WRT54G fried - plugged AC into a DC device. D'oh. Any suggestions?

slade969 hath wroth:

Argh. If you had read carefully, you would have noticed that the WRT54G v1.0 and v1.1 hardware versions are 5VDC only. The others will run on anything between 4.5VDC and 18VDC. It's dead.

Maybe you're lucky. There's a protection diode on the v1.0 and v1.1 versions. See:

It's the black rectangular chip with the polarity stripe, to the right of the SOT regulator package near the power connector. It's probably shorted which may have protected the rest of the circuitry from your destructive tendencies. Look for a bulge or dimple on top of the diode, which usually indicates it has exploded inside.

Welcome to Learn By Destroying(tm). My theory is that you don't really undestand how something works until after you've trashed and repaired it. Have your soldering iron warmed up and ready for action. Incidentally, you're not the first one to have done the overvoltage smoke test. I think your #3 that I know about. Nobody realized that the v1.0 and v1.1 versions were different until fairly recently. Just to make life really interesting, there are apparently two radically different board layouts for the v1.1. One resembles the v1.0 while the other has a wide range regulator similar to the v2.0 and above.

The diode is the most common culprit. There are some other hard to find chips nearby that might also be fried. If that's the case, rip out the diode, regulator, FET's and most everything related to the voltage regulator, and replace it with a more modernish 3.3VDC regulator. That's what the internal chipset runs on.

If you want to try your hand at surface mount repair, I think I can supply the numbers for all the chips in the regulator section. I think the 8 pin SOT regulator is an RT9202. The associated FET's are

9965 or 9985. I have not been able to find sources for any of these.

OK now is a good time to add some info... I also have a WRT54G v.1.0 that I got not working. THe OEM wall-wort was the problem. Here is the schematic of its output ckt...

+---+----|>|----+-----+-----L1-----+----+------->+5V | | | | | | ( +--10R--1n--+ 330uF 220uF +-1K-+--(1)KA431 Vref=2.5V ( 25V 10V | ( 105° 105° 1K | | | | +---------------------+------------+---------+--(2)KA431

The 330uF, 25V input cap is adjacent to the rectifier diode & had overheated. It had a "rounded top".

The wall wort would put out 5V unloaded but dropped to almost nothing with any load.

Unit worked just fine with a good +5V wall wort. (also flashed to dd-wrt)

Here is a partial schematic I had made (ascii art)

+------>7705 chip | Jack1(+)--L2---F1---+-----+-----+-----+-----|>|--->SMPS (-) | | | | 0.188V | +-----+ | | | | 1000uF| _|_ --- --- --- L1 16V --- /_\\ --- --- --- | 105° --- | | | | _|_ _|_ _|_ _|_ _|_ _|_ - - - - - - ID= DS1 CK1 CA3 C3 D1

Device Zss smt smt smt Zss ID 33 chip chip chip 33

"F1" is the yellow thing directly behind the pwr input jack. using the pic Jeff posted...

"DS1" is mostly hidden in that pic - you can see part of it between F1 & L1, the inductor closest to the pcb edge.

The markings on "F1" are...

30V JF34

Its not easily replaceable (soldered in).

Find a friend that knows some electronics & he/she should be able to find the problem & maybe repair it.

Now for some more info on powering this unit. I had traced out the input +5V lead & found that it also goes to some pins of the "radio card". BUT I could not detect any connection to these pins (on the radio card).

What I did discover is that the +5V also goes to a TI 7705AC chip at the front edge of the main pcb & under the radio card. Its a "Supply Voltage Supervisor" chip monitoring the health of the +5V line.

The 7705 chip pdf shows... Vcc: 3.5-18 V Vsense: Recommended Max= 10V

So it seems the resistors in the "Supply Voltage Supervisor" ckt would be the limiting factor in higher voltage (DC) input to unit.

I have run this unit on a 6.5V wall-wort for a few days with no problems.

kc - learned everything I know about wireless from this NG - thanks guys & girls!

PS - Don't try to run a D-Link DI-524 above 5V. I was testing the voltage/current curves & at 6.5V the PA amp shorted. I was able to remove it & bridge the _tiny_ gap (what a pita!!) & the unit still works fine, but _somewhat_ reduced power :)

Kim Clay hath wroth:

I took a better photo. See:

This is my WRT54G v1.1 which might be quite different from your V1.0 version. If yours has the 5 terminal voltage regulator (near bottom center of photo) marked Anachip AC1501-33:

then you have the version that can handle a wide range of input voltages. If you have a tiny chip, as in the above Linksysinfo photo, you have the 5V only version.

Note that the diode has been replaced with a 1N4001. F1 is suppose to be a fuse of sorts, but is really a varistor, which has low resistance when cold, but increases resistance radically when hot. It's doesn't blow up like a fuse and recovers when it cools down.

The what chip? I haven't seen that number in any WRT54G boards that I've played with. It appears that there are a variety of power supply designs, some of which are not distinguished by different hardware version numbers. Oh-oh.

Yep.

I have the 8 pin SOT regulator - RT9202 My pcb looks identical to the pic at

Fi seems to be a PolySwitch of some type.

but I could not xref the number I found on the device

The markings on "F1" are...

30V JF34

TI TL7705 - 5V "Supply Voltage Supervisor" The +5V from the pwr jk runs directly to pins 7 & 8 of this chip.

This chip is really out of place as it is under the front edge of the radio board, way away from the PS section.

Looking at the V1.0 pcb at

the chip would be on the main pcb under the far (front) edge of the radio board, centered about 1cm from the front edge of the pcb - its a "Supply Voltage Supervisor" 8-pin SOT. This pic matches my pcb in every way.

afaik its the only device that uses direct +5V power besides the smps.

Kim Clay hath wroth:

That's the 5V only incantation. Don't try it with higher voltages.

Note that it uses external FET's to increase the output power. Those are what usually blow up.

Yeah, that looks like the part. However, I gave up trying to find the specs based on the "JF34" markings. It kinda looks like a LittelFuse part number, but I'm too lazy to dig any more.

That's a chip that's used to insure that the CPU receives a hard reset when the power supply gets cycled. It's not really part of the power supply system. What I find odd is that it's running on 5V VCC instead of the expected 3.3VDC. That means my suggestion of ripping out the regulator section and running the WRT54G on 3.3VDC isn't going to work. Argh.

Bummer.

Can you post or send me the FCC ID number for your WRT54G v1.0 unit? I might get lucky and find a block diagram on the FCC ID web pile. If you want to dig for yourself, go unto:

and inscribe Grantee Code: Q87 Product Code: -WRT54G Show: 100 records at a time. Note the dash. It's required.

FCC ID PKW-WM54G

1 Applications Were Found That Match the Search Criteria: Grantee Code: PKW Product Code: -WM54G

I dunno if the link will work but the FCC "GenericSearch" found it!

Kim Clay hath wroth:

That's not a WRT54G. It's the WM54G MiniPCI card used inside the WRT54G v1.0 router. Just look at the internal and external photos.

Are you sure that's the number on the router serial number tag? This is getting really weird.

The bottom of the router has 2 stickers - one has the s/n & the mac address & no FCCID #.

another sticker has: FCC ID PKW-WM54G IC ID: 3839A-WM54G

Actually the external sticker on the bottom of my unit that has the FCCID # is the same sticker that in on the RF shield in this pic

My unit has identical stickers on the bottom of its case & on the RF shield internally.

Kim and Jeff,

Many thanks for all the help - especially the photos and schematics.

Perhaps my best bet would be to find a replacement for the polyswitch in F1?

Everything on my board seems to match up with exactly with Kim's - including the PKW-WM54G FCC ID. I'm guessing that since the wireless component on v1.0 is packed onto a detachable mini-PCI card, that's the only component that Cisco was required to register w/ the FCC.

Slade

slade969 hath wroth:

Nope. It's self-resetting and does not blow up permanently like a fuse. You can check it with an ohms-guesser. It should be fairly low resistance between the leads. I would need to find the data sheet to get the exact value (or tear mine apart and measure it).

Nope. The whole unit is suppose to be type certified as a complete assembly. Same with any accessories such as the power supply and antennas. Something is very odd about this type certification.

hmmm... maybe more - First is to find out _what_ is defective...

Root cause of failure was the 12VAC applied to the +5VDC device.

The reversed polarity that the AC provided (the neg 1/2 cycle) was probably shunted to ground by "DS1", possibly shorting it. Thats minor.

the ascii schematic again +------>7705 chip | Jack1(+)--L2---F1---+-----+-----+-----+-----|>|--->SMPS (-) | | | | 0.188V | +-----+ | | | | 1000uF| _|_ --- --- --- L1 16V --- /_\\ --- --- --- | 105° --- | | | | _|_ _|_ _|_ _|_ _|_ _|_ - - - - - - ID= DS1 CK1 CA3 C3 D1

Device Zss smt smt smt Zss ID 33 chip chip chip 33

The really bad part is the positive 1/2 cycle of AC. It went way past the +5V design limit. A 12VAC wall wort will produce peaks of +/- 17V in normal operation.

The wrt54g v1.0 uses a RT9202 "Single Synchronous Buck PWM DC-DC Controller".

From the pdf it is designed to operate with a +5V input. It has a "absolute maximum input voltage" of +7V. Not good if it got +17V. The RT9202 Vcc pin is tied to the +5V input of the wrt54g via a 10R resistor. Not much protection from overvoltages.

Still this chip may have survived if the "DS1" diode shorted fast enough and/or the fuse/PolySwitch opened _very_ quickly.

Find a electronic tech/hobbyist someplace to have a look inside. I would give it (at best) a 50/50 chance it can be repaired easily.

Remember: First is to find out _what_ is defective.

kc

I found a detailed pic of the Version 1 and 2 routers side by side at...

~1.8M

Directly below the yellow "F1" PolySwitch device are the devices I mention on the schematic... DS1 (slightly offset to the pcb edge) CK1 CA3 C3 D1

Next to D1 is the RT9202 switch regulator

The above pic also clearly shows the location of the TI 7705AC chip, as the radio card has been removed. It is at the front edge of the pcb directly inline with the front "guide/retainer" for the radio card.

Another set of good pics can be found at...

including this pic of the FCCID tag on the bottom of the V1 unit

kc

When I was doing FCC certifications, you had to certify an entire system, but if you changed or added an internal board that had its own FCC certification with another board that had its own FCC certification, you stayed in compliance (at least legally). You'd go crazy with certifications if this wasn't allowed.

Of course full advantage was taken of this rule. For example, if you were certifying an add-on card for a PC, you used the best possible PC in terms of emissions (which was usually something from IBM). If you were certifying a PC, you used the best possible add-on cards in terms of emissions, even if that's not what you were going to ship.

The same thing applied to peripherals. They insisted on testing with all the ports in use. If you had a parallel port, you used an HP Thinkjet printer, which had very low emissions as the printer (the labs that did the testing kept these around).

SMS hath wroth:

Yep. Thanks the TI-99/4 and politics for that. If you were really clever, you could certifty the various components of a system individually, even though the entire system would fail. Officially, the FCC didn't like this at all, but with the specs at 20dB (100 times) tighter than they really needed to be, there was plenty of room for creativity and legal hair splitting.

The previously mentioned type certification for just the MiniPCI board is weird. There should have been a certification for the entire WRT54G with the card. If it was a replacement for another device, there should have been a certification for the previous incantation. It may have been under a different number series prior to the Cisco purcahse of Linksys. Digging...

Looks like "Cisco-Linksys" uses 3 different prefixes: Q87 and Q8G after June 2003 (roughly when Cisco bought Linksys), and PKW for April

2001 thru June 2003. Digging through PKW, I find 70 older type certifications, but nothing for the WRT54G v1.0. So, that's not it. There may have been a different OEM other than Linksys involved, but without a clue as their identity, I can'f find it using the totally worthless FCC ID search tools.

Jeff Liebermann hath wroth:

Hmmm... The plot thickens. Linksys was founded in 1988 so the FCC ID data for 1988 thru 2001 should be there. There seems to be a name change involved which, thanks to the nearly worthless FCC search tools, is not easy to trace. Prior to the Cisco purchase, it was known as "The Linksys Group, Inc". However, that doesn't return anything on the FCC Grantee search.

Duz anyone have a really old Linksys product that has an FCC ID prefix other than Q87, Q8G, and PKW handy?

I have here a Linksys "Ethernet 8-Port Workgroup Hub" (10baseT, with a handy 10base2 BNC coax port at one end) labeled

FCC ID: KFYPEH9 MODEL: EWHUB

This is from 1999 or earlier, but I'm not sure of its exact age.

snipped-for-privacy@VictorTangoEleven.net.invalid (Jordan Hazen) hath wroth:

Maybe. KFY is Runtop Inc of Taiwan. Some of their products look very much like the WRT54G v1.0 if the case were replaced. The lights and jacks appear to be in the correct position. Some of the internal components are similar to the WRT54G v1.0, but others are radically different. Hard to tell for sure. No MiniPCI card slot for the wireless, so these are not the same as the WRT54G v1.0.

Photos:

Runtop Inc is also still in business:

Thanks, but sorry, but that's not it.