This is a discussion on computer 1 usb 1<====>usb 2 computer 2 within the Networking/Device Communication forums, part of the General Programming Boards category; Originally Posted by cyberfish You mean both computers running on the same PSU...? It's also not logic voltage. Logic voltage ...
Logic levels are generally current limited by resistors in the chip's outputs... while it's not smart and never successful, connecting logic outputs together is seldom destructive...Originally Posted by cyberfish
However, the USB 5 volt lines are not buffered or fused and the most common result of current races is damage to the USB host devices (which may be on your motherboard!)
Well, I guess that example works. Though, I was actually thinking of even different hardware, (2 computers of the exact some model, running from the same outlet, for (safe) example).
Yeah, I guess it was a bad idea to mention logic voltage. What I mean, is, so long as the cross-voltage stays within both buses' voltage reference range.
What do you mean by never successful? What could one hope to achieve by connecting 2 (assuming constant) outputs?
Then the voltages will most certainly be different.Well, I guess that example works. Though, I was actually thinking of even different hardware, (2 computers of the exact some model, running from the same outlet, for (safe) example).
Transformer has some (non-zero) tolerance, inductors have some tolerance, regulators have some tolerance, and voltage will be affected by what the computer is doing at that time. Even for identical computers.
In the real world, nothing will ever be perfect. We always talk about values with tolerances, and tight tolerance gets expensive, so we always aim for lowest tolerance acceptable.
Tell you what... go visit the Dell assembly plant, or any big computer store that has more than one of exactly the same models... Take some nice careful power supply readings... If you find any two *exactly* the same, it will be a miracle second only to walking on water. I'm not talking about material defects or different models... I am talking about perfectly ordinary and expected manufacturing variations --tolerances-- even in identical make and model.
The only connections that are guaranteed safe are the grounds.
My friend, you should see some of the mindnumbingly stupid things I've seen people do to electronic devices...What do you mean by never successful? What could one hope to achieve by connecting 2 (assuming constant) outputs?
For example: Ham radio operator come technician, wants several receivers on his desk at the same time so he gets coaxial T connectors and hooks them all together... Then he can't figure out why keying up his transmitter blew the inputs on *every receiver* he owned.
For example: A so-called audiophile who connected three different amplifiers to the same speakers, with the logic that "they're never all on at once" and couldn't figure out why he kept blowing output transistors... he had a working impedance of less than .5 ohms...
For example: The rather extreme case of a house fire started because a supposed computer expert put 24 volts into a 16 volt laptop and exploded the batteries.
For example: The CBer who kept blowing fuses every time he hooked up his radio; "The black wire is positive, right?" So, buddy puts in heavier fuses until he finally gets one strong enough to blow out the "idiot" diode in the radio... and sets his dash on fire.
And on and on and on...
Trust me on this... consumers can be extremely stupid people.
Excellent point. Do you think the same applies to microcontrollers or the same model running from the same power source?
Well of course, this is also hew programming (should) work. Nothing new
Power has to go through many many stages before reaching low voltage DC devices in a computer. All of those stages will introduce variations.Excellent point. Do you think the same applies to microcontrollers or the same model running from the same power source?
Yep... even the length of a wire or size of a foil trace can affect it...
Okay, but outside of the computer, when 2 microcontrollers of the same model are running straight from the same power source do this, is it safe then? (I ask because I'm doing this, and I don't want bit in the butt later down the road
If they are both grounded to the same powersupply and exchanging only data signals, yes it's safe.Okay, but outside of the computer, when 2 microcontrollers of the same model are running straight from the same power source do this, is it safe then? (I ask because I'm doing this, and I don't want bit in the butt later down the road
But I would not consider connecting the supply voltages together, especially through a chip or regulator. Any difference in voltage could be damaging, may cause ground currents (which switching power supplies really hate) and might result in system failure.
I'll keep this in mind; thanks.
In electronics in general, you always only want at most 1 driver driving a node, whether it's supply or digital or analog IO. There are rare exceptions, but it's true 98% of the time.
A node is a "point" in your circuit that is connected together, and should be at the same potential. Node (circuits) - Wikipedia, the free encyclopedia
Usually for digital IO, one side will be sending, and one or more sides will be receiving. If multiple parties transmit at the same time, the voltage level on that node will be indeterminate, and you most certainly don't want that.
That's why in protocols where multiple devices are connected to the same bus, like USB, multi-slave SPI, or I2C, there is always some kind of arbitration logic to make sure only 1 party transmit at a time.
Yeah, I wanted to multiplex I/O onto 1 common grounded wire.
I figured that so long as the TX and RX cycles on each side were properly sequenced, each one would have time to receive the peer's bit in each given frame. In the event that both sides are TXing, the indeterminate voltage isn't being read anyway, so it wouldn't matter. After hearing you and Tater, I'm glad I didn't waste any time in implementing this
In a case like that you have two choices... place an OR gate at the receiver to isolate the outputs or in some cases a "diode or"... using two diodes to isolate the respective lines can be used. Just don't be directly connecting outputs to outputs...Yeah, I wanted to multiplex I/O onto 1 common grounded wire.
I figured that so long as the TX and RX cycles on each side were properly sequenced, each one would have time to receive the peer's bit in each given frame. In the event that both sides are TXing, the indeterminate voltage isn't being read anyway, so it wouldn't matter. After hearing you and Tater, I'm glad I didn't waste any time in implementing this
As they say... "When you let all the smoke out, the circuit stops working."
5v + 5v = 10v what happened to voltage doublers ? or voltage multiplying ? or adding two voltages together ? if you tried to read or measure the voltage on the power line of two usb connected together from two separate power supplies from two different computers would you read 10v assuming that the power supplies were 5v ?
what about the circuitry ? voltage regulators ? or clipping diodes like the zener ? that would send any volts over 5v to ground leaving the 5v. so then you would only measure 5v ? assuming both power supplies have the same 5v regulator circuit.
is .125 watt from tolerance example really enough to blow a half watt resistor or half watt regulator ? i doubt any decent designer would use half watt parts in a power supply.
it is the current that blows parts. you can have a potential difference of charges extremely high as long as there is no current flow ie capacitor. same cap with a change of volts current will flow. ie sin or on off . the rating on the component says what it can take.
so the data + and data - are one way ? like full duplex ? the data + transmits while the data - receives ? or is it data + is a control line only two way and the data - is two way data only ?
i can see how that would be a problem for the latter data + as control line switching the lines.
i guess it is how the data - and + work. i will have to read up on that. usb dot org has some good docs but it is not easy to find an answer to some questions. like what will happen if you swap the data + and data - lines.
You'll only get 10V if you connect ground of one side to 5V of the other side, and leave the 5V side floating. This is assuming they are both isolated (not from the same transformer). What does it have to do with this?
Even if you can somehow get 10V on the USB power pin, it will probably just blow up. The USB spec says current cannot be fed into the host. What 0.125W from where? Most regulators are not rated for reverse current. I have no idea what you are talking about. If you want to talk about a specific circuit, draw the circuit. Don't pull numbers out of thin air and start arguing random things. Have you tried logarithm?
D+ and D- are a differential pair, half-duplex. They always carry opposite voltage (read up on differential signalling).
Like I already mentioned in previous post, that is not how it works. If you swap D+ and D- all bits will be flipped, so a high-speed device would be mis-identified as a low-speed device, and vice versa. Then, J and K states will be flipped, which may mess up bit stuffing, etc (USB inserts extra transitions when there's no transition in the data, eg. sending all 0's, to make it easier for receiver to recover clock from the signal). The doc also doesn't say what will happen if a cat chews on the USB cable. It's undefined behaviour if it's not specified.i guess it is how the data - and + work. i will have to read up on that. usb dot org has some good docs but it is not easy to find an answer to some questions. like what will happen if you swap the data + and data - lines.
4Likes
LinkBack URL
About LinkBacks
