Electrical Engineering

7:50 AM

@PlasmaHH gaterosplating.co.uk/Copper-Plating-Kit.php

gaterosplating.co.uk/Replica-Chrome-Plating-Kit.php

caswellplating.com/electroplating-anodizing/…

@Asmyldof yeah, the more or less commercial copper plating all seems to be mostly sulphuric acid based. also, is the uk still in the eu or will it cost extra customs money? ^^

Asmyldof

8:24 AM

@PlasmaHH For now they are still in the VAT free zone for me

KyranF

9:42 AM

can anyone help me understand what this means: "Timebase Accuracy ±50 ppm measured over 1 ms interval" in an oscilloscope datasheet ( this one, page 4 )

PlasmaHH

well, what exactly is there you dont undertand? 50ppm should be clear?

KyranF

the BK Precision 2190D has 100Mhz bandwidth, 1G samples per sec, or 500Ms/s in dual channel mode (which is how i'm using it). Which time value should I be applying the 50ppm to?

how would this reflect in measurements in terms of jitter?

i'm trying to understand how much jitter I am seeing in a signal (time between two rising edges, one on each channel) can be due to oscilloscope timing jitter

Can I assume the sample clock is 1Ghz, and the +-50ppm is that sample clock jitter?

seems like 50ppm of 1Ghz is pretty good, so i guess it's not what i'm seeing.

PlasmaHH

I am slightly confused about the relevance to jitter, afaik such accuracy specifications only tell how much off it can be from its true frequency, not how much it jumps arund within these borders

like, your clock could run at perfect 999.99995MHz with no jitter at all

So if you measure 1ms between two edges, you could be 50ppm off in that measurement, but in theory it could have 0 phase noise and be always that same 50ppm off all the time

I think there is somewhere an eevblog video aobut a pll programming bug in some rigol scopes where dave shows how small after the fix the phase noise really is

KyranF

10:06 AM

well the jitter in question, that I'm trying to check if the scope has anything to do with it, is in the order of +-7 nanoseconds. The scope has a time division down to 2.5 nsec/div, and it appears the accuracy is good enough that a whole bunch of samples taken comparing the difference between two signals leading edges shouldn't show any jitter from the scope

PlasmaHH

Maybe watch the video(s) from dave there, the expected jitter in a moder scope is far less than one division, probably even far less than one pixel

If you have a low phase noise sine wave available you can of course always check yourself

KyranF

yeah

i'll watch the video. good old dave jones, fellow aussie

PlasmaHH

I mean, just think of it, who would buy a scope where the waveform is jumping around on the screen by design?

KyranF

it's a shit $400 scope?

i mean, I definitely wouldn't be worried about this with a $5k-10k+ scope

confidence in your own measurement equipment is an issue hehe

PlasmaHH

even on those that would be ridiculously unusable

KyranF

10:14 AM

well i'm not talking about visible jitter of the waveform on the screen, i'm talking about timing error between two waveforms in the nanoseconds, which have leading edges about 720 nanoseconds apart, on two different channels

so if you do start/stop sampling, have the static waveforms on the screen, and do some t1 - t2 deltas

PlasmaHH

in any price range I would expect that for the smallest timebase settings you can see a stable waveform in a almost stable way. a pixel or two, maybe even three wobbling would be fine, but not multiple divisions

KyranF

trying to establish what is 'real' jitter between the two signals

PlasmaHH

(something different on a 20GHz scope, there I would expect much more visible jitter)

brb, work calls

KyranF

if i had a continuous signal might be able to see if my scope is doing like that rigol in Dave's video. In this case it's two channels of discontinuous signals with large delays between them (720ns) and the +-7ns jitter between the two signals rising edges is the thing i'm trying to measure (and determine if the scope is adding to that or not). There is a lot of stuff between the two reference signals that could be contributing to the jitter that are reasonable enough to make the jitter...

like an LED turn-on/off time, photodiode reaction speeds, TIA and second-stage amplifier phase delays/jitter and finally the comparator on the output with the 6mv hysteresis.

PlasmaHH

10:39 AM

Get some clean signal of known good quality first, verify for your own well sleeping that you can't seeany jitter, then use the scope assuming there isn't any to measure your signal

KyranF

10:52 AM

yeah. Would measuring the same signal on both channels work? Wouldn't have the ~720 nanosecond difference between them but at least cross-channel jitter might become visible

PlasmaHH

they are fed from the same clock, there won't be any jitter differences between the channels.

KyranF

if they are sampled sequentially there could be though right?

i don't think they are sampled concurrently, that's why the 1G/s goes down to 500M/s because it has to do two conversions one after the other on different channels

it would be so low it wouldn't be visible anyway

Asmyldof

11:21 AM

@KyranF That's not jitter, that's total path skew, something completely different

KyranF

@Asmyldof would the skew not be constant?

Asmyldof

The skew in your scope and wires should be, until you get beyond ps

KyranF

right, so i shouldn't see any 'jitter'-like effects from that then

Asmyldof

To be honest, you should not see any scope based "jitter" between signals, because a clock should be highly stable, meaning even if there is jitter, it should not be +100% to -100% between edges

KyranF

indeed, that is the conclusion i've come to

Asmyldof

11:30 AM

Jitter almost never takes big steps within the band of error

TheNoonMoose

12:53 PM

@Asmyldof absolutely, happens all the time

Asmyldof

1:09 PM

@TheNoonMoose Let me know ;-)

TheNoonMoose

oh haha i just scrolled up and saw the part number didn't realize you were trying to sell not buy

Asmyldof

Most of my stock isn't officially deprecated yet, but I do get the impression quite a bit of it soon will, looking at Microchip's product pages

TheNoonMoose

but yeah our whole gig is making sure parts from randos are still god

*good

wow that's an uncommon part, eh?

Asmyldof

What is?

The 8585?

Didn't used to be

I like how this picture broadly ignores the fact either company introduced stuff much before the time the mention is on the opposite side

microchip.com/design-centers/8-bit

(the arrow, right click-catch on the image to always give an error, because Microchip seems to care too much about proprietariness as well... )

TheNoonMoose

ATmega8585?

Marla

1:17 PM

yeah, I googled @Asmyldof 8585 and got : o2xygen.com/photo/HI8586PST/HI8586PST_001.pdf

TheNoonMoose

maybe 8515

if it's the atmega8515 there's tons of stock out there

Asmyldof

3:26 PM

@TheNoonMoose @marla Right, I meant the 8515 and 8535 and such. I do have some stuff very likely to be deprecated from various vendors though

I should have a few MC68000 NIB

Nick Alexeev

3:41 PM

1

Q: SMD soldering: tiny components stick to tweezer point

When soldering SMD components, the 2-pinned ones (like resistors, capacitors, ...) are the hardest in my opinion. I started long time ago with the 1206 format. Then scaled down to 0804, 0603 and finally I'm working with the 0402 ones. I'm facing a new challenge right now. Once I get some flux on ...

Illustrations alone make it worth an upvote.

PlasmaHH

indeed

The Photon

5:26 PM

@KyranF My interpretation is if the scope shows you a rising edge 1 ms after the trigger event, then the actual time that edge happened is between 0.99995 ms and 1.00005 ms after the trigger event. I don't expect this spec to tell you anything about jitter, more about the "long"-term accuracy of the timebase.

Jitter should be a separate spec, and the errors it causes are additive on top of the timebase accuracy spec.

Asmyldof

6:17 PM

zZzZz

K.Mulier

Hi Asmyldof. Thank you for inviting me here

I just checked out your website. It looks like you're a freelance electronics designer, correct?

En u spreekt ook Nederlands!

:D

Asmyldof

@K.Mulier But many others here don't. Though @ThePhoton makes attempts

K.Mulier

:-)

I noticed you live in Netherlands, that makes you my neighbour. I live in Belgium (Leuven). In 3 weeks, I will also start as a freelancer. I'll build a lab too.

Asmyldof

You can see ^^ that a moderator's given you props

K.Mulier

props?

what's props?

Asmyldof

6:30 PM

glossophilia.org/?p=9198

K.Mulier

I still don't get it..

The moderator doesn't like me?

Asmyldof

14

Q: What's the etymology of "props"?

Props can mean compliment / respect / credit, for example: Erika gets props for the great work she did on the music. Wiktionary states that props is: (slang) proper respect or proper recognition for another person; an expression of approval or a special acknowledgment; accolade or accol...

etymology

K.Mulier

Oh, okay, it's a good thing

:D

Thanks :-)

Sorry for the slow understanding...

Asmyldof

I think @NickAlexeev was impressed with the effort invested in the images

K.Mulier

Thanks

Asmyldof

6:33 PM

Anyway, as mentioned in answers, stickiness is mostly a result of dirt.

K.Mulier

Yeah

Asmyldof

Most SMT components aren't magnetic or magnetically susceptible

K.Mulier

It's a big pain - the stickiness

Asmyldof

So it will not change between tweezers

K.Mulier

Hmm

Asmyldof

6:34 PM

Maybe special coatings will help a little, but they're not worth the money

K.Mulier

What do you think about the following product?

neverwet.com/commercial-industrial/index.php

Asmyldof

You're better off having 5 identical $10 ones than one "super coated" $50 one

K.Mulier

"Neverwet Oleophobic Oil repelling treatment"

It's a chemical that makes a material repell oil and water

Asmyldof

You intend to work with oil based flux then?

Haven't heard of any

K.Mulier

Well, to be honest, I have no idea of the chemical composition of my flux

Is it oil or water based in general?

Asmyldof

6:35 PM

Definitely not oil

Some are water based

Most non-liquid could not care less whether your tweezers repel water or not

K.Mulier

These products on www.neverwet.com look good to repel water (some also oil ánd water)

Asmyldof

Equate it to gum

K.Mulier

Hmm..

You think those water and/or oil repellants wouldn't work?

Asmyldof

Rosin based ones just stick to anything. They stick to PTFE and PEEK as well

K.Mulier

Hmm..

Asmyldof

6:37 PM

And non-rosin ones probably eat away at fansypantsy coating

K.Mulier

Too bad

Asmyldof

Aggressive fluxes is one reason I have Fibre ones

K.Mulier

What flux do you use exactly?

Why you choose "agressive fluxes"?

Asmyldof

Almost all between 1.1.1 and 3.3.3

Because sometimes I need to use parts and solders that don't equate to your normal HASL board with SOP8 chips

K.Mulier

Ehm, what does that mean?

Asmyldof

6:39 PM

Non standard

K.Mulier

I've never met a component that couldn't be soldered in my usual ways. What kind of component do you actually mean?

Asmyldof

Not everything is copper, tin and gold

K.Mulier

Oh, okay. But wouldn't that make the "agressive flux" only useful in very specific circumstances - perhaps only 1% of the cases?

Asmyldof

Depends on what kind of work you do

Many people don't need a jack-hammer, but a road-worker would have a sore day without one

What did you measure them with? If it was measured with a true RMS multimeter (most multimeters are, except the super cheap ones), then the value you measured is your RMS value. — Felthry 9 mins ago

Anyone care to enlighten me as to when the market suddenly got flooded with actual true RMS meters?

The Photon

@Asmyldof alleen dat u amusereit weerd

(in case you couldn't guess, I didn't even check that one in Google Translate)

amuseert?

Asmyldof

6:48 PM

Close enough for a non-native

@K.Mulier Re: Flux, soselectronic.com/a_info/resource/pdf/ine/Fluxes_EN.pdf

K.Mulier

Thank you Asmyldof!

Asmyldof

@ThePhoton I'm assuming you were aiming for something proximate to "Alleen om u te amuseren"

No one ever said it was easy

The Photon

@Asmyldof yes, that

Asmyldof

@ThePhoton You happen to have come across a good C-mount colour USB camera?

MickLH

7:21 PM

@Asmyldof I think this person has just gone insane from exposure to reality, and has created a world of delusion where people do reasonable things like spending the extra dollar to actually measure the desired value, instead of measuring a completely different thing and making extremely unreliable assumptions about it having a relationship with the desired measurement.

Since, you know, there's enough food for everyone and the purpose of money (to decide who starves) is obsolete and so spending an extra dollar isn't the difference between life and death but getting a voltage 10% wrong possibly could be

It's a very logical delusional world to create for oneself

Asmyldof

I'm more concerned with the illusion that this person seems to have with everything being printed on every multimeter being true, that might be just as lethal

MickLH

Oh yes I'm sure it is... but only in reality :P

Asmyldof

I dare say the very very vast majority of Multimeters are non-RMS or at best RMS, but certainly not True-RMS

PlasmaHH

@Asmyldof maybe also it means "super cheap" == under 500?

MickLH

I have trouble reconciling the prices of certain marketable features when I know the actual price of a sufficient ADC and RAM to make the measurements

Asmyldof

7:27 PM

@PlasmaHH That;s more in line with what I sort of debated responding with

PlasmaHH

But yeah, there are more useful meters that are true RMS than in the 90s these days

MickLH

So while I agree it's dangerous, I can understand the user's sentiment in choosing to live in that delusion. It's fucking infuriating that any meter would lack a reasonable RMS measurement when sufficient hardware costs less than the enclosure and user interface

At the very least, have a mode that spends a few seconds taking a good RMS reading for a steady wave

Asmyldof

Sampling of signal to include sufficient RMS of square waves, especially up to 10kHz or so isn't "dirt cheap" in comparison. Nor is the extra calibration process

That's going from standard DMM chips to customised embedded.

MickLH

"standard DMM chips" is obscuring the issue

assume a sane world, a reasonable standard is already mass produced

and also real-time measurement isn't really required in most cases, users wouldn't be able to tell the difference with a 10ms sampling period

Asmyldof

@MickLH If you want true RMS you need wave shape information. Wave shape agnostic would be what multimeters tend to call RMS measurements and can be off by as much as 30% on certain wave shapes

MickLH

7:43 PM

I think periodicity isn't too strong an assumption to ask for...

Asmyldof

8:33 PM

@MickLH periodicity, stability, frequency band and non-drift, you mean?

MickLH

yeah yeah all the stuff, upper limit on harmonics too etc

Asmyldof

So not True RMS then

MickLH

I made sure not to say "True RMS"

because something that reads the correct value in the vast majority of cases seems a good enough comprimise to me, better than peak times a factor

if it works on PWM and rectified signals it's probably gonna be just as good as true RMS for the average user

Asmyldof

@MickLH Which is what most labelled "RMS" do

And the whole point was about "The majority are true RMS" and your follow up of it not being that much more expensive these days

MickLH

I was responding off the Stack Exchange question tbh

I can't understand why anyone would build a peak times factor meter today

Asmyldof

8:37 PM

While that's true for Fluke and Keysight, and indeed, they have more and more True RMS and less and less non-so, it's very untrue for any mfg that doesn't do "exotics"

Because the majority of people will buy it and read "RMS" and be happy and it only costs $2 to build

Maybe $10 if the plastic is pretty

MickLH

I'd see if I can sue for false advertising if they sold a peak voltage meter as an RMS meter

A peak voltage meter that's wrong by a factor of sqrt(2) is just a shitty peak meter, not any kind of RMS imo

A heuristic RMS meter would have been fine for the guy with the question

PlasmaHH

If you slap "RMS" onto a meter, people will think it is RMS that is measured, regardless of hte number of * you put behind it and in fine print

MickLH

It's sad that legislation is unskilled. It's really about as bad as writing "non-toxic*" on a bar of soap and then in fine print having "* actually is toxic to ingest"

PlasmaHH

not as fun as the calorie free sweetener that has as much calories as sugar because it basically actually is sugar

among other things it is great that the people of turkey now have easier access to laser hair removal...

The Photon

8:57 PM

Anybody want to play Root Cause Analysis?
I have an LTC2640 DAC, and the output is not equal to the reference voltage times the programmed code divided by 4096.
Any spitballs what is wrong?

PlasmaHH

Your measurements?

How much "not equal" are we talking about?

Asmyldof

Supply out of bounds?

Supply noise?

PlasmaHH

misinterpretation of the datasheet?

Asmyldof

communication failure?

PlasmaHH

what kind of reference anyways?

The Photon

9:03 PM

:37120727 For example, I program it to mid-scale (2048 counts). I measure the reference voltage (external reference) as 4.269 V, output voltage 0.201 V.

@Asmyldof This is the main one I haven't eliminated. Unfortunately it's write-only SPI, so I can't verify that what I wrote is what the chip received.

PlasmaHH

do you have a counts vs. voltage plot?

Asmyldof

@ThePhoton Did you try a linear sweep?

PlasmaHH

did you accidentally treat a 8bit vcersion like a 12 bit one?

The Photon

So it's a bit tricky. The reference is the divided down output voltage of a voltage regulator. The output connects to the feedback pin of the regulator.
I have 4x this circuit on my board, and only two fail.

Also I had no problems on the two Rev 1 boards I built. This is the first Rev 2 board.

PlasmaHH

o_O thats more headscraczhy

and did you like, remove a failing one and replaced it by a known good one?

The Photon

9:07 PM

@PlasmaHH yes

Asmyldof

@ThePhoton 4 on same bus?

And again, did you try a sweep?

The Photon

@Asmyldof Yes.

PlasmaHH

guess a schematic would be helpful then

The Photon

@Asmyldof Not all the way from 0 to 4095.

But...

Power on default is mid-scale.

Marla

Could problem be the load ?

The Photon

9:09 PM

If I start stepping up from there, the regulator output doesn't change for a few steps...then it will randomly change to some other value than what I expect for the code I just programmed....then be stuck for a few more changes...then change again to another random value.

@Marla I checked and the load current is less than 1 mA.

PlasmaHH

and vcc > vref?

The Photon

@PlasmaHH Yes. Also during power up and power down. Protected with a schottky diode.

Asmyldof

Step one: Remove Vref Loopback and connect a fixed reference, see what happens

The Photon

@Asmyldof Reasonable

Asmyldof

Might be bad contacts anywhere, but possibly this batch is more sensitive to something not evident in board one?

(Bad contacts as in dry CS either side, or such. Or dry ref pin, etc)

The Photon

9:12 PM

@Asmyldof I probed the MOSI/CLK/CSn lines and everything is nice and clean.

@Asmyldof Also swapping chips should have cleared that, if the issue was on the chip (as opposed to the driving circuits)

Asmyldof

Had a board with 4 DAC manufactured by a professional, after two hours of fiddling with sweeps I discovered several of the Vref pins were dry

(and other pins not directly influencing anything at that time)

I'm not saying it's dry pins on the DAC perse

or at all

PlasmaHH

When debugging issues with our high performance software I tend to log debug statistics and whatnot in megabytes per second and then start verifying them step by step until I find an anomaly... I woild apply a similar principle here

Asmyldof

Just, learned recently one might want to check whatever is checkable before going in pointless loops

The Photon

Rev 2 board #2 should be here tomorrow.

Asmyldof

@ThePhoton can you sketch the situation visually, vis a vis the reference voltage loop?

The Photon

9:17 PM

@Asmyldof Think using a multiplying DAC as the voltage divider in the feedback path of a buck regulator circuit.

Asmyldof

Any changes between R1 and R2 in either circuit?

Measurements all done by scope?

The Photon

@Asmyldof Not all, but I did check for excessive noise wherever I could think to.

and oscillation

Asmyldof

Oscillation and such was what I was thinking of mostly

Asmyldof

9:33 PM

@ThePhoton Can't see any new risks in the specifics.

But also didn't do a full loop analysis

Because need to do ironing

And stuff

PlasmaHH

ironing?

Asmyldof

Yes

Go on, tell your wife that actually, further research shows that in fact men can do that. I dare you

PlasmaHH

its done in our household like, once a year probably

@ThePhoton I haven't checked the datasheet, but you haven'tl ike, loaded it down or so? maybe a wrong resistor at the output? Tried measuring it with the output disconnected to the rest?

The Photon

9:51 PM

@PlasmaHH Datasheet allows up to +/-10 mA load current, I measured less than 1 mA (by measuring voltage across R40)

Asmyldof

10:24 PM

@PlasmaHH Try every single piece of clothing I wear

Transcript for

Electrical Engineering