Unable to make FV-1 work..

[cloudscapes]

Posted this on the spin semi forums as well, but I figured since there were a lot of people with experience with it here, I'd repost:

***********************************************************************

Hi all,

After prototyping an effect using the FV-1 devboard, I had designed a PCB layout based off of it, confident that as long as I stick to the rules, it'll work.

My PCBs have come in. Populated the first one, but the FV-1 seemed dead. Checked voltages, tested continuity, caps, everything, and when I was confident that I had indeed kept to the rules, populated a second test PCB this time with the bare minimum the FV-1 would need. The crystal, a few caps and resistors, and the EEPROM. I still get nothing.

Here's the kicker. I get no clock signal on the clock pins on either of my two test boards. If I probe the clock pins on my devboard, I do indeed get the 32KHz clock as expected. The other kicker, the clipping led is constantly on as soon as I power on, even though I keep the input pins floating. That doesn't make any sense.

This is my initial layout, the one I thought would work:
http://www.soniccrayonfx.com/private/pitch_shifter_pcb.jpg

And when that didn't work, I populated only this on a fresh board. Same PCB layout as above, but using only the following components. The minimum that would be needed:



This is exactly how it's hooked up for the official dev board. Decoupling caps as close to the chip as possible, ground fills, etc. I left the inputs/outputs floating as all I needed to confirm was that I could get the FV-1 running, by probing the clock pins. But I don't even have a clock signal. Both chips on both boards just seem dead, and I have no idea why. When I populate only these parts, it's isolated from the rest of the board, so no trouble would come from unpopulated parts of this test board.

I solder quickly and cleanly. I'm used to hand-soldering much finer pitch than this, SOIC is as easy as DIP for me. Not a single shorted pin.

What did I do wrong? I just don't see it!

[octfrank]

Couple things to try:

Increase the cap on the crystal, some crystals need a larger value and some need a cap on both legs.

You are running in external EEPROM mode, are you sure the EEPROM was programmed correctly?

Try running internal programs and see if it is an issue with the EEPROM.

[Ice-9]

Exactly what Frank said, try a larger cap of nearer the value of 47p on the oscillator as some will not start correctly with the 15p value. Have you also checked your PCB for any errors ?

[cloudscapes]

Exactly what Frank said, try a larger cap of nearer the value of 47p on the oscillator as some will not start correctly with the 15p value. Have you also checked your PCB for any errors ?

Hey, thanks for the reply.

There was a bit of an exchange on the other forum.
http://www.spinsemi.com/forum/viewtopic.php?t=564

Yes, I did try larger caps. Though maybe I can try even larger than 40pf. I also tried using an external clock source from a function generator.

A few days ago, I thought it might be likely that there was an error on my PCB. Today though, I've gone over it a dozen times, and I just can't find one.

[cloudscapes]

I think the only thing I can do now is order this board I just designed:



and a few more FV-1s. The simplest FV-1 board possible, and go from there. Based off of the dead-simple official devboard schematic and my tests, I thought I could go straight to the finished PCB if I literally copied everything I was using on both the official devboard and the breadboard it was hooked up to, but I guess this was touchier than that.

I'm starting to think both the FV-1 chips I got were defective. There just isn't anything else I can find.

[Ice-9]

I think the only thing I can do now is order this board I just designed:



and a few more FV-1s. The simplest FV-1 board possible, and go from there. Based off of the dead-simple official devboard schematic and my tests, I thought I could go straight to the finished PCB if I literally copied everything I was using on both the official devboard and the breadboard it was hooked up to, but I guess this was touchier than that.

I'm starting to think both the FV-1 chips I got were defective. There just isn't anything else I can find.

That looks all ok, apart from a couple of the values, you have a 100n cap marked from pin 26 to ground which should be ok but I use a 1uf in that place. Also you have what looks like a 10 ohm resistor from pin 26 to 3.3v but again I think a 100 ohm resistor is the correct size here.

[cloudscapes]

I had a dream about this project, where I dreampt that the problem was I had completely omitted the 10/100 ohm resistor on the analog power supply.  I was so embarassed that I made up another reason for it not working. Also, for some reason, the board was home-etched single-sided.

The 10 ohm and 100n parts are as they are on the devboard, so I felt liek I really needed to copy it to get it working. That said, I'll probably socket everything so I'll be able to try different values. I haven't had the time to much work on this this last week.

Not much I can do but wait for this stamp, anyway.

[free electron]

Long shot, but you've got only a few µ1 caps at the output of the 3V3 reg. Try adding one larger bulk capacitor, like 10µ. Maybe the lack of it is causing a power up hiccup as puretube mentioned on the spinsemi forum.

[cloudscapes]

Long shot, but you've got only a few µ1 caps at the output of the 3V3 reg. Try adding one larger bulk capacitor, like 10µ. Maybe the lack of it is causing a power up hiccup as puretube mentioned on the spinsemi forum.

Yeah I had tried a 1uf and a 10uf on the regulator's output. Piggy-backed, as I hadn't initially included it in the layout. Still didn't get anything on either board.

I should be getting the mini test boards this week. And I've already gotten a fresh batch of chips. REALLY looking forward to trying.

[Ice-9]

Long shot, but you've got only a few µ1 caps at the output of the 3V3 reg. Try adding one larger bulk capacitor, like 10µ. Maybe the lack of it is causing a power up hiccup as puretube mentioned on the spinsemi forum.

Yeah I had tried a 1uf and a 10uf on the regulator's output. Piggy-backed, as I hadn't initially included it in the layout. Still didn't get anything on either board.

I should be getting the mini test boards this week. And I've already gotten a fresh batch of chips. REALLY looking forward to trying.

Hi Etienne,
Any news on how you are getting on with the build ?

[cloudscapes]

Not yet, I've been busy + distracted. 

I got the little test boards, even populated one. But haven't yet taken the 20 minutes to clean/fix up a breadboard with it, use good value parts, and just probe to see if I at least get a clock out of the chip. Yes, I'm that lazy.

This week, for sure.

[cloudscapes]

So I wired this up tonight. Fresh FV-1 chips from experimental noize. Exactly as shown on the little test board, except using a 100ohm instead of the 10. Power supply is 3.3v regulator output and 10uf filtering cap fairly close to the board. I have not wired up the inputs or outputs as the goal of this test was just to make sure I can actually boot the chip. I would do this by probing the clock pins to see the 32khz clock. I know this is a good way to see if the chip boots up because I can do it on my working FV-1 devboard.

It still doesn't work. This super simple board with a brand new chip is just dead. Fresh crystal oscillators too. Tried just with the internal programs by pulling T0 down and no EEPROM connected. Tried a variety of oscillator caps ranging from 11pf to 47pf. Tried swapping the refp cap for a larger one too. My soldering is quick and clean, it might as well been wave soldered.

I'm just about done. Either I have been given 3 duds from two different reliable sources, or there is something seriously wrong with my brain/eyesight.

[octfrank]

What xtal are you using? I've typically used Citizen can type and they have always worked. Crystals can be very picky, they can be working but stop as soon as touched by a scope probe. Try connecting everything and passing audio through without probing the xtal pins.

[cloudscapes]

What xtal are you using? I've typically used Citizen can type and they have always worked. Crystals can be very picky, they can be working but stop as soon as touched by a scope probe. Try connecting everything and passing audio through without probing the xtal pins.

sorry took me a while to reply. I've kind of put this aside since I'm at a loss what to do.

no-name brand, which might be an issue. however I did hook up the main clock pin to a function generator with 32khz 3v pk-to-pk sine, which should normally work (it does when I clock AVRs and PICs this way). still a dead FV-1. no audio (input/out caps present). 1v input as test.

I'll hunt down the citizen can crystals.

I really don't know how I can simplify a test board even further. I completely expected these little boards to work, so that I could move up from them and see what I did wrong in the original design.

no ripple/excess noise in my rails

[Digital Larry]

I'm sure you know this, but a crystal oscillator is usually formed by putting the xtal across an invertor.   So there's an "input" which you can drive and an "output" which goes off and drives the rest of the circuitry.  Are you sure you're driving the clock input?  What shows up at the output? (I do realize that Frank said that a scope probe could load it enough to stop it from oscillating).

[cloudscapes]

I'm sure you know this, but a crystal oscillator is usually formed by putting the xtal across an invertor.   So there's an "input" which you can drive and an "output" which goes off and drives the rest of the circuitry.  Are you sure you're driving the clock input?  What shows up at the output? (I do realize that Frank said that a scope probe could load it enough to stop it from oscillating).

The FV-1 datasheet tells me to use X1 as a clock input when not using the crystal, which is what I used. I don't get any sound even when not loading the clock pins with a probe, so I don't really know.

I read somewhere (some thread, somewhere on the spinsemi forums I think) that a full bright clip led even without any sound input may be indicative of a problem with the clock source. Acting kind of like a "there's an xtal issue" status led. Indeed it is lit up when I hook it up, so that's kind of the best info I'm going on.

I haven't hooked up the external eeprom for now, just to remove that as a potential problem. Trying to work it on the internal programs.

[Ice-9]

I'm sure you know this, but a crystal oscillator is usually formed by putting the xtal across an invertor.   So there's an "input" which you can drive and an "output" which goes off and drives the rest of the circuitry.  Are you sure you're driving the clock input?  What shows up at the output? (I do realize that Frank said that a scope probe could load it enough to stop it from oscillating).

The FV-1 datasheet tells me to use X1 as a clock input when not using the crystal, which is what I used. I don't get any sound even when not loading the clock pins with a probe, so I don't really know.

I read somewhere (some thread, somewhere on the spinsemi forums I think) that a full bright clip led even without any sound input may be indicative of a problem with the clock source. Acting kind of like a "there's an xtal issue" status led. Indeed it is lit up when I hook it up, so that's kind of the best info I'm going on.

I haven't hooked up the external eeprom for now, just to remove that as a potential problem. Trying to work it on the internal programs.

So for the clock are you using an external clock generator circuit or are you using a watch xtal ? It is unclear to me which you are using.

[free electron]

however I did hook up the main clock pin to a function generator with 32khz 3v pk-to-pk sine, which should normally work (it does when I clock AVRs and PICs this way). still a dead FV-1. no audio (input/out caps present). 1v input as test.

It's really a square wave you should feed there. Depending on what kind of sig gen that was, the output signal could have been bipolar, did you set the DC bias so it's 0-3V, not +/-1.5V?
One of the reasons why it worked with AVRs and PICs might be the fact that the OSC pins often share functions with ports, etc, which means there are always protection diodes on the inputs, clipping the negative side if there is any.
We don't really know much details about the dedicated XTAL pins on the FV-1, it might be that a sin wave is messing up the internal PLL so it can't lock properly.
Almost all projects i have done with the FV-1 used an externally generated clock signal fed into X1 (X2 left floating), never had any problems with it. In most cases i used MCUs one PWM channel set to desired frequency with duty @ 50%.
Check out LTC1799 - it's a tiny (SOT23-5) clock generator, F set with one resistor only, range 10kHz-30MHz. Not really cheaper as crystal, but you'll be sure the output is a proper clock signal, plus - easily adjustable frequency.

[cloudscapes]

I tried three different clock sources.

1. The 32khz watch crystal. Both ones from smallbear (came with the FV-1, so should have been good) and some I got from elsewhere.

2. Siglent SDG1020 function generator I've used for years. It was set for 0-3(.2)v peak to peak (I wouldn't feed AC into the FV-1). I've used the function gen many times both as high frequency clock sources and as an audio signal gen. I always make sure it's properly setup before I enable the output. I trust it. I set it as a sine wave because when I probe the crystal on my working official FV-1 devboard, a sine is exactly what I get. So I set out to match it exactly. I also tried as a square, but that didn't work. Switched back to sine because, that's what I probe on the working devboard.

3. 3v schmitt trigger from a couple inverters. Wasn't exactly 32khz, but pretty close. I've used these as  clock sources before, though not often. At that point I was willing to try anything. And I know some commercial FV-1 pedals I own (earthquaker afterneath, red panda particle) do something similar with inverters to acheive a variable clock on the FV-1.

I suppose I could try an MCU's PWM output next, at this point I'm not expecting it to work though.

I've looked at both layouts for my failed boards hard enough to make me go blind. I just can't find any obvious errors.

[cloudscapes]

What are the options for commercial mini FV-1 stamps? There's the experimental noize one:
https://www.experimentalnoize.com/products_makers.php

Wondering if there are others. Pretty sure I'll have to resort to that to finish at least this pedal.