Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[claykin]

I mothballed my 825 B1 last year and replaced it with a Netgear WNDR3700 (which works way better).  I now have a use for the 825 B1, but only if I can get it working properly.

I have Comcast HSI and routinely get 20Mb/s+ down and 3Mb/s+ up with my WNDR3700.  My cable modem is a Motorola 5120.

When I use the 825 I cannot get faster than 3Mb/s down and 500Kb/s up.  I've tried several different Dlink firmware versions and its now currently on 2.07.  Tried WAN Auto and hard set to 100Mb.  Tried DNS relay on/off.  Basically tired everything I see others mention here and on other forums.  I even confirmed its not the modem as I tried the 825 at a different location with a Motorola 6120.

This issue occurs regardless of I'm wired or wireless.  LAN wireless and wired speeds are fine.  Its only an issue when I surf.

BTW, I also tried DD-WRT and OpenWRT and had same result.

I think this 825 may be bad.  Thoughts?

[FurryNutz]

Is QoS and Traffic shaping enabled? Try disabled.
The 5 series cable modem don't support Gb WAN speeds, however should still get the faster speeds.

Check cable between Modem and Router, swap out to be sure. Cat6 is recommended.

I'm on a 50/2 ISP here using the 6180 and it works well. Also used on a 6120 too.

Maybe someone can review your router settings with you using teamviewer if your interested. Its safe and secure.

[claykin]

FurryNutz

Thanks for your reply.

I have tried both QOS on and off.

I am aware the 5120 has a 100Mb/s LAN port.  I tried to hard set the WAN port on the 825 to 100Mb/s but that didn't help.

Again, exact same issues with DD-WRT and OpenWRT firmware.  Tried default settings as well as some tweaks receommended by those forum members.  No luck!

I am using Cat5e and tried a few different cables.  Again, my WNDR3700 with the same Cat5e cable has no such problems.

BTW, the Cat6 article you linked to is probably one of the most ignorant cabling articles I've read in a long time.     FYI, Cat5e is "certified" for GigE (1000baseT) and has been since 1999.  The certification is IEEE802.3ab.

Cat6 provides no advantage for a short 6 foot cable (unless you like to be cool and say you have Cat6).  Cat6 advantage begins at longer lengths (over 70m) where you can usually achieve moderately faster GigE throughput in more hostile environments.  Proper Cat6 termination is required to achieve any advantage and its more complicated than 5e to terminate correctly.  I've seen plenty of improper Cat6 terminations in offices.

In case you care, Cat6 is only certified up to GigE (or 1000baseT).  Same as 5e!  If you read IEEE802.3an you'll see that to be future proofed for 10Gbase-T Cat6A is required.  Cat6 "may" work at short lengths for 10Gbase-T but only if the installer follows some very strict guidelines.  Very few installations will meet these guidelines.

No need for a remote session.  Appreciate the offer.  There's no magic in the settings.  Like I said, I've defaulted it, tried every known tweak and then some.  I think this unit has a hardware issue.

I probably won't spend much more time on it.  The 825 wireless range is decent, but not as good as the WNDR3700.  I've also used the newer WNDR4500 and it has even longer range and better throughput than the 3700 with an Intel 2 x 2 client and my iPod touch which is 1 x 1.

[FurryNutz]

Did you happen to follow the FW update process located in the FW update sticky in the FAQ Libaray before you updated?

[claykin]

Problem seems to be resolved.  Read on...

I removed cover to look at PCB.  1 lelon 100uf 25v cap was bulging.  When I went to replace it, also inspected the solder joints around WAN connector and found 2 pins of the WAN connector had cracked solder joints.  Replaced cap (I keep some spare caps handy and happened to have one close - 100uf, 50V, 105C low esr) and touched up poor solder joints.  Powered up and WAN port speed is back to normal.  I seriously think the broken joints were a problem from day 1.

I feel better about this router now, though more testing is required to confirm its repaired.

Still range is not as good as Netgear WNDR3700 (same chipset) or WNDR4500.  WNDR3700 may transmit with slightly higher power or maybe its the antenna engineering in the Netgear that improves things.  Whatever it is, the 3700 has a stronger radio.  Its easily confirmed using inSSIDer and real world testing.  The 4500 is a tough/unfair comparison since its at least a couple generations newer, has discreet power amps which likely help, and its derived from a newer Broadcom chipset.

Anyway, thanks for your thoughts.

Dlink, stop using "value" capacitors!

[FurryNutz]

How did you take appart the 825?

[claykin]

How did you take appart the 825?

Two screws on bottom under rear rubber feet.  Two screws holding LED diffuser in place to remove PCB.  Really very easy.

[FurryNutz]

Kewl, glad you found the issue.

BTW, Been using CAT 6 for years in short and long lengths. IMO, Has better stability over 5e any day. Nothing wrong in using Cat 5e or Cat 6. Cat 6 better interference management as well.
Information about CAT 6 is for avg users and each user can take it or leave it. Theres nothing wrong in using Cat 6 either. 

Enjoy.

[claykin]

I hope i found the issue. Time will tell.

Cat6 is fine. I just didn't want to perpetuate incorrect info claiming 5e is not 1000base-T certified.

Kewl, glad you found the issue.

BTW, Been using CAT 6 for years in short and long lengths. IMO, Has better stability over 5e any day. Nothing wrong in using Cat 5e or Cat 6. Cat 6 better interference management as well.
Information about CAT 6 is for avg users and each user can take it or leave it. Theres nothing wrong in using Cat 6 either. 

Enjoy.

[Patrick533]

Problem seems to be resolved.  Read on...

I removed cover to look at PCB.  1 lelon 100uf 25v cap was bulging.  When I went to replace it, also inspected the solder joints around WAN connector and found 2 pins of the WAN connector had cracked solder joints.  Replaced cap (I keep some spare caps handy and happened to have one close - 100uf, 50V, 105C low esr) and touched up poor solder joints.  Powered up and WAN port speed is back to normal.  I seriously think the broken joints were a problem from day 1.

Welcome to the wonderful world of ROHS/WEEE. Now that they have removed lead from solder, the manufacturer(D-link) has had to raise it's soldering temperature sometimes as much as 300F(148C) (depending on the soldering profile generated at the time and the solder paste used), causing component damage due to overheating, reduced component life and making solder joints much more fragile, especially in the age of SMT. Quite honestly since you seem to be handy with a soldering iron you may want to review the rest of the solder joints, you may just fix that range problem too(resistive solder joint?). A little liquid RMA flux and some good old fashioned 60/40(63/37 too) Kester may work wonders. The last seminar I attended on the NEW solder by Kester a decade ago, they outright said "for you China Lake trained manufacturing professionals", shiny solder joints and pinhole free soldering is a thing of the PAST. To get a descent solder joint, my people were having to run the soldering profiles(IR) so hot the  2 part LPI solder mask(no cheap dry film here) was discoloring. My failure rate post burn in went form 1 in 500 to 1 in 10 after a good thermal/vibe test(shake and bake). Lets not even talk about water soluble flux! I am so glad I got a break from manufacturing in the "NEW AGE"(8 years so far!), LET THE CHINESE HAVE IT! The downside for people like D-link, their negative feedback skyrockets due to manufacturing related issues that NO ONE has resolved, except for the US Military, they prefer LEAD in their electronics and have stated so. When I played with the "NEW" wire solder and water soluble flux, it was more like cake icing, no flow at 900F(482C). Good luck!   

[claykin]

I see you've been around the block a few times with PCB manufacturing.  I actually did same for a video game manufacturer back in the 1990s.  A lot has changed with lead free solder but if I recall correctly lead free melting points aren't quite as extreme as you mentioned.  Typical low temp 63/37 lead solder melted at 183C.  Typical Tin/Silver/Copper solder used today melts at 215C.

Obviously the increased temp requires protection of temp sensitive components and also means more hand touchup for parts that dissipate heat better (i.e. connectors).  The problem with inexpensive manufacturing (i.e. Dlink) is that they don't visually inspect and hand touchup.  This requires people with skills and that costs $$$.

For my 825, I did visually inspect as much of the PCB as I could.  Looks good.  I didn't remove the RF shields or horrible head spreaders DLink chose for the SMDs so I couldn't get a good look at the QFP legs.  Obviously the BGA part(s) are impossible to inspect visually.

I still think the 825 will never have RF range as good as the WNDR3700.  Several web reviews agree with this.

[Patrick533]

I see you've been around the block a few times with PCB manufacturing.  I actually did same for a video game manufacturer back in the 1990s.  A lot has changed with lead free solder but if I recall correctly lead free melting points aren't quite as extreme as you mentioned.  Typical low temp 63/37 lead solder melted at 183C.  Typical Tin/Silver/Copper solder used today melts at 215C.

Obviously the increased temp requires protection of temp sensitive components and also means more hand touchup for parts that dissipate heat better (i.e. connectors).  The problem with inexpensive manufacturing (i.e. Dlink) is that they don't visually inspect and hand touchup.  This requires people with skills and that costs $$$.

For my 825, I did visually inspect as much of the PCB as I could.  Looks good.  I didn't remove the RF shields or horrible head spreaders DLink chose for the SMDs so I couldn't get a good look at the QFP legs.  Obviously the BGA part(s) are impossible to inspect visually.

I still think the 825 will never have RF range as good as the WNDR3700.  Several web reviews agree with this.

Melting vs wetting is a few degrees off. For decades the hand soldering/touch-up temp used to be 700F-800F, with a Metcal iron we were able to get the iron temps down to 600F and still maintain good flow of 63/37 with RMA, which was considered excellent because of the cost of the ASIC's we were using at the time. My company was one of the first sales calls the Metcal team made back in the day. Sorry for not being clear on iron temp vs re-flow temp. My people had to check and log iron temps daily(Was converting from MIL-STD to IPC-610 also to keep costs down, I imagine IPC-610 has been reved 10 times with ROHS or even retired). Back then using a 1/64" tip and soldering 15Mil pitch by hand (for R&D) was unheard of, almost like micro bonding dies by hand back in the day, I used to watch micro-bonding in the lab, looks harder then it really is once you get used to working through optics. The last time I tried the "NEW"(old) solder it had Bismuth in it, as a R&D engineer I was using an adjustable Weller and had to crank the temp up to 900F to get anything that resembled wicking and flow. Of course with that temp you could hit it hard and get out quick reducing the risk to the part. Re-flow of course is different, a nice 6-10 zone oven really helps. Also baking the larger parts(TQFP, etc) to remove moisture before re-flow helped the survival even when they were stored in vapor barriers with desiccant before soldering via re-flow. Using re-flow for production we had to put the large connectors on by hand or the profiles were getting crazy hot on the components. Back then max die temp I believe was 150C, but you had to consider the thermal insulation/conductivity of the package type also. I assume since they got the ROHS/WEEE problems under control that BGA soldering has improved. Sorry to ramble on, one of my last assignments as an engineer was to perfect production and teach the Chinese how to improve quality of the stuff we designed so I could put myself out of work. They paid me off well in the end, but if I would have known they were going to take everything and tank the economy, I would have taught them how to solder at 1400F.    

I think the band duplexer(splitter/combiner) in the DIR-825 was designed poorly, mine seems deaf also so I switched the RF off and use an A/P with much greater receive sensitivity. My mail box is a good clip away and my Android still is pulling 30Mbps with the high power A/P on 2.4Ghz. They should have used separate antennas too, dual/multi band antennas are almost always a compromise. I could not even use my DIR-825 out back where the DIR-655 worked perfectly. So I would indeed say your WNDR3700 has better RF.

Maybe one of the last acts before I put the 825 in the skeet launcher will be to see if I can bypass the duplexer and add a couple of R-SMA's just to see. Might even be worth keeping then! (behind a good firewall)