Arduino SCSI device - Work in Progress

[erichelgeson]

On 4/5/2021 at 12:20 PM, androda said:

here's the prototype version of my PowerBook BlueSCSI device board

Nice! Also thanks for using a Non-Commercial license to keep in the spirit of this project!

[xunker]

On 4/5/2021 at 11:20 AM, androda said:

The two jumpers are for selecting which power source you use (L = logic, M = motor) and term power.  I have found that it doesn't work at all without termination enabled in my PowerBook 180.  Sleep mode is not tested, unsure if it'll work.  This board is definitely harder to put together because of the little surface mount components, but will have better battery life because of active termination vs resistor network termination.

 

 

I ended up doing the same thing  regard to logic or motor power with v1.1 of my board, so I'm glad I'm not the only one. Poring over all the old PowerBook TechNotes and usenet posts makes me *think* that some powerbooks turn motor power on an off to make the drive sleep but I can't find any place where its explicitly stated and what models it might apply to. The motor *should* be switched on/off using the "START/STOP UNIT" SCSI command, but since the drive in my PB145 lacks the "NO SPIN UP" pin as well, it makes me think that feature is not always implemented.

 

I like what you've done with the active termination, I shy'ed away from using a 2.85v regulator it because I didn't want to solder 18 of those 0603 or 0805 resistors .  I'm waiting on another board for my IIGS that'll use an SOIC or SSOP termination ICs instead. I'm especially curious if changing the termination resistance from 110 to 2.5k will help make these work better in problematic Powerbooks.

[androda]

On 4/6/2021 at 7:09 PM, xunker said:

I like what you've done with the active termination, I shy'ed away from using a 2.85v regulator it because I didn't want to solder 18 of those 0603 or 0805 resistors

 

Thanks for the compliment, but my design isn't technically correct yet.  The resistors can't be removed from circuit, which might cause issues in mid-bus usage even with termpower disabled.  It's more suited for end of bus where termination is required.  The next board revision will include some sort of bus switch for connect/disconnect of the term resistors.

 

Termination resistors don't have to be teeny-tiny.  You can use physically larger 110 ohm resistors, the normal through-hole style.  I've found that these little things are actually pretty easy with a solder paste syringe and heat gun.  Just use a pad sized for hand soldering and it works out fine.

 

Are SCSI termination ICs still in production?  What's the part number?

 

As for changing the SCSI termination resistors to 2.5k ohms, that wouldn't produce much termination pull-up current at all.  Only about a milliamp.  The 110 ohm resistance is aimed at producing 24 milliamps of termination pull-up current from a 2.85v termpower source, which is the SCSI standard for termination current.

 

 

[SuperSVGA]

20 hours ago, androda said:

Are SCSI termination ICs still in production?  What's the part number?

 

For something in production, you could try the TL2218-285PWR.

 

I personally use the DS21S07AE (NOS) because I can find it for cheaper, but it requires a few extra capacitors.

[androda]

20 hours ago, SuperSVGA said:

For something in production, you could try the TL2218-285PWR.

 

I personally use the DS21S07AE (NOS) because I can find it for cheaper, but it requires a few extra capacitors.

 

Thanks for the pointer.  Makes me wonder how much longer the 2218 will be manufactured.  How many brand new SCSI devices are being made these days, with everything being SATA and SAS?

[Stephen]

In case it's useful to anyone, here's a low-profile SCSI 50-pin to DB25 adapter board which makes it easier to directly attach the BlueSCSI (and maybe other SCSI devices?) to the external DB25 port. It's all open source so you should be able to modify it however you please!

Edited April 7 by Stephen

[xunker]

22 hours ago, androda said:

Are SCSI termination ICs still in production?  What's the part number?

 

As for changing the SCSI termination resistors to 2.5k ohms, that wouldn't produce much termination pull-up current at all.  Only about a milliamp.  The 110 ohm resistance is aimed at producing 24 milliamps of termination pull-up current from a 2.85v termpower source, which is the SCSI standard for termination current.

 

 

 

I'm using both the UCC5606 and DS21S07A mainly; I think the former is still in production, not sure about the later. If it comes to the worst case, there are many SCSI-2 and Ultra160/360 terminator ICs still in production that that retain SCSI-1/SE compatibility though they are starting to all be shipped as QFN or leadframe. They generally all require at least one capacitor per IC, and then another for decoupling TERMPWR. The UCC56xx family recommends tantalum, and the DS21* require tantalum.

 

The docs for the UCC56xx series say "The 110-Ω termination is used for standard SCSI bus lengths and the 2.5-kΩ termination is typically used in short bus applications", and you probably can't a shorter bus than in a laptop! No idea if it will really improve anything, so it's more of a "try and see" idea at this point.

[Dav]

Dear Fellows,
I would like to present you something I call The Poor Man Blue Scsi (PMBScsi)

Thanks to @erichelgeson, and with hints and moral support of @bibilit.
The idea is to use off-the-shelf components :
- DB25 connector
- STM32 and  micro SD Shield
- A few Dupont wires.

It doesn’t require soldering skills. Here is the details :

 

My two cents :
- My SD shield doesn't work with +3.3v, so  I connected it to +5v. I think that a lot of cheap SD shields suffer from the same problem.
- I sligthly modified the code on line #30, because i can’t access to SD card with SPI_FULL_SPEED (perhaps again because of my cheap SD shield). I put 18 Mhz :
#define SD1_CONFIG SdSpiConfig(PA4, DEDICATED_SPI, SD_SCK_MHZ(18), &SPI)

Tested on Macintosh Classic II.

[mogs]

20 hours ago, Dav said:

- My SD shield doesn't work with +3.3v, so  I connected it to +5v. I think that a lot of cheap SD shields suffer from the same problem.

 

That shield has a voltage regulator and level converter on it, this intended for interfacing to 5V micros. I'm pretty sure that means it will be pushing 5V into pins on the BluePill that are not 5V tolerant.

 

If you search "SparkFun microSD Transflash Breakout" or "wemos Micro SD Card Shield" you'll find a couple of modules that don't include regulators or level converters.

 

[androda]

Tested the updated PowerBook design on a 540 and 520 w/ppc upgrade (my brother's laptops). They booted just fine on the rascsi 7.5.3 image. Will be posting the new design in my repo later today.

 

This morning I finished assembling 8 of the updated design, and will soon be making a trading post thread to list them for sale. No sales website yet, that's another thing I need to do (and I'm pretty bad at web stuff).

[erichelgeson]

On 4/9/2021 at 8:16 AM, Dav said:

I would like to present you something I call The Poor Man Blue Scsi (PMBScsi)

You may be surprised to know I've seen an even more "poorer" implementation - someone wired directly to a 50 pin ribbon cable and soldered the wires directly to a microsd->sd card holder.... If you ever find yourself in that situation I'm happy to send you a board.

[blindowl]

If I want to daisy chain the BlueSCSI (connected externally) with another SCSI device, what are my options then? Are there SCSI splitters?

Edited April 11 by blindowl

[erichelgeson]

@blindowl I would either put it at the end of the chain - or pick up a vintage external case and drop it in that - they usually have a 50 pin to dual db25 connector or similar inside. I have also seen a stacked db25 connector but it was expensive if i remember correctly and seemed a bit flimsy - you'd have to build a mount for it.

[xunker]

On 4/9/2021 at 12:29 PM, erichelgeson said:

and soldered the wires directly to a microsd->sd card holder

 Kind of like this  ?

Microsd-to-sd adapters are cheaper than those SMD microsd sockets, so doing it this way let me validate that I got all the SD card pins correct before potentially wasting a microsd slot on a board design that didn't work.

 

(I actually did the exact thing you mention, too, on a board I got from you; microsd slots hadn't come in the mail yet and I wanted to try out the board so I soldered wires from a microsd-adapter to the pins on the back of the STM32 footprint and then attached it with electrical tape -- still working well a few weeks later!)

[blindowl]

If I connect the BlueSCSI with a DB25 extension cable (1.8 m) my SE/30s just stuck on a flashing ? at startup (ignoring the internal disk). But works with the Classic. I thought it could be a power issue so I hooked it up to USB power, but no change. Any idea what the issue can be? Crappy cable? Do I need to do any kind of modification to run the BlueSCSI on USB power?

Edited Friday at 05:36 PM by blindowl

[mogs]

On 4/12/2021 at 6:40 AM, blindowl said:

If I want to daisy chain the BlueSCSI (connected externally) with another SCSI device, what are my options then? Are there SCSI splitters?

 

It should be possible to use two of the 50pin header to DB25 adapters stacked together with extra long header connectors.

Either as long pins soldered to all three boards, or as stacked sockets.

 

[blindowl]

On 4/16/2021 at 7:34 PM, blindowl said:

If I connect the BlueSCSI with a DB25 extension cable (1.8 m) my SE/30s just stuck on a flashing ? at startup (ignoring the internal disk). But works with the Classic. I thought it could be a power issue so I hooked it up to USB power, but no change. Any idea what the issue can be? Crappy cable? Do I need to do any kind of modification to run the BlueSCSI on USB power?

 

After some trial and error with different setups, my conclusion is that the problem is due to a bad cable. A standard DB25 cable is not good enough. I think I need to get a thicker and better shielded, purpose-built SCSI cable.

Edited yesterday at 11:21 AM by blindowl

[cheesestraws]

20 hours ago, blindowl said:

A standard DB25 cable is not good enough. I think I need to get a thicker and better shielded, purpose-built SCSI cable.

 

That would do it.

[Chopsticks]

22 hours ago, blindowl said:

 

After some trial and error with different setups, my conclusion is that the problem is due to a bad cable. A standard DB25 cable is not good enough. I think I need to get a thicker and better shielded, purpose-built SCSI cable.

Unfortunately it can be hard to find an actual db25 scsi cable, there’s heaps of parallel port and serial port cables for pc that use the db25 connector and they aren’t straight through cables, ive had a bit of trouble finding scsi cables myself in the past and ended up just making some myself