Differences

This shows you the differences between two versions of the page.

--- mcmaster:start [2018/12/24 06:50] – mcmaster
+++ mcmaster:start [2025/08/28 20:26] (current) – mcmaster
@@ Line 4: / Line 4: @@
 All of my images are released under CC BY unless otherwise noted.
+Enjoy this work? Consider [[https://www.patreon.com/mtvre|supporting me on Patreon!]]
 [[mcmaster:wafer|Wafers]]
 Images:
-{{topic>collection_jm}}
 {{topic>collection_mcmaster}}
@@ Line 16: / Line 17: @@
 Relevant equipment list:
-  * pr0nscope: CNC microscope.  Typically BF imaging, but also have DF, polarized, DIC
+  * prawnscope: CNC microscope.  Typically BF imaging, but also have DF, polarized, DIC
   * k2scope: K2 IND CNC confocal microscope
   * brainscope: laser microscope / probe station.  532 nm, 355 nm Nd:YAG
@@ Line 33: / Line 34: @@
 ====== Commissioning ======
+UPDATE: as of 2025 (and really earlier) I no longer accept commissions. However, if you join the siliconprawn discord there is a channel to get some help from others. Email me for a join link
 I got a lot of requests to image stuff, so here's my official policy.
-In short, I focus on chips that are of interest to me.  I do lots of chips for others, but I give priority to people that are able to provide funds.
+TLDR: for a typical chip, I suggest $25-200 donation to get it decapped and imaged for public use. Delayering is extra
+  * e-mail me (hobby/public): JohnDMcMaster @@@ gmail.com
+  * e-mail me (consulting): john @@@ mcmaster.tech Note consulting rates are not the same as hobby rates
+  * [[https://twitter.com/johndmcmaster|DM me on Twitter]]
-In terms of consumables, equipment wear etc, each simple chip costs me about $20 to decap/image (not counting huge equipment NRE).  As of 2014, a professional shop charges around $100 for a simple, low resolution top metal dceap/image.  People give me upwards of $150 per simple chip if they are feeling generous and want to support other projects.  Simple => not implant ROM and preferably does not require delayering.  Not sure what a shop would charge for delayering, but lets say upwards of $100 donation for delayering.  Chips from the 80's should be simple.
+Please provide the following information for each chip:
+  * Can you provide any funding for the work?
+    * Commercial or personal interest?
+    * Decapping involves harsh chemicals, direct expenses, high NRE, and time so contributions are greatly appreciated
+  * Any potential legal issues associated with the chip?
+    * If so, it will be evaluated on a case by case basis
+    * Ex: DRM, crypto can be problematic
+    * Ex: a typical sound chip or microcontroller is not an issue
+    * I will not work on: satellite, cable TV
+  * Clearly identify the chips, especially if you are sending more than one
+    * Ex: specify "344S0061-A" instead of"SWIM", since that's what's printed on it
+  * What deliverables are you looking for?
+    * I assume top metal image at 20x by default
+    * Let me know if you are looking for other services like ROM extraction or delayering
+  * Replaceability/value
+    * Occasionally tings go wrong. How much would it cost to get a second one?
+  * Any previous knowledge? Maybe things people tried that didn't work?
+  * Do you have a particular timeline in mind?
+    * Coordinate with me for best results
+  * By default I retain copyright and publish work CC BY on siliconprawn.org
-Delayering is typically more risky to the chip and myself (toxic chemicals like hydrofluoric acid).  Similar issues with staining and lapping.  Lapping is currently extremely labor intensive and not very well refined (although I can get a small area easily).
+In short, I focus on chips that are of interest to me.  I do lots of chips for others, but I give priority to people that are able to provide funds. This section is written with open source projects as the main audience, but note I also offer services through my embedded security consulting service.
-This is a hobby and I'm not interested in operating as a business.  I'd really like to focus on chips that I can (eventually) release on pr0n, ideally under CC licenses.
+In terms of consumables, equipment wear etc, each simple chip costs me about $25 to decap/image (not counting huge equipment NRE). As of 2014, a professional shop charges around $100 for a simple, low resolution top metal dceap/image.  People give me upwards of $200 per simple chip if they are feeling generous and want to support other projects.  Simple => not implant ROM and preferably does not require delayering.  Not sure what a shop would charge for delayering, but lets say generally 3x the cost of top metal imaging.  Chips from the 80's should be simple.
+Delayering is typically more risky to the chip and myself (toxic chemicals like hydrofluoric acid). Similar issues with staining and lapping.  Lapping is currently extremely labor intensive and not very well refined (although I can get a small area easily).
+I prefer to focus on chips that I can (eventually) release on siliconprawn.org, ideally under CC licenses.
 In general I want to focus on R&D / labwork and not hand digitization.  I do have some limited semi-automatic digitization capabilities for planarized chips though that I'd like to develop (ex: works on PS1 chipset).
-Most of this is targeted towards small chips up to the mid 90's.  If the die is obnoxiously large (ex: 486 CPU) it will be difficult for me to photograph and stitch the entire thing.  Similarly I unfortunately do not have EM capabilities to image fine processes.  180 nm is doable on my equipment but really pushing it (ex: XC2C32A).
+Most of this is targeted towards small chips up to the mid 90's.  If the die is obnoxiously large (ex: 486 CPU) it will be difficult for me to photograph and stitch the entire thing.  Similarly most of my equipment is targeted towards larger processes (say 180+nm), although I do have a small SEM if required. See [[https://siliconprawn.org/map/xilinx/xc2c32a/mcmaster_lap4-04_mit20x/|XC2C32A]] for an example 180 nm chip taken with a relatively low power objective, noting in theory I can get about 11x the detail (NA 0.42 => 1.4).
-Other: most of the work is in decapping, not actual imaging. If you happen to already have a bare die and would be satisfied with a quick / lower quality scan, feel free to send me a few dies. This may actually work out well for you since most of the money is in the microscope, not the decapping supplies.
+Other: most of the work is in decapping, not actual imaging. If you happen to already have a bare, clean die and would be satisfied with a quick / lower quality scan, feel free to send me a few dies. This also can work out well since most of the NRE cost is in the microscope, not the decapping supplies.
-Please provide the following information for each chip:
+Chips are non-returnable. If you need something back, please make an explicit agreement before sending.
-  * Required: markings on the chip
-    * Do not just send me the chip code name and expect me to know which one it corresponds to
-    * Ex: specify "344S0061-A" instead of"SWIM"
+====== Scales ======
-  * Optional: alternate names
-    * Ex: SWIM, Pentium 1 etc
+UPDATE: as of maybe 2024 I have so many microscopes this scheme has collapsed. TLDR its not possible to get the scale information by looking at the picture. Additionally there are infrastructure issues (ex: leaflet render glitch) that have discouraged other routes
-    * These are helpful to add to wiki pages
-  * Required: what you want
+mit20x vs mit20x2:
-    * Full images of every layer so you can digitize it?
+  * mit20x: BH2 microscope, MU800 camera, 0.5x relay lens, Mitutoyo M Plan Apo lens
-      * Do you need the whole die or just a part of it?
+  * mit20x2: BH2 microscope, USB3 camera, 0.75x relay lens?, Mitutoyo M Plan Apo lens
-    * Top metal image for aesthetic purposes?
+    * Select medium resolution 2736 x 1824
-      * Are you hoping for high resolution that can see every wire or just an overview shot?
+    * 20x
-    * ROM dump?
+      * 0.42 @ 800 nm => 952 nm R objective resolution
-      * Suggested method?
+      * 0.42 @ 532 nm => 633 nm G objective resolution
-      * Image mask ROM only?
+      * 0.42 @ 400 nm => 476 nm B objective resolution
-    * Erase fuses / laser magic?
+      * About 820 um wide field of view
-  * Optional: datasheet is a plus, especially if I need to do something invasive
+      * 820 / 2736 = 300 nm / pix raw
-  * If applicable: any previous knowledge of things people tried that didn't work
+        * Cut image by half to sample RGB correctly => 600 nm / pix
-  * Required: replaceability/value
+      * 820 / 2736 / 0.5 = 599 nm / pix ultimate image resolution
-    * Sometimes I can do something more aggressive (faster, sometimes much faster) that might have higher chance of damage
+    * vc60x => 1.4 NA
-    * If they are readily available you should send me several just in case
+      * 1.4 @ 400 nm => 143 nm objective resolution
-      * Especially if you want me to do invasive analysis (ex: delayer)
+      * About 820 / (60/20) => 273 um FOV
-    * TODO: come up with a scale, but here are some rough figures
+      * 273 um / 2736 pix / 0.5 pix/pix => 200 nm / pix ultimate image resolution
-      * Very common: < $5 or can easily get for free
+    * Conclusion: possibly losing slight resolution in blue, but probably good enough
-        * Ex: XC2C32A
-      * Common: can get more if you shell out mild cash
+TODO: link github calibration files
-        * Ex: playstation 1 CPU
-      * Rare: You can probably get more but it might take a while or cost >$100
+^ Scope      ^ Objective  ^ NA  ^ Resolving power (nm)  ^ Camera  ^ nm / pix  ^ Notes               ^
-        * Ex: playstation 4 CPU
+| prawnscope  | mit2x      |     | 5000                  | MU800   | 3550      |                     |
-      * Very rare: $1000 or un-replaceable
+| prawnscope  | mit5x      |     | 2000                  | MU800   | 1420      |                     |
-        * Ex: CSG 4510 R3 (the C65 CPU)
+| prawnscope  | mit10x     |     | 1000                  | MU800   | 710       |                     |
-  * Any other relevant information
+| prawnscope  | mit20x     |     | 700                   | MU800   | 355       |                     |
-  * Can you provide any funding for the work?
+|            | mit20x2    |     |                       | 20MP    |           | New camera setup    |
-    * Commercial or personal interest?
+| prawnscope  | ns50xu     |     |                       | MU800   | 142?      |                     |
-      * I'm not opposed to commercial interest so long as the die images remain free here
+| prawnscope  | nd50x      |     |                       | MU800   | 142?      |                     |
-    * Funded projects are given priority
+| prawnscope  | mit50xn    |     | 700                   | MU800   | 142       |                     |
-    * I am willing to do a lot of stuff for free, but it does involve harsh chemicals, costs me money, and time so anything is appreciated
+| prawnscope  | mit100x    |     | 400                   | MU800   | 71        | FIXME: correct NA?  |
-  * Any legal issues associated with the chip?
-    * Does release need to get restricted?
-      * DRM?  Crypto?
+====== Die archive ======
-    * Ex: a sound chip is no issue
-  * Do you have a particular timeline in mind?
-    * Between work and other requests don't be surprised if it takes me several months to get to it even if its only an hour of work
-    * Coordinate with me for best results
-    * I usually have the most time late Dec / early Jan
-  * By default I retain copyright and publish work CC BY on sipr0n
-    * If you need exclusive copyright, want to be included in CC license, etc make an explicit agreement with me
-Still reading and interested?  Send me an e-mail: JohnDMcMaster @@@ gmail.com
+Chips that were decapped for imaging, but stalled out for one reason or another
+Includes
+  * Konami 054539
+    * Embargoed image data to be released 2022-12-13
+  * Konami 053246
+  * Konami 053247
 ====== References ======