EEVblog #1049 - Mailbag

[jonovid]

EEVblog #1049 - Mailbag


0:00 Isolated Serial UART Kickstarter
https://www.kickstarter.com/projects/902741881/optically-isolated-usb-to-serial-adapter
4:19  TCXO Oven crystal oscillator teardown
17:07 uRad Environmental Monitor
https://www.uradmonitor.com/products/
EEVblog Office Live Data: https://www.uradmonitor.com/?open=820000BE
20:37 Visual VU meter: http://sch-remote.com/
23:25 Random LCD PCB thingo
25:32 Interesting construction industrial controller teardown
30:32 Open Source Servo Motor Controler
https://odriverobotics.com/
38:29 How an inductor works in a light dimmer circuit
43:07 Car Rust Protection BULLSHIT!


apology for any error or missing links   

[Kleinstein]

I like the table of contents 

The Woo-Woo coil for the rust protector might help with EMC at least.

However the bifilar wound coil is probably just as funny  . With just 2 terminals, there is no common mode / differential mode. So this is just a low inductance piece of copper resistor. I don't know who got that funny idea - it does not even save a penny over a correct inductor. 

[Niels]

I have newer heard about Jay Electronique.
But the interesting this is that it conforms to "PL e EN13849 and SIL 3 EN61508" https://en.wikipedia.org/wiki/Safety_integrity_level
This basically means that the MTBFd (Mean time to Failure dangerous) has to be better than 1 out of 10E?7 hour,.
And it can be used in situations where Functional Safety is needed to safeguard human lives, and where a failure of the system could kill a person.
If such a system contains electronics, and/or software, then it has to fulfill EN61508 according to the MD directive. (European, but other parts of the world have similar laws )

This would require a two channel system, with a high DC (diagnostic coverage), basically you need to analyze the effect of a failure any component in the system.
And you need to make sure that any dangerous fault in on channel is detected, and the system is brought into a safe state. Or the fault is so unlikely to occur within its mission time, that it can be covered with the budget of one fault pr. 10E?7 hour.
If the product have a mission time of say 10 years, then it have to be replaced even if it appear to be functional.

You can run both safety channels trough a single radio module that is not SIL rated, provided that you have a system that fail safe in case of 'stuck bit's'.
I.e detects if the radio module is just replaying the same messages, despite having lost the link.

If you have a CPU in the system, then you need to document that you can detect any memory cell or register getting flipped due to a glitch/radiation from space...
This has to be detected within the safe process time.
You either need two CPU's running lock-step (comparing results), or use a safety CPU like the TI TMS570

EN61508 also covers software.
Generally you can't use dynamic allocated objects in the system, and you cant have any unbound loops.
You need to be able to calculate the WCET (Worst Case Execution Time), you basically need to know how many clock cycles a control loop takes, when it runs the longest path.
(A variant of the Turing Halting Problem).

[BU508A]

That thingy at 25:16 is a GDT (Gas Discharge Tube) for surge protection.
https://en.wikipedia.org/wiki/Surge_protector#Gas_discharge_tube_(GDT)



Edit: Image URL corrected

[Barny]

However the bifilar wound coil is probably just as funny  . With just 2 terminals, there is no common mode / differential mode. So this is just a low inductance piece of copper resistor. I don't know who got that funny idea - it does not even save a penny over a correct inductor. 

When I was a little boy, I tried to make a little electro magnet this way.
Strangly the magnet didnt work and got quite hot

[thmjpr]

The Odrive is a great idea, however some ideas:
- Two axis isn't enough, you need three axis for the majority of applications.
- Maybe use the existing number of FETs to form the third axis, there are two NTMFS4935NT1G (30V 14A) for each half-bridge side (28 total). Or find better FETs that don't need paralleling (SISA72DN, AON7522E?).
- The schematic has no notes, some section hints or component choice calculations would be nice (what is LM5109BSD doing for example?).
- Use polarized connectors for each encoder, JST connectors are dirt cheap and you will never plug them in backwards and blow something up.

It would be possible to cut the design down to a single modular axis, fitted at the motor, as done in the nano stepper (see my version here: https://github.com/thmjpr/nano_stepper). But in this case, probably won't work as well as your main control circuitry is more complex.

[TimInCanada]

Ah yes, the electronic auto rust protection scam.  It's still going.

The premise is that by imposing a current on a metal in a corrosive environment, galvanic corrosion can be prevented.  Sacrificial anodes, such as zinc galvanizing on steel, is one solution.  You can also actively impose a current, which is called a impressed current cathodic protection system.  That's used on ships, pipelines, marine structures and such. 

It really does work, provided there is an electrolyte to provide a return path for the current.  So as long as you park your car in the swimming pool, you're protected.

A few years ago at a gathering of my wife's family, someone left their car headlights on and needed a jump start.  I worked the cables.  One uncle pointed to a box in the engine compartment of one of the cars and asked what that was.  I said that was one of those bogus rust protection gimmicks.  Another uncle said "What do you mean bogus? I just paid $600 for that."

He's been a little frosty towards me since, for some reason...   

[reinhardz]

That thingy at 25:16 is a GDT (Gas Discharge Tube) for surge protection.
https://en.wikipedia.org/wiki/Surge_protector#Gas_discharge_tube_(GDT)



Edit: Image URL corrected

And here is the data sheet:



Best regards

reinhardz

[mcdanlj]

The Odrive isn't a completely new idea. The Odrive web site says:

Seriously, a driver that allows this should clearly exist.
But since it didn't, I decided to make one.

https://www.crowdsupply.com/citrus-cnc/tarocco

That's one axis at a time, $95 per axis for controller and motor with encoder. Sadly the discounted 5-packs are no longer available.

The design parameters are clearly different between Odrive and Tarocco, so it's totally not wasted time. The fact that the per-axis price is very similar isn't a huge surprise...

[whit3rd]

That 'inductor' that came from the triac circuit, is NOT intended to have
any particular inductance.   Rather, it is a transmission line, the ferrite
just gives it a longer time delay (and raises the characteristic impedance)
by increasing the henries/meter distributed inductance of the wire pair.

In the circuit as shown, that means it has some resistor-like (characteristic
impedance) character until the reflection from the short circuit makes it
a simple conductor.   The effect is to limit the inrush current to the triac
switch (which has a safe limit on dI/dt).   It's a clever way to match the
thyristor to an unknown load impedance.

[thm_w]

The Odrive isn't a completely new idea. The Odrive web site says:

Seriously, a driver that allows this should clearly exist.
But since it didn't, I decided to make one.

https://www.crowdsupply.com/citrus-cnc/tarocco
That's one axis at a time, $95 per axis for controller and motor with encoder. Sadly the discounted 5-packs are no longer available.

The design parameters are clearly different between Odrive and Tarocco, so it's totally not wasted time. The fact that the per-axis price is very similar isn't a huge surprise...

Its using brushed servo motors that are nowhere near as powerful as the Odrive brushless outrunner motors.
Odrive is a new idea in that respect (but could use some optimization as noted above).

5 A peak current at 24 V
4000 rpm maximum angular speed
3000 g·cm peak torque
30 W peak power
1440 counts per revolution optical encoder

Nano zero stepper is still the lowest overall cost option (motor + controller + encoder), per axis (<$60). With the usual trade-offs of lower cost steppers (wouldn't reach 4,000rpm, etc.).

[GreggD]

You can see a spot of paint inside the Gas Discharge Tubes. That spot is likely slightly radioactive to help start the arc. Some old neon lamps worked that way.

[Neomys Sapiens]

@Barny: The sign you are using as avatar is simply wrong! There was one overrun by a motorist in Bad Ischl last year!

[Brumby]

@Barny: The sign you are using as avatar is simply wrong! There was one overrun by a motorist in Bad Ischl last year!

And that's not the only reported occurrence:
http://www.smh.com.au/world/no-kangaroos-in-austria-at-least-one-is-lost-in-the-snow-20150128-130j3b.html

[SNGLinks]

You can see a spot of paint inside the Gas Discharge Tubes. That spot is likely slightly radioactive to help start the arc. Some old neon lamps worked that way.

When I worked at a NASA satellite tracking station in the 60s we had gas discharge regulators that were radioactive. There was a warning on the boxes not to store more than
a certain amount together