//e rom access speed

8 posts / 0 new
Last post
bmb
Offline
Last seen: 3 years 2 months ago
Joined: Aug 21 2021 - 20:33
Posts: 3
//e rom access speed
AttachmentSize
Image icon 2732-speed.png84.54 KB

Hello Everyone,

I have recently acquired an Apple IIe that I am in the process of bringing back to life.  Currently, it is missing the Video ROM and I plan on burning the rom onto a 2732 eprom chip.  When looking at chips to buy I noticed that there are versions of this chip that have different access speeds.  For example, the datasheet for TI's TMS2732 shows that it comes in both a 300ns and 450ns versions.  I couldn't find what the access time/cycle time version of this chip Apple is usiung for 342-0133-A (unenhanced Video ROM - US version).

I was able to get my hands on an official chip and put it into my ROM programmer to dump/verify and to see if the programmer would give me that information and I wasn't able to make heads or tails.  Included in the screenshot I highlight where 2 different values were given for speed neither of which is consistent with eachother or the datasheet that I linked. 

So, I have 2 questions:

1 - What speed chip should I choose?  Does the access speed matter at all for this application?

2 - How/Where should I have found out the answer to this question.  I seem to be missing some very basic conceptualization of the machine and lack the neccessary skills to solve the problem for myself.  I have a feeling I will run into this type of problem in the future and don't want to have to resort to asking online for help.

Thanks!

 

 

bmb
Offline
Last seen: 3 years 2 months ago
Joined: Aug 21 2021 - 20:33
Posts: 3
Hrm, I linked to the wrong

Hrm, I linked to the wrong datasheet.  The correct one is here -- TMS2732

Offline
Last seen: 3 hours 30 min ago
Joined: Feb 27 2021 - 18:59
Posts: 601
The Video ROM is used to

The Video ROM is used to supply bit patterns to the 74LS166 shift register, that in turn outputs pixels to the screen. A pixel is output every 70 ns, so the ROM is read approximately every 8 pixels or ~500 ns. It needs to be a little faster than this for timing reasons, but 150ns response is more than adequate. I don't think you can buy EPROMs slower than 150ns today.

You can find these timings in a technical book like the original Apple II manual, or the circuit description book by Winston Gayler.

A device programmer is not normally set up to speed-test anything, so you can't get any information on device speed from it. The sum total of the information it is able to get from a device is 16 bits called the "electronic ID", which not every chip has. All the data you see in the programmer software is read from a database of known devices, not from the chip itself. So it is not actually useful and confirms my prejudice against these (engineered down to a price) programmers.

Offline
Last seen: 1 year 10 months ago
Joined: May 27 2009 - 01:37
Posts: 1002
Which of the more expensive

Which of the more expensive not-engineered-to-a-price 'prommers do speed testing?

Offline
Last seen: 3 hours 30 min ago
Joined: Feb 27 2021 - 18:59
Posts: 601
Keatah wrote:Which of the
Keatah wrote:

Which of the more expensive not-engineered-to-a-price 'prommers do speed testing?

Like I said, they are normally not set up for that.

It is possible to speed test memory devices by creating specific data patterns, writing them to the chip, and using matching test vectors to verify behavior, but the processing speed of device programmers is usually insufficient to make all but the crudest speed measurements.

Speed testing memory devices is usually done with a purpose-designed test unit (like a RAMCHECK speed verifier) or a logic analyzer. 

My criticism of "Aliexpress"-style device programmers is that they include features that look good but are useless. The correct programming algorithms for many devices are proprietary; without cooperation of the silicon manufacturer, the unit may claim support but has no way to know if it is correct. Caveat emptor.

Offline
Last seen: 3 years 2 months ago
Joined: Mar 16 2018 - 01:55
Posts: 26
The 300 ns part should be

The 300 ns part should be fine. The 450 ns part is technically too slow for the video ROM. While the cycle time required is 500 ns, the access time has to be faster, because of the Apple's timing of when the ROM address changes and when the ROM output data is loaded into the shift register. The 450 ns part might work in practice, because the parts normally exceed specifications at room temperature. If the part does not provide data fast enough, you will see errors in the video, either solid or twinkling.

 

Robespierre is correct that there is NO automated memory speed tester for EPROMs available for purchase. Even when EPROMs were a going concern, full parametric memory testers were only designed for and purchased by the actual semiconductor companies, and cost hundreds of thousands of dollars. If anyone did try to sell you such a thing for under, say, $10,000, I'd expect it to be some kind of scam, and not a usable tester. As Robespierre said, you _could_ contrive to test EPROM speed with more common test equipment, but it's not an easy task, and the suitable equipment, even used, would cost several thousand dollars, and a lot of your time developing a suitable test procedure.

 

The method everyone uses in practice is to buy the part that has a suitable rated speed, and if it doesn't work, either the part doesn't meet spec (which was unlikely for new parts, but a lot more likely now for parts purchased through third parties), or the system it is being installed into is broken.

bmb
Offline
Last seen: 3 years 2 months ago
Joined: Aug 21 2021 - 20:33
Posts: 3
Apprecuiate everyone's

I appreciate everyone's responses.  

 

Thank you,

B

Offline
Last seen: 16 hours 53 min ago
Joined: Dec 20 2003 - 10:38
Posts: 589
I doubt that you’ll find a

I doubt that you’ll find a too slow EPROM

30 years ago I burned a lot of EPROMs for enhancing IIes and updating cards. Mostly used salvaged EPROMs from old equipment. I never knew the speed on any of the EPROMs and never had one fail.

Log in or register to post comments