"NU180" - A U180 drop-in replacement for the 3458A.

[Zondar]

Simon: Wow, what a sharp eye! Yes, that was the problem. Thank you!

About stray capacitance: One comment concerned the ladder resistors. Those are the 5 resistors in a row to the right of the PGA. There's nothing further I can do to reduce capacitance there.

The through-hole vs. surface mount resistor issue does impact stray capacitance. I've set up the wiring so that either footprint can be removed without any other changes being necessary, but significantly more stray reduction would be possible if either is removed outright. I could also go back to merged surface-mount plus through hole footprints, which would save a small amount of capacitance.

A question I have about stray capacitance regarding the 80k resistors is which side, if either, of the resistors is more sensitive to stray capacitance. That is, between the switches and the resistors, or between the resistors and the pins. At the moment, it favors resistors-to-pins, but I suspect that it's the other side that is more sensitive.

Another thing I should check is whether pins 4, 8 and 64 (+12 Ref.) are shorted on the A3 board or not. If they are already shorted, then I can delete a few short traces to all but Pin 4 or 8 (those are about equal).

(No, I'm not going to move the CPLD to the bottom. I do believe it would be better-off there, but it's too difficult to fabricate under most circumstances.)

I'm still hoping to hear an informed opinion about the main clock line (between inverter and the FF).

[TheDefpom]

Putting parts on the bottom could interfere with IC's which have been socketed on the A3 board, in my case I have a couple of socketed IC's which are close to the bottom of the NU180.

[Zondar]

Yet another round of changes:

* Rationalized (de-scrambled) all CPLD I/O. Requires minor code changes.
* Moved back to merged through-hole and surface-mount 80k resistors to decrease stray capacitance.
* Changed 50k resistor to 2010 size due to off-the-shelf availability (separate through-hole provided too).
* Moved all substantial digital traces to bottom copper, including the clock.
* Deleted top and bottom GND flood to reduce stray capacitance (top, mainly).
* The entire switch signal chain from FF to resistors, etc., and all analog traces (except Sum 1+2) are now top copper only over a continuous ground plane.

To do:

* Figure out how to swap layers, or redo inner Vdd and GND2 planes to improve stackup.
* Never-ending minor details.

Edit: Moved the CPLD region, deleted some unnecessary vias, etc.

[Hydron]

Thanks for all the work on this - I should have a known drifter 3458A on the way to me soon, so I'll build one of these up to try. Has anyone got a spare blank board they could chuck in the post btw? Would save a jlc order (and a little money, though 6L isn't that expensive any more, and jlc generally has some coupons). I'm based in the UK, though I've also got a US address that could be used.

I've already got some SMD 2ppm/C bulk metal foils (4 each of 20k, 40k, 72k and 8k) so should be able to cover the 80k values at least by using 72+8k in series. 50k is more available off the shelf or in a resistor network, so I shouldn't need an expensive custom resistor order at all.

That said, I will also have a go with 15V reference supplies and 50k or 100k resistor networks in place of the 80k parts - I don't really see the downside of one trivial resistor change on the A3 board Vs the potential ratio tracking and cost benefits of using standard resistor arrays. That would likely be a later project though due to design work being required.

[Zondar]

In this round of changes, I focused on the stackup. Here it is at present:

1) Analog signal. (No flood.)
2) Ground plane (GND).
3) GND2 plane, the negative supply to all 4 switch packages (and only those). Includes one trace for AGND2+AGND3.
4) VDD2 plane, the positive supply for all 4 switch packages (and only those). Includes two wide traces for VDD (digital power), which lie along the periphery and do not interfere with VDD2.
5) Ground plane (GND) plus one wide trace for "AGND3."
6) Digital signal (No flood.)

I think the above is an improvement over my previous setup. In particular, VDD2 and GND2 (power and ground for the switches) are now prioritized over digital logic.

My approach is to provide as close to one unified continuous ground plane as possible. However, I note that others have included ground planes selectively, e.g. deleting areas to reduce stray capacitance. I believe this would add value.

I'll make a few more touch-ups, but if someone wants files now, please just ask.

EDIT: Please let me know of any problems or improvements you can see. Thank you.

[Zondar]

I made some final changes and posted the files on the first page of this thread.

It's now been one year since the project started. Aside from correcting errors, this version may be my last for a while. Thank you all for the fun and educational experience of working with you on this great project!

Many ideas are still to be explored, and I encourage everyone to continue the good work!

Thank you again. 

[Hydron]

Any chance of pushing the latest changes to github? Thanks!

[coromonadalix]

i think Zondar made it clear ... other(s) may continue, and push everything to a git repo .... by themselves ?

[IanJ]

Mine will go to my git once I have got it powered up. Working on building up my 2nd pcb soon.

[Zondar]

I uploaded Version 1.1 files (denoted "v11") on github here:

https://github.com/NuZondar/NU180

Report any errors, bring your improvements to people's attention, and have fun! 

[Zucca]

Just wanted to drop a line to thank @Zondar for his immense work and passion invested in this...
Hope to see you back soon here... Stay safe.

[perdrix]

Does anyone (ideally in the UK) have a spare version 1.1 PCB that I could buy from you rather then ordering a batch from China?

Thanks, David (whose 3458A is drifting about 1.2ppm per day).

[Hydron]

I'm hopefully getting a drifty 3458A in the near future, if that comes to pass then I'd go in for a board too, or split the costs of a batch. Don't want to commit to the board fab costs until that's confirmed though.

[IanJ]

I have a bunch of Pcb's for my own version.......it's my 2nd build. My first worked, but I've made some tweaks.
I haven't built one up yet, I still need to order some custom low tempco resistors.
Once I have built one up and tested then I'll have some spare Pcb's (one reserved for TheDefpom and the rest I can make available to UK folks). I'll also publish a BOM etc.

However, I don't have a timeline as yet.

Ian.