Block Erupter USB - Overclocking/ hacking ?

[amigaman]

I just wanted to post my setup, and my results.
Those values are "per stick", and i posted them because no one seems to really know what to expect from those sticks.

Are these values common, do i have excellent/average/bad results?
That was what i wanted to know.

[haxtormoogle]

Just got my crystal oscillator circuit built and installed onto my block erupter.
Without changing the voltage I can get stable output with a 13.5mhz crystal
that is 378 mhz. averaging 350-360 mh/s

Still need to test more.

[amigaman]

Just got my crystal oscillator circuit built and installed onto my block erupter.
Without changing the voltage I can get stable output with a 13.5mhz crystal
that is 378 mhz. averaging 350-360 mh/s

Still need to test more.

1 clock is 1 hash, so 378MHz should give almost 378mhash, minus errors.
I get 335.8 displayed in CGMiner with my stock sticks.
So there is something wrong with yours or you didn't wait long enough...

[r3wt]

Just got my crystal oscillator circuit built and installed onto my block erupter.
Without changing the voltage I can get stable output with a 13.5mhz crystal
that is 378 mhz. averaging 350-360 mh/s

Still need to test more.

1 clock is 1 hash, so 378MHz should give almost 378mhash, minus errors.
I get 335.8 displayed in CGMiner with my stock sticks.
So there is something wrong with yours or you didn't wait long enough...

do your be's ever malfunction at start up? when i restart cgminer, AMU 0 reports 458 MH then it gradually steps down to 335.5 where it settles in a groove and fluctuates about +- .5 m/h. sometimes AMU 1 gets stuck at 400.29 m/h and then crashes and reads "ZOMBIE: 0.000 mh/s  A:0 R: 0 HW: 19

[amigaman]

Zombie and/or High HW is bad comms or power, try another socket.

The extreme fluctuations in the first few seconds seems to be normal, but the avg should stabilize in a few minutes.
My BEs then each stick to 335.5 +- .2, but each one is stable.

[haxtormoogle]

Might be a issue with my clock source setup. Used a 74f04 hex inverter IC to get the crystal oscillating. Might have the wrong capacitance on the crystal. but My block erupter likes to run with a 16 mhz crystal installed and a active mini fan cooling system changed the voltage resistors to 1.2k
Hashing is all over the place between 390-500 mh/s  Im wondering if this is my cgminer settings? sometimes the status led stays on for a bit of time as though it is waiting for work.

[Bluestreak66]

Might be a issue with my clock source setup. Used a 74f04 hex inverter IC to get the crystal oscillating. Might have the wrong capacitance on the crystal. but My block erupter likes to run with a 16 mhz crystal installed and a active mini fan cooling system changed the voltage resistors to 1.2k
Hashing is all over the place between 390-500 mh/s  Im wondering if this is my cgminer settings? sometimes the status led stays on for a bit of time as though it is waiting for work.

What is your error rate? From your post am I reading correctly you replace R1 and R2 with 1.2k resistors? That would be 1.6v correct? I have a 16mhz oscillator installed on the board and am trying tweak the voltage to get it stable.

[sdminer]

interesting post

am also wondering if there is any program out there that can turn these into scrypt hashers.  Is it really something with the components that make them unable to mine scrypt?

[Bluestreak66]

I have it running @ ~440 at 1.48v seems to be stable for now, long term is yet to be seen. I think I'm finished messing with it for now I may play with it some more tomorrow. I'll post pics tomorrow if its still alive. R1 is 1k2 and R2 is 1k5 that calculates to 1.44v.

[amigaman]

interesting post

am also wondering if there is any program out there that can turn these into scrypt hashers.  Is it really something with the components that make them unable to mine scrypt?

Ya, and Nay.
The algorithm is hard coded into the ASIC, and in there is SHA, used for BTC and other SHA based cryptocoins.
LTC and other SCrypt-based coins use another algorithm that is not (yet) available as ASIC, so you're stuck to GPUs.
Thats why mining on GPUs gives a few hundred MHash/s on BTC but only a few kHash/s if SCrypt: The Algorithm is different and not very optimized to be done on Hardware.
SHA is.

[pidge]

The pictures are black X'es.
Only me?

For me no pictures at all.

I've fixed the image links

[amigaman]

ah, yes, now i can see.
Your hub is the same as mine...
Did you modify the insides? because the copper seems not to be able to carry 5+ Amps for me...
Also: Backpowering... My PC doesn't like that.

[pidge]

From memory, the traces on the hub are about 2mm wide, the trace from the DC jack to #1 is carrying 4.5A, then to #2 4A etc.  All that might need to be done is "ring mains" the board, and bridge from the DC jack to the un-populated capacitor pad between #9 and #10, halving the current from DC Jack to #1 to 2.25, the other 2.25A or so half being carried to #9.  Though I may check.

No complaints yet from the PC.  I intend to move the hubs over to my RasPi (it's got 6 USB Eriupters attached, in Unitek hubs), with the PSU powering the RasPi also, and I can tuck everything out of the way.

Edit: And sorry for the de-rail!

[btct22]

Might be a issue with my clock source setup. Used a 74f04 hex inverter IC to get the crystal oscillating. Might have the wrong capacitance on the crystal. but My block erupter likes to run with a 16 mhz crystal installed and a active mini fan cooling system changed the voltage resistors to 1.2k
Hashing is all over the place between 390-500 mh/s  Im wondering if this is my cgminer settings? sometimes the status led stays on for a bit of time as though it is waiting for work.

What is your error rate? From your post am I reading correctly you replace R1 and R2 with 1.2k resistors? That would be 1.6v correct? I have a 16mhz oscillator installed on the board and am trying tweak the voltage to get it stable.

Hi Bluestreak66 could you please point me in the right direction to order the correct 16mhz oscillator for the circuit?  A link to a supplier site or part number would be really helpful.

Thanks!

[Bluestreak66]

Might be a issue with my clock source setup. Used a 74f04 hex inverter IC to get the crystal oscillating. Might have the wrong capacitance on the crystal. but My block erupter likes to run with a 16 mhz crystal installed and a active mini fan cooling system changed the voltage resistors to 1.2k
Hashing is all over the place between 390-500 mh/s  Im wondering if this is my cgminer settings? sometimes the status led stays on for a bit of time as though it is waiting for work.

What is your error rate? From your post am I reading correctly you replace R1 and R2 with 1.2k resistors? That would be 1.6v correct? I have a 16mhz oscillator installed on the board and am trying tweak the voltage to get it stable.

Hi Bluestreak66 could you please point me in the right direction to order the correct 16mhz oscillator for the circuit?  A link to a supplier site or part number would be really helpful.

Thanks!

This is the oscillator I currently have on the board http://www.digikey.com/product-detail/en/636L3C016M00000/CTX691CT-ND/604649 It doesn't seem stable over long periods of time and I get timeouts from the board not communicating after a while. I'm not sure whats causing this, cooling does not seem to be a problem. I think It may be an issue with the on board voltage regulator as I get core voltage fluctuations over time. I haven't had much time to mess with it.

[stellan0r]

About cooling and stuff, a company in Germany makes ceramic double sided tape, I will try to get my hands on some of it.

http://www.kerafol.com/en/innovation-in-thermal-management/products-solutions/adhesive-films.html?amp%3Bamp%3Bamp%3Bamp%3BL=2&amp%3Bamp%3Bamp%3BL=1&amp%3Bamp%3BL=1

About the fridge, last week I followed my gf to a Tupperware party and I did not like it. BUT they have those nice boxes where you can put cheese or meat in the fridge and they have a membrane in the lid which keeps humidity out, but meat and cheese fresh. Maybe one could put a BE in it (on a usb extension cable) and see if it works? Btw the boxes are airtight except for that membrane.

[btct22]

I've got a bigger adjustable power supply here, does anyone know what would be involved in replacing the on-board power with a seperate external supply of ~1.4V ?  Eg cut a track, or remove a component to disconnect the internal 1.4V supply.  That would allow easily adjusting the power with a faster crystal installed.

[Bluestreak66]

I've got a bigger adjustable power supply here, does anyone know what would be involved in replacing the on-board power with a seperate external supply of ~1.4V ?  Eg cut a track, or remove a component to disconnect the internal 1.4V supply.  That would allow easily adjusting the power with a faster crystal installed.

No need for that got it @ 447Mhps stable @ 1.47v, I think it can even handle higher. 

[btct22]

Wow that's awesome!  Do you think it will handle faster crystals than 16mhz with good forced air and finned heatsink cooling?

[Bluestreak66]

Figure I better post some pics so people don't think I'm just talking out my arse

Wow that's awesome!  Do you think it will handle faster crystals than 16mhz with good forced air and finned heatsink cooling?

I think cooling will be the only problem.