Clarification on ILDA voltages

[harrissanford]

mixedgas, schematics would be appreciated!

Stoney3K, hmm, I didn't think about combining them that way. The attachment shows my first thought to doing this without offset trim pots. The switch allows connections in two states.. the top input going to the top output and the bottom input going to the bottom output... or crossed over. I forgot what I bought them for before, but I know they can be found cheap!


Hmm, wonder which way is best.. or if there's another. I saw J4cbo's design which looked nice, but I couldn't find a schematic.

[j4cbo]

You can't combine DACs like that to get more bits; it'll be very non-monotonic when the low-order one rolls over. Feeding one DAC into the reference of the other will give more resolution at the low end of the scale but not throughout. However, 12 bits is plenty for galvos (gives you a 4096x4096 "resolution"); the MCP4922 would be a fine choice for just X and Y.

Going from 0-5v single-ended to +/-5v differential is pretty easy with op amps, but you'll need suitable rails for the op amps (at least +/- 8 volts, unless you use a rail-to-rail part). First generate the inverting output with a gain of -2 centered around Vcc/3, and then flip that around ground to get the non-inverting output.

Or, if you don't want to bother with a negative rail or any op amps at all, you could just produce 0-5v for the inverting output and 5v-0v for the non-inverting output, and deal with having half the output range. It's not optimal, but it'll work just fine.

[harrissanford]

Woao, who else can I summon by saying their name? Thanks for your input j4cb0. I like your ethernet design.

By rollover, you mean when it flips from using the lower dac to the upper dac.. or do you mean when it rolls over the 0v point? I can understand there would be a jump when flipping from using one dac to using the other.. but in my brain I believe I would be OK provided I interspersed the bits to make the voltages. I.e. to make a positive voltage, I would set the lower 2 bits on the first dac.. and the lower 2 bits on the second dac. Then the only challenge is making sure they update at the same time... but if the settle time is small enough, it may not matter.

Does that make sense to you?

[harrissanford]

Another question too. Why is it 16bit audio dacs are cheap.. but standard 16bit dacs are difficult to find? This is why I keep pushing for more bits.. there's got to be a simple means to do it!

[j4cbo]

Oh, I just looked at your diagram - it seems like you'd only ever be setting the two DACs to the same value, which would make it 9 effective bits... there would also be a significant discontinuity when flipping around from one to the other, too.

16-bit delta sigma (http://en.wikipedia.org/wiki/Delta-sigma_modulation) audio DACs are cheap because they're produced in huge quantity; they're designed for low distortion of audio signals but have a noise profile that isn't nearly as good for drawing images. A 16-bit delta sigma audio DAC *probably* wouldn't be any worse than a 12-bit instrumentation DAC, but I wouldn't expect it to be a noticeable improvement either.

[harrissanford]

I would not be setting both dacs to the same value, I would alternate them.



I.e. For 0 to +10v differenc the i/o switch would not be powered (see #4 diagram), to increase the voltages, the first steps after 0 would be
step1: TopDAC: 0x1, BottomDAC: 0x0 ~= 152.587891uV
step2: TopDAC: 0x1, BottomDAC: 0x1 ~= 305.175781uV
step3: TopDAC: 0x2, BottomDAC: 0x1 ~= 457.763672uV
step4: TopDAC: 0x2, BottomDAC: 0x2 ~= 610.351562uV
step5: TopDac: 0x3, BottomDAC: 0x2 ~= 762.939453uV
....


For 0 to -10v difference the i/o switch would be powered (see #4 diagram), the first steps after 0 would be (same procedure)
step1: TopDAC: 0x1, BottomDAC: 0x0 ~= -152.587891uV
step2: TopDAC: 0x1, BottomDAC: 0x1 ~= -305.175781uV
step3: TopDAC: 0x2, BottomDAC: 0x1 ~= -457.763672uV
step4: TopDAC: 0x2, BottomDAC: 0x2 ~= -610.351562uV
step5: TopDac: 0x3, BottomDAC: 0x2 ~= -762.939453uV
....





Does that make more sense? This way I would expect a monotonic step setup.. it may not be linear between each individual step however.. but at such a small resolution, it may not be an issue. Actually, I guess I could see it making everything non-monotonic now that I think about it more It will look like a staircase with lippy stairs if you plotted all the steps. dang!



Those analog switches are pretty fast from my experience. Do you really think the switch delay would be an issue?

[mixedgas]

Just how fast do you intend to excite that crossing scheme? The reason I'm asking is galvo shaft resonances, and the fact that some galvo amps have a pole right around 20-30 Khz...
That is that whine you hear on cheap galvos at rest.

I'm still at work, 12 hour day turning to 13 hours, but I will still scan the doc.

Steve

[harrissanford]

Well, my galvos are 30k galvos. I can't say I've heard that whine which is interesting because ti would be hard to get more chinese than these are. They amplifiers didn't even come with heatsinks! Just a black aluminum plate.

At a max, I suppose I would be toggling it at 15k, but I would design it to go much higher in case I were to really enjoy playing with this set and want to move to some 40/60k galvos.

[simonb]

Hi Harrissanford,

Did you eventually work out how these X & Y Galvo +-10v (or is it 20v) differential Signals work?
I'm having similar problems understanding things!!

My Arduino Due has a pair of 10 bit DACs, which give a 0v to 3.3v unipolar output.
I think I need to connect these to a simple op amp circuit to get the X & Y +ve & -ve outputs, which can go to my Galvo thru the ILDA connector on my 1800mW laser projector.

Did you get yours working?

Best regards,
Si Bond.

[Grix]

Hi Harrissanford,

Did you eventually work out how these X & Y Galvo +-10v (or is it 20v) differential Signals work?
I'm having similar problems understanding things!!

My Arduino Due has a pair of 10 bit DACs, which give a 0v to 3.3v unipolar output.
I think I need to connect these to a simple op amp circuit to get the X & Y +ve & -ve outputs, which can go to my Galvo thru the ILDA connector on my 1800mW laser projector.

Did you get yours working?

Best regards,
Si Bond.

You need to read the datasheet of the microcontroller on the Due a bit more carefully. First of all the DACs are 12-bit not 10-bit, but more importantly they don't output 0V to 3.3V, but 1/6 Vref to 5/6 Vref, in other words 0.55V to 2.75V. This means you need an offset in your amps.
Like I've mentioned to you before you can look at the schematics for the Helios DAC for such a correction amplifier as it uses a microcontroller/DAC in the same family as the one on the Due. Here they are:
offset signal: http://i.imgur.com/A3e5Bf1.png (switch 2.5V for 3.3V source)
main amps: http://i.imgur.com/2CzE9tc.png (R42 and R41 are optional)
The 1.65k resistors need to be adjusted for 3.3V instead of 2.5V reference, just use the feedback formulas to find the resistance needed if you are familiar with opamp circuits