Kurzweil 1000 Expanders
Skip title and menuKurzweil K1000 K1200 Users Group
A support site for Kurzweil 1000 series music keyboards and expanders
K1000 keyboard


Kurzweil 1000 Series Upgrade Chip Information & Chip Data Downloads

Here is information about the chips to use and about the data files you can download and use for preparing (programming/burning) the OS, Setup and Sound chips to upgrade your K1000 module or keyboard. You will need to purchase some chips to program (burn). Information about what kind of chips to use is below. You also need a PROM/EPROM programmer for this and some guts. See the other pages in the Upgrades section for this. It means opening the unit, removing chips, burning PROMS or EPROMS and probably some soldering. Upgrading to many of the various versions of the modules (the HX for instance) is a waste of time. IF you upgrade, do it all the way to one of the Pro's. Save money and time, but with a lot more results. See Process 1 or Process 2 for details.

Burning chips too much for you?  EPromPro or Matt's Basement Arcade can burn your chips.

ROM data for programming your upgrade chips (binary data). These are zip files. Right click the link to download using IE. Other browsers work differently.

Upgrade    Upgrade
   To            From
Pro 1     8 or 23 button module.  Note that the 8 (Eight) button boxes have two possible OS chip sets.  Details in the Download zip file.  Alternate Sound Roms (512k Sound Romsfile1file2

Pro 2     8 or 23 button module. Note that the 8 (Eight) button boxes have two possible OS chip sets.  Details in the Download zip file.  Alternate Sound Roms (512k Sound Romsfile1file2

Pro 3     23 button module.  Alternate Sound Roms (512k Sound Romsfile1file2

Pro 76    K1000SE (and some others) Keyboard

GX         GX 8 button Chip sets - See the readme file in the zip file.

Chip specifics and what to purchase: (Compiled from various user group messages - edited)

You can purchase any of these types of Programmable ROMs but make sure you have a programmer that will support the chips you purchase:
UV Erasable Programmable ROMs - usually called EPROMs.
One Time Programmable (OTP) ROMS - usually call PROMs, some times EPROMs.
Electronically Erasable Programmable ROMS - usually called EEPROMs.

Read this before you purchase chips: http://wiki.xtronics.com/index.php/How_EPROMS_Work

Speed: The original K1000 modules used 200ns chips for the OS and Sample ROMS. You can purchase any chip that will run at that speed or higher speeds. 70ns chips have been used successfully. However, it is important that the two OS chips (modules) are of the same type and speed.

Chip part numbers: Note that every manufacturer uses a slightly different nomenclature and the part number usually indicates the type of chip and the speed of the chip. Search the web using the generic numbers to find the best price and type you want. These data sheets are only a sample of what is available.

Generic numbers:

  • 27c080     1 Meg x 8 Bit - Used for the Sound Samples (modules and keyboards).
  • 27c801     1 Meg x 8 Bit - Alternative for Sound Samples (modules and keyboards)..
  • 27c040, 27C40001 or TC574000   512K x 8 Bit - Alternatives for Sound
        Burn Data is available on the Yahoo User site here:
        512Kx8bit Burn Data
  • Important: Read Note about Sound ROMS below.
  • 27c512     64 K x 8 Bit - Used for the OS/Setups in the Modules.
  • 27c010    128K x 8 Bit - Used for the OS/Setups in the K1000SE/Pro Keyboards.

Data sheets (typical)

  • AT27c080(pdf)     1 Meg x 8 Bit UV Erasable EPROM - Used for the Sound Samples.
  • M27c801(pdf)      1 Meg x 8 Bit UV EPROM and OTP EPROM - Alternative for Sound Samples.
  • AT27c040(pdf)     512K x 8 Bit OTP EPROM - Alternative for Sound Samples.
                             (Burn Data not yet available) Important:
    Read Note about Sound ROMs below.
  • Am27c512(pdf)    64 K x 8 Bit CMOS EPROM - Used for the OS/Setups in the Modules.
  • AT27c010(L)(pdf) 128K x 8 Bit OTP EPROM - Used for the OS/Setups in the K1000SE/Pro Keyboards.

A Q & A discussion from the user group message forum (edited by web master)

Q. [sic]  Do you have to upgrade the OS if you add sample chips and visa versa?

Q. [sic]  Which modules does it make sense to upgrade?

Q. [sic] What kinds of chips can be used? 

A. [editied] If you put more ROM chips in the 1000 modules you MUST update the OS (setup and engine) so it will recognize the presence of those chips. Recognition of the ROM chips is contained in the OS. You can't just pop ROMs inside and expect the OS to know they are there. Likewise you can't just insert an upgraded OS without inserting the ROM chips. You will have voices on the screen with no sounds because the OS can't find the missing chips. The ROMs and the OS in inextricably linked so we went straight for the Pro series modules. The group spent months and months examining this part of the Kurzweil and there was no argument about what to do in the end.
There were several intermediate steps of upgrading that emerged during the model run that were just too numerous to capture and fool with so the K1000 user group did not waste time locating the intermediate upgrades. We decided if we were going to put some chips in the module we would go ahead and upgrade all the way for the exact same price. Example: we did not try to clone an 'SXa' running version 4. Why not spend $12 more for the Pro2? All we needed were a few ROM chips and the OS upgrade was required anyway. Therefore you only have the upgrade data for a full upgrade to a Pro module available from the user group.

You can use any speed chips as long as the paired chips used for the OS are the same speed. Half the OS code is on one chip and half on the other chip so they operate as a pair at the same speed. The split is about every 8 bytes and jump to the other chip to finish and back and forth. The pairs need to be the same speed. For example: the OS setup chips should be the same speed. I have used 70 nanosecond chips in place of the original 200 nanoseconds. At the time of the introduction of the 1000 models the 200 nanosecond chip was as fast as consumer chips went. But fortunately Kurzweil has a "check reply" at the end of each action so the OS chips can't outrun the module itself. Very cool.

You can use erasable chips or one time programmable (OTP) slightly cheaper chips. The OTP require you buy a new blank if there is a burner error. I have used EEPROMS (the correct terminology for the erasables) in the sound ROM slots also without trouble. The problem with the sound ROM EEPROMs is expense and quite rare in the used market. I used them for trouble shooting only. I would recommend using OTP chips for the sound ROMs which is what Kurzweil used originally. If you look at the original sound ROM's they are preprinted with the U-number which indicates they must have subcontracted them out. My discussions with former assembly workers indicates they didn't have any silk screening facilities on site in MA.

A quick check of DigiKey where I buy chips indicates occasionally the chips are becoming legacy products and are special order on some items. Sometimes I return later to the digi-key site and they have a few more back in stock. They have become expensive and fading fast. This is why I stopped doing the chips because chips were expensive, the modules are getting old and people were starting to have "other problems" at the same time and it was becoming impossible to trouble shoot them using only email. I could fix them in minutes if I received them by mail (had the unit), but the users usually had no tools, no expertise, and no money.... that's WHY the user was in the group.... It was an inexpensive alternative and now the modules are getting expensive to maintain.

Chips are here:
AT27C512R-45PI-ND is the part number I used for OS
AT27C080-90PI-ND will work for the sound ROMs

Searching on the digi-key site for "28C" items yields no matches but 27C searches brings up a long list of EPROMs and the UV erasables which are embedded in the list further down.

David in Raleigh


The Long Version of which chips to use for the Sound ROMS.
The 27C080, ST27c801 and other 1Meg X 8 bit (8 Megabits chips) are twice as large as needed for the Sound ROMS but were selected early on in the upgrade project probably because we miss-read the schematics. The schematics indicate the Roms are "1-8Meg". We read this as 1Meg X 8 bit when it really means "any Rom from 1 Megabits to 8 Megabits can work with the design. Kurzweil only used 512K x 8 bit ROMS - 4 Megabit chips. So, what does this mean? First of all we can continue to use the larger chips for now. The data files provided work only with these chips. More info and the data files for the smaller chips (27C040 family) will be added in the future. These smaller chips should also work but only when the data files are ready. 

Here is the rest of the story on the Sound chips:  (Based on users group site discussion - edited by webmaster)

R. (Kurzweil): I don't recall that in the late 80s that 8 meg roms existed yet, except on paper, but engineering knew they would be available in the future, thus engineering made plans for their future use. (To bad they never did!) All 1000 series instruments we designed to handle sound rom chips that could be 1, 2, 4 or 8 megs. And the schematics even say 1-8 meg by each chip. But we never used anything but 4 meggers (512 x 8). The use of different size ROMs is accomplished by address jumpers that are configured to support different size chips. An 8 meg chip needs address 19 (pin 1). If you check the schematics or a board there is no connection between A19 and Arnold Chips or the address buffers for the Arnold Chips. So address 19 is floating, not a good thing to have as we like thing to be high or low mostly. So depending on who's 8 megger you use (to burn Sound data), there could be issue. I have burned several replacement chips using the 27C040. Mostly they are just old obsolete parts now.
So a 27C040 or 27C40001 or TC574000, any will work just fine -that is what is in there!

PS, if one is mixing samples, like acoustic expander with PX, don't forget that the PX DAC filters were rolled off at 10.25K and most of the others expanders at 12.5KHz. You will hear funny things on top with PX samples through a 12.5 filter

Webmaster research: According to the schematics we have it looks like Kurzweil planned to use 1 Meg x 8 chips for expansion. The arnold can address 1Meg addresses (actually up to 8Meg) and they build the boards to address 1 Meg X 8 ROMS - perhaps in case they did lots more samples - which they evidently never did. They designed the board and arnold to use 12 bits of data but each ROM chip (1Meg by 8) only has 8 bits output so they used three ROMS to do the job for each set of samples. For example, they used 8 bits from U50 and 4 bits from U54 for one sample(s) - 8 bits from U58 and the other 8 bits from U54 for the other sample(s). This explains why upgrades come in groups of 3 ROM Chips. Here is what really happened. They used the 512K X 8 chips - never used the 1Meg high address line by setting JP 7 open (not connected to the arnold chip) and JP 5 open (pin 1 floating). This is the way my K1000 SE is set up! FYI, they also designed the 512K address line strappable - perhaps they were not even sure they would need to go to 512K.. According to the schematics, they strapped the 512K line to use the arnold address by setting JP11closed and JP9 open. So, what about using 512K X 8bit versus 1Meg X 8bit chips? 
The chips use pin 1 differently (this is important): On the 27C080, pin 1 is A19 (1 Meg address line). In theory and using good CMOS engineering, it should be held high or low, but is left floating if we use that chip and do not change the straps (this is what we did in the past). On the 27c040, pin 1 is VPP . (VPP is the programming line - set to 12.5+ volts to program the chip). Per the data sheet, VPP can be either 5 volts or 0 volts for normal read operations. VPP is left floating in the production units - probably OK since it does not matter if VPP is hi or low for normal read operations.  Just FYI, the chips also use pin 24 differently. It is OE- on 27c080 and is OE-/VPP on the 27c040.
It is VERY important to set the straps correctly:
For the 27C040:
* JP 7 MUST BE OPEN - Arnold 1 Meg address line not connected!
* JP 5 is an unknown. However, leaving it open obviously works as that is the way all the production units are configured. Closed would connect VPP to + 5 volts - which should also work. The 27c040 data sheet says VPP can be either 0 volts or 5 volts. CMOS logic inputs should be hi or low. Perhaps Kurzweil left it floating to save power.

For the 27C080:
* JP 7 MUST be OPEN - Arnold 1 Meg address line not connected!
* JP 5 SHOULD BE CLOSED! Our upgrades have worked "most of the time" when JP 5 is left OPEN.  This leaves A19 floating on the chip.  I do not know what address this selects (is open a 'true' or 'false'). It may depend on the chip manufacturer. Closing JP 5 will connect A19 to + 5 volts and will firmly select the high copy of the the data we burn onto the chip (see more below). One user had to 'play' with JP 5 in order to make the upgrade work. Now we know why!

Can you MIX 27c080 and 27C040? Yes - normally. We have been mixing chips and did not realize that we were doing so.   The factory installed ROMS are the smaller 27c040 family chips.
To Mix both type chips (Kurzweil 512K X 8 chips and added I Meg X 8 chips):
JP 5 Should be CLOSED. That would select the high address upper copy of the sample data (see below). Many upgrades worked leaving JP 5 open. I guess that in that case each chip would use either the lower or upper copy of the sample data and we would not know.

Using 512k Sound Roms:
A user asks:
If I understand correct, 512Mb eproms (512kx8 - 27c040) could do the job? The data for burning is 1 Mb. Can someone clear this out?

Webmaster: Great question! I looked at the data files used to burn these chips (on this web site). It turns out the data from 00000 to 7FFFF Hex (first 512K bytes - 4Megbits) is the same as data from 80000 to FFFFF (second 512K bytes). So, If the 1Meg X 8bit chips (27c080) are burned using these files, the sample data is simply duplicated at the low and high addresses so turns out you could pull pin 1 (A19) high (or low) and still have the same sample data addressed. How we actually ended up having two copies in the file will remain a guru secret :)  Knowing what we know now, there is no need to burn the 1Meg X 8 chips anymore except - we have some in stock so may as well use them up - and until we get the data files for the smaller chips it will take some special skill to prepare the data for the burning of the smaller chips.

To burn the 27C080 .. chips use the data on the Yaho site (as of March 2010). If there are problems, try closing JP 5. To burn the 27C040 (512K X 8) chips Use the alternate set of data.  (512K data) which is just the first half of the larger files.

512K Burn Data is available on the Yahoo User site here:    512Kx8bit Burn Data