How Many Bad FV-1's In a Batch?

[vigilante397]

I'm fairly new to the FV-1 universe, it's only been 6 months or so, but I love it. It's a great sounding, easy-to-use platform. But I've been extremely disappointed in how many bad chips I'm getting from an authorized distributor. I bought a batch of 10 chips a few months ago from Experimental Noize, the US distributor, and have been putting them in verified circuits with a reflow oven, not even hand-soldering, and so far have had 4 that didn't work and had to be replaced. Normally this wouldn't be a huge deal, but it's kind of an expensive chip to be throwing away 4 out of every 10. Is this typical? Should I be complaining to Experimental Noize?

[pruttelherrie]

I get my FV-1's from Banzai and so far (5 boards) I haven't had a dud yet *crosses fingers*

[Digital Larry]

I only ever built a few FV-1 boards by hand, but:
a) I think they are fairly sensitive to the crystal circuit and loading and even cleaning of flux from the PCB,
b) As Experimental Noize has at least some direct connection with one of the chip's designers, you absolutely should take it up with them.

DL

[cloudscapes]

check the crystal, and flux around the crystal. the FV-1 is DEFINITELY way more sensitive to suboptimal clocks than most chips I've used.

[Ice-9]

Being in the UK my official supplier is ProfusionPLC I have had hundreds of FV-1 from here and have only had one or two  issues which may or may not have been down to a faulty IC.

As said above the spec of the 32khz clock oscillator can cause problems, the capacitor for the crystal in the FV-1 datasheet states 15pf but 33pf works with most crystals.  What problem are you having exactly. for example a squeal at power up is cured 100% with the 33pf cap change. Other problems have there clues as well so if you mention what is wrong with the FV-1's in particular.

Flux used when soldering can also have an effect, for example do not use a water based flux pen.

[vigilante397]

The problem I'm having is that when I blend in the wet signal it's extremely low volume and extremely distorted. Every time this has happened so far it has been fixed by swapping the IC, but I'll have a look at the clock on this board I have on the bench. It would be great if this is my fault and not theirs

[Ice-9]

The problem I'm having is that when I blend in the wet signal it's extremely low volume and extremely distorted. Every time this has happened so far it has been fixed by swapping the IC, but I'll have a look at the clock on this board I have on the bench. It would be great if this is my fault and not theirs

That would not be a clock problem in my opinion, did you reflow the solder joints on the pins before changing the FV-1 to make sure the joints were ok. Maybe give Frank a shout an see what he thinks. 4 out of 10 doesn't sound good.

[octfrank]

That would not be a clock problem in my opinion, did you reflow the solder joints on the pins before changing the FV-1 to make sure the joints were ok. Maybe give Frank a shout an see what he thinks. 4 out of 10 doesn't sound good.

We have seen this as a crystal problem with cheaper crystals, they end up running at a high harmonic of the fundamental. Changing to a quality crystal (i.e. Citizen) has fixed it.

Flux used when soldering can also have an effect, for example do not use a water based flux pen.

^^^^This, water soluble fluxes are nasty, they are corrosive and conductive. We recommend never using them unless you have a proper wash station for PCBs. For anything else use no-clean type flux. This is not just for FV-1, it is for anything you build.

[vigilante397]

I'll try a different crystal on this board and let everyone know how it goes. I don't think flux is an issue, the solder paste I'm using has a no-clean flux. Good to know though. And I have tried re-flowing the joints, no change.

[vigilante397]

UPDATE: I just tried the other model of crystal I have, floating gloriously above the board because it's a different footprint, and still no change. Getting the exact same distorted, low volume sound. I've tried grounding pin 13 to run onboard programs to see if it was an EEPROM problem, still no luck. Any other thoughts?

[octfrank]

Try different values of the capacitor, crystals are sensitive and different manufacturers can have different requirements.

What make/model of crystals are you using?

Post schematic and layout so we can see if something looks off.

FV-1s are 100% tested so while we do expect a small infant mortality rate 40% is way out there. All FV-1s are tested with a crystal, Citizen can style, to make sure the internal crystal buffer works.

[vigilante397]

The crystals are NDK NX3215SA-32.768K-STD-MUS-2, and it looks like they have a load capacitance of 12.5pF, I assumed 10pF would be close enough. I don't think I have any 12.5pF on hand but I may have some 15pF, I'll give that a shot.

Here's my schematic:



And here's the top and bottom of my layout:

[octfrank]

I would try removing C17 and increasing C15 to 22pf

Never tried that NDK part but it does have a higher ESR, the Citizen parts are around 35K and that NDK part is around 70K but honestly not sure how much that would effect things. Crystals are all a bit "magic".

Crystal is nice and close to the pins, some of the layouts I see have the crystal a few inches away from the chip.

Would try to put all routing on upper and make lower as solid a ground plane as possible.

[vigilante397]

Thanks for the tip, I'll give it a try after work today. I would love to not have to re-fab the boards, so hopefully playing with the caps fixes it. I'm using this crystal successfully in two other slightly different FV-1 designs so I'm optimistic that I can solve it on this board.

[vigilante397]

Just tried it, C17 removed and C15 increased to 22pF, still no dice. Same as before, distorted sound at low volume.

[Govmnt_Lacky]

Nathan,

Have you considered building a "test board" so that you can verify the operation of your FV-1 chips prior to using them? This way you KNOW they are good before boarding them.

I cannot remember the packaging type for the FV-1 but, maybe use something like this:

[cloudscapes]

The problem with those test boards is if you have a chip that is extremely sensitive to the proximity of the crystal and decoupling caps as the FV-1 is, I don't see how you can get those components physically close to the chip with the test board (within a couple mm's).

[octfrank]

The problem with those test boards is if you have a chip that is extremely sensitive to the proximity of the crystal and decoupling caps as the FV-1 is, I don't see how you can get those components physically close to the chip with the test board (within a couple mm's).

Actually we use these types of sockets all the time to randomly test parts from batches after the full test. No problem using them.

[Govmnt_Lacky]

Actually we use these types of sockets all the time to randomly test parts from batches after the full test. No problem using them.

Typically, these SOIC test adapters are fairly pricey ($80-100 each) and not really cost effective UNLESS you are building for sale then it would be a great investment.

[octfrank]

Just tried it, C17 removed and C15 increased to 22pF, still no dice. Same as before, distorted sound at low volume.

Hmmm, not sure then. You mentioned the FV-1 is working fine in two other designs so it points to a layout issue. My thoughts are:

Solid ground plane, it is badly broken up and that can cause ground loops.

Single ground plane, it appears you have ground connected to both the upper and lower copper pours. Not good as this invites ground loops and multiple return paths.

If necessary move to a four layer board : top routing, ground plane, power plane, bottom routing

Get rid of any islands and antennas in the pours, can cause problems.