MB85RC64TA FRAM Trouble Writing/Reading [SOLVED]

[ANTALIFE]

Hi all

I have a MB85RC64TA FRAM module that I am trying to write/read to via I2C with an nRF52832. Here is the datasheet for it.

The two commands I am trying to implement for now is "Byte Write" & "Random Read" (see pg8 & pg9 of datasheet). Here is my basic & messy code, all it does is attempt to write 0x30 to register 0x1510 (0x15 0x10) and then read the (hopefully) same value from the same register. But this does not work as all I read is 0x00
NOTE: The WP pin of the IC is always LOW meaning I should be able to write to the IC

When I attach a logic analyser to the SDA/SCL pins I see that all but the read operation works correctly?
So with that said could someone offer some advice?

Cheers, Anton


UPDATE 1: Turns out the x2 MB85RC64TA IC's I had were counterfeit. If I use the same code for the FM24CL64B-G (same memory & command structure) I can easily write and read values. Using my microscope (see below) proves this, as the MB85RC64TA surface looks to be polished and engraved with a laser (something IC manufacturers don't tend to do...)

[iMo]

The writing/reading to/from a fram is identical to w/r with a flash. Mind there is an wr_enable bit required to send when writing, afaik (like i2c_adr_with_wren+address+data)..

Thus your MB85RC64TA_ADDR differs in its LSB bit based on RD or WR operation.

[ANTALIFE]

The writing/reading to/from a fram is identical to w/r with a flash. Mind there is an wr_enable bit required to send when writing, afaik (like i2c_adr_with_wren+address+data)..

Thus your MB85RC64TA_ADDR differs in its LSB bit based on RD or WR operation.

Yup, thats what the logic analyser shows right? Write address is 0xA0, read is 0xA1

[Someone]

One step at a time, can you correctly read the Device ID?

[ANTALIFE]

One step at a time, can you correctly read the Device ID?

Yeah on the non-counterfeit IC 

[wraper]

UPDATE 1: Turns out the x2 MB85RC64TA IC's I had were counterfeit. If I use the same code for the FM24CL64B-G (same memory & command structure) I can easily write and read values. Using my microscope (see below) proves this, as the MB85RC64TA surface looks to be polished and engraved with a laser (something IC manufacturers don't tend to do...)

Relabeled IC's don't look polished. If surface of IC was sanded, it's anything but glossy. IMHO you screw up at writing data. ICs usually can be read the same but writing may have differences.

[ANTALIFE]

UPDATE 1: Turns out the x2 MB85RC64TA IC's I had were counterfeit. If I use the same code for the FM24CL64B-G (same memory & command structure) I can easily write and read values. Using my microscope (see below) proves this, as the MB85RC64TA surface looks to be polished and engraved with a laser (something IC manufacturers don't tend to do...)

Relabeled IC's don't look polished. If surface of IC was sanded, it's anything but glossy. IMHO you screw up at writing data. ICs usually can be read the same but writing may have differences.

Are you sure about that, the write operation (see attached) for both looks pretty much identical to me?

Also it's a bit strange that top & bottom surface of the IC differ in texture, and that's what pg11 of this presentation by CTI says too...

[wraper]

I checked datasheets briefly, look similar. But there are differences such as maximum interface speed, MB85RC64TA is 3.4 times faster. There might be some glitch on I2C interface during transfer which FM24CL64B-G does not notice but MB85RC64TA does. Would be good to check with oscilloscope. Also what do you do with A0-A2 and WP pins? Do you tie them to GND/ Vdd directly or through resistors?
I dunno if it's genuine or not but glossy IC surface is nothing supernatural. Not super common for SO-8 but there are plenty of such ICs. If you had MB85RC64TA from another source which look different, it would be another thing. You could try to rub it with acetone or other solvent. If IC was blacktopped, coating together with marking often comes off.

Here it looks glossy on top and matte on bottom https://lcsc.com/product-detail/FRAM_FUJITSU_MB85RC64TAPNF-G-BDERE1_MB85RC64TAPNF-G-BDERE1_C189804.html

[ANTALIFE]

I checked datasheets briefly, look similar. But there are differences such as maximum interface speed, MB85RC64TA is 3.4 times faster. There might be some glitch on I2C interface during transfer which FM24CL64B-G does not notice but MB85RC64TA does. Would be good to check with oscilloscope. Also what do you do with A0-A2 and WP pins? Do you tie them to GND/ Vdd directly or through resistors?
I dunno if it's genuine or not but glossy IC surface is nothing supernatural. Not super common for SO-8 but there are plenty of such ICs. If you had MB85RC64TA from another source which look different, it would be another thing. You could try to rub it with acetone or other solvent. If IC was blacktopped, coating together with marking often comes off.

Here it looks glossy on top and matte on bottom https://lcsc.com/product-detail/FRAM_FUJITSU_MB85RC64TAPNF-G-BDERE1_MB85RC64TAPNF-G-BDERE1_C189804.html

Yea did see that the MB85RC64TA has a high speed (3MHz) mode but you need to send an extra command before it's activated. When I do get out the logic analyser for the MB85RC64TA, the weird thing is that all commands are acknowledged and match up to what the datasheet says. Also A0-2 & WP are floating, but datasheet says that they have internal pull-downs.

You might be right about IC not being counterfeit, and I don't have another source to compare to. For now will just continue using FM24CL64B-G as it has not had any issues thus far.

[wraper]

 

Also A0-2 & WP are floating, but datasheet says that they have internal pull-downs.

Yes they have pull-downs but I suggest not leave them floating but tie to GND.

Yea did see that the MB85RC64TA has a high speed (3MHz) mode but you need to send an extra command before it's activated.

Yes but it also means that circuit is likely faster regardless of mode enabled.

[SiliconWizard]

When I do get out the logic analyser for the MB85RC64TA, the weird thing is that all commands are acknowledged and match up to what the datasheet says.

If your commands get acknowledged, then it basically means the chip understands them. If the data it puts out is invalid or all zeros whatever command you send, that's quite fishy. A totally borked IC or a problem with communication would probably yield commands to be NACK'ed. To further make sure there is no signal integrity problem, take a loot at the I2C signals with a scope. Logic analyzers are handy, but they don't show anything much if there's any signal integrity issue.

[SiliconWizard]

Relabeled IC's don't look polished.

Well, it may not be a counterfeit, but it could very well be.
The worst counterfeits are not even polished or sanded. They are merely OVERSPRAYED before being marked again, and frankly, this exactly looks like a nasty overspraying job here.
Looking at the chips from their sides would confirm this - you should relatively clearly see a "line" between the top "resprayed" layer and the rest of the package.

[wraper]

Relabeled IC's don't look polished.

Well, it may not be a counterfeit, but it could very well be.
The worst counterfeits are not even polished or sanded. They are merely OVERSPRAYED before being marked again, and frankly, this exactly looks like a nasty overspraying job here.
Looking at the chips from their sides would confirm this - you should relatively clearly see a "line" between the top "resprayed" layer and the rest of the package.

FYI blacktopping does not look glossy as much as I've seen.

[wraper]

Here are 2 another Fujitsu FRAM which are glossy on top and matte on bottom.

https://lcsc.com/product-detail/FRAM_FUJITSU_MB85RS2MTAPNF-G-BDERE1_MB85RS2MTAPNF-G-BDERE1_C189908.html
https://lcsc.com/product-detail/FRAM_FUJITSU_MB85RS128TYPNF-GS-BCERE1_MB85RS128TYPNF-GS-BCERE1_C189906.html