RFI problems on diy headphone amp.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
geraldfryjr

geraldfryjr

Member
Joined 2008
Here is a very interesting post and it is one of the ones I remember although the whole thread deserves a good read,

http://www.diyaudio.com/forums/solid-state/224539-ground-planes-audio-pcbs-legit.html#post3263267

http://campusinterviewindia.com/pcb-layout.html

http://www.picaxeforum.co.uk/archive/index.php/t-12772.html

http://tesla.unh.edu/courses/ece711/refrense_material/Expedition/pcb_overview.pdf

http://www.celectronics.com/seminar/sample/IEEE11-9-05.pdf

http://forum.allaboutcircuits.com/newsgroups/viewtopic.php?t=170

I have been searching through this stuff as a start,

https://www.google.com/search?num=1...33.796.9j1.10.0...0.0...1c.1.7.hp.MeFhtUUJx9c

jer :)
 
Last edited:
A

Amperage

Member
Joined 2013
geraldfryjr said:
Here is a very interesting post and it is one of the ones I remember although the whole thread deserves a good read,

http://www.diyaudio.com/forums/solid-state/224539-ground-planes-audio-pcbs-legit.html#post3263267

I have been searching through this stuff as a start,

https://www.google.com/search?num=1...33.796.9j1.10.0...0.0...1c.1.7.hp.MeFhtUUJx9c

jer :)
Click to expand...

Thanks. I looked through some of the links you posted earlier and am currently reading the analog devices guide on PCB design. I'll make sure to save that guide for future reference. Tons of good stuff in there.

Even if the info doesn't result in a fix for my current problem it's definitely handy to know.
 
qusp

qusp

Member
Joined 2009
yes lord of the board is an excellent paper and if we are getting into loops and EMC, at the risk of burying you in excellent information that will make you rethink everything you are doing, then even the free material Henry Ott gives away is superb. the book, Electromagnetic Compatibility Engineering is essential reference material. Its not light reading (its huge!), or exactly cheap, but very good value IMO. Hottconsultants hes a legend in the industry, for very good reason.
 
C

chris719

Member
Joined 2004
If you have multiple ground planes, a rule of thumb is that they need to be stitched together with vias regularly at 1/20 wavelength of the highest frequency you are "interested" in.

100pF caps are probably not helping much unless they are 0201 or smaller package and you have a low inductance attachment pattern. A few nH of inductance will render them useless at 800 MHz - 2100 MHz.

Cell phones have no audible issues from this sort of thing because they are designed to be RF hardened. Leadless packages help, believe it or not, along with common-mode chokes on inputs and ferrite beads of the right mix on the outputs. Murata sells several whole lines of specialty EMI filters designed for these applications.

I have seen a piezoelectric buzzer on a turned-off device make 217 Hz noise by itself from a GSM phone transmitting at full power placed nearby.
 
A

Amperage

Member
Joined 2013
Well I killed the ground fill since there are too many questions surrounding what it may or may not be doing and created a new single sided layout with star ground. Before I waste more time etching a board and soldering all the components on I figured I'll get some feedback on it first. As an added bonus my ground connection point kind of looks like a evil cat with really long whiskers.

An externally hosted image should be here but it was not working when we last tested it.
 
qusp

qusp

Member
Joined 2009
I think star grounding of this type is a bit of a waste of time. actually ground fill is essential IMO, you just need to use it correctly. as long as the parts have leads, the inputs or outputs have bipolars and you have cables, you still have vulnerability and ground planes are the only way to keep inductance really low at these frequencies. your star ground there, is a huge group of dipole antennas.

yep Chris, the murata stuff is excellent, as are the reading materials. thanks for the backup, I linked/mentioned them earlier. the 3 lead common mode choke + feedthrough cap EMIFIL parts are superb and I think they are the best way to tackle this problem at the input.
 
qusp

qusp

Member
Joined 2009
Amperage said:
Wow.. I was browsing murata's website looking at their noise suppression products and stumbled across this page.. jackpot!

Course for Noise Suppression against LTE Noise | Noise Suppression Techniques | Murata Manufacturing Co., Ltd.

it's information on filtering against the exact type of noise I'm having trouble with.
Click to expand...

erm, isnt that the exact pages I linked you to? hehe ahh no not quite, but its the same parts.

of course they are promoting their parts to solve every problem as you would expect, but thats OK, because I think they are unequaled for the job at hand. common mode + feedthrough cap on input, mixed common mode and differential filter on output. but you need to know what frequencies you are dealing with for any of this really. do you have a scope or access to one?

I think the range for the input is the NFM range
 
Last edited:
A

Amperage

Member
Joined 2013
qusp said:
I think star grounding of this type is a bit of a waste of time. actually ground fill is essential IMO, you just need to use it correctly. as long as the parts have leads, the inputs or outputs have bipolars and you have cables, you still have vulnerability and ground planes are the only way to keep inductance really low at these frequencies. your star ground there, is a huge group of dipole antennas.

yep Chris, the murata stuff is excellent, as are the reading materials. thanks for the backup, I linked/mentioned them earlier. the 3 lead common mode choke + feedthrough cap EMIFIL parts are superb and I think they are the best way to tackle this problem at the input.
Click to expand...

Thanks for the feedback. I did see your earlier post on the chokes but I don't know enough about filter design to order the proper parts. Having found info specifically on LTE noise suppression perhaps they will have some sample layouts with part #'s I can use to spend money more wisely :).

I've been reading so much contradicting material today on star grounding, ground planes and the like I figured I would try almost anything just to see a difference.

I've been soldering components and slicing and patching traces on this little board most of today trying different configurations and nothing short of yanking the op amp outta the socket seemed to help :p. I still like the evil cat though :).

I know one thing for sure though. You guys are a great help. I've been designing and building simple devices for quite a while now without much issue but this is the first time I've had to tackle a high powered RF assassin sitting less than an inch away from something I built.
 
A

Amperage

Member
Joined 2013
qusp said:
erm, isnt that the exact pages I linked you to? hehe ahh no not quite, but its the same parts.

of course they are promoting their parts to solve every problem as you would expect, but thats OK, because I think they are unequaled for the job at hand. common mode + feedthrough cap on input, mixed common mode and differential filter on output. but you need to know what frequencies you are dealing with for any of this really. do you have a scope or access to one?
Click to expand...


Yeah I like Murata components. They seem to be really decent quality. I use their inductors most of the time when I'm making DC/DC converters and they perform very well. I just don't know much about RMI/RFI suppression and didn't want to just start ordering parts till I was sure I had something usable for what I'm doing.

Also I don't have access to a usable o-scope. I have an old heathkit 5mhz scope that's ok for measuring some stuff but a bit long in the tooth for today's technology.
 
C

chris719

Member
Joined 2004
Using the fill is ok, but stitch it to the plane with a grid of vias every 4-5 mm or so. Use direct vias not thermals. These are just EMC suggestions, mind you.

You'd need an RF spectrum analyzer and probes to see this stuff anyway. Try ferrites from the BLM line that have impedance peaks near 1 GHz. Also look into the X2Y capacitors from Johanson or Yageo, they make great input filters and are effective in larger package sizes.
 
A

Amperage

Member
Joined 2013
chris719 said:
Using the fill is ok, but stitch it to the plane with a grid of vias every 4-5 mm or so. Use direct vias not thermals.
Click to expand...

This wouldn't be a problem if I were having boards made at a PCB fab (which eventually I will) but that's a ton of vias to drill out and solder by hand doing home made PCB though :)

http://psearch.murata.com/capacitor/product/GRM033R71C151KA01%23.pdf

Those are the filter caps I'm using Murata 0201 X7R 150pf

and these are the ferrite beads

LI1206H151R-10 Laird-Signal Integrity Products | 240-2402-1-ND | DigiKey

apparently I'm using the wrong beads

[FONT=‚l‚r –¾’©][FONT=‚l‚r –¾’©]
[/FONT][/FONT]
 
Last edited:
C

chris719

Member
Joined 2004
Yeah, it's possible you can improve on those ferrites, but like others said, I would bet you do have something acting like a good antenna. Shorten whatever you can, although at maximum power in close proximity, even really short (way less than 1/4 wavelength) traces can be problematic.
 
A

Amperage

Member
Joined 2013
chris719 said:
Yeah, it's possible you can improve on those ferrites, but like others said, I would bet you do have something acting like a good antenna. Shorten whatever you can, although at maximum power in close proximity, even really short (way less than 1/4 wavelength) traces can be problematic.
Click to expand...

Yeah it's rough when I can't see exactly where the signal is leaking in so I just keep adding filtering components and listen.. nope didn't fix it. lets try something else. I unsoldered the hole thru connectors and components and milled off the other layer I was using for a ground plane just to be sure it wasn't odd capacitance issues with the 2 ground layers. No dice. So now I have only the copper on the component side.

I'm not sure how much shorter I can really make things.. the whole board is already pretty small. like 1.5" x 1.5" and it has to be large enough to physically mount the jacks and volume pot etc.

I've read a number of headphone amp reviews where they commented on radio interference but I keep thinking back to the FiiO E6 and the review that said it's not affected by radio noise from the cell phone. I'm not sure if they tested with LTE signals or not. Mine works ok with 3g it seems. At least nothing I can hear but 4g data is most definitely not going by unnoticed :)

There are pictures of the board available but only one side so I can't make out what they've done layout wise but I'm pretty sure the bottom of the board lays flat on the casing so I doubt they have components on that side and I don't see any chokes or ferrites scattered around the audio signal areas on the visible component side.

The other thing that I keep thinking about is the fact that my amp even when turned off will pass the noise through to equipment I have plugged into the headphone jack until I remove the op amp from the socket. Either the chip itself is picking up the signal or it's acting to bridge diff parts of the circuit to form the working antenna.
 
Last edited:
A

Amperage

Member
Joined 2013
chris719 said:
The op-amp is rectifying the signal, which is to be expected. You have to prevent it from getting in. I don't remember the first few posts off hand, but did you try shielding the op-amp in a sort-of Faraday cage?
Click to expand...

Yeah. in the actual fully assembled amp the whole thing is enclosed in a aluminum shell. I also wrapped a piece of paper around the test board I've been mucking with and wrapped foil around the board and twisted the ends tight so there were no openings except the headphone cable coming out.

It still picked up the radio noise. which it was almost certainly getting in on the headphone cable you would think because everything else was wrapped in foil but I have 150pf caps right from the headphone jack pins to ground and the ferrite beads which I've established aren't perfect by any means. They extend up to the 700mhz band but apparently LTE uses 2ghz as well and the ferrites I'm using do nothing for that but as far as I can tell by this page the E6 isn't using any ferrites at all. They have the component all labeled and I see a solitary inductor hanging out there.

FiiO



If I unwrapped the amp and instead wrapped the foil around my phone the noise stops instantly... and the phone of course loses signal and drops off the network but it will emit some death bursts as it tries to get signal again and none of that gets through. So shield the phone.. all good. Shield the amp and it seems to do nothing.
 
A

Amperage

Member
Joined 2013
Well doing some more reading and it seems that the info I read about the E6 not picking up cell phone RF signals wasn't accurate. I found this thread and the owner says that it picks up noise from his iphone unless he clips it on his shirt.

Well I'm still going to continue trying to beat this but at least I feel a bit better knowing a $20 amp hasn't managed something I've been struggling for the last couple weeks on :)
 
agdr

agdr

Member
Joined 2010
agdraudio.com
Amperage said:
Well I'm still going to continue trying to beat this but at least I feel a bit better knowing a $20 amp hasn't managed something I've been struggling for the last couple weeks on :)
Click to expand...

Kudos on your persistance! A lot of folks have posted about RFI interference problems with amps but you are one of the first to dive into the troubleshooting to track it down.

Everything looks different at 2.5Ghz, of course. Capacitors become inductors and PCB traces turn into transmission lines. I had to think back a bit about how tuners were done way back when on 1gHz ham radio gear and TV UHF tuners. The power lines would always be run through feed-through mica caps like this soldered onto the metal tuner case

http://www.ferroperm.com/gfx/news_images/33812006D%283%29.jpg

to keep the RF from getting back on the power supply leads. I went Googling for what the modern equivalent might be (I'm wayyyy out of date) and came up with what chris719 posted about, these X2Y feed-through chip capacitors:

GSM RFI Suppression with X2Y® EMI Filters (second paragraph has good stuff about audio)

http://www.johansondielectrics.com/x2y-products.html/x2y-for-emi-filtering.html

500X41W224MV4E - JOHANSON DIELECTRICS - CAPACITOR CERAMIC 0.22UF, 50V | Newark

Those really do look like a potential answer. Might wind up requiring a re-think of the whole thing on PC layouts. These feedthrough chip caps right at the inputs, power, and outputs like they show in the first link. I suspect that physically large "lumped" RF filter elements, like the 150pF cap and series resistor, only look like a cap and resistor up to a few hundred mHz. Beyond that it may take something like these feedthrough parts.

There has to be a way to nail it since the audio section (and video camera section) of cell phones are apparently immune. The phone designers have found a way to filter out the RF with a much higher field strength, right there at the emitter.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom