Possible Solution to the Infamous Two-tone Siren [Retired sticky]

swimfast said:

Sadly this didn't work for me... I have a mobile 2400+, and it's so slow at 800mhz, I'm just trying to increase the multiplier so that it runs at stock speed. Anyone think they can help?

Click to expand...

A few posts up it was said that it might be voltage related so i'd find out the highest voltage you can run without it crashing. I could get to 1.85 which is 1.9 in bios...

BTW i did this and it seems to have worked. I'm using 1.88-1.9 idle and it's done prime for a few hours so looking good although will be interesting to see how long it lasts.

Rich Avery said:

I'm using 1.88-1.9 idle and it's done prime for a few hours so looking good although will be interesting to see how long it lasts.

Click to expand...

See, mine would have never been able to last more than 5 minutes at 1.9Vcore... so you may be one of the lucky ones without the problem

That sux!!

When I bought mine I saw this post first and bent the diode back to around 45 degress. I was running 240x10x1.87 for a good few weeks without problems. And then while testing 1gb of ram it started happening but I don't think it was related to going from 512mb to 1gb.
Then took the heatsink of and bent it almost vertical and forced the cpu back on which bent it back down a little... haven't had the two tones since but have had a few ram related crashes. Fingers crossed it won't come back.

It definetely stopped happening when I lowered the voltage to 1.85 pre bending the diode vertically up.. this meant I had to run no more than high 2.3ghz.
Now on 233x10.5x1.88 primed for an hour no errors.
BTW Mobile Barton can do 2.6ghz @1.8vcore so if your board can do 1.8 without the sirens it might be a nice upgrade and a solution.

Well I won't have this problem anymore as I'm getting rid of this late C4xxxx board. Last weekend, I was very lucky to find an early, used NF7-S here in Japan. Got it for $60. I have only seen two used NF7-S Rev. 2.0 in a year here, and the other was a late board so I passed.

This board is an early C3D06505 board with all black IDE/floppy. 'C3' probably means made in 2003. This is one of the first NF7-S Rev. 2.0s with Cr18D chipset. Only missing mosfet is by the AGP. I swapped out the colored slot board with this one and used the exact same CPU. I purposely did not touch the probe as a test. Sure enough, not a single shutdown running between 2300-2400 mhz using 1.875-1.9 volts!

Morever, like the first NF7-S I got 1.5 years ago, this board overvolts big time! The VCORE overvolts by 0.03 or so and the VDIMM is outrageous: 2.9 in the BIOS gives 3.06-3.07. 2.8 gives 2.96. The NF7-S I got 1.5 years ago is identical physically except for a plastic lever. The late colored slot board overvolts too, but way less: runs 2.93-2.96 when set to 2.9 in the BIOS.

This used board I got still had the 5-2003 BIOS, and was filty with dust. Good sign. Board looks unmodded, probably a mom-or-pop rig.

Unmodded with factory BIOS, all three boards overclock about the same: my 2x512 XMS3500 runs 223, 2-2-2-5, CPC enabled. Need 1.7 VDD to get #5 test to run >218-220 or so. Memtest #5 locks >223. I need the L12 mod to get higher, or >1.7 VDD. But these boards have good potential with the L12 mod and modded BIOSs, especially with 3.07 VDIMM on tap! The black slot boards look identical to the pic on TicTac's site of where to measure voltages (only missing mosfet is by the AGP).

I'm putting the colored slot board on the used market here. I don't want to pass the problem on to someone else, but hell, I got the board here and a lot of guys must be stuck with the same problem.

Update: A very strange finding last night. You know this used NF7-S board that was working fine with an XP2100 (which I have since sold?). I put an XP1800 in it and had a severe case of the shutdown/beeps! I bent the probe up 2-3x times, nope. I was running 223x10 @ 1.8 VCORE. I could not even reset the board to default clock as the board would shutdown while in the BIOS (even after pressing the insert key on boot).

On a hunch, I took it out and put an untested XP2100 in there (left the probe alone and kept same BIOS settings). Guess what? No more shutdown problems! WTF??? So far, I can run 223x10 @ 1.65 volts on this chip.

Now the XP2100 I had originally had the shutdown problem with a late, colored slot NF7-S, but not this early black slot board. Take it for what it's worth, but my conclusion:

- The CPU is a major factor: if you got a dud, you will get shutdown problems galore, starting at low voltage obviously. Since I have seen the same shutdown problem on a Jetway NF2 board with the same CPU, there might be credence to this.
- The XP2100s may(?) be immune to the shutdown problem, but mainly on the early boards. The XP2100 I sold maxed at 2.32 gig or so at 1.9 VCORE. Not that much of a dud, but it did give me shutdown problems with a colored slot NF7-S, but not the early black slot boards.
- It does seem for others, bending the probe up helps.

I dunno, maybe someone can pick up on this and run with it.

One thing I like about the 8RDA+ (which does not use a probe behind the CPU): it never has the shutdown issue. In fact, I will take this cruddy XP1800 and stick it on the 8RDA+. I bet I can push it to 2.3-2.4 gig at 1.9+ VCORE with no shutdown issues. I'm gonna test this soon as I get bored with the NF7-S (I flip flop the boards).

Clevor said:

In fact, I will take this cruddy XP1800 and stick it on the 8RDA+. I bet I can push it to 2.3-2.4 gig at 1.9+ VCORE with no shutdown issues.

Click to expand...

I've pushed mine up to almost 2.2Vcore without shut downs.

When I had my Abit I had two different CPUs that ran in it, both having no difference between each other with the same shutdown issue

JPetrillo said:

I've pushed mine up to almost 2.2Vcore without shut downs.

When I had my Abit I had two different CPUs that ran in it, both having no difference between each other with the same shutdown issue

Click to expand...

J, are you sure the VCORE is working up there on the Epox board ? I presume it's a late manufacture board. I hear on the NF3 Epox board, it goes to 2.2 in the BIOS but it really maxes at 1.85. That's why I stick to pre 8-2003 8RDA+s, as they don't have the 1.92 VCORE cutoff. Epox is known for doing this. This is the main thing that keeps me from trying any of their later boards.

Damn...

I'm getting this problem alot.

I just ditched a 2400+ Thornton because the computer would start for 3s and stop leaving me the stupid 2-tone siren. I thought the processor was fried. Maybe it is still but this topic gave me second thoughts. When a fried Athlon XP is installed in the NF7-S what is supposed to happen on power-on?

Anyway, I bought a Barton 2600+ and tried overclocking it (I'm a newbie at this).

195 - 1.85 : Shutdown + Siren at Post
195 - 1.825 : Shutdown + Siren at Post
195 - 1.8 : Didn't load windows, Shutdown + Siren
195 - 1.775 : Didn't load windows, Shutdown + Siren

185 - 1.8 : Didn't load windows, Shutdown + Siren
185 - 1.75 : Shutdown + Siren as soon as windows started (task manager loading appz)
185 - 1.7 : Shutdown + Siren after 3min of Prime95.

I was thinking until now that this siren was normal if the computer was unstable, I was advised to start slowly, at 170 at default voltage (1.65V).

To my surprise I got the mothafookiinnnn siren after 20min of prime 95.

I am sooo pissed and in need for a solution.

When I had my XP1800 in there (and it's a good stepping, a JIHUB or whatever, but uncharacterized), I could not even reset it to default clock once I set it to 223x10 @ 1.8 volts. I could enter the BIOS, but it would shut down there.

Like I said, I kept bending up the probe - no good. I knew this couldn't be it. So I tried another CPU, a second XP2100, and no problems even booting Windows and benching 3D at 1.8 volts. I'm gonna take it to 2.4-2.5 gig tonight, with maybe 1.8-1.9 VCORE, and I don't expect any problems. I had my other one running 2.32 gig at 1.9 volts no problem ('cept on the colored slot board). Most the ones I have hit 2.4 gig at less volts so I sold that sucker (and that fruit board).

Obviously, the better overclocking the CPU, the less problems you will have. I'm thinking maybe AMD put a shutdown thing in their later CPUs, but this should happen on all boards.

Problem is this only happens on mild overclocks or higher, so you can't return the board to the store.

Clevor said:

J, are you sure the VCORE is working up there on the Epox board ? I presume it's a late manufacture board. I hear on the NF3 Epox board, it goes to 2.2 in the BIOS but it really maxes at 1.85. That's why I stick to pre 8-2003 8RDA+s, as they don't have the 1.92 VCORE cutoff. Epox is known for doing this. This is the main thing that keeps me from trying any of their later boards.

Click to expand...

Hell yeah it was getting up there. 2.2Vcore is the MAX setting in the BIOS, when I had it set in the BIOS to 2.2Vcore MBM5 was reading about ~2.15Vcore - somewhere around there, cant remember exactly, but it was close enough to 2.2 not to make that big of a difference!

...UPDATE.

We now have reason to belive that this siren is caused directally by the PSU. What wattage are you guys running?

I am 100% sure of the cause of the siren problem - for me anyway: CPU overheating. While that sounds obvious, it was an arcane issue with me. I always use a temp probe on my CPU. I tape the tip to touch the edge of the core. While the tip of the probe is flat, it is shrinkwrapped about 1" down, which makes it thicker here. This part sits on the PCB of the CPU. This section was preventing my heatsink from seating properly on the core.

The tipoff was when I stuck in my 8RDA+ board, and I noticed temps soaring to 59-60 C watercooled (with water at 25.5 C). 3DMark would lock up even though Memtest was clean, and I could not get my XP2100 above 2200 mhz. I had also seen high temps earlier on the NF7-S. I figured oh well, AMD chips run hot. As a test, I pulled the probe off, and no problems with the siren since, on two NF7-S boards and two CPUs which gave me tons of problems earlier. On one (not-too-good) XP2100 I got to 2455 mhz at 1.98 volts. Heh, if it hasn't sirened yet, it ain't gonna happen.

This shutdown system is a very effective safeguard, as I've experienced the problem maybe two dozen times overall, yet I have not fried a chip or burn't out a board. If you get a steady beep after exiting the BIOS, that's usually the ram.

So if you are having the problem, check to see if your heatsink is seated properly, or get better cooling (though it would have to be pretty bad for this to happen with the sink seated properly). I imagine the stock sink with a real dud of a chip + high volts might cause this to happen.

Bending the probe up probably will help a bit to get the board to read accurate temps. More than likely, you're simply getting better seating of the sink on the reinstall. I don't think it's critical to get the probe to touch the back of the CPU. It's probably independent of any software (i.e. HW Monitor) or BIOS linkage. It delivers dead accurate temps to the board to tell it to shutdown. It just needs to be near the CPU or probably reads air temps in that pocket. And don't trust temps in the BIOS or Windows software! When I saw real high CPU temps of 59-60 C via my probe, the BIOS and AIDA temps read OK!

BTW, doing back-to-back testing, the 8RDA+ is overclocking a tad better than the NF7-S. I got that XP2100 to 2455 using a tad less VCORE. But I max out at 221 fsb on the 8RDA+ versus 223 on the NF7-S, with my 2x512 XMS3500 at 2-2-2-5 with CPC enabled. These are unmodded boards with factory BIOSs (I'm doing baselines before I try mods). When overheated, the 8RDA+ doesn't beep (it doesn't use a probe). It will just freeze on boot.

Sentential said:

...UPDATE.

We now have reason to belive that this siren is caused directally by the PSU. What wattage are you guys running?

Click to expand...

One reason I no longer bump pots is I had two PSUs die recently. One lasted three months and the other two years. What happens is that BOOM! The board will shut down while stressing the CPU. You then have to power off completely (pull the plug), before the ON switch on the case works. It's like a circuit breaker. The distinction here is you get NO SIREN.

I think this shutdown with high/low siren is a universal safeguard, as my Jetway Kocab 18P NF2 board did the same thing (it also has a probe behind the CPU). At least I know now it's not the board. I had the presence of mind to keep the Kocab anyway as it offers VDD from 1.6 to 1.9 and VDIMM to 3.0!

BTW, I'm currently running either an Antec 400W (the old model with two pots), a Sparkle 400W (with two pots), or a Fortron 530W (with 5 pots).

Update: yep, siren problem totally solved in my case. Tested two bad CPUs on two different NF7-S boards in the last two weeks (which sirened earlier). Both need 2.0 VCORE to do 2455 mhz or so (old XP1800/XP2100s). No sign of the siren problem even at 2.0 VCORE, now that my w/c block is seated properly. I know for sure now that Sandra Multi really jacks temps up .

Phfew! These boards are a lot of work. Finally got one early black slot board running 232x10.5, 2-2-2-11 with 2x512 XMS3500, CPC enabled. Needs a modded CPC-On BIOS, the L12 hardmod to 200 fsb, an APIC disabled OS, and around 1.8 VDD through the pencil mod. I have a huge copper insert sink on the NB and mosfet cooling. Man, you NF7-S guys are gluttons for punishment!

Was able to get this done in around two weeks and learned a lot, but with five rigs, now it's time to move on. Got a DFI 250GB NF3 and MSI K8N Neo2 Plat sitting in the corner I'm itching to try. I'm gonna be trying the same NF7-S tricks on the 8RDA+ next time I install it.

I had that siren thing happen to me recently and the little diode trick worked. Don't know why it works, just know it does. No issues with the psu here. I have an antec true 550.

kouch said:

I had that siren thing happen to me recently and the little diode trick worked. Don't know why it works, just know it does. No issues with the psu here. I have an antec true 550.

Click to expand...

Glad to hear it

Well,i did the trick with the diode but unfortunatelly the siren is on again,i just changed my cooling to SLK900a+Tornado 92 at 7v but f@cking siren again,i have even copper TT sinks on my mosfet,very good case cooling but siren again.Could a bad bios flash be the reason for it,even a voltage over 1.5 (default) causes that problem,the funny thing is that i've used my cpu (T-bred 1800+ 0321) at 2420 (1.7v) a few months,then i decided do underclock it for a while and from that momet i can't reach my previous overclock again.Can you help me,using D26 Mantarays XT (TicTac),10x in advance.

If you make it so this blue diode is making better contact than stock, will that raise temps? I have redone this multipule(15+) times, and it seems the better I make that thing contact my CPU, the higher my temps go. Started out getting 44c full load, now I am up to 52c!!!!!!!!!

Is this blue diode just getting closer to the actual CPU temp(rather than socket temp) by pressing against it?

BTW, my board never beeped or anything, just thought I would nip a problem in the bud BEFORE it happened...........didnt work out so well so far....

Hey Sentential, I tried out your solution... but it didn't work for me. I was fricking angry about that crappy alarm, when suddenly I thought of something...

I pushed that diode as low as I could. And it works.

I couldn't get stable with 1.9v, but now I could. Thanks Sentential, at least you gave me an idea on how to fix this problem if pushing down the diode with your CPU didn't work.

Hope this helps for other Abit users.

sirene driving me crazy ! pls help

Bought ABIT NF7 S2 with AMD XP 2600 266fsb and tried all to get rid of this f**ing sirene... have twinmos 512 DDR 400, 64mb Geforce, solid PSU ... no other strange matters, no OC... just want to have a smooth 2133 Mhz system

When I put CPU at 100fsb, all is ok... running at 1600 Mhz. But at 133 which is std, got the 2-tone and system shut off before finishing Windows XP start up. Temp CPU never higher than 40 C

tried all solutions on the net, CPU fan direcly at PSU, changing memory, several times re-installing CPU...nothing helps. Shortly ran ok, than again the sirene...

Removed the CPu, looking at the famous " blue diode " , but cannot see anything in the middle of the socket.... I'm a newbee, pls help me. How can I find this bending amazing solution ?

Thank for you all !!