crasic

general recent conversations

Discussion on question by Heatconomic…

Imported from a comment discussion on law.stackexchange.com/qu...

crasic

Sep 15, 2021 22:00

Mostly I was reacting to the pitch of buying into a business, offering a nominally tangible good for sale, requiring exclusivity (upline), and pushing reselling (downline) of the franchise.

crasic

Sep 15, 2021 22:00

This sounds like, potentially, a MLM scheme.

Discussion on question by Twometer: I…

Imported from a comment discussion on electronics.stackexchang...

crasic

Dec 31, 2020 05:31

One can and does occasionally decode i2c by eye off a scope trace.

crasic

Dec 31, 2020 05:31

Besides the library error, what is different between transaction #10,000 which works and #10,001 which fails

crasic

Dec 31, 2020 05:31

If the hardware is behaving appropriately (as evidenced by appropriate i2c transactions) then the behavior may be coming from somewhere else, up to architecture errors that lead to race conditions, interrupt thrashing, and other "pure software" bugs that manifest as hardware errors. I'm trying to nudge you to divide the problem to improve the question. Outside of a bug or dead chip it may not be clear to any reader where the issue actually is and how this question might apply to them.

crasic

Dec 31, 2020 05:31

A generic library error is not sufficient information and may not even mean anything is wrong except your application code. Do you have ability to inspect the driver code? From the other direction - can you sniff the i2c channel with an oscilloscope to see what the transactions are on the wire? What is the expected behavior?

Discussion on question by crasic: Is …

Imported from a comment discussion on electronics.stackexchang...

crasic

Aug 1, 2020 23:15

@BruceAbbot the part was tested previously independently, and while I would suspect other radiators, the temperature dependence of the mean amplitude (scales as root temp) is confirmed and was surprising as it had not been previously observed, to your point we expected very little thermal noise with the cable being the biggest visible change . Could metal enclosures surrounding such a sensor radiate as sources?

crasic

Aug 1, 2020 23:15

@SpehroPefhany my previous understanding is that switched capacitor circuits will have Johnson-nyquist noise when the charge is transfered after switching, Would that that mean that when n analyzing noise in that application thathe ESR and other parasitic resistances are the primary variables? This is probably a better second question

crasic

Aug 1, 2020 23:15

@SpehroPefhany I am working on getting schematics, but the operating temperature is a titch high :) just unexpected noise given previous testing.

crasic

Aug 1, 2020 23:15

Reference Material Provided

crasic

Aug 1, 2020 23:15

@TimWescott certainly not out of the question that it is another source of noise but it scales correctly with temperature, I have updated the question

crasic

Aug 1, 2020 23:15

@SpehroPefhany Noise mean amplitude scales as the square of the temp, so yes, I mean Johnson-Nyquist. My understanding is that KT noise does appear across a capacitance, which is also Johnson-Nyquist to my understanding, but that may be incorrect.

crasic

Aug 1, 2020 23:15

Thanks folks, I am working on getting a reasonable reference setup :)

Discussion on answer by phoog: I'm a …

Imported from a comment discussion on law.stackexchange.com/qu...

crasic

Mar 19, 2020 22:27

A good reason to ensure compliance is that you will not be able to apply for many government jobs down the road, including volunteer positions at practically every level of government in the USA (city, county, state, federal) .

Discussion on question by user4574: H…

Imported from a comment discussion on electronics.stackexchang...

crasic

Nov 3, 2019 04:19

The ceramic chip cap situation is serious. Point in case when I was forced to use a broker/ reseller I had the caps xsection and SEM.

crasic

Nov 3, 2019 04:19

Anywho, if you find a home for this question I would love to write an answer, this is something I deal with as an EE at a smallish firm (so see more of the production horrorshos) all the time and have some more details to offer . I voted to reopen

crasic

Nov 3, 2019 04:19

If I bought this reel from you today, it would cost me more than $2900 of my time and equipment just to certify for production and that is assuming I had some confidence you treated the reel right. The only way I am entertaining that is if I am contractually obligated to use this part and you are the last holder of inventory on the planet or the other choice is Ali.

crasic

Nov 3, 2019 04:19

Honestly this is why we have our EMS manage component inventory so they can use it for other customers (they write PO to us when they want to) , always get volume pricing because they can split a order between clients, can certify the storage conditions, and have contacts for resellers if the part is not needed at all. A single reel from an unknown entity is basically worthless unless it's a super duper special part. Perhaps just list the thing on eBay for a reasonable price and see if it sells, otherwise just use it up and treat as a learning experience. Talk to your EMS/CM.

Discussion on answer by Ron Trunk: Wh…

Imported from a comment discussion on networkengineering.stack...

crasic

Jun 20, 2019 19:07

And since those features are defined in yet another spec , it is not "Ethernet". They are dictated by numerous other specs and customer requirements, since one cannot define the term specifically, we conclude it has very little real value except jargon. If I say my switch is "industrial" you will have no idea what to expect unless I provide a spec of what to expect. A system designer will need to rely on your extended spec and ensure compatability on their own , or fall back to minimum 802.3 even if your switch is fancy.

crasic

Jun 20, 2019 19:07

@Mike I think there is a mixup in terms and term creep, which usually points to a marketing culprit. There are many specifications and standards targeting ethernet applications in industry. For example, I help write the ethercat specification for my industry.That is an industrial grade ethernet. However, "Ethernet" is not generic, it is specifically defined by 802.3 and technically speaking makes few provisions for industry specific and hardness requirements.

crasic

Jun 20, 2019 19:07

I design industrial ethernet equipment, I'm saying the spec is out of scope for IEEE 802.3 that's it.

crasic

Jun 20, 2019 19:07

Often the industry will specify the hardness requirements and conformance testing. @MikePennington I'm not sure a TSN protocol extension to 802.3 qualifies as "industrial ethernet", even though it is a proposed path for industrial protocols. As far as I am aware the primary way 802.3 facilitates industrial applications on the physical side is through M12 connector spec.

Discussion on answer by Harper: Can I…

Imported from a comment discussion on diy.stackexchange.com/qu...

crasic

Jun 12, 2019 05:18

In my experience, working with MFG in that region of the world. The same folks building "legitimate" products are building these low cost knock-offs on an ad-hoc basis without understanding why their legitimate (Day job) client bothers with all that QA stuff. A cargo-cult of MFG if you will

Discussion on question by rwallace: D…

Imported from a comment discussion on retrocomputing.stackexch...

crasic

Apr 22, 2019 17:37

@BobJarvis Fixed point is used extensively when decimal precision and consistency of operations is preferred over a large dynamic range of number storage

Discussion on answer by John K: Why a…

Imported from a comment discussion on aviation.stackexchange.c...

crasic

Apr 5, 2019 12:37

As a side question, does this also mean the doors would not have rafts?

Discussion on answer by John Deters: …

Imported from a comment discussion on security.stackexchange.c...

crasic

Mar 8, 2019 09:36

@Patrick I think I understand the concept, but at some point you agree one leaves c++ land when there is physicaly manifest compiled machine object code which may be distributed and deployed on many computers and studied by security researchers,, at that point a divide by zero operation in the machine code will trigger a defined exception on certain platforms and at that point in a security context the behavior is defined by the exception handlers as per the thrust of this answer .

crasic

Mar 8, 2019 09:36

@MatthieuM. I only bring this up in this context because exception trapping , including divide by zero cpu exception, has been studied as an exploitable vector on arm systems. Specifically because the root exception vector table is mapped to a known fixed address specific to the architecture, even if kernel base address is randomized.

crasic

Mar 8, 2019 09:36

@MatthieuM. I think I understand your point, however, I think in the particular case of an ALU divide by zero operation, the behavior will be defined (or machine specific/runtime UB). I do see how compiler UB can mean that a statement that would nominally divide by zero may never produce the opcode with divide by 0 even if the input to the function is 0, due to reordering and assumptions about values. I think this is what you are describing?

crasic

Mar 8, 2019 09:36

@MatthieuM. "Undefined Behavior" is equivalent to "Machine and Architecture Specific", for example on arm a divide by zero ALU operation will trigger a well-defined cpu exception and jump immediately to an exception handler which the OS may or may not trap. This behavior can be exploited for other purposes, unsuported opcodes can be emulated by the os via the undefined operation exception, otherwise UB from a strict C++ perspective

General Network Engineering recommend…

Product Recommendations / discussion. This room has different...

crasic

Jan 18, 2019 16:08

@RonMaupin Sounds good, I generally want to ask "Is an industrial network with TSN implementable today" , and "What is the minimum required to implement/deploy TSN in terms of hardware and host requirement"

crasic

Jan 18, 2019 03:11

would a question on the current state of Time Sensitive Networking standards be on-topic?

Discussion on question by heather: Ho…

Imported from a comment discussion on raspberrypi.stackexchang...

crasic

Aug 7, 2018 22:51

but at the end of the day, what you are asking for is a hardware thing (clock) that you want to emulate in software. For this purpose it is better to use dedicated clock hardware, that can inexpensively be attached to the raspberry pi or any embedded controller.

crasic

Aug 7, 2018 22:50

the response is more intrusive software engineering. Starting with "hacks" to get around software layers and have your software sit close to the metal, this can include /dev/mem hacks and thread affinity tricks to minimize swapping and context switch. If that is not enough you progress to kernel drivers and baremetal to seek your performance

crasic

Aug 7, 2018 22:48

but this is unlikely to be improved, and depends on the system conditions

crasic

Aug 7, 2018 22:48

So what @joan is proposing, is to measure this worst case response using user space tools. If it is fast enough for your purposes, Great!

crasic

Aug 7, 2018 22:47

@heather for you benefit it seems you have a large amount of interests, but what you are asking is quite silly in this domain when you think about it. The OS is constantly juggling multiple concurrent processes, swapping their threads of execution off and on the processing core, loading/unloading data. This is a highly optimized illusion that is meant to present aninteractive environment to the user. The system is not designed with any degree of response time to absolute hardware events.

crasic

Aug 7, 2018 21:55

Related EE answer. The hardware limit, regardless of software appears to be on the order of 125MHZ with 1.2V GPIO, achieved by using the hardware clock output function of gpio, software toggling is unlikely to get close to that without baremetal implementation or customer kernel driver . For userspace libraries I would wait for answer from someone who knows the practical limits of the library, there is a lot of jitter involved with context switches.

The Bakery

Main chat room for raspberrypi.stackexchange.com +++ Visit Piv...

crasic

Feb 7, 2018 05:06

@uhoh Went to chat and remembered I started to write an answer!

crasic

Dec 29, 2017 04:00

IoT core is hoping for the market of higher level integrators using off the shelf hardware and embedded OS solutions. Allowing these "IoT" companies, selling high brow consumer products to do so without having embedded domain experts on staff

crasic

Dec 29, 2017 03:58

IoT Core is basically Microsoft getting some skin in the game, in the long run it is very expensive to retain firmware and electrical engineers for every small embedded device.

Discussion between Chetan Bhargava an…

Imported from a comment discussion on raspberrypi.stackexchang...

crasic

Jan 12, 2018 04:15

crasic

Jan 12, 2018 04:15

In any case, to address your actual argument. I disagree for a variety of reasons, the RPI has many uses and users, and prescribing a single use type is shooting yourself in the foot with regards to the future of the platform. For the freelance designer nothing sells for more than Industrial IO tools, it makes the market that sustains hobbyists. One cannot survive long without the other. When the cooperation exists you get inexpensive and extremely robust implementations like:

crasic

Jan 12, 2018 04:05

@ChetanBhargava There clearly was an issue or @Ghanima would not have felt the need to intervene in this chat. Instead of deflecting it would be better to aknowledge and move on. In either case your assumption is wrong, I withdrew the down vote after the system allowed me to following your edit. I hope that you can see some positive and still contribute your expertise to this community. I do not believe your prescriptive attitude is warranted or wise, but my reaction was also innapropriate.

crasic

Jan 11, 2018 17:38

@ChetanBhargava I apologize for any offense I have made, and would like to move on. We can certainly continue to disagree on the nature of the forum more productively. This is, after all, an active living community.

crasic

Jan 11, 2018 17:37

@Ghanima I have attempted to disengage after the initial flame and move the issue to a more productive direction via a meta post, I did not delete my original comments out of transparency, but they were moved to this chat. I hope that we can all just let this go and move on because the badgering is becoming tiresome 3 days in a row.

crasic

Jan 10, 2018 04:00

@ChetanBhargava no need, I opened a question on meta in order to find community consensus on if focusing "DIY" or hobby perspective is actually desired by the community

crasic

Jan 10, 2018 04:00

Lol, I'm sorry you got triggered with my down vote on your answer, but you simply do not know what you are talking about, Being an amateur is no excuse for ignorance :)

crasic

Jan 10, 2018 04:00

What is a PID Controller? Did you read even my answer? A raspberry pi can be a PLC - for example . Being an amateur is no excuse for being ignorant, There are plenty of RPI projects mounted on DIN rails and even specific products to do attach the PI. It is a very clean way to mount components, sensors, and controllers, and simply because you haven't seen it doesn't mean it isn't appropriate for RPI.

crasic

Jan 10, 2018 04:00

Dismissing it as only for "Industrial Infrastructure Architect", is doing yourself, OP, and this stackexchange a huge dis-service. This is how you do such things, and I was attempting to share how it is done "the right way". My approach at RPI.SE is to always provide the correct way of doing something, even if it is beyond what OP is asking for.

crasic

Jan 10, 2018 04:00

@ChetanBhargava What does PID have to do with anything? I will push back on your assertion, this is a rudimentary answer to OP's direct question - how to control a room full of power with a raspberry PI. . If you are talking 8KW of power, aka 40A, using an SSR or Relay directly is asking for trouble, if OP is serious about doing this then there is no other option. Most people also are not attempting to switch 40A load electronically, but those of us who do, have PLC's running at home.