Topic: bad Block Erupter? - page 3. (Read 8316 times)

That's way high HW errors.  You should have around 1%.

Slush's pool is reporting 9.435MHz

while cgminer is showing:
AMU 0: | 290.4M/364.9Mh/s | DA:20 DR:0 HW:106 WU:2.33/m

I'm going to buy an 85°C working temp crystal and replace what's on there and take that out of the equation.

I get those occasionally.... as long as it isn't every second you should be okay.

[2013-07-17 17:14:40] USB: AMU0 read1 buffering 4 extra bytes
 [2013-07-17 17:15:13] Stratum from pool 0 detected new block
 [2013-07-17 17:15:21] USB: AMU0 read1 buffering 4 extra bytes

Have lapped the complex heatsink for better contact with the CP2102 and ASIC tops.  Using thermal compound between the chips and heatsink.  Pre-chilling in AC.  Will shut down bfgminer on the laptop, it having the -S all switch, plug-in the B.E., start bfgminer, plug in fan.  Have an alligator clip to the top of the crystal tho if the heat from the CP2102 is effecting it, this might not improve the situation but instead cause additional heating of the crystal. ...  No joy.  Trying RPi. ... Okay, hashing on RPi via a USB2.0 hub.  Must wait tho as it warms.

First try after repair attempt on the crystal was a bust.  Into hub, short blink of the LED and nothing.  I had used an iron to solder in the crystal.  Now I put the custom protective alligator clip heatsinks in place and while  holding down with the dental pick, heated with hot air.  Cooled in the chill of the AC and tried the hub plug-in test again and got the standard flashings of the LED!  Didn't try to get it hashing yet but looked at the laptops std out and see 'device descriptor read/64, error -71' four time then an instance of 'device not accepting address 40, error -71' then and instance of 'device not accepting address 41, error -71, then finally 'unable to enumerate USB device on port 2'.   So, essentially it's the original failure messages - there's no topside heatsink of any kind.  Replacing the CP2102 is out of the question so I'm left with replacing the crystal.  Meanwhile experience has showed me it will run like this if tucked into the AC outlet.  Will try the two heatsinks and fan.

You rebuilding it?

Wow... hope you get it working.

Keep us posted.

Made a complex heatsink that made contact with both the CP2102 and the ASIC. Also, the heatsink was mounted to two standoffs atop which sat a small 12v fan.  I was undecided whether to power the fan with a wall wart or step-up module powered from the USB 5v. I went with a 9 volt wall wart then changed up to that thru the step-up module for high fan speed.  It was only a little more capable of communicating with the RPi or a laptop via a USB2.0 hub than when it only had a TO-220 heatsink and chilled.  A few mods to the heatsink then it stopped hashing and couldn't get brought back to life.  I looked into what an 11.0592MHz crystal would be used for.  One application online said the frequency was right for the interrupt for multiple processes at some common frequency another said it related to baud rate in communication processes.  So, since it was positioned in what looked like a funky fashion, I decided to remove it and replace it.  Taking two large alligator clips I trimmed them down with a Dremel until each could clamp either side of the crystal protecting the adjacent chips from heat as well as some caps.  I applied light pressure with a dental pick to the top of the crystal while heating with a hot air desoldering tool.  The top pair of pads freed first but the lower pad pair ripped off the device.  Too much pressure, too little heating time.  So, looking up replacement crystals I find only one brand in a Digikey catalog 2010 that was the correct package dimensions and had an operating temperature of 125°C as it was for automotive applications.  Unfortunately not available in 11.0592MHz.   Searching Digikey online couldn't find the frequency at 125°C.  Other brands offered the frequency but with a max of 85°C.  I recall I found the temperature in the area of the CP2102 and 11.0592MHz crystal to be around 250°F with an IR meter.  That's 121°C in the neighborhood of a crystal who's max working temp is likely 85° and that is involved in com baud rate.  So, the heat from the CP2102 is perhaps destabilizing the crystal.  Before buying another crystal I'm trying to repair the present.  Curiously only pins 1 and 3 of the 4 pad crystal have any function.  This might explain the funky positioning, getting max access to the only 2 functional pins.

Lastly went back to the TO-220 heatsink mounted atop the B.E.  Anodized so it doesn't conduct, a reduced Teflon washer as a spacer off the board so it lays flat, heatsink compound, in free air in a 75° room without a fan, it lasted 5 minutes.  This was using the CyberPower USB2.0 hub rather than the new USB3.0.  I don't think the USB3.0 hub feels the signal is good enough to pass.  I'll try it again tomorrow.

The new USB3.0 hub with the room temperature B.E. was plugged into the booted Inspiron 5150 which was at a linux root command prompt.  The new hub was turned on and immediately there scrolled on the screen 'unable to enumerate device' messages and this with the TO-220 heatsink still strapped to the B.E to the surface of the CP2012.  I unplugged the B.E. and the scrolling stopped.  Will try and measure the voltages on Vdd and REGIN (USB voltage in) and will attempt a reset pulling 9 low.

So, the CP2102 among other things takes the USB 5 volts and generates 3.3 volts that it can source to other components.  250°F is hot.  the datasheet says its temp range is -40° - +85°C (185°F) although under absolute maximum ratings, ambient temperature under bias max is 125°C (257°F).

When the AC went to fan only last night as the home cooled for the night, the green LED came on as it stopped hashing.  I took a TO-220 heatsink, put on a dab of heatsink compound, and strapped it to the CP2102 on the B.E. board and returned it to the AC air duct.  It's been running ever since - 11  hours.

So, its heat might be due to high USB voltage that needs to be knocked down to 3.3 volts by the CP2102 or internal switching supply inefficiency or excessive load on the 3.3 volts.  I looked at the poor positioning of the crystal next to the CP2102 and what looks like excessive solder - if it's shorting to the metal case that would load it down.

The new USB3.0 hub is sitting at the post office.  Must get it now and check if it improves the heating problem.

Just looked at the cp2102 datasheet and it's a USB 2.0 device.  Must look tomorrow if there's a suffix.

If it is the CP2012 heating due to incoming slew rate of the data, since this type of failure is apparently common in the USB3.0 world, somebody can make a good buck selling a small signal conditioner to be placed between a USB port and a USB3.0 device.

You could have a poor solder connection some where on the board.
If you can find some thing to magnify with, look at all the solder connections and make there sure good.
You might be able to find out if it is a bad solder connection by vibrating the board while it is running in the AC.
Also I think you can buy cold spray from Radio Shack. Just cool components one at a time and see which one makes it work.

Took it out of the AC and it failed in 60 seconds.  I wonder if the problem may be the USB2.0 rise time.  Lower slew rate causing heat as occurs in poorly designed switching supplies.  Popped it back in the AC and restarted - running okay.  Wondering if the input to the USB3.0 hub will be too fast for my USB2.0 machines.

I suppose Schmitt triggers would be asking too much.

Right.  It flashes when it finds a share and when it is sent new work.  Otherwise, it should be off.

Correct

So, when it's running properly the green LED is off except for when sending a burst of data.