morning

Yup

@Asmyldof thank you! that certainly helps

@Asmyldof Have you ever heard of short answers? :-p

I have

They bore me

It really needs more extension as well, though

Very limited amount of knowledge transferred

@Christoph one thing I forgot to add (might edit later) is the need to know what you're doing with regards to power conversion. Never look at "This converter will do 98%", but look at graphs. For example a 500mA regulator might be 94% efficient, but if it spends 99% of its time regulating a 0.5mA uC, you can bet it'll not be much more than 50%

If you'd need to post-regulate the uC, you'd be better of having a dual-mode one, or just two separate ones, or even just LDO'ing the uC all together.

There are regulators with two operating modes, usually PWM and PFM, that have a peak at 100mA and 1mA, or 500mA and 5mA, etc, they may stay efficient enough in a lower power domain.

But, if you have solar power a plenty, that won't matter that much at that end, i.e. regulating the sun just for the uC, that's allowed to be inefficient, because then you're only using 1% of available power anyway

@Asmyldof yes, I'm aware of the efficiency issues. By post-regulating the uC you mean inserting a 3.3V reg between 4.25V and uC? I will also have to protect the uC's inputs from overvoltage, but zeners will probably do that job well enough. I did a quick search for protection ICs, but they either don't seem to fit into the scheme in your schematic or simply don't have the right overvoltage level. I might also be misinterpreting the datsheets.

The connection from GPIO to M2's gate will also need some more protection if the GPIO is not 5V tolerant and has a low V_GS,th

@Christoph Correct. The extra diode in my drawing, D1 I believe, already protects the uC

Just connect it to your uC VCC

It will not allow that node to go above uC VCC + diode. With a BAT54 and <1mA that'll be 0.15V

most cases

If your uC drains only 0.1mA for long times, you may need to reconsider though

Since then you'll be upping the VCC that way

With post-regulation I mean any kind of regulation before the uC

@Asmyldof The µC will draw less than 0.1 mA for long times (I hope), but hy would I have to reconsider the diode (?) in that case? The extra V_diode will be even lower, so VCC + diode is closer to VCC. How is that upping my VCC then?

@Christoph The protection diode, when the solar cell is at maximum as drawn, will dump up to 1mA into your VCC from the solar panel. So if your uC only uses 0.1mA and nothing else adds about 1mA to that, then your VCC will charge up slowly

Ah that...of course! If I have a post-reg, though, I could use a zener to protect the GPIO from too high M2 gate voltage

@Christoph only if you then have sufficient triggering headroom left on the most

Else the best way is just chuck in a SOT323 NMOST as well

tiny little grain of sugar that'll fix everything

@Asmyldof yes that would work as well. Regarding the "Solar Buck": it's now fixed to 4.25V and the cell protection IC has to limit the current to some pre-defined, fixed value. It acts as an LDO while charging. Correct?

that probably needs clarification: "It acts as an LDO" should be "The protection IC acts as an LDO"

teardown of a 100ghz scope drool

@Christoph no, not really. The protection IC just cuts off. Those little boards you can order have all the circuitry to act as a voltage and current fuse.

The buck has to be current limited to 75% of battery capacity, so that it drops its output voltage depending on the cell voltage.

sort of a CC CV buck

@PlasmaHH I guess after lunch I know what I'm youtubing

because the fortune cookie will tell you?

@Asmyldof When I build a CC/CV solar buck, I can also make it controllable from the uC and have full control over what's happening, within the limits of the battery protection IC. How such a regulator should be built is another problem, but I can make that a separate question (which is probably more relevant to others). I've done some simulations for such a regulator over the weekend, without great success, but somehow promising

the BQ24xxx parts from TI do look good for battery protection and charging

@Christoph TI makes a lot of chips for various power applications. CV/CC buck modules exist as well, SMT plastic molded packages, but they will probably be expensive

Lunch now :-)

enjoy

20 100ghz channels... wtf are peopling doing with these things...

Eh

Parametrisation, characterisation, etc

@jippie 40Wp, Poly, incl shipment and import etc, should be about € 90,-. I'll let you know if I go ahead with the order, but would you like me to consider an extra unit?

but with 20 channels...

I work for a chip manufacturer that wanted me to design a modular system to build up to 1024 full quadrant sourcemetering channels in one set-up for their waver prober and lab setups... so 20 is a tiny amount

that would many orders of magnitude lower bandwidth for a smu

but 20 channels with that bandwidth... it is not as if you could just use a bed of nails for whatever stuff you do and run through that... you need these 1mm coax stuff and maybe waveguides and whatnot...

Yes, but not everybody makes chips that can be DC tested....

I am not saying everybody should have one. I'm just saying that a market of 1000's of units definitely exists

1000's of units = business case

for your smu? or a 20 channel 100ghz scope?

For mu 16ch SMU a market of 100's of 1000's exists, but they're not selling them

my*

If you need to measure all functionality on a 5GHz chip, you will want at least 10 ~ 12 channels that do up to 20GHz to see deformations. If you have a chiplab building them, that's 4 test setups

Shit adds up

20g and 100g are quite a significant distance

Building a waver-prober that is matched, clean and operated only in a noise free environment is much cheaper than 12 separate 50GHz devices that can't even relate to each other

20G and 100G isn't even an order of magnitude... If I can think of a reason for 20GHz of the top of my head with only a factor 4 budget, I'm sure there's people making things with 20 probe points that need at least 60GHz for good pictures

If that's a mainline production as well, that's easily 20 units sold

If you have to buy a reactor of € 2mln, a twinscan of € 8 mln, buying a cable bundle and waver probe of € 1mln to be able to use your € 0.5mln 20channel scope is peanuts

in that range orders of magnitude dont count much... no one in the industry is currently able to have adcs that have a higher bandwidth than 36ghz

20 channels t 100gzh would be more like 20mio

Also fine

dont know if you get a discount after a certain number of channels...

Anything you have to buy once in such a process can cost up to € 50mln if it's proven tech

define proven

Someone shows that a sine at max freqency can be rendered correctly on the screen

That's proven enough

hehe

It's also why spinning up new shit is so goddamn expensive...

fun thing is that if you had that 20 channel ting, you could also run it in a 60 channel 36ghz mode...

One the one side there's people going "Oh, we'll make billions of chips later, so buying a machine that does 2mln of work can easily cost 20mln at ASML" and on the other side people get invoices of 250k for their test batch

On the one side*

measuring 4ps rise time waveforms, yay...

Is that photon speed yet?

light travels 1.2mm in 4ps ... a bit more than across the connector they use

Hm

Would have been a very very flat cathode ray then

+tube

:-D

hm, its red, it must have to do something with fire alarms

Decided to leave the tweet with the shit pictures after all

It's one of the many incarnations of a stack-up of SMUs

one of the many many things missing in my home office ... at lest even a single channel smu... ^^

Continued working on the firmware as employee and now had to make a box for localised testing rigs with a special control board equipped for directer communication with PC

I got to keep the board I built by hand

First ever unit

Other than that I have some tools that can add up to an SMU of pretty good accuracy, but after pre-calibrating the calibration cell to 0.005% value, the card I got at home now is the most accurate device I have for the purpose

@Asmyldof Apart from the approximate size we were discussing, do you have a more detailed size indication perhaps?

@Asmyldof what does it do?

@jippie chat.stackexchange.com/transcript/message/22587718#22587718

Poly should be within 710mm x 530mm x 30mm with 4,6 kg

@Asmyldof what is a stack-up of SMU's?

Cost estimation can go up by 5% or down by 15% depending on which country it first lands in after departure in China

Source Meter Units

4 quadrant force and measure system

16 = number of channels per board

@Asmyldof /me shrugs

4 = number of boards

+current +voltage = most well known quadrant

@Asmyldof I have an appointment in a few minutes, I'll check this evening.

-current - voltage = second most well known quadrant

but the other two, +voltage -current and -voltage +current are interesting for some companies as well

@Asmyldof solar panel manufacturers :-p

doing what exactly?

Enojy

@jippie and laser-chip manufacturers

hm, I wonder if there are any small 12V power supplies which you screw into an E14 socket

Used to be

12V 80mA old-school transformer

Just as E14 sockets

All got outlawed

Still have some sockets

outlawed? hm...

I need some power inside my fridge...

Yeah, something with dead eople

people*

:-D the office ordered Ice-cream!

diverting delivery to here...

too late

dang

reviving time machine project

:-P

No electronics inside the fridge?

the light bulb ^^

It's not that hard to smash an old E14 and solder&glue a wire safely onto its base

Add a wire with a EU-2-pin and buy a € 7 12V 1A supply and you can also add LED lights

sure, though for insurance reasons I prefer not to do things myself with mains

?

if anything goes up in flames, no one pays...

My house is still fully insured

doing ^^ would be the least of it

Also need a battery, though, since the light goes off when you close the door

lets talk further about it after you burned it down ;)

11 years and counting

try harder?

It also features a welding transformer, averagely 15liters in isopropanol, 10 liters of acetone and usually a lot of oil and gasoline-type products

so nothing to worry about

You are aware what fuses are for, right?

And correct working practise, and all that

have you ever seen the image of my gfci?

If my house goes up by something after the fuse, I'm insured unless they can prove wilful neglect, which they can't

I'm not very snoopy when it comes to online personas

If you have 20k rep, I read your wrong answer twice before I moan about it, that's about all

I tend to have not much luck with electronic devices from time to time...

and if something I tinkered with goes up in flames, and the find out, then no matter how good it was built, or if it even was the fault, the insurance won't pay

there was a funny little thing when I was younger in the apartment of my mom, where at the meter of their neighbours downstairs something arced over, melting the whole box, they had to pay because the tenant before my mom had run a wire through that box into the basement, which was not the fault, but was damaged during the incident.

Well, you see, there's 2 things:

1. ^^ has not happened to me in 30 years of prodding things with tools, starting with screw drivers, progressing to soldering irons around 5-6, and moving on to probe-pens at about 12

2. I have lawyers

:-)

I have lawyers too, but it doesnt work out if you dont have enough money to bribe the judge... "your cable, your fault, I have no idea bout electricity, but that was the only thing not within code the police says"

I have made fires, mind you, and exploded a few caps, but the fire was always intentional and only 2 of the exploded caps ever were by accident

It's not hard to get a friend at the KEMA to write a 3 page forensic report explaining why the clearly molten piece of copper I added myself at the point of origin of the fire was not the cause of, but caused by.

fires in the lab (and I stopped conuting how often I set stuff on fire or blew it up in the last two decades) are quite different to waking up in the middle of the night by the smoke alarm to find your tv on fire

Again, unless you balls up everything rigorously, even 0.2mm2 wire will blow a 16A mains fuse before it glows

@Asmyldof as I said, it depends o the judge. "I don't understand all of your reasonings from both sides, but since you both agree that it was not up to code, it is your fault"

You just need better friends

:-P

that say hello to the judge on the golf court? ^^

That work at whatever the organisation is that approves installations for mains use in Germany, is that also TuV?

but anyways, I am not particularily lucky with electronics stuff, let alone mains voltage... that gfci above, I have never met anyone who saw one failing in that way

TÜV checks the construction details, but as long as you are a certified electrician you can tack your sticket onto anything I guess ^^

Maybe I'm just careful, or maybe I just stock proper materials

Or maybe one causes the other

YEah, but then if a senior level test engineer at TuV writes whatever you did was up to spec, that'd be good gravy for a judge, right?

oh I am sure had this gfci be installed anywhere else, it would still function ^^

You might wanne stop poking things with sharpened iron sticks.

Could help

@Asmyldof it certainly increases the chances, but if the judge is grumpy and doesnt like you...

Nobody doesn't like me until the 2nd or 3rd week

No small claims judge will have that time for me

after that they hate you?

Almost universally

I seem to be able to speed that up to 2-3 minutes

Hmm

I can too sometimes

But it takes some effort

I think it's better if I do not answer this electronics.stackexchange.com/questions/178900/…

I stopped reading after the first sentence, boring, likely dup

My answer would start: Are you sure you should have installed Altium?

good old religions...

?

pcb design tool choice is mostly about religion

Ah right

Well, "my company" supports nearly anything anyone has installed.

autocad?

OrCAD, Altium and Eagle natively, so to speak, and all other with 1 week of half-pay buffer time

But I do like Altium for a variety of projects

There obviously are stupidities and flaws in there, but the amount of time you spend on learning to work with or around them is low by comparison

wasnt it them who setup a new online thing for cloud community sharing editing stuff? like github for circuits?

And once you do it's actually quite "easy" (again by comparison) to make a well thought through 12 layer board

CircuitMaker ?

I think that's the name

yeah that could beit

Haven't tried it yet. In public beta now

On the EEVBlog people were tearing it down, no end, so they decided to, of course keep it free after the 30 day initial period

I should get an account and publish pcbs that look great and useful with a tiny "do not use" note that will blow up

It was this: | | close to being dead before it was born

LOL

What you could do: Flood the libraries and projects with general names that are pointless

That'll potentially kill it still, because naming, wait for it.... is GLOBAL!

eh, hhuh, waddyamean?

You call your design "USB-Interface-Board" and nobody will ever be able to use that name

:-)

"For improved searchability and sharing"

oh, that is nice, where can I get an account again to upload pointless generic names?

Or as I put it: For imroved being pissed off the first time you want to use it

its like getting a gmail address ... ok, just wahc a random number at the end

Might be that because of people face-palming endlessly on the blogforum that they are already changing it

Well, they could, for example, make it so in the library everything gets appended to {username}:

But, they didn't think of that yet

So quite possible their master library is now fucked up for eternity

On account of public beta

everything should internally have a uid and people should be able to just change the name

Doesn't matter how they fix it, but project names and local components should be internal to a user, not global.

They may have fixed the components thing, but that's also fun to fuck up

Just start "drawing components" with popular type names and add them to the global directory... "oops, did I short all the pins?"

oops, xchanged data and address?

They're really inviting a whirlpool of sloppy diarrhoea in their system as it stands now.

well, whatever it is, you will always find people that are into it...

South Park taught me that in the case of whirlpools of shit it'll be Germans....

>.<

dont forget the change in capacity due to the flux

:-D

I love it when people stalk me

well "stalk"

Real stalking is going to get people physically hurt at some point

just wait a whiel

frown

Are you going to send lady folk after me?

Rule of large numbers, at least one will be interesting

Soooo... you have my address, I have cops and lawyers... I see an experiment

:-)

Just had to tweet the paste question

I never got the hang of this twitter thing

I only do on and off

one bit tweets?

Can't

Even if I tweet 0 and 1 I'm tweeting bytes

Let alone packaging and pointers and code cleanup

hmpf, why is no one giving away some bigger solar panels for almost free... want to experiment a bit, dont want to spend any money ;)

Is €90 for 40Wp almost free?

more or less the exact opposite...

Not exact....

Local stores/importers charge € 190 per

for a 40Wat panel?

Which is why, regardless of insane shipping and import, I asked my dedicated supplier of mini-cells to find me some larger ones

Reliable ones, yes

Anyway. Bye

@Asmyldof They can fix it by crowdsourcing: Anybody who wants to can edit your designs.

;)

@Asmyldof hmm not sure if I agree with you, unless you haven't shared specific properties of the panel. 40Wp, slightly different sizes. First duckduckgo hit gave me a 75€ including shipping. Local webshop.

is it just me, but are a lot of questions from people using FPGAs in applications where a microcontroller would suit them just fine?

@whatsisname Maybe just people learning how to use FPGAs and not ready to use them on a problem big enough to need an FPGA yet

how dare they?

alo?

anybody out there~

'eyo

@jippie I admit. I made the mistake of never searching hobby stuff. My searches for weather resistant mounting with low-iron safety glass. (i.e. 40Wp is what it delivers with 48W of cells, in stead of it has 40W in cells and on a good day does 30W -- I don't know what the hobby bobbies have, though). You guys better off buying hobby bobby € 60,- then

I think my customer prefers if a bird can't destroy it, on account of 20year run time.

@crasic Hey

About bed time, tbh

Just finished washing the motocyclette

@Asmyldof I'm thinking of doing something drastic with eagle component management

I pumped everything as xml snippets into a sql database, need to write a small utility to generate "per-project" libraries instead of the mess of system-wide and local libraries

Also, we can support different versions and equivalent export options for altium and kicad