Re: Preamp tests

[Gerard Visser <gvisser@indiana.edu>]

Hall D Electronics:

Hi Yves,
	I have here 20 capacitors, 330 pF 2.5 kV NP0-type ceramic in an 1808 
size surface mount package, Kemet P/N C1808C331JZGACTU. This (or 
equivalent from AVX or Johanson) is the type I would strongly recommend 
to use in any new design. It _may_ be possible to use X7R-type ceramic 
but it is usually not a good idea from a noise point of view.
	The above capacitors can be ordered from Digi-Key, stock number is 
399-4832-1-ND for cut tape packaging. They have stock. The price is 
rather high if you only buy a small quantity... So my mind was to try 
them out when I come visit you and then you can buy a large quantity on 
a reel, suitable for production of some new HV boards. But if you want 
to try some real soon, order some from DigiKey you can have them in 
~2days usually.
	The coupling capacitor being too small should (I expect) make 
_absolutely_ no difference to any noise pickup or oscillation problems. 
So, no need to change the coupling capacitor with a view to fixing that.
	On the other hand the bypass caps should also be larger, for these I 
have ordered and received 3.3 nF 3 kV X7R-type ceramic in an 1812 size 
surface mount package, Johanson P/N 302S43W332KV4E, DigiKey stock number 
709-1056-1-ND. Again this is probably what I would recommend to use in 
any new design. (But a larger value might be dictated by test results.)
	Surface mount high-voltage ceramic capacitors are much smaller for the 
same capacitance, primarily because leaded high voltage ceramic 
capacitors usually use a traditional single-layer construction, whereas 
the surface mount capacitors are usually multilayer. They are quite 
reliable and easy to work with when professionally reflow soldered and 
properly cleaned, but you do have to use extreme care when hand 
soldering. Certainly they cannot be hand soldered as a production method.
	About the signal connecting cable, I do believe still that coax is 
unnecessary _if_ the connections are short. At present your connections 
are not short. The coax shields should be AC grounded - but of course it 
needs to be a low impedance ground or else this will just be a source of 
massive crosstalk. [In particular I think just the 33pF bypass cap on 
that line is not low enough impedance, but you can try it. The line 
itself may be too skinny, we'll have to see. Either can be corrected by 
some hacks to the existing HV board design.]
	A 2-layer board should probably be just fine for the new HV coupling 
board design.
	I suggest as step 1.5 route the input signals directly away from the 
preamp output cable. This is probably best done my modifications to the 
way the preamp interposer board connects to your existing HV board. Also 
improve the ground connection from HV board to the endplate (it should 
be as short as the signal connections, both running together, and should 
be as short as possible, and preferably of somewhat low impedance at up 
to ~10MHz, i.e., use a copper strip, braid, or at least wire gauge 16 or 
less.)
	Thanks,

	Gerard

Yves Van Haarlem wrote:
> Hi Gerard,
> 
> I asked McCracken how we came to this design: the plan of CMU was to use
> coax cable to bring the signal from the sense wire to the HV board, the
> plan was to connect the cable shield to the HV board ground. There was a
> electronics meeting in Indiana where Paul Smith suggested to connect the
> cable shielding to the HV and not to the ground. Gerard opposed the idea
> of using coax cable. The decision on which capacitors to use came after a
> small calculation done at CMU.
> 
> I decided not to connect the cable shielding to ground when bringing the
> HV board outside the CDC end plenum because we saw no difference (using
> the old preAMP) with/without connection.
> 
> Plan:
>         step 1: I just started (together with Mike) to collect the
> material to connect the cable shielding to the HV line on the HV board.
> After that we will test this.
> 
>         step 2: Change the capacitors on the board with the ones Gerard
> proposes and test -> Gerard, what type/value do you propose; i remember
> you saying we should also using a different type.
> 
>         -- My hope is that this can be done in a short period of time and
> that we can take data afterwards --
> 
>         step 3: first iteration in redisigning the board (shrink it, new
> orientation preAMp, maybe two ground planes, ...)
> 
> 
> Cheers,
>         Yves