What's new
By:

Naomi 2 netboot issue, failure at first reset

ekorz

ekorz

Multi Boyz 4 Pi
Immortal
Multi Boyz
Joined
Aug 21, 2016
Messages
5,329
Reaction score
8,195
Location
Boston, MA
I have a netbooting Naomi 2 that is suffering from an odd and intermittent issue. It's connected to a raspberry pi running standard piforcetools. It's running Bios 23606c-multi.ic27 "Rev C" and a Zero PIC on the dimm.

I'm used to this process:
Turn it all on, Naomi 2 successfully boots to "Checking Network" screen.
Ping net dimm from the pi, success
Select game on pi
Naomi 2 resets, returning to "Now Loading" screen
Game transfer complete
Naomi 2 resets, checks memory (...yes I can disable this now, just isolating variables so I kept it in the script for now ;) )
Naomi 2 resets again, finally booting the game

My problem is that this unit *sometimes* fails to recover from the initial reset, after I select a game (bolded above). When it fails, it doesn't make it to the "Now Loading" screen, it just stays black. I'd say failure is 4/5 boots right now.

Things I've tried
PSU is a Lindbergh, voltages are goooood at the connectors
Pi voltage is from the naomi cn4, it's also fine
Changing the target on the pi and the IP address of the Naomi 2
Tried the pi, SD Card, and Net Dimm on a Naomi 1 (they all work fine)
Tried a few Ethernet cables, crossover and not crossover (both work on the naomi1)

Any ideas what to check?
 
I had similar issue, I had to adjust the voltage under load and test the voltage on the actual naomi because it would drop significantly from 3.3 and 5.0 under load, when it was spot on T the connectors
 
Check the supercap on the n2 mobo and rip it off! Has no use and can do serious harm. Sometimes makes all good again. Also like @Anselmo is saying check the mobo underload voltage as they are verry picky with a dimm installed
 
Thanks! Alright, I pulled that supercap (shame on me for forgetting) and bumped the 3.3v up a couple percent. I couldn't find a good test point on the motherboard so if you have one lemme know. But first try booted up well! I'll keep testing.
 
Whoa, you can just tear the supercap off, and there won't be any problems? It's C29, right? I'll have to do that on my Naomis.
 
@Ajax yes you can remove it

Ok, seems stable after bumping it up and doing a few resets. Anyone got a good test point on the Naomi2 itself for 3.3v though?
 
ekorz said:
I couldn't find a good test point on the motherboard so if you have one lemme know.
Click to expand...
remove the metal shield piece from the net-dimm put the negative probe on the ethernet port and the positive probe on one of the giant pads for the crystal near the ethernet port, this should get you a 3.3V test point there is an equally easy 5V test point near by as well though I don't remember where off the top of my head. you can confirm these buy checking continuity between the points and the main power connector with the unit off and unplugged.

I've been meaning to make a how-to post with pictures for this for a while now, just haven't got around to doing it.

basically the NAOMI mobo uses only 0.6A of 5V but the mobo sucks down about 2.4A of 3.3V and the Net-DIMM uses another 2A of 3.3V on top of that. so you're trying to deliver almost 4.5A of 3.3V total and that needs to be stable in the DIMM board for the RAM.

speaking of RAM, it couldn't hurt to reseat the DIMMs or swap them out with some good working ones.
 
  • Like
Reactions: AMS
ekorz said:
@Ajax yes you can remove it

Ok, seems stable after bumping it up and doing a few resets. Anyone got a good test point on the Naomi2 itself for 3.3v though?
Click to expand...
What I do is to insert a multimetter at the rear of the JVS connector while the unit is running to have the exact measurement on load. Ideally you should never drop below 3.25V and never go above 3.35V (on load), below is unstable, above risks frying the fairchild buffer chip on the dimm board.
 
8fNHUC0_d.jpg


@twistedsymphony that one with V25.000 on it? None of those four points tone for me. Bottom left one beeps for a second but then cuts off like it’s tied to a cap or something. The Ethernet port outside metal doesn’t tie to ground that I can find either. Interestingly the top of the v25.000 is metal and THAT is ground.

I’ve heard 3.3v circuits are troublesome for cheap multimeters...
 
@mathieulh in my case 3.35v at the jvs connector doesn’t seem to be enough. 3.41 is giving me stability. Which is the Fairchild Buffer chip? I’d like to take the 3.3v reading there if it is practical to see how much drop I have.
 
ekorz said:
@mathieulh in my case 3.35v at the jvs connector doesn’t seem to be enough. 3.41 is giving me stability. Which is the Fairchild Buffer chip? I’d like to take the 3.3v reading there if it is practical to see how much drop I have.
Click to expand...
I think he's taking that reading on the rear while it's under load, are you doing the same?
 
ekorz said:
@mathieulh in my case 3.35v at the edge connector doesn’t seem to be enough. 3.41 is giving me stability. Which is the Fairchild Buffer chip? I’d like to take the 3.3v reading there if it is practical to see how much drop I have.
Click to expand...
Are you reading these measurements on load? That's way too much if that's the case even on a Naomi2, the fairchild buffer chip is the chip at IC12S, unfortunately it's a ticking time bomb that will eventually wear out (even at nominal voltage levels) the more you use it, the higher the voltage, the faster it dies, the buffer chips in the naomi have a similar issue (they don't die over time, unlike the fairchild one on the DIMM, but running them at more than 5.15V on load will eventually wear them out.)

Eventually when IC12S fails you will either be unable to boot while the dimm board is inserted or you will get an error 22 message and firmware will be reported as 0.00.

Since this is super fine pitch SMD, replacing it requires some skills that go (way) above your average soldering skills, more so considering it lies right on the other side of a BGA chip, which you can break the joints of if you attempt to remove this SMD using hot air. (I personally use a very low melting point alloy to do it instead).

I bought a "junk" netdimm from Yahoo Auctions (in Japan), for roughly €50 which came with a dead buffer chip, this is how I fared replacing it: https://twitter.com/Mathieulh/status/1304436466194493443 (keep in mind that I am rather good at this).
 
mathieulh said:
Are you reading these measurements on load?
Click to expand...
Anselmo said:
I think he's taking that reading on the rear while it's under load, are you doing the same?
Click to expand...
absolutely, under load

Thanks for the photo! Looks like IC12S is on the bottom board so I can't take voltage off it while the power is on. Hopefully I can find another test point on the top to make sure I'm within tolerance, which the datasheet does at least show as 3.6v. I'm feeling fine sending 3.41v through the connector, all the way through the motherboard, up to the dimm, especially when there is reported (and seemingly experienced) drop-off. Still I'd like to measure it exactly so I don't overwork anything.

failing that I’ll just test to find the lowest voltage at which it works, I guess. Maybe signs of death...
 
Last edited:
So, I did some measurement, and it turns out you can actually go below 3.25, I tested both my DIMM boards at roughly 3.22 while connected to a communication board and both are stable.

i81Er27.jpg
 
mathieulh said:
So, I did some measurement, and it turns out you can actually go below 3.25, I tested both my DIMM boards at roughly 3.22 while connected to a communication board and both are stable.
Click to expand...
Mine boots with less voltage... it's the reset after selecting a game where it freezes/fails
 
ekorz said:
mathieulh said:
So, I did some measurement, and it turns out you can actually go below 3.25, I tested both my DIMM boards at roughly 3.22 while connected to a communication board and both are stable.
Click to expand...
Mine boots with less voltage... it's the reset after selecting a game where it freezes/fails
Click to expand...
I've got no issue, either at boot, reboot, ingame or doing the dimm test.

Are you sure of your PSU? Is it stable? Does what's the ripple tolerance on it?
 
mathieulh said:
Are you sure of your PSU? Is it stable? Does what's the ripple tolerance on it?
Click to expand...
It's a lindbergh psu, but I have no idea how to test ripple tolerance. I suppose I could recap it but there are no voltage swings that I am seeing.
 
ekorz said:
mathieulh said:
Are you sure of your PSU? Is it stable? Does what's the ripple tolerance on it?
Click to expand...
It's a lindbergh psu, but I have no idea how to test ripple tolerance. I suppose I could recap it but there are no voltage swings that I am seeing.
Click to expand...
You wouldn't see ripples unless you use a scope. I suspect a faulty capacitor in either the PSU or the NAOMI (if so, probably on the 3.3V rail), it's hard to tell unless it's visible or you use an ESR meter, at this point you might as well recap the whole thing though.
 
You must log in or register to reply here.
Back
Top