I need some time to figure out what the question means.
Yet again, I don't understand the question. Which is E-B, and which B-C? What is shown is effectively pairs of opposing diodes, with different breakdown voltages. Which is which?
First thing I'm going to do is stretch the schematic to get rid of as many crossing lines as possible, separating the rows and columns, so as to see what belongs where.
Are the 10e20 resistors there to make Spice behave? I hope they are ;-)
Gimme time, and I'll get back to you.
--
"For a successful technology, reality must take precedence
over public relations, for nature cannot be fooled."
(Richard Feynman)
"Equipment Smarter Than Operator" ;-) ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Puts those nodes into the "select list". Doesn't automatically display, as in PSpice. ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
U0101, U0102, U0104, U0105, U0201, U0202, U0204, U0205, U0401, U0402, U0404, U0405, U0501, U0502, U0504, and U0505 all draw about +44pA when I1 is set to 2.0nA. Likewise, U0103, U0203, U0403, U0503, U0301, U0302, U0304, and U0305 all draw about -177pA with the same drive. The net total "bypassing" of U0303 is -712pA and the current thru U0303 is about -1.3nA; adding those together gives about -2.0nA; the total I1 drive SO all is accounted for. Isn't it nice that 1+1=2?
So, pick one of the parts in the +44pA set; what is the current path? Likewise, pick one of the parts in the -177pA set; what is the current path?
See the subckt model which is 2 diodes back to back clearly labeled and specified as to voltage. An equivalent might be a transistor with a floating base (node 3 in the subcircuit).
Yes; sorry about that, i had no idea that the Codatron(TM) part schematic would be so large; that was added last. And the "editing" capabilities leave a lot to be desired for separating things.
Yep; i put them in first, then i put in the "grid" lines.
I've done a bit of work, redrawn a symbol that's more compact, and redrawn the matrix to reflect matrix numbering conventions - (X row column).
To make it simpler, I ditched the current source in favor of a voltage source with series R, then did an op point analysis, schematic attached.
There are sneak circuits from row 3 and column 3, apparently due to "unused" elements leaking, but not enough to worry about at this stage Worst case 1.7nA, whilst we have 500 microamps in the DUT. 430pA in more remote devices.
I think it's time to make a better device model, incorporating temp coefficients, and real leakage figures. The Berkeley Spice diode models probably don't adequately reflect the real thing.
*codatron2.asy - goes in /lib/sym/special functions. Version 4 SymbolType CELL LINE Normal -8 18 22 -2 LINE Normal 22 -2 23 30 LINE Normal -8 18 23 30 LINE Normal 0 0 6 8 LINE Normal 25 51 42 36 LINE Normal 25 51 25 51 LINE Normal 23 30 25 51 LINE Normal 42 36 23 30 LINE Normal 49 64 34 43 WINDOW 0 24 -14 Left 2 SYMATTR Prefix X SYMATTR Description Proprietary shunt regulator SYMATTR Value Codatron PIN 0 0 NONE 0 PINATTR PinName + PINATTR SpiceOrder 1 PIN 48 64 NONE 0 PINATTR PinName - PINATTR SpiceOrder 2
********************
Codatron.ASY: Version 4 SymbolType CELL LINE Normal 0 -16 0 0 LINE Normal 0 0 0 0 LINE Normal 0 0 0 0 LINE Normal 12 -5 0 0 LINE Normal 12 4 12 -5 LINE Normal 0 0 12 4 LINE Normal 20 0 0 0 LINE Normal 17 12 22 16 LINE Normal 17 16 22 16 LINE Normal 17 20 22 16 LINE Normal 17 12 17 20 LINE Normal 18 8 20 0 LINE Normal 17 16 8 16 LINE Normal 32 16 22 16 LINE Normal 8 8 18 8 LINE Normal 8 16 8 8 WINDOW 0 -34 0 Right 2 WINDOW 3 2 -29 Right 2 SYMATTR Value Codatron SYMATTR Prefix X SYMATTR Description Bipolar NPN transistor, open base PIN 32 16 NONE 0 PINATTR PinName Cathode PINATTR SpiceOrder 2 PIN 0 -16 NONE 8 PINATTR PinName Anode PINATTR SpiceOrder 1
** Prelim analysis: Follow a "typical" path, line 103 ("ground), up thru U0203, left (line 202) then thru U0201, down the 101 line to and left thru U0301, up and to the right (line 203) to I1 source. Three other paths from U0203 are via U0202, U0204 and U0205. Those four currents eventually get combined online 203. Likewise, there are four currents via U0101, U0201, U0401 and U0501. There are 16 devices that have about 44pA going "backwards" subtracting from 8 devices that have about 177pA going "forwards" adding to the major (desired) current thru U0303.
That is as far as i am now; not too bad preliminary. The fun starts: 1) ways to decrease or prevent most if not all of those "sneak" current flows, and 2) see if there is a simple non-switching way to do that, especially at all temperatures up to 204C.
In view of the fact that, whichever device is chosen to test, the results are the same (try it), does it matter about currents in the offline devices that are better than four orders of magnitude below the DUT current? Your test point voltage will be the same for all devices, except bad ones. I'd say the method is proved.
Why the current source, rather than a voltage source? Voltage sources are easier to implement.
The 10e20 resistors aren't necessary. the simulation will run without.
an .op analysis will show where all the currents are, and their relative magnitudes in one stroke.
Are you sure that the device model accurately represents the behavior of a real device? It's relying heavily on the default parameters of the Berkeley Spice diode model (there are two possible diode device models in LTSpice), see the manual.
If you want to do temperature sweeps, you need some temperature parameters in the device model, they are set at zero, unless you specify.
Here's what I get from my .op analysis, using a 4000V 300K source, the DUT is U33 (d:u33:z1 and d:u33:z2).
Yep; I knw that it does not matter which device to choose, all of the other paths ("on the line" and "off the line") act the same.
Ahh...that is the problem! What if there is a bad device - it can give an out-of spec reading for good ones - resulting in a lot of down time especially if testing is done at (say) 185C. The idea is to have something "close" to mass production / testing due to large quantities, and at minimal cost (hence the matrix idea).
This is a zener and it is impossible to test one at specified currents if a voltage source is used; the zener voltage can be programmed (hence the pun in the device tradename) from 50V to 2500V and was designed as a solid-state replacement to the Victoreen Corotron (a gas-filled glass envelope shunt regulator nominally used with photomultipliers). See our datasheet:
Not too bad; but 4KV with 300K limiting can give 13mA short circuit current; also note * that d:u33:z1 is incorrect; i flagged d:u33:z2 as being reasonable.
Try it, I did, one bad device has no perceptible effect on input current, unless it's the DUT itself, short circuit, open circuit, bad parameters, it doesn't matter.
I figured that.
WHAT??? I do it with a curve tracer all the time.
the zener voltage can be programmed (hence
I thought it might be shaped like a fish, to go downhole easily ;-)
I've read your literature, and prov. pat. app.
I *still* think the device model is inadequate, Read the manual section on diode models.
Better still, do some measurements on a real device, plot some curves, and make a behavioral model that reflects those measurements.
--
"For a successful technology, reality must take precedence
over public relations, for nature cannot be fooled."
(Richard Feynman)
For an array 10x14 the sneak currents are huge at 204C and problems with bad devices can create an undesirable reading at the DUT. Problem devices can 1) be effectively shorted (bad results), 2) be very intermittent or arc (again bad), or 3) be effectively open (OK).
The reference to using a curve tracer is, in a sense, not relevant. One slowly increases the applied voltage to the Izt or specified current point and reads the voltage. In effect, this is what i do by forcing a known current and reading the voltage; works perfectly with devices programmed (so far) from 400V to 2000V. I test at 20uA, 25uA, 30uA, 100uA, 200uA and 500uA; accuracy and repeatability seems to be in the region of 0.1% Typical test temperatures are 20C, 100C and 185C, but -70C, -30C, 0C,
25C, 55C (for the zero TC units that Sandia wanted) and 200C, 204C, 210C have also been used on occasion.
For this preliminary "matrix" testing scheme it did not matter much. Found too much interaction, so am going "back" to "each on its own" testing, reducing the PCB array from 10Hx14V to 9Hx11V. That means 99 red Teflon wires plus one black Teflon wire from inside the oven going thru / past the door to the outside world where there will be a connector array for the test plug. ((anyone want the "excess" black wire?))
The datasheet shows the curve of real devices. The knee is rather sharp even in the nanoamp region, and there is no negative resistance or oscillations or other zener-like funny-business all the way up to beyond the max spec pulse current rating.
Do not know how to make a model based on (the) data.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.