SV: Re: [sdiy] ARGHHH - PIC detunes VCOs depending on RFI *only* if USB cable plugged in

[Eric Brombaugh]

Karl,

Well, too bad those suggestions didn't work - my motto is to always 
start with the easy stuff...

The fact that touching the line cord changes things really bothers me - 
that still makes me think EMI/EMC which means you might need some good 
test equipment to track this down.

Adding RF beads to the USB data lines isn't what I meant (sorry) - I 
meant that you should filter the power supply lines into your USB-CV 
module - remember to put caps on both sides of the bead - that way you 
protect the module from the PS and the PS from the module.

Yes - the grounds between the PC and synth need to be hooked up to 
eliminate common-mode junk on the USB lines. Putting filtering/beads on 
the ground might help, but the impedance could mess things up too.

Are you using single-point (star) grounding in your synth supply 
scheme? If not then ground loops within your synth can cause the 
modules to talk to eachother.

I'm still wondering about ground loops, etc. If you hook your 
multimeter between the synth and PC USB ground pins, do you see much 
voltage difference? Hook a resistor between the two grounds - how much 
does the voltage change? Use ohms law to figure out the ground loop 
current.

Hang the multimeter on the synth power supply and plug in the USB - do 
you see any change in the rails?

If the DC view doesn't help, the next trick is to look at it with a 
good wideband 'scope. Check for HF noise on the supplies - does it 
change when you plug in the USB? If you don't have a 'scope then try 
changing the filtering on the supplies - does that change the amount of 
junk on the VCOs?

That's all I can think of for now... Good luck & thanks for sending the 
link to your site.

Eric

On Feb 12, 2006, at 10:31 PM, Karl Ekdahl wrote:

> Eric,
>
> see my earlier e-mail about the PIC USB
> schems/firmware etc.
>
> Putting the synthesizer and the computer / adding
> ferrite ring to the USB cable didn't change the
> behaviour.
>
> I added RF-Beads to D+ and D- along with caps to gnd.
> Any caps above 1nF made the USB not working anymore,
> not too strange but i noticed that whenever i plug in
> the USB cable the pitch jumps and starts wobbling.
> This became more apparent while using too big caps
> since the wobble quit as soon as the computer decided
> it couldn't communicate with the device. This leads me
> to think that the data being constantly transmitted to
> the device (polling data) is causing the interference.
> Nevertheless, the RF-Beads (i have two little black
> dealies without any markings that i'm fairly sure are
> beads... hehe) and caps didn't do anything. I am
> currently running the device from the synth-psu so
> that shouldn't be a problem. Should i put a bead on
> the GND coming from the computer perhaps? As i
> understand the USB ground and the PIC gnd (my internal
> gnd) *has* to be connected for the USB to work at all.
> Is there an alternative way of connecting the grounds
> if that seems to be the problem??
>
> thanks
>
> Karl
>
> --- Eric Brombaugh <ebrombaugh at earthlink.net> skrev:
>
>> Hmmm...
>>
>> A few thoughts:
>>
>> 1) Wow! You've got a circuit based on a USB PIC!
>> I've got an Olimex
>> board with an 18F2250 that I'd like to play with,
>> but there doesn't
>> seem to be much in the way of documentation on how
>> to drive the USB
>> hardware. Care to share your code?
>>
>> 2) Your problem sounds like an EMC/EMI problem. Any
>> time you get
>> problems from touching the (insulated!) power cord,
>> that sounds like
>> some sort of radiated interference or ground loops
>> between the synth
>> and PC through the USB cables.
>>
>> Some diagnostics:
>> a) Have you tried plugging both PC and synth into
>> the same outlet?
>> b) try some big ferrite clamps on the USB cable to
>> attenuate RF on the
>> USB cable
>> c) Ferrite beads and small (high-freq) caps on the
>> power into your
>> USB->CV circuit to filter out possible RF leaking in
>> from your PC.
>>
>> That's the first thing I'd try.
>>
>> Eric
>>
>>
>> On Feb 12, 2006, at 7:29 PM, Karl Ekdahl wrote:
>>
>>> Hrrfggg... yes, so, i have a USB->CV based around
>> a
>>> PIC18F2455. My VCOs detune when i lift or come
>> close
>>> to the main powercable, but this *only* happends
>> if
>>> the USB cord is plugged in to the PIC board. This
>>> happends wheter the actual v/oct from the pic
>> board is
>>> connected to the VCOs or not, so it seems to be
>> going
>>> through the powerlines... The firmware in the PIC
>> is
>>> allways running, usb or not, hence any possible
>>> interference from the x-tal can be ruled out (i
>>> guess). The PIC was first running on USB-power,
>> but i
>>> tried connecting it to the internal +5v rail
>> instead
>>> and that seemed to fix the problem - until i
>> actually
>>> soldered the cables in place. There might be
>> something
>>> wrong with the internal 5v supply (since i
>> "designed"
>>> it myself), it looks like this:
>>>
>>>                 43ohm 2W
>>> +9v o---------o---xxx-----oC
>>>         2.7K              /
>>> +9v o---xxx---o---->|---(  ) <- NPN 2N2219A
>>>               |           \__ 5.4v Out
>>>  6.2v Zener-> A            E
>>>               |
>>>              ---
>>>               -
>>>
>>> My thought was that 9/2700=3mA, down to 6.2, down
>>> about two voltage drops again through the diode
>> and
>>> the NPN. And then given worst case hFE i can get
>>> 75*3=225mA though the resistor only allows
>> 9/43=209mA
>>> which should be plenty for the pic to do its job.
>>> Right? Did i miss anything?
>>>
>>> I don't have a clue wheter it's the powersupply or
>>> not, but *something* weird is going on...
>>>
>>> thanks
>>>
>>> Karl
>>>
>>
>>
>