Fun with encoders

Hi all,

I just caught up on all the posts on encoder read heads. Interesting
stuff. I thought I would throw a few thing in the mix. I used to work
for HP, and did a few projects with the OED division that makes the
encoder heads. The reason that they never did any read heads that are
logical (ie 250dpi) is that OED sold 90% of its product to one
customer--the printer divisions of HP. And they generally wanted all
their pitches some multiple of 300 (goes with the DPI of the printer).
Yes, almost every HP inkjet printer has a linear encoder and read head.
It is mounted along the carriage axis. Next time you go to the computer
store, look and you will see it. Almost all of these are 150 lpi pitch
units.

Given there are some ambitious people who are talking about making their
own head from scratch, let me tell you about a trick that was done on
the HEDS modules that made them much higher resolution. The modules
normally read from the opto sensors and pipe the signal to a comparitor.
For those not familiar with comparitors, this device will take the
linear signal from the sensors and convert it to either TTL high or low
based on a preset trip point. The trick was to crack the HEDS case open
and jump the signal from the sensor over the comparitor so as to get a
linear signal that was sinusoidal in nature (I don't know how to do
this, so don't ask. Maybe one of you brilliant EE types could pull this
operation off). So now instead of just getting an ON signal and an OFF
signal, you get a current that rises and falls as the encoder line
travels through the read grate mask. Now, if you pipe this signal into
an analog to digital converter, you can pick up position information
when you are between the lines (this technique is not new, most high end
encoders do this). This worked very well. On a 200 lpi system we were
able to get 6 bits (63 counts) of position data between each line with
this crude setup. That equals 12600 LPI without quad decoding on a $30
encoder head. And it worked very well--so well in fact that it made it
into several HP products. So why hasn't HP put this cool part on the
market outside HP? Well, the engineers at OED tried. They were working
on a version that would have given you 12 bits or more of resolution
between each count at the time I left HP. Then, soon after I left some
brilliant manager somewhere decided to shut down OED! All the engineers
either lost their jobs or went to other divisions. Needless to say,
that encoder head project was canceled. I hear that they transferred
R&D control to the plant in Asia that manufactures the HEDS parts, but I
have no idea if they will ever pick up the ball on this part and bring
it to market. As an aside, I do know an engineer at another low cost
encoder manufacturer that is talking about bringing out a product just
like this sometime next year, but he told me he is getting pressure from
marketing to can it as they don't think they will ever be able to sell
enough of them to make it worth while.

I am currently working on a project that uses the relatively new HP HEDR
encoder module. This is a REFLECTIVE read head, as opposed to the more
typical TRANSMISSIVE read head. I found all the talk on having linear
strips made very interesting as I have been getting quotes to do just
this for the last few weeks. NO, a typical mylar printer cannot do
these accurately. You need to go to shops that specialize in high
accuracy lithography. I have several companies lined up to do this
right now. The actual printing of the strips is fairly cheap. But the
cheapest setup charge I have found is $600. There is a company out of
germany call PWB-Rulahtec that will make samples of any resolution you
want for $250. But that is still out of hobby $ range, isn't it.


-Mike