NEC 3520A strange burning behaviour



Hi all,

I’ve just got a NEC 3520A and flashed to LD3520_1UF firmware. The writer plays DVDs and CDs fine, its just burning seems to be an issue. I have an Athlon 1400 on Asus A7A266 mboard with 512MB running Win2K-SP4.

When I try to burn a 4GB ISO to my Verbatim 16x (MCC004) discs at 16x in Nero the read memory buffer drops to 50% and the writer buffer seems to go empty quite a lot, presumably meaning my PC can’t get the data to the writer buffer quick enough, which I know will kick in the buffer underrun… I also noticed Nero was using up all my free memory, so I changed to using Alcohol, and then decided to burn the same ISO to a disc at the lowest speed of 2.4x, which has produced the included performance graph. Should the dips from 2.4x to nearly 1x be happening? Surely my PC is quick enough to move data at 3324KB/s and thus the graph should be just a smooth line right? I benchmarked my primary HDD, it can do up to 60MB/s, and it is defragged fully… I’m convinced the I/O on my motherboard is to blame for this strange behaviour.

In short, if I attempt to burn a disc at 16x the buffer underrun is always kicking in and disc quality scans in cd-dvd speed have been as low as 40% in one case, burning at 2.4x seems to produce discs of quality 97%.

I’m just lost at to why my PC cannot burn at higher speeds? and DMA is indeed enabled.

Any ideas, pointers to look out for or time for a new PC?


For the buffer jump issue read

The burn quality problems at 16x write speed are probably because 16x is a little too fast for your discs.


Have you checked that DMA is enabled for your drives? You may be running in PIO mode. That is usually the main cause of buffer underrun protection kicking in. for more information.


Yeah, DMA is enabled, and checked in Nero InfoTool. This is real weird, even my old LiteOn 2410 is only able to burn at 8x in this system without kicking in buffer underrun protection… I think my PC has ‘issues’.

I’m going to try my 3520 in my old P3 system, when I fix the cpu fan, and report back perhaps… I’m thinking its a problem with my current Athlon system. I wish there was a tool to test inter-bus-device IO, as in transfer rates between IDE channels…?

Here is a screenshot from Alcohol @ 16x… not too good judging by the graph at the end of the burn :s



Maybe you need to update your MB I/o Drivers or Bios etc…


Which IDE drivers do you have running? You should try the standard Microsoft drivers as they tend to be the best for burning.


Hi, Well I did a test burn on another PC (pentium 3), but it didn’t have enough cpu power to burn anything above 8x smoothly… So, I came to the conclusion the overall problem is lack of CPU power…

I therefore put the drive in a PC with plenty of power - a P4 3.4Ghz i865PE 1GB RAM system… and ran CD-DVD Speed 3.8 to check the CPU usage on that system… it shows 63% at 8x, is this right?? Surely this means at 16x the drive is using 126% of the CPU?? So, this must mean that it would be impossible to burn at even 8x smoothly since you would only have 37% of CPU left for moving the data from the HD/RAM to the CD burner?

Am I reading this right?

As for my IDE drivers on my Athlon system, I am using the MS ones now after removing the Ali Magik IDE Bus Master driver. I will test that out later, but I’m slowly getting the impression that you can’t burn smoothly to DVD due to lack of CPU power?

Is this extreme CPU use normal for DVD burning? Other drives?



Oops, I’m confusing myself here… the above graph is for ‘reading’ a DVD+R, and comparing this to screenshots of other drives reading DVD+R discs the high cpu usage seems the norm, much higher than reading a dvd-rom disc. A lite-on managed to hit 98% reading a DVD+R @ 8x. Strange.

I guess when you are actually burning a disc, you are not reading it at the same time, so the cpu usage for the actual writing would be lower.

Having said that when I was burning with Alcohol @ 16x the cpu usage was continuously 100%, which I still find bizarre since we’re only talking about 21MB/sec transfer to an IDE device, which is running at DMA-33, and its an ISO image, so one large file, perfect for streaming surely?


Thanks anyway!


I can’t see much wrong with your graph. Here is the same test carried out on my AMD XP3200+ system, and you can see there isn’t a hell of a lot of difference. In fact, my CPU usage is a shade higher than yours.

Also, I could burn easily at 16x on my Athlon 1.4 running at 1.5, even whilst browsing so maybe your motherboard causes a bit of a bottleneck somehow.


I used to have an Athlon 1800+ with 768mb of ram and had no problems at 8X burns would use about 50% of processor. You can open Windows task manager and watch it during burn to see your CPU usage. I am now using a AMD 64bit 3000+ system and cpu usage is down to around 25%.