• Hello, Guest! Welcome back, and be sure to check out this post for more info about the recent service interruption and migration.

Missing SE/30 SCSI Controller Chip

cannfoddr

Well-known member
I picked up a 'spares' SE/30 logic board from ebay - OK I didnt look closely enough at the pictures, live end learn, and discovered once I started working on it aht the SCSI controller chip is GONE!

Looking at another board I can see an NCR 53C80 p/n 609-3400257 a few google searches are not turning up much.

I have been able to find some NCR 53C80-40 609-3400456 which seem to have the same packaging and look like they would fit, but no idea if they would work

Any thoughts from the experts?
 

LaPorta

Well-known member
I purchased two or three a few years ago as spares (can't remember where from, eBay probably). I can check them tonight and get some photos.
 

bdurbrow

Well-known member
According to the datasheet; the 53C80-40 is just a higher speed version of the 53C80. As long as they are the same form-factor, I think it will work without a problem.
 

cannfoddr

Well-known member
I believe that these SCSI chips are still available for sale from Mouser electronics --I bought one brand new last summer.


There is a datasheet PDF accessible through the link above --it should have a lot of useful info about the SCSI chips.
Thanks for this - I didn't realise that 53C80 has different manufacturers. I will add one to my component order for this logic board repair
 

chillin

Well-known member
According to the datasheet; the 53C80-40 is just a higher speed version of the 53C80. As long as they are the same form-factor, I think it will work without a problem.
The SCSI-1 bus on the SE/30 is a crushing bottleneck. Last few weeks I have been trying to think of a way to increase its bandwidth. I had RAID in mind. That bus will take 6 or 7 (I thought SCSI was limited to 6, but I saw something that claimed the SE/30 SCSI bus maxed at 7) storage devices, and a 7 device RAID 0 could provide 7 times faster read/write speeds (granted, that may only be 35MB/s, but I'd take it). I am, however, unaware of any RAID card for SE/30 PDS, or IIsi, etc., nor am I aware of any softRAID from that era. My hope was maybe NetBSD had a software RAID available in the low version numbers.

But then I stumble in here and see someone is talking about replacing the SCSI chip, and that some chips (I assume all are still SCSI-1) are faster than others. Has there been any effort to increase the storage bandwidth on SE/30? Could it be as simple as replacing the SCSI-1 controller chip with (dare I dream?) a SCSI-2 chip of the same size and pinout (if extant)? Since cannfoddr's issue has resolved, may I please have some 68kmla minds on what possibilities may exist towards increasing storage bandwidth in an SE/30?
 

bdurbrow

Well-known member
Faster in this context translates to the chip being able to support a higher clock rate. However, the SE/30 does not supply that higher clock; and I think that it wouldn't be easy to get it to do so. It's been months since I looked at that datasheet, but IIRC the difference was only a few Mhz between the different versions.

Something modern and solid-state like a SCSI2SD should be able to saturate the built-in SCSI bus and have essentially zero seek time.

Building a faster PDS card is do-able; but would be a non-trivial project (you'd have to design the hardware; write firmware for the CPLD and possibly FPGA chips; possibly write firmware for an ARM microcontroller that handles the low-level interface to a SD card or some other flash chip; and write a Mac driver to allow it to be bootable).
 

chillin

Well-known member
Thank you for the reply. Good to learn SD2SCSI saturates bus bandwidth, though I suppose that means there could be no r/w speed gain using more than one bussed SD2SCSI in a hypothetical softRAID 0.

I am not clear on RaSCSI. I know it emulates SCSI, but I don't understand what actual the physical storage devices are with RaSCSI. Initially I thought it was a way to mount and read old SCSI drives using just the RaSCSI with no Mac, effectively a way of getting to HFS data on a naked HDD without a Mac, or even with a Mac available, still useful when one has a stack of naked SCSI drives with no idea what's on them and no motivation to open the Mac up and temporarily install HDD. I thought this improbable, but pretty neat. But I just saw a video of an SE/30 booting off a RaSCSI, and now I'm utterly confused, because RaSCSI seems now a more complicated method duplicating SD2SCSI functionality. I suppose RaSCSI slightly predates SD2SCSI.

Obviously SD2SCSI uses SD or micro SD cards. Does RaSCSI only use the Raspberry Pi's onboard micro SD storage? or is there some other configuration that includes physical HDD? afaik, there's no Pi with a SATA bus. I'd appreciate a brief explanation or a link to a complete and accurate explanation of what RaSCSI is (development thread is too looong).

Let me get my SE/30 SCSI facts straight because my investigation is old and my memory not so good.

My original understanding was that the SE/30 internal SCSI bus was fast SCSI, SCSI-2 "narrow", max bandwidth 10MB/s, but the external SCSI port bandwidth maxed at 5MB/s (SCSI-1 narrow). But now I'm not sure that is correct... no SE/30 SCSI HDD benchmark I have seen exceeds 1.5MB/s, and somewhere I read that all the SCSI bus in SE/30 was SCSI-1 narrow, 5MB/s. If the internal bus is fast SCSI 10MB/s, there may be a moderately attractive benefit to RAID 0 with 1990s HDD, but if it's all 5MB/s narrow SCSI bus speed in SE/30, there's half the room for increasing bandwidth. I guess what I need to know is if it is clear there is no old narrow or fast SCSI spinning disk that can compete with the bandwidth of SD2SCSI or RaSCSI. SD2SCSI seem to use fast SCSI, SCSI-2 "narrow," which would explain bus saturation if the SE/30 internal SCSI bus is SCSI-1.

(unrelated, recently I realized how much SATA III sucks, maxing at 600MB/s, which USB 3.0 already exceeded in 2008, while SATA III only appeared in 2009, and yet SATA III is still the ubiquitous onboard storage bus on modern logic boards 12 years later and somehow still competing against PCIe SSD, which itself appeared in 2007 and yet is still the exception rather than industry standard... but I suppose I have a long time to wait before 68kmla engineers fix SATA III, if ever, due to it having nothing to do with any 68k machine... /rant)
 

trag

Well-known member
The company "Logic" claims that their 53C80 is executes SCSI transactions more efficiently and yields higher bandwidth. It's tempting to get some and install them them on old Macs to test that out.
 

Attachments

  • Logic 53C80.pdf
    334.4 KB · Views: 5

Fizzbinn

Well-known member
The company "Logic" claims that their 53C80 is executes SCSI transactions more efficiently and yields higher bandwidth. It's tempting to get some and install them them on old Macs to test that out.
That would be too cool if it worked! At some point Apple started using their own Combo? chip instead of stock NCR SCSI chips, I guess this could be a potential for Macs before then, Mac II series?
 
Last edited:

trag

Well-known member
Yes, pretty much the Mac II series. The PowerBooks used a combo 53C80/85C30, which is also apparently on the LC - 85C80 from AMD, I think.

I think the Quadras used a 53C96, except the AV machines which used the AM79C950 which is SCSI/Serial and ethernet in one chip. That chip was used through the X500/X600 machines.

Interesting thing about the 85C80 (if I have that number right) is that the data sheet shows that it is a 53C80 and an 85C30 just sharing most of their input pins. It could literally be replaced by the separate chips wire properly. If there's physical space for it.

Which means that if the Logic 53C80 provides performance improvements, it might be possible to wedge it into those machines using the 85C80 as well.
 

chillin

Well-known member
from the Logic 53C80.pdf
Asynchronous Transfer Rate Up to 4 Mbytes/sec
offering up to a 2.5x perform- ance improvement, 10x power reduc- tion,

Also seems to answer my question about bus speed... has to be SCSI-1, 5MB/s.

Except I just read this at Low End Mac
In 1987 Apple introduced the Mac SE and Mac II with improved SCSI implementations based on the newer SCSI-2 specification.
...which again has me believing the SE/30 internal SCSI bus is Fast SCSI-2, 10MB/s.

By whatever method, I want that 10MB/s. 1-5MB/s is no way to live.
 

cheesestraws

Well-known member
LEM, to quote, contains "much that is apocryphal, or at least wildly inaccurate". I would believe the datasheet for the chip.
 

Fizzbinn

Well-known member
Except I just read this at Low End Mac
...which again has me believing the SE/30 internal SCSI bus is Fast SCSI-2, 10MB/s.

While the SCSI-2 specification defined modes with faster than 5MBps operation (Fast 10MBps, Wide 10MBps, Fast+Wide 20MBps) it also improved/formalized regular narrow 5MBps SCSI. Being a SSCI-2 device doesn’t mean you necessarily support faster speeds.

I don’t think Apple offered a Mac with built in Fast SCSI until Power Macintosh:
 
Last edited:

max1zzz

Well-known member
The company "Logic" claims that their 53C80 is executes SCSI transactions more efficiently and yields higher bandwidth. It's tempting to get some and install them them on old Macs to test that out.
Huh, I didn't know that. I actually have a logic 5380 in my reproduction SE board (It was the cheapest NOS 5380 on ebay at the time!) If I can find the time one evening I'll try and do some disk benchmarks and see if they preform any differently to the AMD 5380's
 

ymk

Well-known member
Being a SSCI-2 device doesn’t mean you necessarily support faster speeds.

Right. SCSI 1/2 are collections of standards and no device implements all of them.

By whatever method, I want that 10MB/s. 1-5MB/s is no way to live.

Trying to improve the I/O performance of a 16MHz SE/30 is probably a lost cause.

JDW benchmarked SE/30 disk performance, showing significant gains with 030 and 040 accelerators.

Timing the duplication of a large file on a ramdisk could reasonably measure the CPU/RAM bottleneck.
 

chillin

Well-known member
I don’t think Apple offered a Mac with built in Fast SCSI until Power Macintosh
But for the WGS 95, with the fast Apple SCSI card, not built-in, PDS, yet OEM.

Thanks for the links Fizzbinn... stealing that rather upsetting first page of the table from the second link. I'm having trouble believing it. The SE/30 internal SCSI-2 bus is limited to 1.25MB/s by the 53C80? Or that's just a real world expectation?

Trying to improve the I/O performance of a 16MHz SE/30 is probably a lost cause.
I'm beginning to see that. :\

JDW benchmarked SE/30 disk performance, showing significant gains with 030 and 040 accelerators.
Ah... so it isn't the SCSI chip that is limiting, but the 16MHz 68030. That makes sense. 10BaseT ethernet also saturates that processor.
 

Attachments

  • 2AAD945E-EC91-4AE3-A8FD-91965DE592AB.jpeg
    2AAD945E-EC91-4AE3-A8FD-91965DE592AB.jpeg
    461.3 KB · Views: 9
  • 8A22737E-E44F-4B5D-AEC2-D74964FC9F5B.jpeg
    8A22737E-E44F-4B5D-AEC2-D74964FC9F5B.jpeg
    502.4 KB · Views: 10
  • 29B22D5D-6EBF-46F1-AC4B-68A0871D76A5.jpeg
    29B22D5D-6EBF-46F1-AC4B-68A0871D76A5.jpeg
    870.8 KB · Views: 11
  • 08717678-CF2A-4FA7-95FB-1ECC2A020E95.jpeg
    08717678-CF2A-4FA7-95FB-1ECC2A020E95.jpeg
    367.1 KB · Views: 10
Last edited:

TPope

Active member
The SCSI-1 bus on the SE/30 is a crushing bottleneck. Last few weeks I have been trying to think of a way to increase its bandwidth. I had RAID in mind. That bus will take 6 or 7 (I thought SCSI was limited to 6, but I saw something that claimed the SE/30 SCSI bus maxed at 7) storage devices, and a 7 device RAID 0 could provide 7 times faster read/write speeds (granted, that may only be 35MB/s, but I'd take it). I am, however, unaware of any RAID card for SE/30 PDS, or IIsi, etc., nor am I aware of any softRAID from that era. My hope was maybe NetBSD had a software RAID available in the low version numbers.

But then I stumble in here and see someone is talking about replacing the SCSI chip, and that some chips (I assume all are still SCSI-1) are faster than others. Has there been any effort to increase the storage bandwidth on SE/30? Could it be as simple as replacing the SCSI-1 controller chip with (dare I dream?) a SCSI-2 chip of the same size and pinout (if extant)? Since cannfoddr's issue has resolved, may I please have some 68kmla minds on what possibilities may exist towards increasing storage bandwidth in an SE/30?
Hey Chillin,

Check Wikipedia for SCSI2 for definitions. The only products I ever heard of SCSI acceleration was Jackhammer and OTTO's SEIV. But I don't I know how you want to use them they may be NuBus only. Once I had a SCSI accelerated PLUS with a TOTAL SYSTEMS Gemini so I know they exist. Good luck

TPOPE
 

chillin

Well-known member
Thanks TPope.

Unlike the Total Systems Gemini 68030 Accelerator for SE, the ATTO Silicon Express IV and FWB Jackhammer for NuBus are not accelerators, per se, but distinct and faster SCSI busses run through NuBus.

OP cannfoddr was missing the SCSI controller chip, and bdurbrow noted a faster version of the 53C80 SCSI controller, and I fixated on that and inquired if increasing the SE/30 SCSI bandwidth could be as simple as replacing the SCSI controller chip.

But, as ymk has suggested, it seems the issue with the SE/30 SCSI bus speed has more to do with the 16MHz 68030 than the SCSI-2 controller: increase the processor speed, the SCSI bandwidth will also increase, at least within SCSI-2 limits.

Though it would be pretty great to have a SE/30 PDS to NuBus adapter, I don't think the issue is with developing an adapter, but with fitting a NuBus card inside a compact mac. There are smaller NuBus cards, but the SCSI cards tend to be giants.
 
Top