Unable to make FV-1 work..

[free electron]

I just took my old FV-1 dev board and cut the trace supplying the clock to the DSP. Result: peak led blinks randomly to stay permanent on after awhile.
Does the peak led go off when you supply the clock into the X1?

Can't really see anything wrong with your layout...

[octfrank]

What are the voltages on the pins? Did you use a flux to solder with and if so what type? Pictures of the failed boards?

[cloudscapes]

Picture of one of the failed boards.
http://c1.staticflickr.com/1/702/22141088093_bb44b6b594_o.jpg
The bits of fluff are q-tip stuff, from cleaning.

I'll do more later this week (voltages, other boards, etc). The other boards are messier, from all the crystal/passive swaps and desoldering, etc.

Peak led is always on. As soon as there's power applied to the circuit.

Continuity is all fine, no shorts, no cold solder joints. I use flux, kester flux pen, water soluble, ph neutral. Same flux I use for all my other stuff. I'm used to hand soldering much 0.5mm pitch stuff of higher density, so it would be pretty embarassing if I couldn't solder SOIC. I check all pins with a 10x eye glass. They're all good.

[octfrank]

Try another board but don't use a water soluble flux, just a little left on a board can cause problems with crystals.

[cloudscapes]

Just got only one board left, so I need to be sure. Would it cause problems with an external clock source such as function gen or cmos clock source?

I've used this flux on so many other projects, including boards with a variety of crystals and clock sources.

It's this pen:
https://www.sparkfun.com/products/retired/8967

Which flux do you use?

[snap]

Maybe a breadboard issue? It happens! (especially with multipin-units stuck in there). Needs a few more continuity-tests.

[amz-fx]

Peak led is always on. As soon as there's power applied to the circuit.

I have had this happen when the pcb was powered up when the FV-1 is set for external memory and no memory chip connected.

If you are not using the 24L32 memory, make sure the INT/EXT pin is set properly.

regards, Jack

[cloudscapes]

If it's a breadboard issue, it doesn't apply to my first two failed boards, since I was testing those without a breadboard. No shorts on the breadboard for the rows I'm using.

I've tried setting it to use both itnernal and external memory (with the memory hooked up of course). Either by setting T0 high or low.

[cloudscapes]

Forgot to mention. On my latest failed board, made a point to check the temperature of the chip when I powered it on for the first time. It's cold.

I'm going to try with different flux as soon as I've found flux that isn't... incompatible? No idea what to look for though. Never had any previous problems with it. Still doubtful that'll even fix my problem.

I find it bewildering that others have gotten the FV-1 to work with success on single-sided home-etched boards, and crstals/caps placed farther away from the chip. But I can't get mine to work after 4 boards/chips, with double sided boards, ground fills, ideal crystal/cap placement, etc. It's like everything I learned when working with 32bit microcontrollers is out the window. It's tremendously frustrating.

Even if I get this specific project working, I highly doubt I'll stick with the FV-1. It's not worth the effort. I should stick to easier DSPs, ARM, PIC32s, etc. It's a shame because the spincad designer is really really great.

[octfrank]

It's this pen:
https://www.sparkfun.com/products/retired/8967

Which flux do you use?

I use rosin based flux, water soluble are corrosive and conductive. Any board where a water soluble flux is used really needs to go through a proper pcb washer after soldering, hand washing is not enough as even just a little residue will cause problems with xtals.

[cloudscapes]

I'll try two things, drench the cleaner failed board with evaporating flux remover  and see if it revives it.

Then do a new board with just the solder's own flux. Because it's not that fine lead pitch, I might get away with it.

[anotherjim]

This must be extremely frustrating for you.

One thing I can suggest, and it's tedious to do, is...
If you have a working module with similar connections (I don't know how different your is from the Dev board).
Measure resistance to ground and Vcc for every pin on both and compare. A dumb thing to do I know, but it can show up an anomaly and point the way. You might even find something from the values of a bare unused chip. You need a needle test probe too.

[slacker]

I think the flux thing is probably a red herring, a bit of stray capacitance or resistance might stop an xtal working but I can't see how it's going to bother a sig gen or cmos oscillator. Probably a stupid question but on your layout both legs of the crystal are only connected to the FV-1 pins and not to ground? I can't tell from the picture, the pads look the same as the cap next to the crystal where presumably one leg is connected to ground.
It's a long shot but what are you getting the 3.3 Volts from? Your power supply's not oscillating or unable to provide enough juice, it needs about 70ma.
Unfortunately I can't think of anything else to suggest, in my experience they just work and most things I've made the FV-1 is stuck on a SOIC to DIP adapter board socketed on to vero, so far from ideal layouts. I have always used 10uF on REFP from the guitar amp schematic rather than the 1uf on the appnote but that shouldn't matter.

[cloudscapes]

I suppose next I could go raw SOIC to DIP adapter, nothing custom, like you did. I generally don't do that for microcontrollers and other high speed parts though, since I want to have caps as physically close to the chip as possible. And generic adapters tend to not have ground fills, which I also try and do on my custom boards.

For xtal grounding, only x2 is decoupled to ground with a cap (tried values between 11pf and 47pf, if memory serves). x1 is connected to xtal and nothing else. No cap. It's exactly as in the datasheet.

For kicks, I measured the resistance of a thin drop of the solder flux. Probes about 1mm apart, measured around 15-20meg. Yeah it's not a terribly useful or scientific test.

[octfrank]

I think the flux thing is probably a red herring, a bit of stray capacitance or resistance might stop an xtal working but I can't see how it's going to bother a sig gen or cmos oscillator.

I've seen it happen twice with water soluble, once on an FV-1 based board and once on a board with a different processor. In both cases even driving the boards with different clocks didn't work, so it may be rare but I have seen it happen.

[Ice-9]

I think the flux thing is probably a red herring, a bit of stray capacitance or resistance might stop an xtal working but I can't see how it's going to bother a sig gen or cmos oscillator.

I've seen it happen twice with water soluble, once on an FV-1 based board and once on a board with a different processor. In both cases even driving the boards with different clocks didn't work, so it may be rare but I have seen it happen.

I would have thought that very unlikey on Cloudscapes 3 boards to be reason for not working. I can agree maybe on one of the boards that the water based flux could show a problem but on all them ? I would have at least expected one to at least show signs of working.

All the same, maybe give your boards a spray with some solvent pcb cleaner.

Your circuit and pcb layout seems to look good but I am also assuming the pins that should go to ground actually do go to ground on the underside of the pcb. I can't see that for sure from the top of the pcb.

Not a lot can go wrong wih the basic circuit you are using, is your breadboard working ? have you tried wiring it up without breadboard ? is your 3.3v reg working ?

[Digital Larry]

Once upon a time when I was still working as a hardware engineer I did a board layout and generated the films with the "generate thermals" turned off.  Supposedly it was a quick board fix so nobody checked the film.  We had 200 boards made, and then the production manager came to my cubicle.  "ahhh there's a problem with those new boards you redesigned".  I went out there and saw it immediately.  Needless to say, the production folks who had to fix it were not too fond of me after that.  They had to scrape solder mask and add solder blobs on about 20 points per board.

Not long after that I went into project management!

Can you put up a clean shot of both sides of your board without any silkscreen?

[cloudscapes]

sorry for short reply, distracted

[Digital Larry]

Sorry if you've already answered this, but are you 100% sure your vias have continuity?  Does the board draw any current at all?

[cloudscapes]

Yup, I continuity tested everything.

Haven't had the time to do anything today, sorry. aside from ordering some SOIC to DIP adapters. I'll do more in a couple days, including measuring current. the clip LED is always lit so it'll draw some