Search content
Sort by

Showing 9 of 9 results by PupTentacle
Post
Topic
Board Hardware
Re: GekkoScience has a new stickminer that does 300+GH
by
PupTentacle
on 06/01/2022, 00:17:39 UTC
Good link regarding USB-C PD & data transfer as well as how the various USB ports (A, B & C) are used.
Something I did not know about how USB in general does things.
Quote
The point is that because USB is a master/slave protocol the functions of two USB-C ports on a USB-C dock has to be split up to allow for more than one master. The laptop is the master of the data portion and the power brick is master of the power portion. It's rare to see more than two USB-C ports on a dock because the USB protocol doesn't really allow for more. If a dock has more than two USB-C ports then the other USB-C ports will have some limits on the port functions to avoid having to deal with more than one master, or the dock is using a more complex protocol like Thunderbolt that is not master/slave but peer/peer.
So still sounds like that a PD port is only to be used for power or data but not both at the same time.

As Spock would say "Interesting".

So if I read that right, if PD is divorced from the master/slave (by supplying power from say a barrel connector) one could use the 4 wire USB 3.0 master/slave data protocol & achieve a higher data rate than USB 2.0. That would also allow for higher PD than 100W. Effectively limited only by the current handling of the barrel connector.

This makes sense as many USB 3.0 to SATAIII adapters have external wall warts to supply power to the SATA power port.

Did I glean that correctly?
Or am I all FUBAR & in need to re-read that reference.

Pup
Post
Topic
Board Hardware
Re: GekkoScience has a new stickminer that does 300+GH
by
PupTentacle
on 05/01/2022, 23:51:50 UTC

Remember any intermediary steps will have their own power losses, and efficiency percentages cascade. Two 90% stages in series still results in a 19% power loss, and if you 20V into 5V you still have the problem of trying to get reasonable efficiency out of an 8% duty cycle switcher. I'd be more likely to take 20V down directly with a half-bridge forward converter and custom wound transformer especially if it was a one-off. That's how a lot of your good active-PFC supplies handle bucking 350+VDC down to 12VDC at a hundred or more amps.

Good advice, I just hate hand winding transformers/inductors. But yeah, losses in stages multiply versus being additive.

The eleven projects can largely be subdivided into a few project families with a decent amount of overlap. Eventually I'll be fully excised from the assembly line and can focus more on R&D. That's the dream anyway.

My ardent hope is for your emancipation from the tedium of manufacturing (which pays the bills).

You hinted in a previous post about a new miner project that doesn't rely on BM hash chips. Per chance w/ a chip that's not made in the Pacific Rim? This is something I'd be VERY interested in. And in whatever impotent capacity willing to contribute to.

In the interim if you're not gonna' do a second run of the F I'd be interested in buying some of your leftover PCB's hanging around. Not for resale just for my personal consumption.

As always, a pleasure.
Pup
Post
Topic
Board Hardware
Re: GekkoScience has a new stickminer that does 300+GH
by
PupTentacle
on 03/01/2022, 13:09:12 UTC

Kano's driver is very unlikely to work natively with the BM1398. The chips enumerate themselves on the bus with a distinct ID per generation, which the driver uses to distinguish which chip (and how many of them) it's talking to. It won't recognize the 1398's response and so won't initialize it for use.

The operative word in your reply is "natively". While not having dug into Kano's driver to deeply it shouldn't be too difficult to add those ID's. Was more concerned that BM dramatically changed their comm protocol.

Temperature sensing using the innards of the BM1397 also won't work natively. Bitmain implemented a register set that communicates with an external temp sensor over I2C using some otherwise unneeded pins, but this has to be built into the driver protocol. It is true that the two temp-sense-diode pins could be interfaced to any temp sensor chip and handled externally.

Your response answers the question, for me, as to how BM reads the 4 external board temp sensors. Thanks.

"... nor were there any reasonable hubs that would do the 20V spec."
Having used a variety of USB hubs over the years concur that, by & large, they're all crap (save yours) for current handling. I was assuming I'd have to build my own w/ some beefy copper bus's for PD & the USB C connectors.

Additionally, providing a higher voltage into the device would tend to *decrease* your conversion efficiency, unless you used something like a forward converter with a transformer to help balance out the duty cycle. The problem isn't the power into the main regulator, it's the power out. 400mV 40A bucks are a rare breed. I've looked into this and it's sorta possible but I have something like eleven design projects already in the pipeline so I'm probably not gonna do it anytime soon.

Good point on the decrease in efficiency w/ a larger VDC I/O spread. Kinda' assumed there would need to be an intermediary step down located close to the chip. Like 20V to 5V or 20V to 3.3V. Just use the 20V rail for PD to the port. The device plugged into the port would need to do the step down. Like what is used to do Li battery / Supercap charging in USB connected devices (cell phones, tablets, etc.).

In my experience, low V bucks (1.8V & 0.8V) are the realm of CPU/GPU power supplies & are usually seen as multiple units in parallel appropriately sized for the demands of the device. Which obviously would not fit in the Compac X form factor.

Eleven projects? You're a braver/smarter man than me. I wouldn't be able to juggle that many projects.

Thanks for all the clarification.
Pup
Post
Topic
Board Hardware
Re: GekkoScience has a new stickminer that does 300+GH
by
PupTentacle
on 02/01/2022, 15:53:20 UTC
Sure looks like a power supply decoupling cap to me since there is a polarity mark on the largest component on the backside of the hash chip PCB (2 pin component).
Right - makes sense, thanks guys! I just assumed diode since it has the distinct polarity mark which diodes also have. But it seems a cap makes more sense in this part of the circuit.
I'm surprised your works without the cap in place, @Biffa! Then I'll see if I can get it working simply without that.

Would still be great to hear what component this is (specifications) so it can be replaced if it is missing or blows / melts like in that case.

One could de-solder it, test it, & determine it' value.
So if it screwed up it could be replaced.
PITA for sure, but . . .
Post
Topic
Board Hardware
Re: GekkoScience has a new stickminer that does 300+GH
by
PupTentacle
on 02/01/2022, 14:56:05 UTC
Does anyone know the model number and / or parameters of the diode that is directly behind the mining chip (like on the opposite side of the PCB)?
I recently got my hands on a Compac F where that diode is missing. Would be sick to get it working!
Sidehack will have to chime in to confirm but I believe it is a capacitor -- not a diode. Somewhere further back in the thread there is talk about measuring the Vcore across it and it was referred to as a cap..

Sure looks like a power supply decoupling cap to me since there is a polarity mark on the largest component on the backside of the hash chip PCB (2 pin component).
Post
Topic
Board Hardware
Re: GekkoScience has a new stickminer that does 300+GH
by
PupTentacle
on 02/01/2022, 14:44:15 UTC
1397 and 1398 mostly look to be pin-compatible, but the documentation I have says they're not entirely so. 1398 operates on a much different V/I which makes it fairly non-suitable for single-chip operations. I've looked into it, and the main regulator engineering would be difficult and probably pretty inefficient. A series string like we did for the R606 would be a lot easier to manage. In any case, a direct chip swap from 1397 won't work.

Kinda' what I figured I hear. Butcha' don't know till you ask.

That accounts for what Bitmain calls "voltage domains" when troubleshooting hash chip chains on their boards.
And I can see the board engineering constraints on trying to maintain packaging on your USB Compac form factor.
But if one was willing to deviate from that form factor a tad .  .  .

Anyway, assuming one could supply the appropriate V/I to the chip do you believe Kano's 1397 driver would communicate w/ a 1398?
Is that something you've played with?

On a different note, I found these breakout PCB's for 1397/1398 prototyping & bought 5 of them (& D/L'ed the gerber files): https://www.pcbway.com/project/shareproject/Socket_BM1397.html
Figured I solder them up so I could socket them into a larger board to ring out the signals (multi channel o-scope) & test w/ Kano's driver merged into cgminer for testing/tweaking.

Was also thinking that if I swapped out the USB A for a USB C connector and fed it 20V @ 3A per USB C 3.2 / 3.1 specs it might make the main LDO engineering simpler to implement w/ "off the shelf" components. That also would decrease the current handling on a USB connector.

W/ a USB C 3.2 connector an APW 9 (14.5V-21V @ 170+ A or equivalent) could be used to supply trons to the USB power bus in a USB hub versus being limited to the 5V @ 3A USB A standard.

Curious to hear your thoughts.

Pup
Post
Topic
Board Hardware
Re: GekkoScience has a new stickminer that does 300+GH
by
PupTentacle
on 02/01/2022, 14:09:40 UTC
Cgminer does have provisions for temp monitoring, fan control, shut down, etc. but -- there needs to be a temp monitor of some sort to feed that info into cgminer and none of the Compacs have that. You could always look for RasPi projects that deal with temp control, mount a temp sensor to the heatsink and use it to simply switch off power to the sticks.

Stavroski, NotFuzzyWarm's solution is without a doubt the simplest way to garner temp monitoring.
In theory, if one could breakout pins 21-24 on the BM1397 hash chip one could garner a finer measurement of what one might call TJunc.

Please reference attached image: https://www.zeusbtc.com/Upload/image/202108/16280626681136747.JPG

Having said that I have no clue whether that is an analog or digital signals.
But it appears that Bitmain included temperature sensing into the chip/die.

Further, Bitmain uses the Ti TMP451AIDQFR temperature sensor on their 17 series hash boards to measure board temperature. Which are currently unavailable from any supplier anywhere in the USA (But may be found on eBay or Aliexpress). The data sheet may be found here: https://www.ti.com/lit/ds/symlink/tmp451.pdf?ts=1641066272554&ref_url=https%253A%252F%252Fwww.google.com%252F

This uses a simple 2 wire serial interface which on the Pi is easy to interface with. Soooo, if one epoxy'ed one to the heatsink . . . .

Hope this helps & provides some possibilities.

Pup
Post
Topic
Board Hardware
Re: GekkoScience has a new stickminer that does 300+GH
by
PupTentacle
on 30/12/2021, 16:12:04 UTC
Quote
If one replaced the 1397 w/ a 1398 on the Compac F would your driver for CGminer work?
They appear pin compatible. Or are the comms too different?
I highly doubt a simple 'drop-in' replacement is possible. Looking at older chips data sheets it is obvious that Bitmain plays around with different timing specs for the internal circuits - none of them are the same. Since BM no longer releases the data sheets Sidehack had to manually debug the signals/commands and timing for the 1397 and then supplied that info to Kano to work into the driver.

Hey NotfuzzyWarm,

Long time no talk, the old, now defunct, PlanetCrypto here (forgot password & lost email account when we shut it down 5+ years ago).

Kinda' assumed there was a LOT of reverse engineering going on between the 2.
And likely only those 2, outta' the community, coulda' figured it out, given BM's tight lipped agenda & complete absence of documentation. (They don't even teach theory of operation to their repair techs in their school).

Currently running some 17's & 19's, have a hot air rework station & test gear to fix'em when they break & have a small inventory of BIN 1 1397's & 1398's (& have a source for more). Mostly retired & was looking for a winter project to play w/ on these long cold MN nights. Upgrading an F seemed like a nice challenge. I know w/ esoteric cooling the 1397's can be pushed to 900+ Mhz w/ CoreV tweaks. The 1398's w/ the same heat dissipation & tweaks, in theory, should be able to exceed 1TH/s each.

Back in the day, we did some prototyping w/ 2 phase immersion cooling w/ Novec engineered fluids. So was thinking 'bout building some small immersion tanks to handle the heat. Like we did w/ S3 boards: https://www.youtube.com/watch?v=fStHV_Q-x7s

Anyway, thanks for the response & be safe out there.
Pup
Post
Topic
Board Hardware
Re: GekkoScience has a new stickminer that does 300+GH
by
PupTentacle
on 30/12/2021, 02:27:21 UTC
For linux I compile and build it on Ubuntu.
While that doesn't really matter too much, ubuntu is very simple to install.

Hi, oh smarter one than I,

If one replaced the 1397 w/ a 1398 on the Compac F would your driver for CGminer work?
They appear pin compatible. Or are the comms too different?

Just curious & thinking outside the box, a little.

Pup