Electronics > Projects, Designs, and Technical Stuff
EMF pickup from amplifier in I2C line causing glitches. (Now with scope trace!)
<< < (21/31) > >>
Starlord:

--- Quote from: T3sl4co1l on July 10, 2016, 02:58:56 pm ---Why in the hell... would anyone buy a $3 LTC chip to do what four resistors and a pair of transistors can do for pennies?

A classic current source circuit, tuned for 1-10mA perhaps, will do the same job. :)

--- End quote ---

Would a 3mA current source really do a much better job than a resistor sized to allow 3mA to flow?

Why doesn't the I2C spec suggest this?  It suggests buffers and switched pullups and other things:
http://www.nxp.com/documents/user_manual/UM10204.pdf



I mean, I think I see how what you're suggesting might work.  I'm guessing that the reason the rise time is curved in that manner is because as the voltage rises, the difference between the pull up voltage and the voltage on the data line becomes smaller, and as a result less current flows, and the signal rises more and more slowly as it approaches equilibrium.

This isn't something I had ever considered before, but it seems to make sense.

That being said, since I would need two of these for each data line, that's four transistors and eight resistors, and with PCB space at a premium I'd be lucky to find room for two transistors, let alone all the rest.

https://en.wikipedia.org/wiki/Current_source

I wonder if that single jfet with resistor would do the job?

Or, I could perhaps find an inexpensive chip that acts as a constant current source?  I've been using one chip that's the size of a single FET that requires only one external current set resistor and provides a constant current supply, which I use for high power LEDs.  I suspect some may exist that are designed for lower currents.  I guess I'll have a look on Digikey.
 
2N3055:

--- Quote from: T3sl4co1l on July 10, 2016, 02:58:56 pm ---Why in the hell... would anyone buy a $3 LTC chip to do what four resistors and a pair of transistors can do for pennies?

A classic current source circuit, tuned for 1-10mA perhaps, will do the same job. :)

Tim

--- End quote ---

Tim, guy is a beginner, and not an expert... He admits that and  asks for help.. :-+..
Active pull-up IS a current source... You're right.  :-+
But... I did not suggest he should or must take that chip... I made a point, and Linear has good datasheets and application notes, and I thought it would be a good place for him to get acquainted with the concept... I also said there are many alternatives, including the discrete current source circuit... :-/O

But if I where in hurry, running out of time and was making one of prototype, I WOULD just slap on one of these guys in the circuit, removing pull-up resistors and got it done with. This chip (and all alike ) is characterised for this task and will be pretty much quickest way to do it...

If you ask me how to do it on a large scale  where each penny counts... Well, I wouldn't do anything alike what he did....
I would drive leds differently (i know I do on my products), would make amplifier differently... I wouldn't trust I2C bus to do what he wants to do in a first place..

It all depends on many factors... :-//

Take care !! :)

2N3055:

--- Quote from: Starlord on July 10, 2016, 08:05:03 pm ---
Would a 3mA current source really do a much better job than a resistor sized to allow 3mA to flow?

Why doesn't the I2C spec suggest this?  It suggests buffers and switched pullups and other things:
http://www.nxp.com/documents/user_manual/UM10204.pdf



I mean, I think I see how what you're suggesting might work.  I'm guessing that the reason the rise time is curved in that manner is because as the voltage rises, the difference between the pull up voltage and the voltage on the data line becomes smaller, and as a result less current flows, and the signal rises more and more slowly as it approaches equilibrium.

This isn't something I had ever considered before, but it seems to make sense.

That being said, since I would need two of these for each data line, that's four transistors and eight resistors, and with PCB space at a premium I'd be lucky to find room for two transistors, let alone all the rest.

https://en.wikipedia.org/wiki/Current_source

I wonder if that single jfet with resistor would do the job?

Or, I could perhaps find an inexpensive chip that acts as a constant current source?  I've been using one chip that's the size of a single FET that requires only one external current set resistor and provides a constant current supply, which I use for high power LEDs.  I suspect some may exist that are designed for lower currents.  I guess I'll have a look on Digikey.

--- End quote ---

You guessed correctly... If you look at the shape of that signal, the front slope IS the curve of capacitor charging trough resistor..
If you run it with a current source you will have linear(straight) voltage raise... How fast, depending on current... But a simple current source won't do for this.
Look what datasheet says :

..."During positive bus transitions, the LTC4311 provides slew limited pull-up currents to quickly slew the I2C or SMBus lines to the bus pull-up voltage.
During negative transitions or steady DC levels, the currents are disabled to improve negative slew rate, and improve low state noise margins. "...

So it behaves differently when voltage goes up or down... So, current source, with voltage clamping and also with a circuit that "knows" to charge capacitor with big current (fast and linear rise time), and to go high impedance what chips on a bus are pulling down, so it doesn't violate their specs and have fastest possible fall time...

Of course it makes sense, it is being used in many forms for many years on other more robust data protocols...   :-+

Original I2C doesn't suggest this because I2C wasn't intended to do what you try  do... and also Philips won't endorse other manufactures... :clap:
You have to learn that datasheets and all materials published by manufacturers are written as much by marketing and legal department as they are by engineers..
Pretty much very correct most of the time, but not complete truth and info.. Many things unsaid and incomplete.

Buffers are also very good, and those you found would possibly solve your problems...

I suggested this active pull-up because it doesn't need ANY CHANGE to your circuit... You connect chip in parallel to existing 10K pull-p resistors and try if it helped..
I would suggest you to try exactly that as a troubleshooting tool, as no other changes are needed..
If it helps it means there is hope. :-+. If it doesn't, you need linear audio amp  :scared:

If it looks better, than you do a small redesign and use those Ti buffers, they look very good and might work too....

Take care!

Howardlong:
A quick aside, I2C is used on VGA, DVI and HDMI cables for EDID and HDCP. It's also one of the main sources of failure in the HDMI world, somewhat ironic considering the speed of the uncompressed digital video is of the order of 100,000 times faster.
T3sl4co1l:

--- Quote from: janoc on July 10, 2016, 07:32:16 pm ---That sadly assumes that the OP knows what they are doing. Right now I only see a level of ignorance that is only dwarfed by the arrogance of the "designer".
--- End quote ---

Geez, I know I'm frank, but that's just being mean...


--- Quote ---I would still ditch the whole I2C idea over the long wire and put there I2C<->RS485 converters.  However, seeing the "arguments" being put forward about the EMI not having to care about certification, I have my doubts that the OP will be able to pull that off.
--- End quote ---

I think this was originally suggested (in a past thread), so it's kind of redundant..


--- Quote ---Cause RFI on an airport frequency and you will have a van with a directional antenna in front of your house before you realize that you even have the gizmo on. Heck, people got located and fined for using a stupid baby monitor that was causing RFI at an airport frequency!

--- End quote ---

Anecdote: I worked several years at an industrial power conversion company, which literally was directly across from the airport.  Never had a call in history.

Probably the main 480V transformer, steel building, and generally-not-screaming-fast electronics, avoided putting anything in ATC range.  There was even a tech on the shop floor who preferred listening to AM radio, which was only ever intermittently interrupted by operating power supplies.  (Except for certain hijinx, but I will admit to none of it, I swear.)

Tim
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod