COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	.require "let.pub" source file
C00009 ENDMK
C⊗;
.require "let.pub" source file;
.<<
.FONT 1 "basl30"; TURN ON "%";
.FONT 2 "BASI30";
.FONT 3 "basb30";
.FONT 4 "sta200";
.FONT 5 "ngb25";
.FONT 6 "SUP";
.>>
∂AIL  Mr. Art Hausman, President↓Ampex Corporation↓401 Broadway
↓Redwood City, California 94063∞

Dear Mr. Hausman:

	We wish to express our concern about what appears to be a lack of
quality control in a computer disk system that we recently ordered from
your company.  We initiated a search for a replacement for our IBM 3330-11
system several months ago and selected your product on the basis
of price and field performance of the early installations.

Based on your quoted delivery date of June 20, we arranged to release
our IBM equipment a month and a half later.  This planned overlap has
now been eaten up by slightly late delivery and chronic malfunctions
in your equipment.  We are now faced with the prospect of having to
extend the IBM lease, at substantial expense, and not having any
clear idea of when we can get out of this bind.  At the present time,
the control unit is not sufficiently reliable to sustain operation of
our system.  We wish to call the following specific facts to your
attention. 

	1.  We have discovered several marginalities in the control unit,
which, up until a week ago, caused it to be totally unusable.  In
fact, one of the diagnostics (A4) still does not run.  It took Ampex
field engineers many hours of board shuffling to find a combination of seven
cards that would make the control unit run at all.  Despite all efforts,
the control unit does not perform reliably.

	2.  Many of the circuits have the appearance of having been
translated from ECL logic to TTL without adequate consideration of
the effect of changing the logic family.  Where ECL has complementary
outputs, TTL requires extra inverters and has slightly longer gate
delays, so that simple substitution of TTL circuits causes critical
timing chains to malfunction.  Also, we have informed your
representatives that we believe that the cause of one of the problems
is the incorrect copying of a signal name from the original IBM
circuit. 

	3.  One symptom of shortcomings in the design is the narrow noise margins:
our controller malfunctions consistently when the power supply
voltage is adjusted to the "normal" TTL level of 5 volts, but works
somewhat better when the voltage is lowered to 4.85. 

	4.  Read amplifier cards turn out to be critical in determining error
rates.  Some cards give 100 times the error rate of others.  These rates move
from drive to drive as the cards are moved.  Ten new cards supplied by the
factory produced only one more good card with low error rates.  This means that
production quality control is bad and that the factory tests are inadequate.
Random fixes like replacing ten percent resistors by one percent resistors were
tried by your service man at factory suggestion and produced insufficient
improvement suggesting that they don't yet understand the problem.  If you are
producing mostly inadequate read amplifiers now, can't test them and don't
understand the problem, then you are risking catastrophe as well as incurring
costs in producing unusable cards.  A slight change in your production process
may cause you to produce totally useless boards.

	5.  One of the five disk drives that was delivered here (and presumably
had passed factory inspection) was so mechanically unstable that doing a seek
would cause it to be unable to read any data previously written and to write
files that could not be read later.  It was removed and, to date, has not been
replaced.

	We were, and are, attracted by your lower price than IBM's, but we are quite
unhappy about the amount of time our engineers have had to spend tracking down
your bugs.  We hope, Mr. Hausman, that you will give this matter careful attention. 

.sgn

.begin nofill; skip 5;


cc:  Hersche Allen, Les Earnest, Ralph Gorin, Ted Panofsky, Jeff Rubin,
	Rainer Schulz (IMSSS), Richard Highfield (Purchasing)

Stanley Mantell
Vice President and General Manager
Data Products Division
Ampex Corporation
Redwood City, California 94063

Robert M. Trick
Northwest Regional Manager
Data Products Division
Ampex Corportion
465 Maude Avenue
Sunnyvale, Calif. 94086
.end