Re: What's with the DPRB?

neat,how many layers? Did you do the layout?

martin

Six. The circuitry is pretty simple. All the fast stuff is on layer 1, and the miscellaneous signals are on 6. Ground is 2. The other layers are all power pours.

Three of us worked on the pcb layout. I did the fast stuff... still tweaking it in fact. My cad guy did the basic board setup and the pours (he's great at pours) and will do the formal release and gerbers. One of my engineers designed and layed out the power supplies on the right. Another one of my guys worked out the optical receiver stuff.

My production manager is also a great mechanical designer. We looked around for standard boxes, but nothing was right, so he designed this to be custom fabbed from sheet metal. It will be cheaper than a standard enclosure, which seems silly. The pic is from SolidWorks.

The front and back will have polycarb label/overlays. Nowadays they're digital printed in unlimited colors (not screened) and laser cut (not steel rule die) so we can go crazy on graphics.

The next spin, once we get this to work, will dump the switches and trimpots for a uP with dacs, and add time trims per output.

The goofy looking run upper-left is an impedance test trace. I'll TDR it on a bare board to see how things came out. The two square SMA's are for experimenting with plane impedances and noise, just for fun.

John

Thanks, can I ask what the 2 versions of Lombard Street are, going to S2, S3, timing would be my first guess

martin

Trying to equalize prop delays. This is better:

ftp://66.117.156.8/P730_16.gif

It's almost done. Tweaking trace lengths gets tiring after a while.

John

It's hard to see what's going on, with all those horizontal lines in your .gif image.

Those are unreal/artifact slits in the power pours that let you sort of see how the layers are arranged. I think they're colorful. Actually, the gif resolution is just about good enough to figure out what we're doing.

Show us one of your boards.

John

OK, something simple, a pcb with an Analog Devices DDS to 16 outputs (via relays) for a Smithsonian Observatory gravity experiment. ftp://ftp.rowland.org/pub/hill/pics/RIS-564_pcb.gif This was designed before I discovered the appeal of mounting parts on both sides of the pcb.

Here is a piece of junk I made

I always have a piece of stip board with a zif socket for an 8051 around, this one still works after 10 years

martin

Hey, why is nobody using rounded traces anymore? I mean, CAD can do that automatically these days.

Interesting. I always like to see other peoples' styles and stuff. The Christmas Tree primary color scheme is a bit startling; we prefer tasteful pastels.

Are the RY things relays? We use "K".

We try to avoid putting a lot of parts on the bottom, but often it's unavoidable. Having all the bottom parts be the same value, ie bypasses, isn't too bad. We haven't done bga's on both sides, but I have seen it done. Shudder.

Hey, we just got one of the newish Stanford Research clock generators, the 2 GHz thing. It looks pretty good, and we got the prbs option to test our fanout boxes. What's really unusual is that their instrument manuals come with complete schematics and BOMs, and a component-level theory of operation. The schematics are nicely commented with design notes and explanations. You don't see that much any more.

John

Looks Victorian. Soooo last millenium.

John

But in high-G environments you have to ;-)

K for kontactor?

The best I've seen was V for transistors. That was really retro.

I've used some of their stuff in the Bay Area. What really surprised me was that their prices were not sky-high, in fact they were quite modest.

Ugly. This is nice:

Six-range Low-noise zero V drift STM preamp, Vbias control, both referenced to STM ground (strangely a feature not found in any commercial amp, which is why I designed this thing in the first place).

robert