Topic: Block Erupter USB - Overclocking/ hacking ? - page 28. (Read 168765 times)

Keep up the impressive work!

Another thing to keep in mind is that the original heatspreader is attached with a silicone thermal pad. Those are not really known for their amazing heat conductance. Of course they are relatively good compared to say, air, but they are are pitiful compared to silver and alumina based thin layers of thermal paste or compound.

Also instead of modifying the USB plug that is soldered on, it might be easier to get a spare socket or the socket from a male-female USB extender cable.

I had the exact same setup as above on both BLock Erupters with the solder mask on. I can tell you there is a massive difference in temp. I only did this to one so I can compare them. They both have the same heatsink from intel motherboards. The bare copper transfers heat much more efficiently. The solder mask also has a cross hatched pastern on it so it effectively insulates the pcb by not allowing direct contact. I noticed this while sanding it off. I'm going to desolder the usb plug off the board and use the data lines from a usb cable then power the board from a external supply. I may bolt this to a water block and see how far I can crank it up. I may even pull off the core voltage supply regulator and substitute my own off- board. One thing I would like to know is where does the 3.3 volt supply come from on the board? Measuring the voltages on the board I get 4.73 on the usb and header pins, 3.4 on the vcc and 1.07 on the regulator output.

Edit: On further inspection it appears the silabs chip has and internal regular the outputs 3.3 externally.

I don't believe it's the missing color layer that improved your thermal conductivity, i believe it's more from not using the original screws.
The layer would be ~30-50µm thick, the 0,5mm silicone layer has around 0,5-1 k/W, a layer ten times thinner could not make that much difference even if it is simple lacquer, don't you think?
Also, it has perfect fit on the pcb side, because it was liquid when applied, and the other side has less bumps than your now blank copper side, i think.

Mine look like they bend from the pressure of the silicone layer, increasing resistance. A Screw in the middle would help

Anyone knows if epoxy or CA will withstand the temps?

If anyone needs a heatsink for the USB Block Erupter's ASICs, the guy who sold me several Raspberry Pi heatsink kits, is now selling only the heatsink 8x8x6 mm, good for the BE100 chip:

http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=121148027445

Damn I want to sand one of mine down now and maybe polish it flat with a dremel. Will see how I go for time tomorrow!

So I removed the solder mask from the entire back side of the pcb with some 1000 grit sandpaper. The difference is incredible! I have had it running now for ~35 min and the heat sink is now as warm as the top pcb, the pcb is not hot any more! Looks like it's time for some overclocking!







I used what I had on my desk rubber bands and chipset heatsink   But it works! I'm only using the thermal sheet between the pcb and heat sink. Power is going though some of the vias and will short to the ground plane if it contacts the heat sink. If I can find one of my ir thermometers I will post some temps.

Edit: I just ordered some oscillators and 0805 resistors, I'll post back when I get some results. 

Thats what I had determined from my research as well. Have you measure the difference in heat transfer without solder mask? I had planed on over clocking with a 14mhz oscillator but ran into a brick wall with the cooling. I tried a variety of cooling solutions. At one point I removed the head spreader and insulation sheet, put artic silver directly on the pcb and bolted it to a intel cpu cooler. The heat transfer though the pcb  was the issue, the cooler stayed cool but the top pcb layer around the chip was hot. If you say that it can be improved by removing the solder mask the I may need to try an remove it. Acetone will not work its used on boards to clean up hand soldered areas so the solder mask is resistant to it. If it's acrylic based standard off the shelf paint stripper can be applied with a Q-tip or brush. Sand paper work well also. The schematic and data sheet for the voltage regulator is posted in one of the first posts. Its a standard voltage regulator that is set by a voltage divider and compared to an internal reference. Let me know if you get any farther with this. I had all but given up on the idea, but I may have another go with it if the temps can be brought down at stock frequency.

Unfortunately heat transfer rate and absolute temperature differential are not the same things and do not have the same impact on cooling performance. It could be in the freezer at -18 deg C but as soon as you apply power to it, that power has to go somewhere as heat through all the materials that surround the ASIC die. It's entirely likely that the die is still sitting at 50+ deg C.

I'd be willing to bet that you could remove more heat effectively with watercooling at ambient temps than running a half-assed sub-ambient cooled system in a freezer at -18 deg C, even if it was actively cooled inside the freezer (with fans).

At sub-ambient temps condensation is a real threat to electronics and I'd be willing to bet if you just put an erupter on a USB extender cable* and slide it into your freezer it will be destroyed within seconds. If you went to the effort of sealing the overclocked erupter in an almost perfectly dry isolation vessel, with heatsinks on both the top of the chip and on the PCB underneath, with a fan inside it to circulate the air, you might have good enough heat transfer to cool the chip better than with just a crude watercooling setup. But generally in that setup you are going to be limited by two things: The thermal conductivity of air is very poor, and the heat must still pass through the isolation vessel itself.

The difference with water cooling is that you have a very high heat capacity medium (water) flowing past a metal surfaces (waterblock) at a high flow rate. The metal itself being copper (or brass in my idea above) or aluminium in most cases also has exceptional heat conductivity that is orders of magnitude better than air.

On the other hand, if you had large enough heatsinks removing the heat into the freezer and the metal directly conducted heat from the ASIC chip top/bottom then you would get superior cooling. But I think it would be a huge pain in the ass. It's times like this where die-mounted temperature monitoring like in CPU and GPU would be great, but considering how much the AM ASIC chips cost it's to be expected that they don't have this feature.

If the erupters had built in temperature monitoring on the ASIC die I would have already tried putting on in the freezer just to see what kind of temps can be obtained. However because monitoring the external chip temp is almost meaningless there is little incentive to try. Don't even get me started on actually trying to get high accuracy reproducible measurements of surface temps with thermocouples or IR pyrometry...

I suppose some good news is that as far as I can see none of the components on the erupters are highly susceptible to damage from sub-ambient temperatures. As many people have found out trying to do 'extreme overclocks' on graphics cards, the electrolytic caps don't enjoy being frozen solid. It changes their electrical properties too much. The spec sheets of the major components do state minimum operating temps and many of them are -25 deg C although I wonder if the manufactures actually do failure temperature tests to determine that. I could understand them testing maximum temperature ratings of course, but minimum temp is a bit unusual in my mind and probably not worth their time.

If you wanted to submerge the unit in oil for your testing it will probably be alright as the only component on the circuit that could conceivably be affected is the choke, because the epoxy could be broken down and the magnet wire insulation degraded over time. Which can and does lead to some failures in large oil cooled power transformers.

*FYI, I have already tested running erupters off USB extender cables. I am currently running two erupters each on a 2 meter USB 3.0 rated cable, on a netbook. Interestingly I measure no voltage drop over the 2 meters and the netbook happily outputs 5.06 volts both at its hub sockets and also at the end of the cables while under load. USB 2.0 cables are usually rated to much lower current carrying limits so those might not be able to handle it. Then again 2.5 watts on a unmodified erupter is pretty small amounts of power to begin with.

Just put it in the freezer, laptop on the top of the freezer, very thin usb cable. Cooling done.

I was thinking about heat removal from these little things and because of their size it presents some unique problems. If you are serious about trying to reach such high frequencies then I might suggest watercooling.

But of course, where to find such a small waterblock??? Well I was thinking that the best idea might simply be to buy some copper or brass plumbing fittings from a hardware store, which has a surface around the right size to sit over the PCB, by resting the side of the nut against the ASIC. I was thinking maybe the nut of a screw threaded fitting, then you can just run the coolant through some matching pipe that is joined to the fitting. You could repeat this on both sides of the ASIC since the underside will remove quite a lot of heat too (ie. in the standard design most heat is sinked through the base of the chip anyway).

Of course it probably isn't practical by any means... but if you really want to say "I DID IT!"...

http://i.imgur.com/Di9TtVG.jpg

Ok Ive done quite alot of work today on overclocking

Pin outs updated with more info

Highres
http://wtfmoogle.com/wp-content/uploads/2013/07/bepinouts2.0.jpg
http://wtfmoogle.com/wp-content/uploads/2013/07/bepinouts2.0LR.jpg

Got some info from friedcat today

"The multipliers of our chips are fixed (28x), so you could only change the oscillator frequency to achieve higher internal frequency. The oscillator of choice (12MHz) is fixed so overclocking will involve re-soldering of a new oscillator with higher frequency. Also, the resistor controlling AOZ1021's output voltage needs change.

Best regards,

friedcat"

So I am working up a order for the parts needed to overclock My Target speed is 400mhz.
It is not a simple crystal, you need a oscillator to do the overclocking.
Still Need to reverse the power section of the asic to determine how to increase the voltage.
I also have a programmable oscillator on its way so I can fine tune the speed of the asic (up to 800mhz if it can take it but i doubt it :p )

nice to see some brave exploration.

You might try using acetone to remove the solder mask, perhaps test on a salvage/junk pcb first.
And of course use proper safety measures - acetone is pretty nasty stuff.

As far as firmware - it would seem Sir Friedcat is the only possible source.

EDIT: Ha - after a quick look at your blog, I'm guessing you can handle a little acetone 

So.. I can finally post here
Started working on overclocking and abusing the Block Erupter

From what info I have found
"Each chip's rated frequency is 336MHz at 1.05V
The chips work stable and well at 392MH/s at 1.15V. Further overclocking needs proper handling of heat and power supply."

So I started looking into how to boost things up.

Simple start... the soldermask on the back of the board... this will prevent good heat transfer. I am planning on polishing my boards up better
but simply scratching the solder mask off carefully under a microscope

http://wtfmoogle.com/wp-content/uploads/2013/07/draw_0719-191.jpg
 could be polished up better.

Same can be said about the asic.
http://wtfmoogle.com/wp-content/uploads/2013/07/draw_0719-51.jpg
very porous and could be smoothed out for better heat transfer to a heat sink.

I started to reverse engineer the board completely.
High Res warning
http://wtfmoogle.com/wp-content/uploads/2013/07/bepinouts1.0.jpg

I have a feeling the speed of the asic is controlled by the firmware in the attiny
I would kill for a full detailed schematic or chip spec of the asic.
Can anyone get me one? Firmware of the attiny would be nice too

On updates on actual overclocking?

I'm interested.  Please do open source it.

I wrote a Node.js program to push the data to Librato Metrics (or any statsd backend like Graphite). I can open source it if there's interest.

What app are you using for monitoring?

FYI these little guys fit perfectly on the ASIC chips: http://www.amazon.com/gp/product/B007XACV8O



I have no idea how well they're working though. I also added a fan at the same time, and as far as I can tell I'm still getting the same number of hardware errors.

The orange line shows the temp of the Raspberry Pi so you can see when I added the fan and heatsinks around 12:00 yesterday. Green is hashrate of the two miners combined (not sure why it drifts but it doesn't seem correlated to temperature). Blue is hardware errors.



Here's just the hardware errors since I setup the metrics:

You can cool almost anything you can think of with peltiers, but the problem is they are horribly inefficient devices. You'd be using a lot of power to cool a 2.5 watt device (+some extra from the OC) and that means more heat to remove. You'd have to cool the hot side of the peltier with a large heatsink, and you'd be better of just putting that heatsink onto the miner directly. From my experience with them, to get their rated temperature differential (perhaps 30 degrees) with no thermal load on the cold side, you have to run them with their rated power which is usually 50+ watts. So your 2.5+ watt heat load is now 52.5 watts or more.

Also because peltiers are capable of sub-ambient cooling, you run the risk of condensing water on the miner if you lower its temp below the dew point of the ambient air: pretty easy to do if suddenly the miner becomes idle and is no longer giving off much heat (you'd probably want dynamic temp control for the peltier itself, using a furnace controller programmed in cooling mode instead of heating mode - most Eurotherm models and some cheaper brands support this - or get a 'peltier temperature controller' off ebay).

A better option for overclocked miners (if anyone bothered to make one) would be to watercool them. Although they're terribly designed, you can get small waterblocks from ebay (read: China) for <$10. If you had a lot of miners to cool they could be attached to a larger watercooled plate, there are several commercial solutions to do this, or you could approach a machine shop. You can even make waterblocks yourself with a router and drill press and a bit of swearing.