EMRFD Message Archive 14235

Message Date From Subject
14235 2017-08-30 19:55:31 vk3hn I2C clock noise getting into RF circuits

Hello all,

Not exactly EMRFD material but increasingly an integral part of homebrewing receivers and transcievers, I am considering running I2C on two wires from an Arduino controller (si5351 and LCD) to a remote PCF8574 I2C 1 of 8 demux (for band switching a BPF bank).  I2C clock runs at 100kHz.  Will the I2C clock noise get into the receiver mixer and IF? 

The I2C will run a length of about 10cm.  I could use shielded twin core cable. 

The alternative is to put the PCF8574  close to the Arduino, eliminating the long I2C wires, but necessitating a 10cm run of 6 conductor ribbon cable carrying the P0-P5 demuxed lines to the 2N7000 relay switches.  

The former arrangement suits my physical layout better.   

Does anyone have experience of this?

Paul VK3HN. 

14236 2017-08-30 21:59:46 Clark Martin Re: I2C clock noise getting into RF circuits
You could slow the I2C down to 10 KBPS.  And remember, the SDA and SCL lines would only be active when accessing the devices on the bus.  So sitting there just listening there shouldn’t be any impact.

Clark Martin

14237 2017-08-30 22:25:00 Will Kimber Re: I2C clock noise getting into RF circuits

If you go to the Bitx29@groups.io group for the bitx40 which uses that
scheme there is much information. Mostly the noise emanates from the
nano and it requires much bypassing but some have had noise when the LCD


14238 2017-08-31 09:16:50 vk3hn Re: I2C clock noise getting into RF circuits
Thanks Clark, setting the I2C to 10kHz is a good idea.

You are right that the SCL only has clock on it when the bus is active.  However the si5351 is on the same bus, so it will get a clock on it every time the dial is touched.    
14239 2017-08-31 11:41:30 vk3hn Re: I2C clock noise getting into RF circuits
Thanks Will, I look at BITX20@groups.io.  I have a prototype receiver that already uses an Arduino, si5351 and LCD and is noise free.  My query concerns running the I2C lines away from the VFO module, up to the back of the chassis and near a (sensitive) BPF board at the receiver front end.  

14240 2017-08-31 15:13:21 Clark Martin Re: I2C clock noise getting into RF circuits
Run them on a ribbon cable with alternate wires being ground

+3.3 or +5 (optional)

Clark Martin

14241 2017-08-31 21:50:23 AncelB Re: I2C clock noise getting into RF circuits
I2C lines @ 5Vp will produce about 2.5mW of RF based on a 4.7K loading
(pull up), plus square wave harmonics. So running it at a slow rate
won't stop the HF capable edge harmonics.

Then of course there is the converse, nearby RF getting into the I2C
lines and messing up the comms. with the highish 4.7K pull up impedances.

Best option is to rout the i2c signal via twisted pair and run a couple
turns of it through a lossy ferrite core with a good permeability. Then
the harmonics are blunted and you have a common mode choke for incoming

If your lines are long I'd put 4.7K pullups on either side of the
signal run to combat RFI.

14242 2017-08-31 22:30:25 Jim Strohm Re: I2C clock noise getting into RF circuits

In my decades as a technical writer and occasional builder of digital hardware, I've never heard of I2C being the cause of noise in nearby circuits.  That's not to say it COULDN'T -- but it's not something that is a topic in my source documents, nor a topic in tech support discussions.

Of course, we're doing some fairly cutting-edge stuff nowadays with digital-enhanced radios, even in our home shops.  So if this were a problem, here is where we'd see it, and so this DOES count as a valid EMRFD topic.

Of course, I'd be more worried about RF getting into I2C than the other way around.

And speaking of "getting into" -- my bunnies got into my bookcase and flung my EMRFD CD onto the floor, then had their bunny ways on it.  I washed it off, and hope it's still usable.

At least they didn't take a mega-byte out of it....


PS My next batch of QSL cards will have bunnies on them, assuming I can get them to pose next to a Morse key and a warm final.

14243 2017-08-31 23:36:12 vk3hn Re: I2C clock noise getting into RF circuits
Jim (N6OTQ),

Thanks for your comments, yes, perhaps I am being over careful, and you might be right that a greater danger is my RF getting into the I2C bus and pulling the si5351 and over the place.  Maybe I'll just build it this weekend and see what happens.  The proof is in the pudding.  

Regarding your bunnies, and their bunny-like behaviour, shame on them for choosing your EMRFD CD for that purpose.  They could have chosen a far more appropriate place.  The instruction manual for your 2kilowatt linear, perhaps.  Or better still, they could have done their bunny-work inside your 2kw linear, while it was powered on.  Then, when they were well done,  you could have taken a mega-byte out of them!

73    Paul VK3HN   8^)
14244 2017-09-01 09:30:33 Jim Strohm Re: I2C clock noise getting into RF circuits

For what it's worth, I have an old free-standing Motorola VHF lo-band repeater in the house.  The 300-lb P/S is out and sitting on a little dolly, and the bunnies frequently nap in the open space in the bottom of the rack.

So basically they are already living inside my 2KW amp.

But they're too cute to eat!


14245 2017-09-01 19:40:35 Ashhar Farhan Re: I2C clock noise getting into RF circuits
Are you sure it is the i2c line? It could be a ground loop injecting the CPU noise into the rf/af board.
Can you disconnect the i2c lines leaving the ground and power lines in and then check?
- f

14248 2017-09-03 10:42:57 adwsail@gmail.com Re: I2C clock noise getting into RF circuits
From my limited experience Asher is right. I've seen a lot of noise on the power supply to the nano that finds its way into everything. After trying twisted pairs, (small improvement), inductors, ( little to no improvement) and various decoupling techniques I wound up with a series resistor on the 5v line to the nano. The value must be determined by trial and error. Mine turned out to be 72ohms but on a duplicate setup to get the same level of noise suppression it required 56 ohms. I'm not an EE so ymmv and I won't attempt to explain why it works but it does. 
14250 2017-09-03 10:59:02 Dana Myers Re: I2C clock noise getting into RF circuits
14251 2017-09-03 11:26:43 Bill Carver Re: I2C clock noise getting into RF circuits
In my RX preselector, an unamplified small antenna signal environment,  I have a MEGA2560 board with an SPI interface to a Harris RF592-02 preselector and parallel interfaces to a large 2x16 LCD, a bunch of SMA relays and a step attenuator. In addition to the preselector assembly, there are two low noise amplifiers to compensate for preselector loss and an 8 way output splitter.  There is NO RFI from the MEGA.

As Dana points out, Arduino code might well be sampling the switch and knob repeatedly and producing EMI even when there is nothing external happening.  In my preselector the MEGA board is interrupt drive: always ready to respond to a pusbutton, or the knob/2 phase encoder, but not generating any signals on I/O lines until that happens. My program is written in C in Atmel Studio environment (same C compiler as Arduino) and is interrupt driven. The MEGA chip only talks to things when a button is pushed or knob is turned. It could have been put to sleep, but that wasn't necessary.

I'm running an Orion-II board radio now, last item is 20 RX bandpass filters. A nano will periodically request FA and FB from HPSDR and change  RX1 and RX2 filters with parallel wires when needed. I'll post a message if I have EMI problems from the nano, but I'm not expecting any.


14254 2017-09-03 19:38:27 AD7ZU Re: I2C clock noise getting into RF circuits
I have a project in the works utilizing multiple I2C interfaces.  So far haven't seen any indication of noise into the Rx

but here are the precautions I used:
1. I2C connections via ribbon cables with alternating grounded conductors the clock and data lines. ie:  gnd - sda - gnd- scl- gnd
2. 2k pull up resistors on both scl and sda lines to 3.3v supply
3. 51 ohm series resistors in both scl and sda data lines at the output of a I2C mux chip, I am using 4 I2C interfaces in the design. 

there are several app notes describing adding caps to the lines to form a low pass filter  ~20ns time constant  but I did not implement that and have had no issues so far.  The general idea is to slow the edges
14256 2017-09-03 21:14:45 vk3hn Re: I2C clock noise getting into RF circuits
OK, thanks to Randy AD7ZU, Bill W7AAZ, Dana  K6JQ, adwsail@gmail.com, Ashar VU2ESE, Ancel, Clark KK6ISP, Will, Jim N6OTQ (for the bunny stories, although I must admit I still cannot be sure whether the bunnies are metaphorical or real) for your helpful comments on the reality of I2C noise getting into sensitive RF circuitry.

I must confess that I have not built my I2C controlled PCF8574 demux for filter switching yet, so as of today there is no noise problem -- I posted to solicit advice about such things in time to influence how I made up one of my boards.  

Given that most  people don't think it is a problem, I'll place the PCF8574 down on the filter board, with relay drivers, relays and BPFs, employing some of the methods people have described if it proves to be noisy.  Judging from people's responses, I don't expect it to be.    

Thanks for your collective wisdom.  

73 Paul VK3HN.