v1.0rc2 schematic review

hartytp commented 4 years ago

@jordens have you completed your review of the schematics? I'll do mine now.

hartytp commented 4 years ago

[ ] Cosmetic: why do R170 and R171 have the value "undefined" rather than just being DNPd in this variant with a red cross? Doesn't that screw the bom/make populating them later on harder (since the value they should have in a populated variant is not shown)
Maybe I'm being blind, but I don't see any red user LEDs.
Cosmetic: why is the power budget an excel file instead of the pretty looking annotations we normally have? Doesn't really matter, I suppose, but nice to keep all the designs consistent and pretty!
[ ] +-12V regulator set resistors look an order of magnitude too high to me.

hartytp commented 4 years ago

[ ] The LT3045 and LT3094 require a minimum of 10uF input capacitance and recommend low ESR capacitors. Are you sure the 1uF ceramic + 10uF electrolytic will do, or should we replace the 1uF with 10uF ceramic?

hartytp commented 4 years ago

NB I know we've had this discussion before, but the LT3045 set capacitor is recommended to be ceramic. I guess we don't care in this case as we're not relying on this to be massively high stability, but worth noting

hartytp commented 4 years ago

[ ] for the reference buffers, the set capacitors should be ceramic as recommended by LTC

hartytp commented 4 years ago

[ ] I believe you got the Ilim wrong for the LT3094

jordens commented 4 years ago

I'm done with my part schematics review.

If you search the PDF for "RED" you'll get plenty of meaningful hits. LD1D and LD1C are red.

The power budget obviously needs to be taken with a grain of salt. (a) An opamp doesn't simultaneously load both rails with maximum current. (b) A large part of that power would be dissipated elsewhere (c) All channels at maximum current is not within the target use case. Baseline power consumption is probably more like 4-5 W.

hartytp commented 4 years ago

@gkasprow not that it really matters, but why are we not using the data sheet recommended values for the passives on IC18?

hartytp commented 4 years ago

If you search the PDF for "RED" you'll get plenty of meaningful hits. LD1D and LD1C are red.

Aah, thanks, I'd mised LED_RED.

The power budget obviously needs to be taken with a grain of salt. (a) An opamp doesn't simultaneously load both rails with maximum current. (b) A large part of that power would be dissipated elsewhere (c) All channels at maximum current is not within the target use case. Baseline power consumption is probably more like 4-5 W.

Yes, these power budgets always end up being very pessimistic. Still, they're useful to have...

hartytp commented 4 years ago

@gkasprow not that it really matters, but why are we not using the data sheet recommended values for the passives on IC18?

Same for IC90

hartytp commented 4 years ago

@gkasprow if you still have problems with the layout, I'm fine scrapping the CMCs.

hartytp commented 4 years ago

Comments about the LT3045 apply to ref buffers as well (not sure any ILIMs are set correctly)

hartytp commented 4 years ago

@dnadlinger are you happy with the current filter implementation? Didn't we agree to DNP C3?

hartytp commented 4 years ago

[ ] I don't think this annotation is correct any longer

hartytp commented 4 years ago

[ ] these annotations are incorrect (x4 gain and +-2*Vref)

hartytp commented 4 years ago

[ ] Is this compensation capacitor optimized for the current OpAmp choice?

hartytp commented 4 years ago

[ ] Do we need to increase P5V0A to 5V1 or something to ensure that the DAC Vdd is always higher than Vref (even when taking component tolerances into account)? Before doing this, please confirm that there is nothing else on that rail that can't handle the change in voltage.

dnadlinger commented 4 years ago

@hartytp: The issues are you pointed out are valid. As for C3, it does (in SPICE-land) buy you 16 dB extra attenuation at 4 MHz, so I'm 50:50 on whether to populate it initially to see how bad drifts and channel-channel spread really are vs. doing so by hand on a single board.

hartytp commented 4 years ago

hmm...or can we just scrap P5V0A and use only P5V5A instead?

hartytp commented 4 years ago

(maybe not if there are sequencing issues)?

hartytp commented 4 years ago

As for C3, it does (in SPICE-land) buy you 16 dB extra attenuation at 4 MHz, so I'm 50:50 on whether to populate it initially to see how bad drifts and channel-channel spread really are vs. doing so by hand on a single board.

Okay, let's keep it for the initial version.

hartytp commented 4 years ago

what P1V0?

hartytp commented 4 years ago

what is P3V3A? I don't see a source for it? Or, am I missing something?

hartytp commented 4 years ago

Should IC12 EN be tied to P5V5A PG?

hartytp commented 4 years ago

We don't need D22A since there is no way of powering Fastino without the EEM connector

hartytp commented 4 years ago

Shouldn't L6 be the same as Stabilizer?

hartytp commented 4 years ago

Sequencing isn't correct (mentions rails that don't exist)

hartytp commented 4 years ago

Sorry, let me break these down into easier issues for triaging...

hartytp commented 4 years ago

Okay @gkasprow I've finished my review. Looks good, only some minor fixes.

Any chance we can send this to manufacture on Monday?

gkasprow commented 4 years ago

I will do my best to send it on Monday. The same with Stabilizer.

hartytp commented 4 years ago

Amazing! Thanks!

hartytp commented 4 years ago

The design looks good by the way! I'm very happy with the way it's turned out

sinara-hw / Fastino

v1.0rc2 schematic review #24