"New Drives" for MIO

[+bob1200xl]

... for some reason, I thought you were working with new, high-speed drives. Yes, that is the 80-pin connector you want. It looks like it will terminate properly.

 

Biggest problem is that the new Cheetah LC drives do not have jumpers on them. Old ones do, not new ones. You have to get an LW drive if you want jumpers - a 68-pin device. I'm going to try an old Cheetah on one of my 256K MIOs and see if we really need termination. Parity is a different story.

 

Bob

 

 

MonoPrice.com has an adaptor (#268) with 16 bit termination for $8. Looks like it would work OK if you were the last device on the 50 pin buss. I don't know how you mix a random set of devices.

 

Bob

 

Hi Bob-

 

I've never seen an 80-pin drive, but I presume that they have male connectors (w/pins)? If you click "enlarge" at the Monoprice.com listing and then look at the reverse side of the adapter, does that look like it mates to an 80-pin drive ok? Pardon my ignorance, but I've never dealt with anything but 50-pin SCSI.

 

So this device would plug directly the back of the drive, then I could use a regular IDC-50 cable to go to the MIO?

 

Thanks,

Larry

[+Larry]

... for some reason, I thought you were working with new, high-speed drives. Yes, that is the 80-pin connector you want. It looks like it will terminate properly.

 

Biggest problem is that the new Cheetah LC drives do not have jumpers on them. Old ones do, not new ones. You have to get an LW drive if you want jumpers - a 68-pin device. I'm going to try an old Cheetah on one of my 256K MIOs and see if we really need termination. Parity is a different story.

 

Bob

 

Hi Bob & kurtm-

 

I just learned what the SCA really is -- thanks! I really had wondered what the odd number of pins were; 68 versus 80.

 

Bob, if you could get out your MIO and try a couple of drives, that would be absolutely great!

 

-Larry

[+Larry]

... for some reason, I thought you were working with new, high-speed drives. Yes, that is the 80-pin connector you want. It looks like it will terminate properly.

 

Biggest problem is that the new Cheetah LC drives do not have jumpers on them. Old ones do, not new ones. You have to get an LW drive if you want jumpers - a 68-pin device. I'm going to try an old Cheetah on one of my 256K MIOs and see if we really need termination. Parity is a different story.

 

Bob

 

I should have added to the previous post (but can't edit now)...

 

It's not that I don't want to use newer drives, but lack of knowledge about them has been quite a hindrance. If this style of adapter works, I can foresee using my SUN 68-pin case with the 80-pin drive and still connecting to the 50-pin MIO. BTW, I do use SATA, so I'm at least up-to-date there!

 

-Larry

[+bf2k+]

I do have an IBM 80 pin SCA drive but it cannot provide termination power to the SCSI bus, so I'll have to do WareRat's MIO trmpwr mod before I can try it. I have the components to do the mod - maybe tomorrow.

 

All of my tests have been done with Narrow drives to this point...

Edited May 14, 2009 by bf2k+

[+Larry]

Bf2k+ kindly provided this RWTEST info:

 

SDX 4.42 - ST32430N

 

RWTEST.COM

 

256b

w: 11415.4566

r: 22312.029

a: 16863.7428

 

512b

w: 22964.4275

r: 40905.3866

a: 31934.907

 

Note -- slightly faster than the original MIO with MFM (256-byte) and 1.1 firmware; slower than the ST15150N and ST31055N and Warerat's 80-pin drives.

 

-Larry

[+Larry]

High-byte termination?

 

I understand the need for high-byte termination since our MIO's are 8-bit (Narrow) devices, but it seems logical that any 68- to 50-pin adapter does not carry the added (Wide) byte and it's parity bit. Under this situation, do we even need to be concerned about the high-byte termination since it is effectively truncated? Or are there other considerations?

 

Edit: found a source that says these are simply adapters and do not provide high-byte termination -- the high byte simply floats.

http://www.mrichter.com/cdr/files/68_50_n.htm

 

This seems like an important point, since as far as I can tell, 68-pin drives are widely available, but when I check with suppliers of 68- to 50-pin adapters, they simply say (to the effect) "this is an adapter not a terminator."

 

This also relates to the question to Warerat in the "New MIO production run" thread, since studying his pictures, I couldn't see any high-byte termination method other than the passive 68- to 50-pin adapter. Perhaps I'm missing something in the pics or perhaps the SUN enclosure provides the termination, powered from the bus from the MIO?

 

Or perhaps it is provided in the cable?

 

Edit 2: After looking at many search results, I found this adapter with termination, and it looks like what we need. For example, this would plug into the back of the SUN enclosure's 68-pin female connectors and provide the 50-pin IDC ribbon cable path to the MIO.

http://www.ramelectronics.net/computer-par.../prodSM058.html

 

-Larry

Edited May 17, 2009 by Larry

[+warerat]

This also relates to the question to Warerat in the "New MIO production run" thread, since studying his pictures, I couldn't see any high-byte termination method other than the passive 68- to 50-pin adapter. Perhaps I'm missing something in the pics or perhaps the SUN enclosure provides the termination, powered from the bus from the MIO?

 

You can't see it from the picture but my adapter is a Datamate SP5000-5068-13 I pulled off some old Compaq gear. It isn't just a converter-- it has resistor networks inside that terminate the high byte. I'm not saying this is the one you need to use, it's just one of many that does what needs to be done. The whole concept of high-byte termination is so when you connect a 16-bit device to an 8-bit bus, the upper data bus won't float and pick up noise. Termination in this case means the upper byte of the data bus and its parity line is tied high. I'm pretty sure I already posted a link about this. See here: http://www.scsita.org/aboutscsi/SCSI_Termi...Tutorial.html#4

 

You really need to refer to this document. It explains all the caveats of connecting different types of devices, specifically sections 4 and 6 are relevant to connecting newer devices to the MIO.

 

Or perhaps it is provided in the cable?

 

Nope, regular cable.

[+Larry]

You can't see it from the picture but my adapter is a Datamate SP5000-5068-13 I pulled off some old Compaq gear. It isn't just a converter-- it has resistor networks inside that terminate the high byte. I'm not saying this is the one you need to use, it's just one of many that does what needs to be done. The whole concept of high-byte termination is so when you connect a 16-bit device to an 8-bit bus, the upper data bus won't float and pick up noise. Termination in this case means the upper byte of the data bus and its parity line is tied high. I'm pretty sure I already posted a link about this. See here: http://www.scsita.org/aboutscsi/SCSI_Termi...Tutorial.html#4

 

You really need to refer to this document. It explains all the caveats of connecting different types of devices, specifically sections 4 and 6 are relevant to connecting newer devices to the MIO.

(snip...)

 

Thanks, Warerat-

 

After looking at a bunch of 68- to 50-pin adapters, I found a couple last night that looked like the passive adapters, but did contain the resistor network (like yours). I also got a nice reply from a seller on eBay who clearly understood SCSI connections, although he unfortunately didn't sell what I needed.

 

I'll study the docs in the link you provided. I do understand the need for HBT, and have found several other good docs about connecting Wide and Narrow devices. Slowly, but surely, I'm getting a handle on this newer SCSI stuff. I've got some parts on order -- we'll see how much I've learned!

 

-Larry

[+bf2k+]

I'll study the docs in the link you provided. I do understand the need for HBT, and have found several other good docs about connecting Wide and Narrow devices. Slowly, but surely, I'm getting a handle on this newer SCSI stuff. I've got some parts on order -- we'll see how much I've learned!

 

-Larry

 

Even with all this, isn't it true that we still have to provide trm power to the SCSI bus? If the drive can't do it, the MIO has to... correct?

 

In all of my 'N' drive testing, I use that jumper on the drive since I have not modified the MIO yet...

 

On the few LC and W drives I have messed with, I have had zero success. I suspect I have not gotten the HBT done correctly and have not provided trm power to bus.

Edited May 19, 2009 by bf2k+

[+Larry]

While I am waiting for more drives & adapters to test with the MIO, I hooked up my trusty old MFM drives with the Adaptec bridge-board. These are low-level formatted to 256-bytes/sector. I have both the ST251 and ST125 and used with the original 1.1 firmware. The 251's are actually quieter than the 125's -- wouldn't have thought that, but both my 125's have a distinct "whine" to them.

 

Anyway, I ran drac030's RWTEST on both and got very similar numbers -- not sterling, but definitely better than my old ST157N.

 

Writing 11152

Reading 11228

Average 11190

 

Hey, MEetalguy66 -- sometime could you run RWTEST on your setup using the older embedded SCSI drives that are formatted to 256 bytes/sector. Just curious if you see a speed penalty having them "half-sector formatted." With the ST157N and the 1.4Bx firmware, there is a significant penalty, only doing RWTEST of around 6000 bytes/sec. on both Reading and Writing.

 

-Larry

[+bf2k+]

OK one more...

 

ST15230N

 

RWTEST

512b/s

w: 37758.8184

r: 40905.3866

a: 39332.1025

 

256b/s

w: 20777.3392

r: 22830.9134

a: 21804.1263

 

This drive was a breeze to setup and configure and very quiet so I thought it was going to be a great one, but the drive appears to have TONS of bad sectors on it (or some type of problem that looks like bad sectors). Oh well... maybe a LL format would fix it.

[+Larry]

What is the verdict on external terminators?

 

I don't want to modify my existing MIO to provide termination power, so (in general) do add-on terminators provide a workable solution for drives without termination jumpers?

 

Thanks,

Larry

[+Larry]

OK, this AM I got a "new drive" working with the MIO & 1.4B3. It is an ST118273LC -- same model as one of Warerat's drives. My setup is a little different, so I'll post the method.

 

First, the jumpers: Parity Disable and Force Single Ended are jumpered; no others.

 

I'm using an 80 to 68/50 pin adapter of the type previously mentioned by bob1200xl. The seller of this stated that this does provide high-byte termination in Narrow mode.

 

Since the MIO is 50-pin and all my cables are 50-pin ribbon, I'm using the 50-pin portion of the adapter with an (apx) 3-foot cable to the MIO. (Currently no other terminators on the line).

 

I had trouble initially getting this setup to work reliably, but then I noticed that the back side of the 50-pin portion of the adapter was perilously close to the drive -- probably touching it part of the time. So I put a strip of electrical tape on those contacts and tried again.

 

This time, it was successful and stable.

 

My throughput is a little lower than I expected, but still very acceptable. (This is a 7200 rpm, 512 byte/sector Barracuda drive.)

Writing: 11384

Reading: 13237

Average: 12311

 

That is using MyDos (which is a bit slower than Sparta). This is almost identical to the throughput using my old MFM drive with 256 bytes/sector. Another indicator is the time it takes to sector copy my HD image to the new partition using my Action! sector copy backup program. This took 23 minutes and 9 seconds versus just a bit under 23 minutes with the original setup. But the bonus is the this drive is much quieter than the old ST125 which has a distinct high-pitched "whine" that I find particularly annoying.

 

I will more than likely add some type of termination to the drive just because it is good practice, but at least for now, things are looking good! Hooray!!!

 

-Larry

[+bf2k+]

Edit 2: After looking at many search results, I found this adapter with termination, and it looks like what we need. For example, this would plug into the back of the SUN enclosure's 68-pin female connectors and provide the 50-pin IDC ribbon cable path to the MIO.

http://www.ramelectronics.net/computer-par.../prodSM058.html

 

-Larry

 

Is this the one you are using?

 

Are you using an additional 'active' terminator as described in the link above?

 

I also have one of these drives and would like to get it working.

[+Larry]

Addendum for the previous post on the ST118273LC:

 

One difference that I've noticed is a longer power-up time with the newer drive. It seems to go through a number of gyrations before the "busy" LED goes off after powering the drive. I suspect it is waiting for some signal from the interface or doing calibrations and self-diagnostics, but it does take close to 30-seconds to complete from a cold start. If you don't wait, the MIO will not recognize the drive, and if you try to boot the Atari, you will end up at the self-test screen.

 

With this drive, you must power the drive first and wait for the busy light to go out. Then power the MIO (busy light will flash on for a couple of seconds). Then power the computer.

Following this routine, it boots properly every time.

 

-Larry

[+Larry]

No, it is an adapter just like this:

http://cgi.ebay.com/SCA-80-PIN-TO-68-50-PI...93%3A1|294%3A50

 

That was not my seller, but it was another Hong Kong seller.

 

There is currently no other terminator on the cable or at the drive. Also, there are no jumpers set on the adapter -- only the two on the drive.

 

Also, if you scroll down on the eBay page (link), you can see the solder pins protruding on the adapter. These were (I think) contacting the edge of the drive before the electrical tape was added. Hard to say if that was the problem, but the tape is cheap insurance.

 

-Larry

 

 

Edit 2: After looking at many search results, I found this adapter with termination, and it looks like what we need. For example, this would plug into the back of the SUN enclosure's 68-pin female connectors and provide the 50-pin IDC ribbon cable path to the MIO.

http://www.ramelectronics.net/computer-par.../prodSM058.html

 

-Larry

 

Is this the one you are using?

 

Are you using an additional 'active' terminator as described in the link above?

 

I also have one of these drives and would like to get it working.

Edited July 5, 2009 by Larry

[+Larry]

Here's an updated installation.

 

I have mounted the ST118273LC in a SUN HD enclosure of the same type as Warerat used. (Imitation is the sincerest... or alternatively, why re-invent the wheel? ) This model has the 80-pin SCA internal connector and dual 68-pin external connectors. Although I have a 68-pin cable ordered (with active termination), I still only have 50-pin cables. So I used a 68- to 50-pin adapter on the input port of the enclosure. The system works fine -- just like the other 80- to 68-/50- pin adapter method.

 

As soon as I get my termination power mod completed and cable received, I'll report what difference that makes in the RWTEST results.

 

*I'm having trouble understanding one thing about this whole termination situation -- these are non-parity systems, so how would the drive or controller know if it got bad data?*

 

-Larry

[+Larry]

Here's an interesting one, I think.

 

I have several ST118273LC drives.

 

Everything else being equal, here are two sets of speeds from RWTEST.COM (3.5):

 

"Drive #1" (50-pin ribbon cable)

Writing: 11384

Reading: 13237

Average: 12311

 

"Drive #2" (50-pin ribbon cable)

Writing: 9648

Reading: 12626

Average: 11137

 

"Drive #2" (68-pin ribbon cable)

Writing: 9601

Reading: 11448

Average: 10525

 

Edit:

Here's Drive #2 using an apx. 8" 50-pin cable

Writing: 9648

Reading: 14384 <<---

Average: 12016

 

Again, repeatable differences and enough here to notice the difference during boot-up, etc.

End Edit:

 

These aren't huge differences, but the results are very repeatable. Perhaps the firmware is slightly different in the two drives. (?) Then there is the difference between the 68-pin and 50-pin cables (each with it's required adapter.) Again, not a great difference, but the difference is real and repeatable.

 

Last, if you look at bf2k+'s speeds, he gets several thousand bytes/sec. faster than my speeds using same drive (ST31055N), same dos. Strange. I have another MIO which I'll test soon and add that to the mix. That one has the termination mod, so we'll see if that changes things.

 

-Larry

Edited August 13, 2009 by Larry

[MEtalGuy66]

YEah.. If youd get your termination correct, you"d get at least 3 times those numbers.. Warerat has repeated his speed results on many different drives.. Proper termination is the key factor..

[+Larry]

YEah.. If youd get your termination correct, you"d get at least 3 times those numbers.. Warerat has repeated his speed results on many different drives.. Proper termination is the key factor..

 

Hi Ken-

 

I certainly hope you are correct, but...

 

Can you or Warerat speculate how termination can be a factor on a non-parity system?

 

How can bfk2+ and I run the same drive/cpu/cable/Dos (ST31055N) terminated in the same way and get significantly different results?

 

Hopefully, I'm going to be able to test the ST118273LC very soon with the termination mod in place.

 

OTOH, I'm intrigued that the read (but not write) speed went up so much when I went to an ultra-short cable (apx 8") between MIO and Drive. Maybe there are "soft errors" that are being corrected in the drive and that is slowing things down. (?) That would seem plausible for data *to* the drive, but not *from* the drive. Confusing -- to me, at least.

 

Anyway, as soon as I can, I'll test the system(s) as you and Warerat have suggested.

 

-Larry

[MEtalGuy66]

It's simple.. You have a termination issue.. Therefore, one or more lines is "floating" and does not alwayse produce the correct "state" from one device to the other.. Any time you are talking about data transfer, certainly longer cables introduce more opportunity for inductive noise, additional capacitance on the circuit, etc.. This does not change the fact that if the bus was 100% terminated correctly, you could use a cable of any length (within SCSI specs) and have far fewer bus errors and thus much higher transfer rates. This is all assuming that the drive you are using does not have compatability issues with the MIo in geenral.. If BF2k+ is getting similar throughput to Warerat's, and you are using the same exact model drive, then I suspect that not to be the case..

 

I'll tell you what we could do.. I havent shipped your MIO back to you yet.. You mail me the drive in question, and I'll make it work at full speed.. Then I'll send you back the drive, the MIO, and the cable, set up exactly the way it needs to be..

[+warerat]

Can you or Warerat speculate how termination can be a factor on a non-parity system?

 

Termination has to do with the signal quality. If the drive is marginally terminated, it may or may not work (if the cable is really short then your chances increase that it will work). Termination affects not only the signal quality but also the delay and settle time of the signals. When the drive selection occurs (asserting SEL- and negated target ID on the bus) there's a delay between the time the target ID sees the select request from the host (MIO) and a delay between the drive asserts its acknowledgement. The termination (or lack of) can affect those times and affect how quickly the drive responds.

 

SCSI parity is practically useless on the Atari (BB or MIO) because neither the hardware or software is providing a mechanism to recover or detect a parity error. The only practical use is to allow a newer parity-only drive to be used. If you're familiar with all these 'LS280 hacks, all they are doing is sending the parity bit to the drive. If the drive hardware really did detect an error, it lets the host know through a message mechanism that neither the BB or MIO implement in hardware. Remember, these are minimal interfaces to SCSI.

 

Termination is always a factor-- it is part of the physical design of the SCSI bus. If your signal path is bad, parity or non-parity, it's not going to work as expected. Termination isn't optional. Parity is.

Edited August 14, 2009 by warerat

[+Larry]

It's simple.. You have a termination issue.. Therefore, one or more lines is "floating" and does not alwayse produce the correct "state" from one device to the other.. Any time you are talking about data transfer, certainly longer cables introduce more opportunity for inductive noise, additional capacitance on the circuit, etc.. This does not change the fact that if the bus was 100% terminated correctly, you could use a cable of any length (within SCSI specs) and have far fewer bus errors and thus much higher transfer rates. This is all assuming that the drive you are using does not have compatability issues with the MIo in geenral.. If BF2k+ is getting similar throughput to Warerat's, and you are using the same exact model drive, then I suspect that not to be the case..

 

I'll tell you what we could do.. I havent shipped your MIO back to you yet.. You mail me the drive in question, and I'll make it work at full speed.. Then I'll send you back the drive, the MIO, and the cable, set up exactly the way it needs to be..

 

Hi Ken and Warerat-

 

Thanks for the explanations and kind offer, but I'd rather not ship out more parts. The sooner I get the mod'ed MIO back, the sooner I can test it.

 

If the term power mod works as you state, then that is that. Warerat has laid out a clear path.

 

-Larry

Edited August 15, 2009 by Larry

[+bf2k+]

I'll tell you what we could do.. I havent shipped your MIO back to you yet.. You mail me the drive in question, and I'll make it work at full speed.. Then I'll send you back the drive, the MIO, and the cable, set up exactly the way it needs to be..

 

I have the same drive... since I was going to ship you my MG66 MIO this week, you want me to ship the drive too?

[MEtalGuy66]

I'll tell you what we could do.. I havent shipped your MIO back to you yet.. You mail me the drive in question, and I'll make it work at full speed.. Then I'll send you back the drive, the MIO, and the cable, set up exactly the way it needs to be..

 

I have the same drive... since I was going to ship you my MG66 MIO this week, you want me to ship the drive too?

 

Sure.. Sounds great.. But to be honest, Im not touching any more projects until after Ramadaan (which starts thursday or friday, depending on the visibility of the moon). It's a month long.. So, you can go ahead and ship it, and Ill let you know when I get it, but it wont get touched till Ramadaan is over..