(Please Pardon the Gratuitious Advertisement Above)
((In return for posting these banners on my site, my banner gets displayed on other sites, too)) 

Dave's CXA2075 page!

Well, people keep writing me for information on the Sony CXA2075, which we designed as a replacement for the CXA1645 which you may recognize as the RGB to NTSC/PAL video encoder in the Sony Playstation and SEGA Genesis (I think.  I didn't work for SEGA ;)

Before I go any further, I should say that I do not represent SONY corporation in any manner whatsoever, and am little more than a fan of their products.  Any information presented here is in no way guaranteed to be accurate or even remotely correct.  I haven't had contact with Sony for years and in any case, they are solely responsible for their own products.  This should probably be obvious, but Sony is not bound by any information presented here.  This information should be viewed as little more than a historical record circa 1996 of the CXA2075 prototype.  And even for that, it's not guaranteed to be accurate.  (Thought I'd state that again, in case you missed it :)

If you look at Sony Semiconductor USA's product page, you'll see there's a listing for the CXA2075, but no datasheet available on the web.  Heck, I don't even know if they make the chip available, but I suppose they do.

As I've so far gotten about a half dozen requests for info on the CXA2075, I've decided to summarize on this page.  If you have any questions, feel free to drop me a note (plus I'd really love to hear if anybody uses it successfully!! :)

Keep in mind I haven't worked for SONY since 1996, and the document I present here is NOT OFFICIAL!!!  It's the original datasheet I wrote.  It may or may not be accurate, I recommend you contact SONY for an official datasheet before attempting to use the CXA2075, but this might give you some idea about the chip....

(Not that I'm biased, but I think it's a kick butt part.  Our design goal was to move the chip to a cheaper process (with only 50MHz PNP devices (it's a bipolar part, btw) and to create 60MHz RBG outputs, remove all the external precision circuitry, cut the circuitry by half to reduce the die size and add some new features (like a built in filter).  The goal was to create the ultimate in performance and the lowest cost part available (especially system cost).  On my datasheet we only claimed >30MHz because worst case simulations only guaranteed that (Yeah, but how often are you running the dang thing at -20C ?!?  (Yes, that's a minus).  Typical performance we got on our test chips was above 60MHz.  Be sure to go back and read my disclaimer before you run off and design your multimillion dollar product to run at 60MHz...

Anyway, I think we managed to accomplish everything we intended.  The only thing was that the part consumed more power than SEGA wished, the last I heard the fusable link they used to power the chip plus some other circuitry would not cover the power required by the CXA2075 (well, all that power for those incredibly fast slew rates had to come from somewhere!!)

Anyway, here's what I could find:

My copy of the original CXA2075 datasheet (now in pdf form) (I don't know what changes (if any) were made to the final chip)
(Also, disappointingly, I could only find the first 5 pages, plus my NTSC application note (now in pdf form).  I can't find my "notes on operation" page.  Most of the notes are similar to the 1645, but 8 is different (can't remember the proper R value for a PAL trap using the internal trap!!) and I had added a couple more notes.  I might have the additional notes because I have some Excel spreadsheets with some more information on them which I can't read since I don't have Excel on my Mac :)

My old copy of Sony's CXA1645 datasheet (I couldn't find Sony's on the web, strangely)

Here's a typical letter and response:

> Hello from Italy
> A little questìon
> Is cxa1645 = cxa2075 ?

Oh, no!  The cxa2075 is vastly superior!  (Not that I'm biased since I
designed a large proportion of the 2075 ;)

Our design goal for the 2075 was to increase the performance of the output
drivers to handle 60MHz signals (enough to drive high resolutions on the RGB
outputs) (up from like 12MHz or 15MHz or something) remove all precision
external components, cut the die area in half and transfer the chip to a
cheaper process to save costs.  We did all that and even added a couple
features.  (like a better on-chip chroma trap filter :)

The 2075 is basically pin compatible with the 1645, however.  We made the
precision part connections into no connects for the most part so it could be
dropped in directly to a 1645 design and be cheaper and improve performance
without modification.  (However, unless you had a reall good trap filter
(which had to be external on the 1645) you should probably disconnect it and
use our internal one, which turns on automatically when no external circuit is detected)
(I did the trap switch and the high speed drivers, as well as a bunch of other
stuff, but I was most proud of the drivers :)

There's ONE little catch, however.  In order to attain the ultra-high
performance, we had to burn more power.  This is not normally a problem (and I
don't know about most of the playstations) but the Sega that used the 1645
didn't have quite enough power budget (they used a fusable wire to supply
current to a few circuits) and our 2075 was over the power limit they had set.
 I left Sony about that time after providing some suggestions for reducing
power for them so I don't know what they did with it at that point.