<tnt>
I used them a couple of times, always great results and very reasonable prices.
<azonenberg>
Yes, i was considering using them for some of my upcoming projects
<azonenberg>
but havent actually done any business with them yet so cant comment on results
<tnt>
What's really nice is they also do things like pressfit stand-offs and other stuff like that.
<Degi>
Oh nice, how are the prices like?
<tnt>
I paid like ~ 50 eur for some custom front/back panels (qty 1) for a classic hammond instrument case with a few holes. ( like 5 spdif tx/rx + bevelled screw holes on one, and 3 stand-offs and some led holes + screw holes on the other ).
<azonenberg>
The other thing I want to explore long term is building a resistive probe that has a 50 ohm termination to Vddq at/near the tip
<azonenberg>
behind the attenuator
<azonenberg>
The use case is to avoid shifting the DC bias on a DDR RAM signal
<azonenberg>
actually i'd be better off with a 500 ohm to vddq at the tip i think?
<azonenberg>
anyway yeah, basically i want to counteract the 500 ohm loading to ground with an equal pull to vddq
<azonenberg>
the thinking being that current sunk from the line is less of a big deal than shifting the DC offset on a SSTL signal
<azonenberg>
I dont know if it will be necessary but i think i should at least spend some time simulationg
<someone-else>
split termination with 2 100ohm resistors could also work
<someone-else>
might even be easier than providing Vddq source with freq response flat to 10ghz
<someone-else>
probe can then even have a Vddq input to connect to the board's source
<someone-else>
I meant Vtt in some of the places above
<azonenberg>
that was my plan, vddq input from the dut
<azonenberg>
and its more complicated than that because the output of the probe is 50 ohm to the scope
<azonenberg>
so there's 50 ohms to ground behind the resistive attenuator stage
bvernoux has joined #scopehal
<azonenberg>
the goal is to avoid introducing a dc bias into the dut pulling it away from vtt
<azonenberg>
so if you add a second current path it can't be to vtt, it has to be to vddq to produce a thevenin equivalent of a load to vtt after including the 50 ohms to ground at the scope
<someone-else>
yep, I meant split 2x 100ohm termination to gnd and vddq to produce vtt dc level
bvernoux has quit [Client Quit]
<azonenberg>
you cant have 100 to ground though
<azonenberg>
you already have 50
<azonenberg>
you can add parallel resistance to ground
<azonenberg>
but you can't go above 50 total
<someone-else>
why not?
<azonenberg>
because the scope has a 50 ohm termination in it
<azonenberg>
you can't remove that loading
<azonenberg>
You can however put the terminator in the middle of, rather than at the end of, your tip resistors
<someone-else>
attenuation factor would be 20:1 instead of 10:1 but otherwise it will work fine I think
<azonenberg>
Say 400 series then a 50 up to vddq, 50 series, then 50 to ground at the scope
<azonenberg>
100 up to vddq*
<someone-else>
ah, you also want to subtract the dc offset
<azonenberg>
Yes that is the whole point
<azonenberg>
i want the tip to be biased at vddq/2
<azonenberg>
this would be specifically for probing SSTL signals
<azonenberg>
without shifting the dc levels
<someone-else>
but that's possible without *subtracting* the vtt bias
<someone-else>
we can just pass it through to the scope
<someone-else>
ah, I made a mistake in my math - for this split termination would need to be to vddq*2
<someone-else>
or probe could have a single 50ohm resistor to vddq while scope will have 50ohm to gnd
<azonenberg>
Exactly my point
<azonenberg>
you cant have <50 to ground so you need 50 to vddq
<azonenberg>
Unless you are OK with changing the attenuation
<azonenberg>
in which case i think you can do some tricks in which you add the pullup before rather than after the attenuator
<azonenberg>
hypothetical but hard to actually build example: 500 to vddq at the tip
<azonenberg>
then 450 series, and 50 to ground at the scope
<azonenberg>
now you have 500 up and 500 down, so thevenin equivalent of a 250 to vtt
<someone-else>
sure
<someone-else>
20:1 version might have better SI though due to being source-terminated
<azonenberg>
more realistically you'd have a value between 500 and 50 somewhere in the resistor string
<azonenberg>
anyway, just an idea - the ground terminated version will come first
<someone-else>
I think adding a dac to make this adjustable could also make it useful for diff serial data probing
<someone-else>
to avoid upsetting common-mode voltage
<azonenberg>
That would make sense to do at the diff amp though. not in the tip
<azonenberg>
this immediate use case is for single ended passive probing