NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
Solved
WNDR3700v1 wireless throughput half of DIR-825
Hi, First, am I remembering correctly when I think there used to be a dedicated subforum for the 3700? If it is here, I don't see it. It's been years, though; I have had this router a long time...
Then it's definitely a hardware failure.
That's the plan! I ordered the new capacitors today, so they should be here within a week or so. Thanks for pointing me toward hardware-- I was focused so much on the often-buggy firmware that I could not see the forest for the trees. Once I was thinking hardware, my mind immediately went to capacitors-- that was why I opened the unit. I don't think there is anything else I would be able to diagnose or fix inside the unit.
What a pain cheap capacitors are. I also have a Cooler Master PSU and an LG Flatron monitor that have partly or totally failed, and upon inspection were found to have bulged capacitors. The PSU is not really worth fixing as it was a lower efficiency unit without PFC anyway, but I will also try to fix the monitor (I ordered the parts for it too). The people over at badcaps.net are helpful for all kinds of capacitor issues.
The four caps for the 3700 and the five for the monitor, in premium quality items with low ESR, cost under six dollars, and that's the retail price for a very small number of them. It would have been a lot less for Netgear or LG to buy those instead of the Ltec and Su'scon caps, and the devices would last longer and make for happier customers. Sure, they made it past warranty, but customers expect to be able to use things they buy longer than that, and having an item die on them before they think it should have makes it more likely they won't buy that brand in the future. Since capacitors are the most likely (as far as I can tell) components to fail from normal use of a properly-designed electronic device, does it really save money in the long term to use cheapo caps rather than good ones?
I know when I was in the market for a new PSU (active-PFC, gold or better efficiency, modular), it was top-tier caps across the board or nothing. Inspecting the capacitors and other components in a PSU is the norm for all of the good tech sites that do reviews on them, but that level of attention has not filtered down to reviewers of things like routers, unfortunately. If top tier (Japanese, usually) capacitors were a selling point as they are for PSUs, motherboards, etc., I am sure Netgear would have used them. I'd bet that all of the other consumer router makers use cheapo caps too; this is not a Netgear thing. It's a consumer electronics thing.
I will post again after the replacement is done!
I'm excited to know the outcome!
It worked!
I got the new caps today and put them in. It was easier than I expected, but I definitely need a desoldering iron or solder sucker.
As soon as I got it all put together, it started right up, and when I tested the wireless speed, it was back up to where it should have been, ranging from 13 to 15 MB/s for the transfer of a 4 GB file.
Now I just get to worry that I somehow messed something up and it will fail again, but there's nothing I'm aware of that I messed up.
Here is a pic of the new caps installed:
...and it was well worth the effort, too; the DIR-825 can't compete with the WNDR3700. I know these are old routers, but the 3700 is AFAIK still in production in its fifth or so revision, and the D-Link was only recently EOL'd, so maybe it's still relevant to someone other than me.
I tried using an online game, one that I had QOS and WISH rules set for maximum priority, in the DIR-825. I then sent a large file (4 gb) from my laptop wirelessly to my sorta-NAS PC while I watched the latency on my main PC (so three PCs total involved here). The file copy only affected the main PC insofar as it was loading the router and testing its ability to route packets from WAN to that PC while loaded.
The latency in the game went up and up (it's an average latency over some period of time displayed in game), causing a noticeable lag in the game. This is precisely the kind of thing QoS is supposed to prevent, but it was happening anyway, even though I had QoS on and WISH on in the D-Link. VOIP would have also been affected negatively.
I tried disabling WISH but leaving QoS on; I tried putting QoS on and WISH off; I tried putting them both off. I tried changing QOS settings. Some attempts were worse than others, but all of them increased latency at least 300 ms beyond the usual. And that's with the router only having to handle 15 MB/s or so, well below the 100MB/s of gigabit to gigabit transfers, although the wired connection would not have to process the WPA2-AES encryption either.
After the new caps were put in, I tried it on the (again fully functioning) 3700. The entire file copied without even 1 ms increase in latency in game. Not even the slightest hint that something was happening could be seen in game.
I know the WNDR3700 has twice the clock rate in its Atheros CPU as compared to the Ubicom in the DIR-825 A1 (no idea how they compete in instructions per cycle; I could not find any info about that)... but the B revision of the 825 has the same CPU and clock rate as the 3700, and according to benchmarks, the 3700 still stomps it in throughput. I have no idea why-- I cannot imagine that the firmware on the D-link would be SO bad as to harm the performance that much, but who knows? At least the rev B can use the same DD-WRT as the 3700, which may improve performance. The A1, with its Ubicom cpu, cannot accept any alternative firmware that I know about, so the performance I described was all I was going to get.
I have thought about putting DD-WRT on my 3700 (again), but it is working so well... why fix what is not broken? Until there is some feature I want that is not in the stock firmware, or until I discover a bug that affects me, I can't think of a good reason to do that.
That's awesome!
It cost you so much less than buying another brand new one.