PDA

View Full Version : IBM PC 63 watt supply and 3.5" MFM drives



cr1901
August 9th, 2013, 12:40 PM
I'm aware from reading previous posts on this forum that the original IBM PC power supply cannot support a full height MFM drive reliably- at least without the risk of causing unacceptable load. However, does anyone have any experience with trying a 3.5" drive such as (in my case), the Kalok Octagon? I would venture to guess that by the time the Octagon came out, power consumption of hard drives had decreased appreciably.

And I suppose I could try it- IBM power supplies in my experience, can take quite a bit of abuse.

According to this page, http://www.mfarris.com/hard_drives/kalok/kalok_kl330.html, the Kalok Octagon uses approximately 11 Watts.

glitch
August 9th, 2013, 12:47 PM
I used a HardCard workalike with one of my 5150s that had the original 63 Watt supply for years. It used a 3.5" FH stepper actuated 8-bit IDE drive, so probably not the lowest power 3.5" drive out there. Nowadays I'd go for an XT-IDE card with an industrial Flash module (IDE SSD).

Stone
August 9th, 2013, 01:27 PM
I've got a Teac SD-340 42MB IDE drive that only uses 5.1 Watts if that's of interest.

cr1901
August 9th, 2013, 01:32 PM
I've got a Teac SD-340 42MB IDE drive that only uses 5.1 Watts if that's of interest.
Don't have an 8-bit IDE controller... hence why the title says '3.5 MFM drives' :P

paul
August 9th, 2013, 02:09 PM
At only 11 watts I think there is a good chance there is enough capacity given that the PS rating refers to the continuous DC output. However I would bet the hard disk startup draw is about twice that, but that may not be an issue either as the PS would simply shut down if it can't support the load.

cr1901
August 9th, 2013, 02:13 PM
In any case, I've run it continuously for the past 30 minutes and performed a number of random read burn-in tests... the power supply doesn't appear to be complaining, so I guess the 63 watt power supply CAN support certain hard drives!

vwestlife
August 9th, 2013, 02:25 PM
Good luck finding a 3.5" MFM drive that actually still works! The Miniscribe and Kalok drives were built using cheap components with loose tolerances. The few Miniscribe drives that I have left all suffer from stiction -- the heads get stuck to the disk platters, and I have to rotate the interrupter to break them free before turning them on, or else they'll just blink an error code and won't spin up. And my last Kalok Octagon drive had read/write errors which got so bad that I couldn't even low-level format it anymore.

My late model 5150 PC, made in 1986, already came from the factory with a 130-watt XT power supply, so I put in a good ol' reliable 5.25" half height Seagate ST-225 drive, which just hums right along and works flawlessly. Even if the machine sits unused for a year or two, I can pull it out, fire it up, and the ST-225 will spin right up and work fine. I haven't even needed to low-level format again it since I first installed it in 1997 or so.

cr1901
August 9th, 2013, 03:16 PM
Good luck finding a 3.5" MFM drive that actually still works! The Miniscribe and Kalok drives were built using cheap components with loose tolerances. The few Miniscribe drives that I have left all suffer from stiction -- the heads get stuck to the disk platters, and I have to rotate the interrupter to break them free before turning them on, or else they'll just blink an error code and won't spin up. And my last Kalok Octagon drive had read/write errors which got so bad that I couldn't even low-level format it anymore.

My late model 5150 PC, made in 1986, already came from the factory with a 130-watt XT power supply, so I put in a good ol' reliable 5.25" half height Seagate ST-225 drive, which just hums right along and works flawlessly. Even if the machine sits unused for a year or two, I can pull it out, fire it up, and the ST-225 will spin right up and work fine. I haven't even needed to low-level format again it since I first installed it in 1997 or so.

My Kalok Octagon works just fine with no bad sectors other than those listed on the defects table (came with my XT). :) The components look like they're of middling quality, but not low quality- a mix of surface mount discrete components and ICs as well as through hole. Also RLL 30 MB- both standard for late 80's? I found out that you can confuse it by running it component-side up, and it will not recalibrate properly, though...

I also have a working Computer Memories hard drive for the PC AT that hasn't had any issues for the past 2.5 years. I've been fairly fortunate with early hard drives considered unreliable.

TTL monitors, on the other hand... not so much luck. I suppose that's a tradeoff in Murphy's Law of conservation.

Chuck(G)
August 9th, 2013, 03:24 PM
I've got an 80 MB hard card with an ST11R controller and a 3.5" drive all mounted on the same bracket. Works fine.

Why a 3.5" IDE drive and not just a simple CF, by the way?

Stone
August 9th, 2013, 03:54 PM
Why a 3.5" IDE drive and not just a simple CF, by the way?It's an MFM, not IDE... I already got blasted for that one! :-)

cr1901
August 9th, 2013, 03:54 PM
I've got an 80 MB hard card with an ST11R controller and a 3.5" drive all mounted on the same bracket. Works fine.

Why a 3.5" IDE drive and not just a simple CF, by the way?

Because I don't have an 8-bit ISA or 16-bit ISA/8-bit tolerant IDE controller card. Stone gets it :P! It is in fact as 3.5" MFM (RLL, actually) drive... I did not make a mistake on the thread title.

Chuck(G)
August 9th, 2013, 04:36 PM
Ah, okay, just checking.

However, finding a working 3.5" MFM drive is not going to be wasy. They weren't all that good back then.

k2x4b524[
August 9th, 2013, 08:16 PM
yea, don't knock the kalok until you've heard one chirp :P sucker works pretty good still. every drive manufacturer that made bad drives all the time, out of those, some are pretty decent. But if you find a NIB old hard drive, i'd be sketchy. the 63 watt should be able to handle 3.5" hdds ok, same with hard cards, anyone know how well the half height 5.25" drives work with it?

SpidersWeb
August 9th, 2013, 09:51 PM
However, finding a working 3.5" MFM drive is not going to be wasy. They weren't all that good back then.
Yep. I needed one for my 286 laptop, so had to go hunting. I had a 1 in 4 success rate with the easy to find ones, Rodime/Miniscribe/Fuji etc

But no matter the brand, unless its $5, always make sure you ask the seller if it spins up before buying one.
I was finding many have mechanical faults like heads that have not just stuck to the platters - but have "become one" with them.

As for "chirp of the year" - NEC D5126 wins hands down for me ;) I have a few, never miss a beat.

vwestlife
August 9th, 2013, 10:29 PM
anyone know how well the half height 5.25" drives work with it?

My original Tandy 1000 worked perfectly fine with an ST-225 (5.25" half height) drive, and it has a 54 watt power supply! However, that power supply may have better handling of the high current draw spike upon power-up, which is really what killed the 5150 power supplies, not continuous current draw. I believe it was an Astec power supply, which are generally very good quality, at least in my experience.

cr1901
August 9th, 2013, 10:30 PM
Ah, okay, just checking.

However, finding a working 3.5" MFM drive is not going to be wasy. They weren't all that good back then.
I happen to have a working one sitting on its to the left of my laptop :P. I tested it with the power supply (also to my left for easy access) earlier today and it works fine.


yea, don't knock the kalok until you've heard one chirp :P sucker works pretty good still. every drive manufacturer that made bad drives all the time, out of those, some are pretty decent. But if you find a NIB old hard drive, i'd be sketchy. the 63 watt should be able to handle 3.5" hdds ok, same with hard cards, anyone know how well the half height 5.25" drives work with it?

It certainly makes a peculiar seeking sound. NIB? I'm not sure I'm willing to push my luck with a 5.25" half height drive. As far as I'm aware, however, the Kalok and MFM 3.5" drives in general are also half-height... modern 3.5" drives are third-height.

cr1901
August 9th, 2013, 10:33 PM
However, that power supply may have better handling of the high current draw spike upon power-up, which is really what killed the 5150 power supplies, not continuous current draw.

Oh, hard drives outright killed the IBM power supplies? I thought the 63-watt supplies would tolerate the load, but there would be glitches/sag in the output voltage that would cause hardware components to respond as if a failure occurred (perhaps even the power good line).

vwestlife
August 10th, 2013, 09:06 AM
Oh, hard drives outright killed the IBM power supplies? I thought the 63-watt supplies would tolerate the load, but there would be glitches/sag in the output voltage that would cause hardware components to respond as if a failure occurred (perhaps even the power good line).

I had a 5160 power supply die on me, and take out a 486DLC motherboard along with it. So regardless of how these IBM power supplies fail, I'm glad that I have an aftermarket 200-watt PS in my 5150 now.

pearce_jj
August 10th, 2013, 11:49 AM
Probably the 63W PSU issue is more to do with 5.25" drives - the ST412 requires 4.5A on the 12V rail at spin-up (http://bitsavers.trailing-edge.com/pdf/seagate/ST412_OEMmanual_Apr82.pdf) and 25W steady-state draw. Later, especially 3.5", drives draw considerably less.

Agent Orange
August 10th, 2013, 12:24 PM
Only recently retired my ST-225 from my 1000SX with the advent of the Hargle XT-IDE controller. Never had any problems with the SX power supply as it currently runs a 3.5" & 5.25' floppy's, as well as IDE & SCSI hard drives.

Chuck(G)
August 10th, 2013, 12:29 PM
Probably the 63W PSU issue is more to do with 5.25" drives - the ST412 requires 4.5A on the 12V rail at spin-up (http://bitsavers.trailing-edge.com/pdf/seagate/ST412_OEMmanual_Apr82.pdf) and 25W steady-state draw. Later, especially 3.5", drives draw considerably less.

If you dump those full-height Tandon floppy drives for something like a half-height Teac, you'll also decrease the load on the +12.

cr1901
August 10th, 2013, 06:15 PM
My Kalok Octagon works just fine with no bad sectors other than those listed on the defects table (came with my XT). :) The components look like they're of middling quality, but not low quality- a mix of surface mount discrete components and ICs as well as through hole. Also RLL 30 MB- both standard for late 80's? I found out that you can confuse it by running it component-side up, and it will not recalibrate properly, though...

I also have a working Computer Memories hard drive for the PC AT that hasn't had any issues for the past 2.5 years. I've been fairly fortunate with early hard drives considered unreliable.

TTL monitors, on the other hand... not so much luck. I suppose that's a tradeoff in Murphy's Law of conservation.

Well, it appears I ate my words tonight... the Kalok Octagon can't seem to recalibrate to track 0... I suppose I'll try a half height 5.25" drive... unfortunately, I don't have an RLL-certified drive, and all I have is RLL controlers... this oughta be a great idea!

And now my Microscience HH-825 drive has failed too... unlike the Kalok, however, this one never responded to any command I gave it. A half height drive appears to work fine with the original PC power supply as well, for those who are interested.

SpidersWeb
August 10th, 2013, 07:31 PM
cr1901 - if it's scanning the drive on power up, e.g. slowly seeking to the last track then pulling back and the process repeats - it probably just needs a low level format. The Microscience could be a cabling issue.

Stone
August 11th, 2013, 02:21 AM
... I suppose I'll try a half height 5.25" drive... unfortunately, I don't have an RLL-certified drive, and all I have is RLL controlers... this oughta be a great idea!I've got a Seagate ST-238R in excellent condition if that's of interest.

cr1901
August 11th, 2013, 04:58 AM
cr1901 - if it's scanning the drive on power up, e.g. slowly seeking to the last track then pulling back and the process repeats - it probably just needs a low level format. The Microscience could be a cabling issue.
Both return the same error from Seagate BIOS and SpeedStor... "0x07 Drive parameter activity failed."... whatever that means. In any case, I have an old system that's in need of testing, I'll at least try the Microscience drive again and see if it loaded the power supply unacceptably...


I've got a Seagate ST-238R in excellent condition if that's of interest.
It appears my Seagate controller accepts both MFM and RLL drives and didn't know. Nevertheless, I may accept your offer anyway when I have some time to make a trip.

Stone
August 11th, 2013, 05:03 AM
FWIW, I've also got a ST-225 that's available.

Chuck(G)
August 11th, 2013, 09:38 AM
You'll please pardon me for saying it again--Kalok drives were garbage. It may be cheaper for you to get an XTIDE kit or assembled board in the long run.

cr1901
August 11th, 2013, 10:59 AM
You'll please pardon me for saying it again--Kalok drives were garbage. It may be cheaper for you to get an XTIDE kit or assembled board in the long run.

You never said that specifically Kalok drives were garbage. Just that 3.5" MFM drives weren't made very well :P. But I suppose you were implying that.

I just found it a bit appropriate that it failed as soon as I said it was fine. The components didn't seem like they were cheaply chosen or anything... was there any particular design choice or choices that cause(d) Kaloks to fail?

Does XTIDE replace the INT 0x13 BIOS services in the same way as an MFM BIOS does (in reference to my BIOS clone project)?

pearce_jj
August 11th, 2013, 01:01 PM
A question for the XTIDE Universal BIOS folk I guess :) But anyway, with the universal BIOS the XTIDE is fully bootable and without any size limitations.

Of course, I'll mentioned my own incarnation of it - see here (http://www.lo-tech.co.uk/lo-tech-isa-compactflash-board/).

k2x4b524[
August 11th, 2013, 01:53 PM
Hey now, don't jynx my kalok drive :P


You never said that specifically Kalok drives were garbage. Just that 3.5" MFM drives weren't made very well :P. But I suppose you were implying that.

I just found it a bit appropriate that it failed as soon as I said it was fine. The components didn't seem like they were cheaply chosen or anything... was there any particular design choice or choices that cause(d) Kaloks to fail?

Does XTIDE replace the INT 0x13 BIOS services in the same way as an MFM BIOS does (in reference to my BIOS clone project)?

Chuck(G)
August 11th, 2013, 02:40 PM
You never said that specifically Kalok drives were garbage. Just that 3.5" MFM drives weren't made very well :P. But I suppose you were implying that.

This forum can be your guide. I quote from 2012:


and a (gasp!) Kalok Octagon KL-320 hard drive. It has read errors all over the place, so it may be headed for the trash can, just like every other Kalok drive I've ever encountered or heard of before.

India in the early 90s was not exactly the hub of precision electronics manufacturing.

One option to look into is the XT-CF2 card. Very convenient--there's a slot in the mounting bracket into which one inserts a CF card. Very convenient.

And like the XTIDE versions, it comes with its own flashable BIOS support. I rather like it.

In general, 3.5" MFM drives were none too good--by the time the manufacturers

vwestlife
August 11th, 2013, 04:01 PM
It appears my Seagate controller accepts both MFM and RLL drives and didn't know.

MFM and RLL drives are mechanically identical. RLL drives were simply tested and certified by the manufacturer to meet a higher data reliability standard than MFM drives, because RLL squeezes 26 sectors per track onto the disk while MFM uses only 17 sectors per track, thus RLL increases the data density being stored on the magnetic surface of the platters.

The Seagate ST-11R controller card was designed for RLL, but you can tell it to format the drive with only 17 sectors per track. Whether that actually switches it into MFM mode, or if it's just only formatting the first 17 out of RLL's 26 sectors per track and leaving the rest blank, I have no idea. Maybe a program like SpinRite could do a surface scan of the disk and tell you how the sectors are laid out on the tracks.

Chuck(G)
August 11th, 2013, 04:12 PM
I suspect that the ST11R simply spaces the sectors out with bigger gaps. Given the intricacies in recovering data optimized for a 7.5MHz clock rate rather than 5 MHz (normal MFM), I'd be surprised if the ST11R, being a low-end 8-bit controller was capable of switching clock rates. I can dig into my WD literature if you're curious, however.

Regardless, a simple probe with a 'scope on the READ DATA line would tell you all you needed to know.

Stone
August 11th, 2013, 04:18 PM
The LLF is written with the number of sectors per track that the controller calls for, e.g., 17 or 26. There are no blank sectors.

glitch
August 11th, 2013, 04:58 PM
One option to look into is the XT-CF2 card.

You can also PM me about having an XT-IDE Rev 2 assembled, with or without an industrial Flash module. I've got a few blank boards left. Still working with a local supplier on getting parts kits put together!

Chuck(G)
August 11th, 2013, 06:59 PM
The LLF is written with the number of sectors per track that the controller calls for, e.g., 17 or 26. There are no blank sectors.

Stone, I don't think anyone said anything about blank sectors, just increasing the gap between sectors.

cr1901
August 11th, 2013, 10:14 PM
Stone, I don't think anyone said anything about blank sectors, just increasing the gap between sectors.



The Seagate ST-11R controller card was designed for RLL, but you can tell it to format the drive with only 17 sectors per track. Whether that actually switches it into MFM mode, or if it's just only formatting the first 17 out of RLL's 26 sectors per track and leaving the rest blank, I have no idea. Maybe a program like SpinRite could do a surface scan of the disk and tell you how the sectors are laid out on the tracks.

I believe this is text in bold is what Stone is referring to. That said, if that is true, then it's possible that my Microscience drive is fine and is just refusing to accept a Low-Level Format from the controller. It worked fine a year ago, and these were considered high-quality drives so I've read. Find it a bit odd for it to just fail like that.

Chuck(G)
August 11th, 2013, 11:10 PM
There's not that much difference between low-level formatting a floppy and a hard drive, aside from address marks and data transfer speed. I can format a high-density 3.5" drive with 9 512-byte sectors per track--and no blanks (whatever that means--I'm not certain).

After the last sector is laid down during formatting, the controller just outputs fill bytes (with no sector IDs) until the leading edge of the index comes around.

You can verify this with some simple timing (although the buffered nature of PC disk controllers will complicate things a bit).

But basically, the idea is that you format 17 sectors and measure the time it takes to read sectors 1 and 2 and then sectors 17 and 1 on the same track. Observe that you'll have to take into account controller latency, interleave and other such variables, but the difference should be obvious.

MFM and RLL drives are basically dumb-as-a-stump devices. They don't have much intelligence other than that required to provide buffered seek (and the ST506 didn't even have that). There are no commands to accept or reject. The controller asserts WRITE GATE and outputs to WRITE DATA, just like a floppy. The drive knows nothing about encoding or track layouts. Heck, you don't even have to output nicely formatted data--you can basically shove anything that conforms to the bandpass of the drive at the thing.

Even though the maximum frequency required of the read/write channel on the drive is the same for FM, MFM and (2,7) RLL, the accuracy demanded of the actual pulse placement within the data cell window becomes more critical as the effective data rate increases. That is, MFM can encode twice as much data as FM, and (2,7) RLL can encode three times as much per time unit, they also require that the drive be able to more accurately resolve the time a flux transition occurs. This can be affected by the read/write channel electronics, head design, coating characteristics and even speed stability of the motor.

That's really all there is to it, even if I'm oversimplifying a bit.

cr1901
August 11th, 2013, 11:26 PM
I was under the impression that 'blanks' meant "a (possibly set) number of particles on the disc surface whose magnetic flux transitions that are not assigned any particular meaning to the floppy drive controller or hard disk controller..." i.e. what I've heard you refer to as the 'slop' between sectors.

But then again, my understanding of floppy geometry is pretty rudimentary. I don't even know for sure what a drive head actually does on the molecular level to the disc surface.

Chuck(G)
August 11th, 2013, 11:45 PM
I was under the impression that 'blanks' meant "a (possibly set) number of particles on the disc surface whose magnetic flux transitions that are not assigned any particular meaning to the floppy drive controller or hard disk controller..." i.e. what I've heard you refer to as the 'slop' between sectors.

But then again, my understanding of floppy geometry is pretty rudimentary. I don't even know for sure what a drive head actually does on the molecular level to the disc surface.

Digital recording is a magnetic-saturation type of thing. Every time the magnetization direction changes, a pulse appears at the read head. Note that you get a pulse, but it doesn't matter if you're going from north to south magnetization or the reverse. It's customary to write filler between sectors and headers for those sectors; this helps the data recovery circuitry stay in sync with the original write clock. So the usual scheme is to write some pattern from the end of the last sector to index.