Topic: Modifying a USB hub for extra power and "I did it!" showoff thread - page 2. (Read 6395 times)
September 20, 2015, 10:18:46 AM
I had good luck modding cheapest hubs I could find at the time. Shelf board with all the hubs attached is available if someone is interested, make an offer, its heavy.
September 20, 2015, 04:30:24 AM
I found this 5 volt meanwell
https://www.trcelectronics.com/ecomm/pdf/lrs75.pdf so this one = yes
should power a 10 port hub 24/7/365 all 10 sticks at about 1.1 amps
86.5% eff very good for a 5 volt psu.
I have 2 of these
https://www.trcelectronics.com/ecomm/pdf/lrs150.pdf this one = yes
on order they are a direct replacement for the psu in my 2 studhubs from klintay
https://www.trcelectronics.com/ecomm/pdf/nes100.pdf = no
The eff will jump from 80 to 86.5 the watts will jump from 100 to 150
this below one will do 32 sticks and is 87.5% eff
https://www.trcelectronics.com/ecomm/pdf/lrs200.pdf and this one = yes
https://www.trcelectronics.com/ecomm/pdf/lrs75.pdf so this one = yes
should power a 10 port hub 24/7/365 all 10 sticks at about 1.1 amps
86.5% eff very good for a 5 volt psu.
I have 2 of these
https://www.trcelectronics.com/ecomm/pdf/lrs150.pdf this one = yes
on order they are a direct replacement for the psu in my 2 studhubs from klintay
https://www.trcelectronics.com/ecomm/pdf/nes100.pdf = no
The eff will jump from 80 to 86.5 the watts will jump from 100 to 150
this below one will do 32 sticks and is 87.5% eff
https://www.trcelectronics.com/ecomm/pdf/lrs200.pdf and this one = yes
September 19, 2015, 08:01:43 PM
That eliminates some components - specifically whatever technology used to step the voltage down from 12V to 5V. On the other hand, it introduces the need for a 5V power supply which are not as commonly available (computer PSUs and PSUs meant for building into devices - like a good chunk of MeanWell's range - aside). There's also no guarantee there aren't other components in there that would still limit the per-port or overall device current capability. Example: say the hub can use either the power supplied over the USB cable, or the power supplied via external power, and uses a bit of logic and some solid state component that acts as a 'switch' between the two. That 'switch' may well impose a limit.
Again, check specs or stick to hubs people have already successfully used
Again, check specs or stick to hubs people have already successfully used
September 19, 2015, 05:31:28 PM
I'm with Mikestang, kind of got the jest of it, that for a 12V it boils down to the 12V---->5V as how good it is.
So I guess it's better to do a hub that takes 5V straight, like the one AJRGale did and take the conversion from 12V to 5V out?
So I guess it's better to do a hub that takes 5V straight, like the one AJRGale did and take the conversion from 12V to 5V out?
September 17, 2015, 01:04:16 PM
If things are still a bit confusing, I would highly recommend sticking to some of the suggested hubs that people have already tested and are known to work well
Thanks for the explanation, most of it got through my thick skull. You started to lose me with the components, but I think I get it.
So upon further inquiries it appears that my hub is not going to be good for overclocking stick miners (post). I'm going to pick up one of those usb testers and play around with this hub (afterall, playing around is most of the fun), but if I'm going to be running multiple sticks overclocked I'm going to have to look for new hardware.
September 16, 2015, 07:00:15 PM
After playing with a RPI and trying to get a powered hub I can say all hubs are not equal. Very safe thing to say.
In a perfect world any usb 2 hub would work with RPI perfectly. But it so very very different not all are the same.
In a perfect world any usb 2 hub would work with RPI perfectly. But it so very very different not all are the same.
September 16, 2015, 05:39:28 PM
The 12V, 5A is typically the electrical rating for a hub taking external power as the type of power supply that you should connect: i.e. 12V, and capable of handling at least 5A. That doesn't mean the hub will always draw 5A, but the hub itself is designed such that it at least theoretically can - though this is often based on datasheets, rather than actual measurements (i.e. using test equipment to force a particular current draw and then see if the components get too hot / trigger an internal shutoff, etc.).
The differences between hubs is in what components they use to regulate the voltage down to 5V and how it is laid out electrically. If you're familiar with typical small indicator LEDs, then you might consider a small 1/8th Watt resistor to 'regulate' current for an LED. But replace that small LED with one used for lighting purposes, and that 1/8th Watt resistor will get way too hot which may change characteristics or even cause the resistor to just burn up. So you'd have to replace it with, say, an 0.5 Watt resistor. Or, better yet, you'll use a linear voltage regulator that can handle maybe 2A. Better yet, a switching regulator with components that can handle 5A, etc. It's that choice of components that determines how much current the hub itself can handle.
( The electrical layout also matters - specifically trace widths. A very narrow trace can't handle nearly as much current as a nice wide trace or a trace that has been tinned over. Take for example a typical bit of wire, a small bicycle bulb, and a 9V battery and you can light up that bulb no problem. Use steel wool, on the other hand, and the steel wool will ignite, as each steel wool fiber has a very low current carrying capacity. )
So what makes a hub beefy is how much current it can handle. Hubs slightly complicate things because the current on its input (e.g. that 5A) does not necessarily translate in how much it can handle per port. Let's say it has 4 ports and each can handle 2A; 4*2A = 8A = exceeds what it can handle on its input. Alternatively, say each port can only actually handle 1A, then it's all good and well that the engineers specced its design to 5A and the people slapping stickers onto the thing and putting print on the packages copy that, but as each port tops out at 1A that's a bit moot - and there's definitely no drawing 5A from a single port.
But say you need 2A and each port can only deliver 1A: that's when you can use a Y cable (or other such constructions) to basically divide the load (2A) across two ports (2A / 2 = 1A). On a cheapo hub, that's not much of a problem - on things like a laptop, I'd be more cautious; no point in potentially 'ruining' a $500 device if you can instead mess around with a $15 one.
That's all somewhat unrelated to USB 3.0 vs 2.0. USB 3.0 is supposed to be able to deliver up to 900mA, USB 2.0 up to 500mA . However, a USB 2.0 hub may well be able to deliver 3A on a single port. As long as it's a minimum of 500mA, they're okay to slap a USB 2.0 label on it. In short, I wouldn't rely on USB 2 vs USB 3 in terms of assuming current capability, and instead look up the exact specs - specifically: How much current can it deliver on a single port, and how much current can it handle total. The input voltage is something you can ignore in terms of the hub itself - it's only something you need to know in order to figure out what voltage rating power supply to look for / hook up.
If things are still a bit confusing, I would highly recommend sticking to some of the suggested hubs that people have already tested and are known to work well
The differences between hubs is in what components they use to regulate the voltage down to 5V and how it is laid out electrically. If you're familiar with typical small indicator LEDs, then you might consider a small 1/8th Watt resistor to 'regulate' current for an LED. But replace that small LED with one used for lighting purposes, and that 1/8th Watt resistor will get way too hot which may change characteristics or even cause the resistor to just burn up. So you'd have to replace it with, say, an 0.5 Watt resistor. Or, better yet, you'll use a linear voltage regulator that can handle maybe 2A. Better yet, a switching regulator with components that can handle 5A, etc. It's that choice of components that determines how much current the hub itself can handle.
( The electrical layout also matters - specifically trace widths. A very narrow trace can't handle nearly as much current as a nice wide trace or a trace that has been tinned over. Take for example a typical bit of wire, a small bicycle bulb, and a 9V battery and you can light up that bulb no problem. Use steel wool, on the other hand, and the steel wool will ignite, as each steel wool fiber has a very low current carrying capacity. )
So what makes a hub beefy is how much current it can handle. Hubs slightly complicate things because the current on its input (e.g. that 5A) does not necessarily translate in how much it can handle per port. Let's say it has 4 ports and each can handle 2A; 4*2A = 8A = exceeds what it can handle on its input. Alternatively, say each port can only actually handle 1A, then it's all good and well that the engineers specced its design to 5A and the people slapping stickers onto the thing and putting print on the packages copy that, but as each port tops out at 1A that's a bit moot - and there's definitely no drawing 5A from a single port.
But say you need 2A and each port can only deliver 1A: that's when you can use a Y cable (or other such constructions) to basically divide the load (2A) across two ports (2A / 2 = 1A). On a cheapo hub, that's not much of a problem - on things like a laptop, I'd be more cautious; no point in potentially 'ruining' a $500 device if you can instead mess around with a $15 one.
That's all somewhat unrelated to USB 3.0 vs 2.0. USB 3.0 is supposed to be able to deliver up to 900mA, USB 2.0 up to 500mA . However, a USB 2.0 hub may well be able to deliver 3A on a single port. As long as it's a minimum of 500mA, they're okay to slap a USB 2.0 label on it. In short, I wouldn't rely on USB 2 vs USB 3 in terms of assuming current capability, and instead look up the exact specs - specifically: How much current can it deliver on a single port, and how much current can it handle total. The input voltage is something you can ignore in terms of the hub itself - it's only something you need to know in order to figure out what voltage rating power supply to look for / hook up.
If things are still a bit confusing, I would highly recommend sticking to some of the suggested hubs that people have already tested and are known to work well
September 16, 2015, 04:56:53 PM
The power going into the hub is 12V at up to 5A.
But that gets regulated internally to 5V, right? I think that's what you wrote in response to me in your stick thread. I guess that's one disconnect I'm struggling with. Internal ratings, external rating, usb port ratings... Were those numbers (12V @ 5A) determined from manufacturer literature or experimentally? If every hub regulated voltage down internally to 5V where do the differences between hubs come into play?Most folks will be running beefy hubs, or using Y cables to pull power from two ports at the same time for one stick.
Here's where I get lost. What makes a hub "beefy"? This is directly related my questioning above.
Would a single usb 3.0 port be essentially equivalent to a y cable on two 2.0 ports?
September 16, 2015, 04:05:52 PM
The power going into the hub is 12V at up to 5A. This means a maximum of 60W. Derate that a bit for internal conversions, you're looking at about 50W total usable power from your ports. If your ports can safely push 1A each (fairly beefy), you can draw 5W from each port (5V times 1A) which means you could run ten sticks safely (10 x 5W = 50W) from the available power supply.
Most folks will be running beefy hubs, or using Y cables to pull power from two ports at the same time for one stick.
Most folks will be running beefy hubs, or using Y cables to pull power from two ports at the same time for one stick.
September 16, 2015, 03:54:01 PM
So can I assume that any one I see running a compac stick at more than ~8GH (source) is using a hub they modified? Or could they be using a "beefy" usb 2.0 hub that just happens to be able to handle more current?
Since usb 3.0 can provide 0.9A at each port can I assume a usb 3.0 hub would be able to run the compac sticks at a higher frequency than a 2.0 hub?
Just not all hubs are equal. Best example is from old day's: https://bitcointalksearch.org/topic/which-usb-hub-to-use-with-block-erupters-nanofury-nf1-bpmc-red-fury-ant-u1-253749
Look at that thread it shows some hubs and power that they have from ac. They very GREATLY.
I'm not understanding the "Power from AC" column - 12V @ 5A for example, does that mean the entire hub is rated for that draw? Each port? What am I reading? 
Since usb 3.0 can provide 0.9A at each port can I assume a usb 3.0 hub would be able to run the compac sticks at a higher frequency than a 2.0 hub?
Just not all hubs are equal. Best example is from old day's: https://bitcointalksearch.org/topic/which-usb-hub-to-use-with-block-erupters-nanofury-nf1-bpmc-red-fury-ant-u1-253749
Look at that thread it shows some hubs and power that they have from ac. They very GREATLY.
September 16, 2015, 03:43:33 PM
In terms of easy over clock :
freq 250 uses 1.1 amps does 13.7gh uses about .33-34 watts a gh likes a fan but can be run without the fan. you can use this link
freq 425 uses 2.6 amps does 23.3 gh uses about .58 watts a gh and must have a fan
should let you go to 1 amp maybe 1.2 amps, still waiting for these to test them.
I would think that with the y adapter this hub can do 4 sticks at freq 250 giving you 4.4amps at 5 volts about 22 watts for about 66gh . comes with a 3 amp 12 volt brick which has 36 watt max and you are pulling 22 watts. may not be able to do this once again I have these on hand and will test once the y-adapters come.
freq 250 uses 1.1 amps does 13.7gh uses about .33-34 watts a gh likes a fan but can be run without the fan. you can use this link
freq 425 uses 2.6 amps does 23.3 gh uses about .58 watts a gh and must have a fan
should let you go to 1 amp maybe 1.2 amps, still waiting for these to test them.
I would think that with the y adapter this hub can do 4 sticks at freq 250 giving you 4.4amps at 5 volts about 22 watts for about 66gh . comes with a 3 amp 12 volt brick which has 36 watt max and you are pulling 22 watts. may not be able to do this once again I have these on hand and will test once the y-adapters come.
If a usb 2.0 port only provides 0.5A how are you guys seeing 1.1 amps or 2.6 amps supplied to a single stick? Are these all done on modified hubs?
snip
Just not all hubs are equal. Best example is from old day's: https://bitcointalksearch.org/topic/which-usb-hub-to-use-with-block-erupters-nanofury-nf1-bpmc-red-fury-ant-u1-253749
Look at that thread it shows some hubs and power that they have from ac. They very GREATLY.
September 16, 2015, 03:26:32 PM
In terms of easy over clock :
freq 250 uses 1.1 amps does 13.7gh uses about .33-34 watts a gh likes a fan but can be run without the fan. you can use this link
freq 425 uses 2.6 amps does 23.3 gh uses about .58 watts a gh and must have a fan
should let you go to 1 amp maybe 1.2 amps, still waiting for these to test them.
I would think that with the y adapter this hub can do 4 sticks at freq 250 giving you 4.4amps at 5 volts about 22 watts for about 66gh . comes with a 3 amp 12 volt brick which has 36 watt max and you are pulling 22 watts. may not be able to do this once again I have these on hand and will test once the y-adapters come.
freq 250 uses 1.1 amps does 13.7gh uses about .33-34 watts a gh likes a fan but can be run without the fan. you can use this link
freq 425 uses 2.6 amps does 23.3 gh uses about .58 watts a gh and must have a fan
should let you go to 1 amp maybe 1.2 amps, still waiting for these to test them.
I would think that with the y adapter this hub can do 4 sticks at freq 250 giving you 4.4amps at 5 volts about 22 watts for about 66gh . comes with a 3 amp 12 volt brick which has 36 watt max and you are pulling 22 watts. may not be able to do this once again I have these on hand and will test once the y-adapters come.
If a usb 2.0 port only provides 0.5A how are you guys seeing 1.1 amps or 2.6 amps supplied to a single stick? Are these all done on modified hubs?
I'm trying to wrap my limited electrical engineering neurons around what I'm going to be able to do with my 4 compacs when they show up, e.g., how far can I push them safely on an unmodified usb 2.0 hub? So far I've gotten as far as 0.5A @ 5V = 2.5watts (theoretical) max power available to a normal usb port. I want to run 4 sticks on one hub and know what frequency I can push them to without smoking my hub.
I'm having as much difficulty formulating my question(s) as I am in trying to understand what it is I'm questioning, ha!
August 24, 2015, 01:36:41 AM
Thanks AJRGale, not sure if I'm going this route or going with the Y cables, yet.
Did you disconnect the 5V+ (or are you going to cut the wire) in the cable from the hub to computer/Pi/?, so you don't get back current?
Just noticed in the pics you posted and the pics from ebay listing, your's is missing 2 of the 3 filter capacitors, so who knows what else is missing in that hub.
Hate for you to see smoke from the computer or whatever it's hooked up to!
Did you disconnect the 5V+ (or are you going to cut the wire) in the cable from the hub to computer/Pi/?, so you don't get back current?
Just noticed in the pics you posted and the pics from ebay listing, your's is missing 2 of the 3 filter capacitors, so who knows what else is missing in that hub.
Hate for you to see smoke from the computer or whatever it's hooked up to!
if I was going to use it to power a Rpi, almost unnecessary, same if I used the same system for data and power, but if I have a PSU seperate from the system, like a laptop, I may need to chop something, i don't know. But I don't think 5V won't harm it may keep a led lit up on the system. just as long as you plug the data in 1st, then plug the PSU in, whilst its off. When every connection is secure, power on the PSU then the laptop. then when everything is evened out, plug the miners in.
If I was going to chop anything, it would be the 5V from the host, since the psu power is what is feeding the miners and the Hub chipsets.
As for the filter caps, I'd have to find some thats 5+v and about 1000nF and mount them the wrong way
August 23, 2015, 01:42:32 PM
Thanks AJRGale, not sure if I'm going this route or going with the Y cables, yet.
Did you disconnect the 5V+ (or are you going to cut the wire) in the cable from the hub to computer/Pi/?, so you don't get back current?
Just noticed in the pics you posted and the pics from ebay listing, your's is missing 2 of the 3 filter capacitors, so who knows what else is missing in that hub.
Hate for you to see smoke from the computer or whatever it's hooked up to!
Did you disconnect the 5V+ (or are you going to cut the wire) in the cable from the hub to computer/Pi/?, so you don't get back current?
Just noticed in the pics you posted and the pics from ebay listing, your's is missing 2 of the 3 filter capacitors, so who knows what else is missing in that hub.
Hate for you to see smoke from the computer or whatever it's hooked up to!
August 20, 2015, 06:11:34 PM
Here's another thought. Use the Y-cable, but connect the "Power Only" cable directly to the supply for the hub. That would take the hub traces out of the power distribution for the most part, thought it would be fairly messy cabling wise.
yeah but the black data cable carries power need to be careful.
this y cable is a simple play n plug that is letting me do freq 250 at 1.1 amps brainlessly. (good for those that are plug n play tech level )
they are mining here but I seem to be getting some roll over from an s-3 in my friends office. so it reads well over 28-30gb
http://solo.ckpool.org/users/1JdC6Xg3ajT3rge3FgPNSYYFpmf53Vbtje
August 20, 2015, 05:09:31 PM
Here's another thought. Use the Y-cable, but connect the "Power Only" cable directly to the supply for the hub. That would take the hub traces out of the power distribution for the most part, thought it would be fairly messy cabling wise.
August 20, 2015, 04:10:40 PM
okay the y-cables are in.
so far so good they do allow me to run about 1.1 amps and freq 250 hash rate = 13.5 very few errors will post more since this has been running under 2 hours.
so far so good they do allow me to run about 1.1 amps and freq 250 hash rate = 13.5 very few errors will post more since this has been running under 2 hours.
So the Y cable is a "Only power" and "data" plugs?
I can test that tomorrow.
I will lower freq to 150
On first usb y- cable I will pull the plug that has a red tag(guessing a power only)
On the second usd y-cable I will pull the unmarked black plug (guessing data/power cable)
The first one should not drop the usb since I am only pulling the power booster.
In theory the second one should drop the usb.
I will photo everything tomorrow.
Nice, I'll be waiting! could just do a normal thumb drive test, same results.
all photos and results are here
https://bitcointalksearch.org/topic/m.12194638
looks like the red plug is power booster the black plug is power and data.
nice cable for around 2 dollars
August 19, 2015, 11:54:14 PM
okay the y-cables are in.
so far so good they do allow me to run about 1.1 amps and freq 250 hash rate = 13.5 very few errors will post more since this has been running under 2 hours.
so far so good they do allow me to run about 1.1 amps and freq 250 hash rate = 13.5 very few errors will post more since this has been running under 2 hours.
So the Y cable is a "Only power" and "data" plugs?
I can test that tomorrow.
I will lower freq to 150
On first usb y- cable I will pull the plug that has a red tag(guessing a power only)
On the second usd y-cable I will pull the unmarked black plug (guessing data/power cable)
The first one should not drop the usb since I am only pulling the power booster.
In theory the second one should drop the usb.
I will photo everything tomorrow.
Nice, I'll be waiting! could just do a normal thumb drive test, same results.
August 19, 2015, 11:30:11 PM
okay the y-cables are in.
so far so good they do allow me to run about 1.1 amps and freq 250 hash rate = 13.5 very few errors will post more since this has been running under 2 hours.
so far so good they do allow me to run about 1.1 amps and freq 250 hash rate = 13.5 very few errors will post more since this has been running under 2 hours.
So the Y cable is a "Only power" and "data" plugs?
I can test that tomorrow.
I will lower freq to 150
On first usb y- cable I will pull the plug that has a red tag(guessing a power only)
On the second usd y-cable I will pull the unmarked black plug (guessing data/power cable)
The first one should not drop the usb since I am only pulling the power booster.
In theory the second one should drop the usb.
I will photo everything tomorrow.
August 19, 2015, 11:23:34 PM
okay the y-cables are in.
so far so good they do allow me to run about 1.1 amps and freq 250 hash rate = 13.5 very few errors will post more since this has been running under 2 hours.
so far so good they do allow me to run about 1.1 amps and freq 250 hash rate = 13.5 very few errors will post more since this has been running under 2 hours.
So the Y cable is a "Only power" and "data" plugs?
Jump to: