controller, because new options allow complete support of these disk types in hardware.
dual-ported memory
Contrary to it's predecessors, the new Catweasel MK4 can pass the data that it is currently reading from a disk to the computer while the read access is running. This allows realtime emulation, and errorfree function of copy-protected software on emulators.
more flexible read- and write operations
In addition to working on whole tracks, which made all previous Catweasel models so flexible, tracks can now also be accesed in part very precisely. Should this become necessary for compatibility or speed reasons, the Catweasel MK4 is perfectly prepared.
extensive timer-functions
Although most operating systems already offer timer-functions in software, you cannot always rely on them. The most recent example are the timing-problems that occur with Hyperthreading-processors and Windows operating systems. Since all timers are running independantly in the hardware of the Catweasel MK4, nothing can go wrong in this regard.
all events can trigger an interrupt (IRQ)
Together with the hardware-timer functions, this is the best solution for multitasking operating systems. The driver software does not have to check regularly if the controller needs attention, which reduces the processor load.
Improvements on the keyboard interface
In addition to Amiga-keyboards, PS/2 devices can now also be connected. Not only keyboards, but also PS/2 mice are supported. The keyboard controller can now trigger IRQs, and for those customers who want to continue using their favourite combination of PS/2 mouse and keyboard on USB-only computers, the Catweasel MK4 has two connectors of this kind.
Improvements on the joystick ports
Amiga mice supported in hardware
Amiga mice only have minimal electronics that always pass the movements of the device to the computer in realtime. Classic Amiga computers have hardware-support for interpretation of these signals in the chipset, and this support has now been added to the Catweasel. Theoretically, using Amiga mice was already possible with the Catweasel MK3, but this required a software effort that was not justifiable. With the new hardware, the software effort is reduced to a minimum.
every signal can be programmed as output
The digital joystick ports of the 8-bit computers of the 80s were mostly usable in two directions, they were not only inputs, but also programmable as data outputs. We're following this tradition, and also present this possibility for the Catweasel MK4.
compatible with CD32 pads
The game controllers of the Amiga CD32 can now also be used on the Catweasel. A special capability of the classic Amigas (and therefore also of the CD32) made these pads exclusive for this computer, if connected to other computers, not all buttons of the pad could be used. Technically speaking: Even the potentiometer-pins of the digital joystick ports can be programmed as outputs on the Catweasel MK4.
Improvements on the SID audio part
DC-DC converter eliminates noise
On the Catweasel MK3, it was possible that noises from 3D-graphics cards or high-speed harddrives were coupled into the 12V-power supply of the SID audio part. This cannot happen any more on the Catweasel MK4, because a DC-DC converter is an insuperable obstacle for such noises.
cycle-exact control
In addition to the known programming that's compatible with the Catweasel MK3, the MK4 has a sophisticated script-language for SID control. This lets the programmer define the exact time for data that's being written into the SID chips. To make sample playback sound exactly like on a real C64, the data rate to the SID chip must be kept at a constant rate. This is accomplished with Fifo memory that's big enough to maintain the datarate even under high processor load conditions.
Digiboost for new SID versions
As opposed to the 'classic SID' 6581, the newer SID-chips 8580 and 6582 cannot playback samples any more. This option, which is also called 'the fourth voice', is replaced by two sigma-delta converters on the Catweasel MK4, so the fourth channel is also audible with the newer SID versions. Since the filter properties and the sound of mixed waveforms of all SID versions have their supporters, this should make the decision for the right chip a little easier.
Filter capacitors selectable
Commodore has defined three different capacitor values for the filters of the SIDs during the years that this chip has been produced. The result was that the same chip sounded differently if used in different computers. To bring the sound as close as possible to what you are used to, the filter capacitors can be chosen with a few jumpers.
precise clocking
The Catweasel MK3 used the commodore-chip 8701 to recreate the exact same clock. Since our stock of this chip is empty with the Catweasel MK3 being sold out, we have cloned it on the main logic chip of the Catweasel MK4: The exact base frequency is generated with crystals that have been made especially for us. By division and multiplication according to the specifications of the C64 schematics from 1982, we managed to replace the 8701, which is not made any more. Even the slight difference between PAL and NTSC computers is software-selectable!
two SIDs for stereo sound
You'll have twice the SID pleaseure after installing a second SID chip. Every SID has it's own selection of filter capacitors, and SIDs of all versions can be mixed.
Technology improvements
compatible with 3.3V and 5V PCI slots
Even though PC boards with 3.3V PCI slots are not yet widely available, the Catweasel is prepared for it. The roadmap of the PCI special interst group plans to abandon 5V PCI slots within forseeable time, and the Catweasel is perfectly suited for that date. Local generation of the 3.3V power also ensures proper function on early PCI motherboards that do not comply with the ATX standard.
two DMA interfaces
In addition to processor-based data transfer, the Catweasel MK4 can excahnge data with the main system through two low-speed DMA channels: The first goes throught he PCI slot, and it has a capacity of about 8K per second and direction. The second uses the direct connection to the onboard-floppy controller, and the speed is up to 100K per second.
low power consumption
The Catweasel MK4 makes use of the latest FPGA technology with 2.5V core voltage. This reduces the power consumption of the new controller to a fraction of what the Catweasel MK3 used. This also reduces heat generation a lot.
re-configurable logic
The FPGA on the Catweasel MK4 is completely re-configurable by the drivers. This means that a hardware update can be done through the internet! Should we find a disk format that cannot be handled with the current hardware, the core of the Catweasel can be 're-wired' to address the problem. The controller doesn't even have to be taken out of the computer for ths update!
drivers for many operating systems
The Catweasel MK4 is delivered with drivers for Windows 98(se)/ME/XP/2000, Amiga OS4, and for Mac OS X at a later date. MorphOS drivers are available for a surcharge.
The Catweasel MK4 will be available starting october 2004.
The target retail price is 99,- EUR.
|
|
List of all comments to this article |
Announcement: Technical data of the new Catweasel MK4 : Comment 425 of 469 | ANN.lu |
Posted by Stefan Burström on 21-Jul-2004 13:29 GMT | In reply to Comment 384 (Johan Rönnblom): >They do? Please be careful before someone lumps you together with Bill
Indeed. I havn't seen any actions lately to get either MOS nor Pegasoses
on the market by bPlan. Howver, I have seen huge efforts by Genesi on this.
If you search the archives I am quite sure you will find BBRV's post explaining
how this constelation all fits together.
>McEwen just because you happened to make the browser that's shipped
>with AmigaOS4.
As long as he keeps his mouth shut regarding IBrowse, I don't have anything
to complain about. But as soon as he starts making claims about IBrowse
or sell it based on merits I don't aggree with you can be very sure I will
react. So, so far your lump argument doesn't really work on me.
>I agree, if we're talking about the average margin across the line.
Good, at least we agree on something.
>Yes. RadioShack sell Nokia products. Genesi sell bPlan products. RadioShack
>are not the only resellers of Nokia products for sure, but neither are
>Genesi the only resellers for bPlan products.
Well, how do you explain the change of Pegasos supply when Genesis main source
of income was cut (Thendic-France)? The connections are far too obvious not
to be speculated on. However, if RadioShack lost its main customer, I hardly
think that would be noticed by Nokias sales.
>So in a discussion about whether Pegasos production is subsidised or not, I
>think we can leave Genesi out. It's a different topic, interesting as it
>may be.
Well, interestingly enough, they have been the entity with money, so that's
why I left them in there.
>What do you base this on? We know that bPlan existed before Genesi. Of
>course things would have been different without Genesi, that's for sure.
Yeah, for one thing, the Pegasos boards would not have been in the same
state as they are today since bPlan would have to do alot more design house
work to keep afloat, that's for sure.
But, sure, let me reprase it. bPlan as we know it today would have looked
_alot_ different.
>And interestingly, bPlan start to ship directly to dealers, when previously
>all resales went through Genesi. I don't really think this supports your
>"symbiosis" theory.
Since you don't know how bPlan is finaced, you cannot prove anything with that
argument. What if bPlan is partially financed with VC money coming from
Genesi?
>> However, I was looking at the total BOM of the product.
>Which I guess is a natural approach considering your field of expertise,
>but it does have its weaknesses.
I approached the BOM simply because that would be the easiest one for you
to understand and also the one that would favour your opinion the most.
You do know what BOM stands for, right?
>>> Microvias between layer 1-2 and 7-8 to be able to route the BGA's.
>> There's no layer 7-8..
> Duh! You don't understand how funny you are!
>No. There is no layer 7-8, period. Obviously neither of us know the exact
>design of this PCB. You made a guess, but you were wrong.
Sigh! Either you are this stupid or you are deliberately trying not to
understand what I am talking about. There is an othermost layer on the
bottom end of a PCB. Period!!! If it is layer 8 or 6 is irrelevant. It is
still an RCC layer if RCC is used between layer 1-2. Besides, I made guess
about 8 layers, but checked the prices for 6 layers just to stay on the
safe side. So, yes, saying 8 layers was wrong, but the rest of what you
wrote is just clueless junk written by someone who doesn't have any clue about
PCB design.
>Let's say I just took your word for it? You seem to consider yourself quite
>an authority here, so I guess that should satisfy you?
Well at the same time you didn't understand that a PCB has 2 outermost layers,
one on each side and used that as an argument against me.
>And why would you do this for each run? I was under the impression that you
>were talking about per-unit costs rather than the entire investment.
It is not even sure that the same PCB manufacturer was used for the different
runs. Besides, I am quite sure that there was changes to the PCB between the
different runs just to improve producability and increase the yeild. So the
CAM preparation is most likely included for every run.
Maybe we're talking about different subjects here, but from my perspective,
the interesting thing is the marginal profit/loss. That is, the profit/loss
of producing 1 extra board, given that everything else stays the same.
I mean, of course it's true that if the only Pegasos boards ever produced
are those made and sold for MorphOS users and the odd Linux geek hardware
fan, it's not going to repay the investment. To do that, they need higher
volumes, for sure. But to me at least, the suggestion that sales would be
"subsidised" means more than just that these sales are in a way riding on
the back of an antecipated larger production. The interesting question for
me is whether making these boards helps or hurts bPlans finances. I find it
more likely than not that it helps their finances, even if it cannot run
their business in the long run.
> Right! I have seen many pictures from Phase5 but I never saw that they
> had their own PCB manufacturing too.
>No, indeed not, but I wouldn't be surprised if they do part or most of the
>work of assuring the paths are ok etc themselves.
They don't do the CAM preparation themselves. That's not how the flow looks
like when you do this. Sometimes you do the panel design yourself, but that
is not very common either. (We didn't start doing that until just recently)
>But anyway, do you know the cost of the other components involved, besides
>the PCB? I do admit that I have no clue what PPC CPUs cost, for example. It
>seems those who know aren't allowed to tell.
I seem to recall numbers in the range of 150-200 USD for 1k quantities quite
some time back. Checked some of the connectors at digi-key, they could run
for around 1-5 USD / piece, and some of them are not easy to source so they
can get quite expensive if you have to ordewr them through digi-key
>Now this is *pure* speculation, I have absolutely *no* info about this
>other than the respective pricing of the units and the well known history
>of the Peg 1, but may it not be that a certain quantity of 600MHz G3 chips
>were bought, intended for production of the Pegasos 1, before it was
>discontinued due to the decision to change northbridge?
This could be true. Note however, that these G3 cpus were bought 2 years ago
then, when wern't exactly cheap.
>Of course, if my theory is correct, we shouldn't see any more 600MHz G3
>units made now, as in that case these chips are likely to be used up by
>now.
I'd guess that the G3's today are alot cheaper to source, so that could
solve the problem. It still wouldn't prove your theory right or wrong.
[CyberstormPPC]
>> is technically more advanced (and thus more expensive to make) than the
> Lol! Care to tell us why it is technically more advanced? For a start, it
> doesn't contain a 133 MHz SDRAM bus which isn't childsplay to design.
More layers, higher component density. Well actually I intended to refer to
the BlizzardPPC, not the CyberStormPPC. You know, components on both sides,
etc. But I take your word that they probably aren't using exactly the same
software as in 1997. I certainly wouldn't have guessed on that anyway, I
just pointed out that none of us can accurately guess how much money that
would be needed for this. And anyway, we're talking about one-time
investments, not per-unit costs. You seem to like it both ways, whenever I
point to how margins can be cut, you insist on discussing only the BOM. But
when you find costs to add, you seem less strict. :-)
Anyway, what do we have now for the BOM? Say $80 + ?? + ?? + ??, or do you
have a better guess? Doesn't really allow us to draw any conclusions, now
does it?
I adopt a different strategy. I assume that the AmigaOne is not produced at
a loss. This may of course be wrong, who knows what MAI might do to keep
their only known customer. But really, I don't expect it to be sold below
production cost at any stage.
Now, without knowing about the details, we can easily see that the A1 will
be more expensive to produce. You're probably better than me at telling how
much. Then we need to add the higher cost due to licensing fees, MAI's
involvement, shipping them across the world, perhaps back and forth when
there are problems (these are not unknown for the A1..), and Eyetech. And
we also know (at least I do) that the dealers' margins are higher on the A1
- which IMO suggests that the margins are also higher upwards the chain.
Looking at it this way, The price difference between the A1 and the Pegasos
seems quite realistic to me.
Then of course there is a third way of looking at it. Personally, I would
be quite baffled if someone decided to sell boards below production cost at
these circumstances. I don't claim that everyone's rational, but when
things look just too crazy to be true, it is often the case that it's not
true.
> Reality check please! We are talking of the same components, the same
> PCBs, the same connectors here. What the Japanses manufacturers managed
> to do was to replace the whole BOM making it cheaper.
That is only part of the story. They also minimised spoilage, slimmed down
organisation, reduced storage time for parts, and made production more
efficient in many other ways. What must be noted here is that the materials
cost is not the only way to improve, rather I'd say it's the most difficult
area in most cases, probably including this one.
> As I said, the Pegasos 2 is another story and it has a different price.
Hmm, I think it's a very similar story, and it has a very similar price (G3
version).
> Well, based on the above, isn't it rather obviuous that I have done my
> homework alot more through than you?
Nope. Out of the two verifiable figures you gave (number of layers, and
size of the board) both were significantly wrong. And both these figures
Sorry, the size of the board was based on another MicroATX board. And they
number of layers isn't significantly wrong if you ask as PCB designer. Esp.
since I based the cost on a 6 layer PCB. But due to your recent nitpicking
I am not surprised you take this as an argument that you are right instead
of understanding that you should get a clue what you are talking about
instead.
>shouldn't be too hard to find out, especially the board size should be easy
>to find for anyone who's heard about google.
Well, I did find the MicroATX board using Google, smartass!
>Based on these *verified* overestimates, and your general display of a bias
>here, I don't give your overall guess much credit. And yes it can hardly be
>called anything but a guess, because you've given only a small part of the
>calculations necessary to employ your method in order to get a proper
>estimate of the total price.
So by you failing to understand the basics of PCB desin, you figure I am
wrong? Sigh, well it is impossible to argue with you anyway, I should have
figured that from the beginning. But fwiw, I calculated the PCB cost now
just because it was the easiest one to do with most chance of getting
accurate prices even 2 years back. For the NB, SB, Ethernet Phy, AC97 etc.
it would have been much harder to find the accurate numbers which I started
to look up way back.
>Well of course I know you won't accept any story that does not condemn BBRV
>as Satan incarnated as anything but apologism, but hey, that's the price of
>being balanced.
You, balanced? Give me a break! If you were this balanced, then you should
know better and stay out of discussions you obviously have no clue about.
rgds,
Stefan |
|
List of all comments to this article (continued) |
|
- User Menu
-
- About ANN archives
- The ANN archives is powered by #AmigaZeux. It was updated daily (news last: 22-Oct-2004; comments last: 18-May-2005).
ANN.lu was created, previously owned and maintained by Christian Kemp, www.ckemp.com.
- Contribute
- Not possible at this time!
- Search ANN archives
- Advanced search
- Hosting
- ANN.lu was hosted by Dreamhost. Sign up through this link, mention "ckemp" as referrer and he will get a 10% commission on any account you purchase.
Please show your appreciation for any past, present and future work on ANN.lu by making a contribution via PayPal.
|