Avalon ASIC users thread - page 172.

ProfMac

legendary

Activity: 1246

Merit: 1002

Quote from: senseless on June 01, 2013, 11:24:33 AM

Quote from: ?? on ??

Quote from: coastermonger on June 01, 2013, 11:15:20 AM

Just as a cursory reader of this thread it might help if someone posted a picture of the before and after job of what the fuse looks like when it is cut. Or at least a photo with a very clear arrow pointing to whichever fuse is the culprit.

Is this "fuse" thing specific to Batch 2 units? I've never seen anyone discussing it before Batch 2 started shipping, now it's suddenly a big deal...

It's specific to the v1.5 control module. From what I understand they've already fixed all new batch 2 units shipping out. So if you receive a unit on or after May 29-June 1 it probably already has the fuse removed.

This is evidence that the team is tuning the design.
I myself am glad to know this.

WinTame2012

sr. member

Activity: 315

Merit: 250

Official sponsor of Microsoft Corp.

Quote from: drakness on June 02, 2013, 08:30:00 AM

I removed wifi on my batch #2's on Thursday evening... have not cut/removed fused F1, yet. Wanted to report that the NEC usb hub chip is room temperature, to the touch (during hashing).

Leaving it alone, for now. Anyone else besides btclvr try the mod?

Batch#2 was shipped w/o antenna...
I've catched a thought last night that chip overheating could take a place when TP-LINK have enabled Wi-Fi module but missing antenna. If TP-LINK implements dynamic power calibration for Wi-Fi PHY lacking of antenna could force it drawing much more power hence USB-hub chip overheating. This is just my deduction though.

spiccioli

legendary

Activity: 1379

Merit: 1003

nec sine labore

Quote from: drakness on June 02, 2013, 08:30:00 AM

I removed wifi on my batch #2's on Thursday evening... have not cut/removed fused F1, yet. Wanted to report that the NEC usb hub chip is room temperature, to the touch (during hashing).

Leaving it alone, for now. Anyone else besides btclvr try the mod?

Drakness,

There are two options on the GUI: disable wifi and remove wifi, which one did you choose?

I've disabled it friday evening and since then my units have been hashing without problems. In a couple of days I'll stop one and open it to see if chips on the control board show signs of overheating.

spiccioli

drakness

newbie

Activity: 33

Merit: 0

I removed wifi on my batch #2's on Thursday evening... have not cut/removed fused F1, yet. Wanted to report that the NEC usb hub chip is room temperature, to the touch (during hashing).

Leaving it alone, for now. Anyone else besides btclvr try the mod?

spiccioli

legendary

Activity: 1379

Merit: 1003

nec sine labore

Quote from: btclvr on June 01, 2013, 12:33:30 PM

Quote from: abracadabra on June 01, 2013, 08:49:13 AM

Quote from: btclvr on May 31, 2013, 11:09:54 PM

Quote from: drakness on May 31, 2013, 12:31:57 PM

Anyone with a successful removal of fuse F1, yet?

Yes, I removed it from my Avalon and all appears to be working again. The fuse would not pop completely off the pads but enough of it has been removed to open the circuit. It was more of a cut than a removal. A better soldering station might have helped.

Had you noticed any overheating of the NEC usb hub chip?

There were no signs of damage from heat and the Avalon has been stable before and after the mod was performed. The component temperature was not measured.

Btclvr,
are you, were you using wifi ?

spiccioli

btclvr

newbie

Activity: 29

Merit: 0

Quote from: abracadabra on June 01, 2013, 08:49:13 AM

Quote from: btclvr on May 31, 2013, 11:09:54 PM

Quote from: drakness on May 31, 2013, 12:31:57 PM

Anyone with a successful removal of fuse F1, yet?

Yes, I removed it from my Avalon and all appears to be working again. The fuse would not pop completely off the pads but enough of it has been removed to open the circuit. It was more of a cut than a removal. A better soldering station might have helped.

Had you noticed any overheating of the NEC usb hub chip?

There were no signs of damage from heat and the Avalon has been stable before and after the mod was performed. The component temperature was not measured.

uhoh

hero member

Activity: 742

Merit: 500

Circle gets the Square

Quote from: coastermonger on June 01, 2013, 11:15:20 AM

Just as a cursory reader of this thread it might help if someone posted a picture of the before and after job of what the fuse looks like when it is cut. Or at least a photo with a very clear arrow pointing to whichever fuse is the culprit.

senseless

hero member

Activity: 1118

Merit: 541

Quote from: ?? on ??

Quote from: coastermonger on June 01, 2013, 11:15:20 AM

Just as a cursory reader of this thread it might help if someone posted a picture of the before and after job of what the fuse looks like when it is cut. Or at least a photo with a very clear arrow pointing to whichever fuse is the culprit.

Is this "fuse" thing specific to Batch 2 units? I've never seen anyone discussing it before Batch 2 started shipping, now it's suddenly a big deal...

It's specific to the v1.5 control module. From what I understand they've already fixed all new batch 2 units shipping out. So if you receive a unit on or after May 29-June 1 it probably already has the fuse removed.

coastermonger

sr. member

Activity: 367

Merit: 250

Find me at Bitrated

Just as a cursory reader of this thread it might help if someone posted a picture of the before and after job of what the fuse looks like when it is cut. Or at least a photo with a very clear arrow pointing to whichever fuse is the culprit.

uhoh

hero member

Activity: 742

Merit: 500

Circle gets the Square

Mine wasn't even warm, but I removed it anyway...better safe than sorry.

abracadabra

hero member

Activity: 956

Merit: 1001

Quote from: btclvr on May 31, 2013, 11:09:54 PM

Quote from: drakness on May 31, 2013, 12:31:57 PM

Anyone with a successful removal of fuse F1, yet?

Yes, I removed it from my Avalon and all appears to be working again. The fuse would not pop completely off the pads but enough of it has been removed to open the circuit. It was more of a cut than a removal. A better soldering station might have helped.

Had you noticed any overheating of the NEC usb hub chip?

btclvr

newbie

Activity: 29

Merit: 0

Quote from: drakness on May 31, 2013, 12:31:57 PM

Anyone with a successful removal of fuse F1, yet?

Yes, I removed it from my Avalon and all appears to be working again. The fuse would not pop completely off the pads but enough of it has been removed to open the circuit. It was more of a cut than a removal. A better soldering station might have helped.

drakness

newbie

Activity: 33

Merit: 0

Anyone with a successful removal of fuse F1, yet?

senseless

hero member

Activity: 1118

Merit: 541

Quote from: mrb on May 31, 2013, 12:06:53 PM

Quote from: senseless on May 31, 2013, 11:56:29 AM

I finally got altium viewer installed and looked at the board designs. Removing J1 would remove all power to the USB. But you would need to power the tp-link by some other means

You are reading it wrong

F1 powers USB-A1, nothing is connected there. The tp-link is connected to USB-B1.

First time reading one of these things, I'm going blind! What a horribly tedious job designing this stuff must be.

At least it's an easy fix for anyone who's cloning the avalon design. Should just be able to delete the one pathway that leads back to the usb-hub from the usb-a1, no?

mrb

legendary

Activity: 1512

Merit: 1028

Quote from: senseless on May 31, 2013, 11:56:29 AM

I finally got altium viewer installed and looked at the board designs. Removing J1 would remove all power to the USB. But you would need to power the tp-link by some other means

You are reading it wrong

F1 powers USB-A1, nothing is connected there. The tp-link is connected to USB-B1.

mrb

legendary

Activity: 1512

Merit: 1028

Quote from: senseless on May 31, 2013, 02:18:51 AM

Thank you for the explanation. Another question that I haven't been able to find an answer for. Why does removing the fuse matter if there is nothing plugged into the port?

Because the fuse supplies 5V to both USB-A1 (unused) and to the USB hub (which is supposed to get 3.3V).
Removing it makes USB-A1 unpowered (we don't care) and stops the overvoltage of the USB hub (which is okay with 0V/GND on this pin).

Quote from: spiccioli on May 31, 2013, 06:30:06 AM

From the link to github I read BitSyncom answer where he says: "Thanks, this is true, will cause the chip to be about ~20C higher than avg, which is still fine [...]"

So, adding a small heat sink could be a "simpler" solution, isn't it?

With all due respect to Yifu, I think he is too negligent here. I don't think it is "fine". We all saw senseless' picture of a melting chip package. And nobody has enough data to proves whether a heatsink is sufficient or not. Even if it helps, the chip's pin input is still being overvolted, so it is still reducing its operating life...

senseless

hero member

Activity: 1118

Merit: 541

Quote from: jgarzik on May 31, 2013, 09:20:53 AM

Disabling wifi was always a good idea.

Hard line is far more reliable in general, rather than suffer a wifi drop at a bad moment (submitting a difficulty 64 hash). It also uses less memory.

I wasn't using it, it just wasn't disabled.

Quote from: drakness on May 31, 2013, 07:48:15 AM

Bump senseless's question.

If F1 is the power line to USB-A1... if nothing is plugged into USB-A1, how is this causing too much power to be drawn from the control board? Perhaps that is why Bitsyncom said "this is fine"?

I finally got altium viewer installed and looked at the board designs. The fuse definitely needs to be removed.

I thought maybe you could just remove the J1 jumper, but it looks like the usb-hub chip will still be overvolted unless the fuse is removed.

jgarzik

legendary

Activity: 1596

Merit: 1100

Disabling wifi was always a good idea.

Hard line is far more reliable in general, rather than suffer a wifi drop at a bad moment (submitting a difficulty 64 hash). It also uses less memory.

drakness

newbie

Activity: 33

Merit: 0

Bump senseless's question.

If F1 is the power line to USB-A1... if nothing is plugged into USB-A1, how is this causing too much power to be drawn from the control board? Perhaps that is why Bitsyncom said "this is fine"?

spiccioli

legendary

Activity: 1379

Merit: 1003

nec sine labore

Quote from: mrb on May 30, 2013, 11:46:34 PM

The description that the TP-LINK is "drawing too much power" is actually incorrect.

The USB hub chip does not overheat because the TP-LINK draws too much power. The TP-LINK is powered via a 5V line that does not even come from the USB hub chip.

Instead, the hub chip overheats because of a flaw in the control unit v1.5 design which I found while poring over the schematics: they accidentally feed 5V to one of the pins of the chip instead of 3.3V. This overvoltage is what causes the chip to overheat. I pinged Yifu who confirmed it, and I filed a bug quoting his response: https://github.com/BitSyncom/avalon-ref/issues/5

Removing the fuse F1 merely cuts voltage to this pin. It leaves the USB type B port (which powers the TP-LINK) completely unaffected.

From the link to github I read BitSyncom answer where he says: "Thanks, this is true, will cause the chip to be about ~20C higher than avg, which is still fine [...]"

So, adding a small heat sink could be a "simpler" solution, isn't it?

It seems that removing F1 is the correct thing to do, but it can be done, not it HAS to be done.

Am I reading this correctly?

spiccioli

Topic: Avalon ASIC users thread - page 172. (Read 438600 times)