Trying to determine EPROM access speed

[Circlotron]

I have these Chinese EPROMs that are rebranded and second hand. e.g. some are not completely erased, and when you rub off the new paint it has Chinese writing on the ceramic body. Despite this, I have been using them for quite a while and generally they are okay for my purpose. They are (falsely) labelled as ST and 100nS access time. Actual access time is somewhere between 1.25 and 2.5 uS. If I try and read it 1.25 uS after addressing it doesn't work but after 2.5uS it does.

Decided to rig up something to try and measure the actual access time. I simply toggled A0 address line and compared that edge with one of the data lines that happened to change state. Lo and behold, the data changed 20nS after the address! No EPROM is that fast, especially these ones. What is a good way to measure the access time?

[srb1954]

I have these Chinese EPROMs that are rebranded and second hand. e.g. some are not completely erased, and when you rub off the new paint it has Chinese writing on the ceramic body. Despite this, I have been using them for quite a while and generally they are okay for my purpose. They are (falsely) labelled as ST and 100nS access time. Actual access time is somewhere between 1.25 and 2.5 uS. If I try and read it 1.25 uS after addressing it doesn't work but after 2.5uS it does.

Decided to rig up something to try and measure the actual access time. I simply toggled A0 address line and compared that edge with one of the data lines that happened to change state. Lo and behold, the data changed 20nS after the address! No EPROM is that fast, especially these ones. What is a good way to measure the access time?

What is your set-up for measuring the access time at between 1.25us and 2.5us?

I don't think EPROMs were as slow as that since at least the mid 1970s.

[Terry Bites]

check "AC MEASUREMENT CONDITIONS" in a trustworthy datsheet

[DavidAlfa]

Another vague question: Unknown eeprom model (Because there's only one eeprom model in the market, right?), no details at all...

Anyways 100ns spec is worst-case scenario, temperature, voltage, whatever conditions, will affect it.
Also you might get 20ns in *some* data pins, but maybe not all of them will be fully settled down.

[Circlotron]

What is your set-up for measuring the access time at between 1.25us and 2.5us?

Set the address then 5 cycles later (5 x 250nS) read the data. No good.
10 cycles later reads okay.

[Circlotron]

Another vague question: Unknown eeprom model (Because there's only one eeprom model in the market, right?), no details at all...

EPROM, not EEPROM.
That's right, no data. It's some oddball rebadged Chinese fake rebadged as ST. M27C801. 1M x 8.

[amyk]

Since it has an exposed die, you could look for markings on it to determine what it really is.