Need help with my design: oscillator problems

[NogginBoink]

I am designing and building my own light controller board. (I'm doing this mainly for the fun and challenge of it.) If I get it working, I'll be happy to share the details with the forum.

Part of my design uses a 4MHz oscillator as a timing reference. I've scoped the output of the oscillator and I'm a little concerned that the logic high value is 'fuzzy' on the 'scope. Does this mean that the output of the oscillator isn't always going to the full 3.3V V+ rail? At the moment, the output of the oscillator isn't connected to anything. The oscillator I'm using is an Abracon ASFL1-4.000MHZ-EC-T. (datasheet: http://www.abracon.com/Oscillators/asf1-L-L1.pdf)

Should I be concerned about this? (The oscillator will be an input to an Altera Cyclone FPGA.)

I apologize for the blurry picture; one hand on the 'scope probe, the camera in the other hand, with a 2-second exposure, just wasn't possible to get a clear pic.

[FireGod]

I would not worry about it. I have had one oscillator that "kinda" worked. I added a pull up resistor to it and it was OK, all the others I have used were OK.

[P. Short]

There is something funny there. It might just be the scope, but I'd take a closer look at the power to the oscillator, as well as making sure that the oscillator output isn't shorted to something or fighting with a PIC I/O pin that is in output mode.

--

Phil

[NogginBoink]

OK, this is interesting... look at the waveform of the +3V3 power rail. Those dips happen to be about as often as the zero crossing detector fires. (Bottom trace is GND; I need a second scope probe!)

I am using the zero crossing detector circuit described here: http://www.christmasinshirley.com/forum/viewtopic.php?t=927&highlight=zero+crossing

The 'scope is at 1v/div. Even at the low point, the voltage is still >2V, but that's still way wrong!

If there are any analog guys out there... does anything in my zero crossing circuit suggest what could cause this waveform on the power supply rail?

[NogginBoink]

Here are two traces of the output of the zero crossing detector at different timebases on the scope. (I now know why the waveform falls off at the end of the zc pulse!)

If anyone has any ideas what might be causing this I'd love to hear it!

(I will of course start tinkering with the ZC circuit but I am pretty weak on the analog side. Hopefully some Jedi master here will know the right black magic and be able to offer a suggestion. Note that the original ZC circuit was designed for a 5V circuit and I am using it unmodified in a 3.3V circuit.)

Thanks, all!

[NogginBoink]

Hm... I've removed R8 from my board, which effectively takes the zero crossing detector out of the circuit.

I'm still getting the odd dips in the 3.3V power supply rail.

My power supply is pretty simple: a 5V transformer to a bridge rectifier feeding a 3.3v regulator that feeds a 1.5v regulator. 10uF electrolytic across the bridge rectifier + and -/GND, 10uF electrolytic across the output of the 3.3V regulator and ground, and a 22uF electrolytic between the 1.5V regulator output and ground. Currently there's no load on the 1.5V regulator, and only the oscillator and the FPGA config device on the 3.3V regulator (as well as a power LED to ensure a minimum load for regulator stability.)

Other than that, all I have on the board right now are passives, the 4MHZ oscillator, and the FPGA configuration device. However, since the oscillator problems (which now appear to be symptoms of the power supply problem) appeared before I soldered the FPGA configruation device to the board.

Here are two new traces. The top trace is between GND and the + output of the bridge rectifier. The bottom trace is between GND and one side of the transformer secondary.

Something is definitely wrong here. But I have no clue what that something might be. The DC output should be at least CLOSE to a straight line, since it's got that 10uF cap on it, right? The xformer to GND should be a sine wave, right?

[P. Short]

You need to make the input capacitor to the 3.3V regulator larger.

Since I don't know what 3.3V regulator you are using, I'll assume that it starts to drop out at 5V.

With a 5V transformer, the peak voltage should be about 7V (since the transformer is light loaded, I'll assume 7.5V). That means that the voltage can only droop 2.5V before you run into trouble.

So the maximum current that you can draw is C*deltaV/deltaT.

With C = 10 uF = 10^-5F, deltaV = 2.5V, and deltaT = 8.33 mS = 8.33*10^-3S, this comes to

10^-5 * 2.5 / 8.33*10^-3, or 3 mA.

So it's not surprising that you see those displays.

If you are using a true FPGA (as opposed to a CPLD), the current can get quite high (perhaps 100 mA, or so), so I'd increase the capacitor to a much larger value (perhaps 470 uF, or larger), or increase the transformer voltage, if possible.

[NogginBoink]

I'm using a Triad FS10-600 transformer (http://rocky.digikey.com/scripts/ProductInfo.dll?Site=US&V=237&M=FS10-600) and an ST Micro LD1085V33 regulator (http://www.st.com/stonline/products/literature/ds/6738/ld1085xx33.pdf) which has a dropout of 1.3V, so 4.6V input should be sufficient, shouldn't it?

I have a beefier transformer. I'll see if I can hook that up and see if that works. (Unfortunately the footprint won't fit on the board, but I can implement some creative engineering to test.)

I'd like to plan to be able to draw an amp from the 3.3v rail (128 channels times 5ma/opto plus FPGA plus some buffer room). What size transformer and capacitor across the rectifier would you recommend? Would moving to the FS12-500 transformer (6.3V@1A) help? That has the same footprint as the current transformer so I could put it on the board.

(I'll do the math myself after a few cups of coffee but your input is sure appreciated.)

Thanks, Phil!

[P. Short]

One cup of coffee should suffice.

Anyway, if you use the 6.3V, 1A transformer, then (fully loaded) the peak output voltage should be around 6.3V * 1.414 = 8.9V. Using 1N5817 Schottky diodes in a full wave bridge rectifier configuration, the peak voltage going into the capacitor is about 8.1V, and so the allowable voltage droop is 3.5V or so.

So, the capacitor should be 1A * .00833S/3.5V, or .0024F, or 2400 uF. Bear in mind that this is a nominal value, as the incoming AC may sag a bit if you have a big display, and the capacitance of electrolytic caps decreases at low temps.

--

Phil

[NogginBoink]

Wow... 2400uF caps are a bit hard to find, and a bit pricey! :)

[P. Short]

You don't need anything fancy, something like mouser p/n 140-XRL10V3300-RC should work fine.

--

Phil