PerStor Controllers

[Ira Baxter]

orion at nuchat (Roland Dunkerley III) writes:

>I am currently in the process of putting together a
>machine, and have seen the ads for the PerStor Disk
>controllers claiming to give 190% of the storage on a
>standard MFM drive.  The drives I plan to use are a
>Seagate ST-251, and an ST-4096.  If this gadget works
>it would give me 228Mb instead of the 120 that these
>drives provide, with a 9Mbit/s transfer rate instead of
>5.  My question is have any of you used the PerStor
>controllers, do they really work as advertised on
>standard drives with no problems?

I evaluated a Perstor PS180-16F controller, and its immediate
predecessor (I don't remember the model number) with both ISC 1.0.5
and 2.0.2, as well as dull ol' DOS.  The short analysis: it does work,
but I decided against it.

It was rather awhile ago, so I may have the details a bit fuzzy, but
this is what I remember.  A special set of installation programs are
required to set up the Perstor.  I found these programs to be rather
inferior in maturity, but they did work; typical problems were menus
that offered (at different times!) choice of obviously garbage disk
configurations, and a program to choose-the-best-interleave that
couldn't compute the transfer rates correctly... so you had to
determine the best interleave (I remember it being 3) yourself.  I was
pretty incensed that such trivial errors were present in a supposedly
mature product; perhaps these stupidities have been fixed by now.  One
must use the formatter provided by these programs rather than the ISC
formatter.  I don't remember any tools provided by Perstor to mark the
manufacturer-supplied bad block list, so they have to be re-discovered
(if re-discoverable) by the ISC surface-analysis process.  Since the
controller isn't very fast, the additional drive capacity provided by
the Perstor actually makes this more of a nuisance rather than less.
This business about 9Mb/sec transfer rate is max burst rather than
sustainable; there isn't any track buffer, and with a best interleave
of 3, the best sustainable transfer rate was about 250Kb/sec; this
wasn't much better than my MFM WD1003 controller that I wanted to
replace.  Last, but not least, the first controller they sent me would
format but produced zillions of errors during surface analysis;
Perstor finally replaced it with the PS180-16F and the problems went
away.  They claimed the problem was "marginal VCO" chips from their
chip vendor, but I think the real problem is that the entire design is
marginal (pushing the drive right to the edge of what it can do), and
their software support didn't give me a very good feeling about their
quality control.

The up side: the controller does work, and with a Maxtor 2190 (MFM
rated for 150Mb) one gets an awesome 290+Mb of real capacity.  Both
1.0.6, 2.0.2, and DOS 3.10 (with a 32Mb partition) seemed to work fine
for the day or two that I ran them.

>                                  What technology do
>they use to accomplish this?

The folk at Perstor are pretty closed-mouthed about this.  I suppose
one could guess by inspecting the chips on the board fairly carefully,
which I haven't done.  Considering the trouble I had with the
predecessor board, which was quoted as being due to the "marginal VCO"
chips by the factory tech, I would guess they simply run a higher bit
rate at the drive, and count on their 56-bit ECC to save them.  That
doesn't leave one with a warm feeling about the reliability of the
inner tracks.


>[stuff about potential configuration deleted]
All of my remarks are, I think, OS independent.

>   thanx in advance,
>    Roland Pleasant Dunkerley III, K.S.C., NonD

I finally chose to use a WD1006SRV2, which is RLL compatible, on my
Maxtor.  It doesn't double your capacity, but it does raise it
signifcantly, and with the track buffering, one gets around 500Kb/sec
sustained transfer rate.  ... and its cheaper than the Perstor (your
cost may vary).  The bad news is the WD1006SRV2 has its own peculiar,
but quite rare, lock-up problems, see the ongoing thread in this newsgroup.
I have been using the WD1006SRV2 for several months now, and except
for the rare hiccup, I'm very happy with.  Now, if we can only resolve
the hiccup...

Anybody want a very slightly used Perstor controller, cheap?
--
Ira Baxter