Okay so long as it doesn't expire.

soy; 2; 0.62, 1MNFGC5yp3JFCmuA6S1LUBoHG5bFuqTszE

abc7425b1bb92ab329bfb9fcf8afce164c29f3e013999c2f93e1128c4b1ce43e


So the USPO after asking everything but my blood type want's to know what day it's to be shipped on.  What should I tell them?

soy

Knock wood it's been hashing all afternoon like a champ.

It seemed positioned as perfectly as I can get it by hand.  The clothespin with screw to hold it in place had to be modified.  Looked perfect after soldering.  Tried connecting but got comms errors.  Heated again, tried again, now hashing at 333.4MH/s.  Apparently success!

soy

Checked the voltage feeding the resistor/LED and found 5vdc so I figured either the CP2102 internal switching supply was dead or the mispositioned chip was shorting pins 6 & 7. 

Started heating the chip to remove, saw a very nice flux bubbling, better than yesterday, and decided to just test the voltage out of pin 6 again.  Now I was getting a very stable 3.5v at the resistor/LED! 

The original CP2102 was allowing the voltage at pin 6 to migrate as high as 3.9v.  Since when pin 6 is at 5v flashing LED appears okay tho marginally brighter, connect failures happened.  Whether the 3.9v out of the original CP2102 was a problem we don't know. 

So, I cleaned, reassembled and tested with the stable 3.5v out of pin 6.  Proper LED flashing but failures to connect, comm errors from AM0 on cgminer.  Bummer.

The IC has been again removed from the BE and after lunch I'll try remounting it. 

I hope the CP2102 doesn't need any programming. 

Only having purchased a single CP2102, if it isn't working today I'll order one or two more.  I'm sure it can tolerate some heating but repeated mount, un-mount, remount can't be good. 

Noticeably, the original CP2102 while attempting to connect got very hot, this isn't heating but that may change with remounting.

Unknown if any other chips on board fried when the hub saw 12vdc.

CP2102 replacement failed.  Will try again tomorrow.  Darn thing shifted on me.  Looks like I'm not yet eligible for a B.E. coupon.  I don't expect they're real tempted to issue coupons to those who paid 1BTC + shipping when they can offer those to anyone who will buy a back order B.E. at 0.6BTC + shipping.

Does this count?  https://docs.google.com/file/d/0B_DLmNkCkkVpWlpfa3BUVm9iZDQ/edit?usp=sharing

I don't quite understand the question.  I was getting failure to enumerate messages.  The changes were attempts to get stable operation, wasn't "upgrade"ing.

  Ah well, moot.  I devised a heatsink and fan.  Moot because I changed the crystal, then added a startup network to the crystal, then added a tantalum capacitor to the USB power in.  The close proximity fan caused dust to build up in lees and that degraded performance after a while.  The startup network worked well, the B.E. was running on a not-approved Cyber Power hub as the room climbed into the mid-80°'s F.  Usually the B.E. would shut down if the ambient got to 79°+ registering impossibly high hash rates on the minepeon cgminer.  I decided to buy the D-Link DUB-H7.  It arrived today.  I also had purchased a Seagate Plus Backup for my main system which I'm going to use today.  Both were open.  The wall warts have identical plugs tho the Seagate Plus is 12v while the D-Link DUB-H7 is 5v.  I noted to friends that the cheap 7 port hubs on ebay, external 5v plug same as the D-Link, hardwired to the upstream port and that 12v wall wart plugs fit perfectly (bought a L-C meter that came with one).  Don't know if the D-Link upstream port is hard wired to the voltage in plug but it certainly goes to the downstream port.  That gremlin that danced around on the wing of an aircraft on Twilight Zone must have guided my hand as I fired up the block erupter on the D-Link hub using the 12v wall wart.  The CP2102 quickly fried.  My B.E. is dead. 

soy

PS: A replacement CP2102 has been ordered, should ship Monday and arrive later in the week.  With luck it was the only chip that fried.  The D-Link hub is fine.  My just short of 65 year old eyes and reflexes aren't what's needed to remove and replace a CP2102 so odds are less than even money it will succeed. 

I'm not buying it.  Sealed after testing at the production facility and sold without color choice.  No valid reason to open the box.  With a big back order line and anxious buyers, breaking the seal sounds unreal not to mention the question of what kind of testing could possibly be done for what period of time given the numbers after which there'd still be back orders and opened boxes of BE's that were iffy or returns.  Nah, they'd be in and out and seals opened by the lucky customers.

I mentioned in an early post in this thread that the Block Erupter arrived and the small box had a seal that had been opened.  The box was unsealed.  I only own a single Block Erupter.  It's still not right.  I imagine it makes perfect sense to send returned bad B.E.s to customers who only get one.  No way to compare performance, is there BTCGuild.

I see the pads at the end of the B.E. and expect the USB 5v is there.  Wonder if one would see any advantage to bringing out that 5 volts from a number of B.E.'s and powering them as a group.  Would less current on the USB supply 5 volts improve communication?  Although I see some USB2.0 control chips are output to individual chips for each downstream port so the 5 volts might not be common downstream.

Researching the crystal, there is a discussion about its "start-up" capacitor values tho it's an unrelated application.  The range of start-up caps was 18pf-33pf depending on which manufacturer's datasheet.  So, I made a small start-up network using 18pf caps and added it to the Block Erupter.

https://docs.google.com/file/d/0B_DLmNkCkkVpaE54VGd3dW50Wk0/edit?usp=sharing

There was no improvement with startup and possibly was more difficult to get running.  I see when the B.E. shuts down from ambient temperature >80°F, cgminer reports the hashrate goes into the gigahertz range.  I suspect the crystal but it could be a parasitic oscillation of some kind.  It remains to be seen if the added start-up network will improve functionality when ambient temperature rises above 80°F.

A replacement crystal arrived.  It has an upper packaging of the black material like an IC rather than metal.  I'd wonder about the metal, picking up heat from the CP2102.  The B.E. has been running well on a RPi but hasn't connected to bfgminer on the Vaio I use with zTex.  I first connected the BE to the USB2.0 hub is assess if the new crystal was soldered correctly.  The LED lit and flashed so yes.  There was a single failure to enumerate message.  The BE had no heatsink, top or bottom, but I had no intent of hashing yet.  I reassembled the heatsinks and tested again.  I now was getting clean USB connects to the Vaio, no error messages at all.  I changed the -S option on bfgminer from only the single port for the zTex to -S all, and tried both.  Bfgminer would hang.  I was seeing no indication bfgminer registered the presence of the BE.  

Putting the BE on RPi again, there was a single comm error and it was hashing.  Ambient is 73.4°F and excessive errors occur when ambient approaches 80°F so it should be fine.  It had accumulated 20 HW errors early but that has been unchanged for 25 of the 30+ minutes it's been hashing at ~334MH/s.

I tried the BE with the _USB3.0_ hub to the Vaio and quickly got a page of "unable to enumerate USB device" so changing the crystal isn't the solution.  I had determined the 11.xxxMHz was a crystal and not a resonator (which includes a pair of capacitors) by its size.  I'm using a 2010 Digikey catalog so maybe the 2012 show resonators in a 5mm x 3.2mm four pin package but this crystal behaves like the original so....  The resonators were larger.  Looking at the BFL Jalapeno schematics, they use an FT2232 and its datasheet shows it using a 6MHz resonator OR a 6MHz crystal with attendant external capacitors.  I don't see an appropriate capacitor pair for the BE 11.xxxMHz crystal.

So, it's leave it with the new crystal and put it back on the RPi.  

On another subject, I wonder if MinePeon startum+tcp can source to another system rather than run startum+tcp on two systems or if my best bet is to put the --no-stratum switch in cgminer.conf and use the same switch with bfgminer then run stratum-miner-proxy.

https://docs.google.com/file/d/0B_DLmNkCkkVpbWVWZkZvdlNFVzQ/edit?usp=sharing

https://docs.google.com/file/d/0B_DLmNkCkkVpLTducXJsVWwxSVk/edit?usp=sharing

Maybe try this: shut down the fan, start it up so it fails to connect and gets hot.  Let it sit getting hot for some minutes.  Then when it's not going to get hotter, apply light pressure to the top of the CP2102 and shut down the power allowing it to cool while applying pressure.

If the hottest part of the CP2102 is at the internal bottom facing the ground plane and if it's suppose to be soldered but wasn't heated enough, since the temperature above the CP2102 can reach 250°F+, then the area near the ground plane is hotter, perhaps hot enough to liquify the solder.  If so, after cooling and having pressure applied topside while cooling, the next time it fires up the ground plane can draw away the CP2102 heat more effectively.

That or the BE gets killed.

What's your ambient?

You might try a heatsink designed for a TO-220 transistor.  That's what worked well to cool the CP2012 while under a fan - I needed to modify a Teflon washer of the kind typically found in a TO-3 transistor mounting kit.  It's fairly large so I had to cut a 90° notch so it would fit then had to lap it sandpaper to get the  height right.  Then you need to get a #3x.50-10 screw to hold it in place and a dab of heatsink compound.  It doesn't sit square and in fact is 45° off the line so it doesn't see interference from the USB connector.

I wonder if the CP2012 needs to be soldered to the board at its center so it could cool better but I'm not going to expose it to high enough heat to test the theory.  I suspect the problem is more closely related to the 11.xxxx crystal.  Note there is very little information on the crystal aside from the frequency.  The BE manufacturer may have gone to cheaper crystals that can't take much heat.

Right, the 11.xxxxMHz crystal. 

Wonder if a simple barrier of some kind between the CP2102 and the crystal would improve communication.  Not metal naturally, a thin piece of Styrofoam as a wall sitting on double sided tape would show if the approach would help.

The CP2102 datasheet indicates the chip is a USB2.0, interface I believe.

It's the cgminer indication of the BE stats I believe:

AMU 1: | 333.3M/334.9Mh/s | DA:392 DR:1 HW:2 WU: 4.45/m

The room temp dropped to below 80°F, the B.E. had been disconnected from cgminer and the miner sitting idle.  I flipped the BE upside down so the attached fan would be drawing air from below thru the steel bars of the rack, unplugged and hotplugged the BE, it came up and has been hashing okay.  I think the 80°F temp is the key.  Still, the failure to connect to anything but the RPi is unacceptable.  I've ordered two crystals, one good for 85°C at the correct frequency and one good for 125°C at 12.000MHz.  What the heck right?

 AMU 1: | 335.1M/334.6Mh/s | DA:165 DR:0 HW:2 WU: 4.71/m