Test Equipment Anonymous (TEA) group therapy thread

[bd139]

Yeah that's why I've avoided it so far. It's CH340 or feck around with that. If I'm shipping 1000 units (first round target for this) then I'm going to save $80 total. Not sure if it's worth it for the software dev cost. 10000-50000 units, yes it is, but I'm not getting my hopes up yet 

[Martin.M]

I set the 3400A aside until I find a donor unit that won't make me feel like I am spending $50 for a plug socket. Wandered into the storeroom and pulled the next thing out of the pile. Guess what it is from this photograph:



Yes that is a 0.009975mfd, 400V electrolytic capacitor. One of two in this unit.  The big capacitor above it has the negative lead removed and the entire case is soldered, along its length, directly to the printed circuit board.

It powers up and generates and appropriate signal (more or less), too, which was a surprise as it came out of the last "here take this stuff I need to get it out of my garage" trip.

Med or one of the other tek-addicts will probably recognize it.

that is a holy grail capacitor from "Tek Selected", very low temp. drift (factory aged), not polarized but they are almost healthy, 50 years is no age for them.
0.009975mfd  is Tek language, translated to us it is a .01µF or 10nF SQ, + 0% -0,1%, or simply: in parallel with a 0-30pF variable ceramic C it is a perfect 10nF oscillator C.

[mnementh]

Yeah, but the CH340 real/fake ratio is like... 10/90... they're as bad as FTDI now. How can anybody produce anything with them when even the official channels are known to pipe reject and outright counterfeit product back into the supply chain? 

@med - Thanks for the compliment! 

mnem

[Mr. Scram]

Combined IQ of all involved of about 150. With the chimp scoring 70. What a bunch of morons. 

Also very fake.

[Kosmic]

Combined IQ of all involved of about 150. With the chimp scoring 70. What a bunch of morons. 

Also very fake.

Probably part of the marketing campaing for "planet of the apes" or something.

[bd139]

Yeah, but the CH340 real/fake ratio is like... 10/90... they're as bad as FTDI now. How can anybody produce anything with them when even the official channels are known to pipe reject and outright counterfeit product back into the supply chain? 

@med - Thanks for the compliment! 

mnem

Ah fuck didn't know that. SiLabs CP2102 is plan C  (there is a plan D too)

[Kosmic]

Yeah, but the CH340 real/fake ratio is like... 10/90... they're as bad as FTDI now. How can anybody produce anything with them when even the official channels are known to pipe reject and outright counterfeit product back into the supply chain? 

@med - Thanks for the compliment! 

mnem

Ah fuck didn't know that. SiLabs CP2102 is plan C  (there is a plan D too)

STM32 has USB integrated you know 

[Cubdriver]

There is a nice thing on it's way to me.
And it contains some of them:



And one of them:



  

So you snagged that one, huh?  I'd been resisting it for some time now.  Congrats!!!

-Pat

[Martin.M]

there are more obsolete Nixies in the world,
by example:  the "s" "ms" "Hz" "kHz" Nixie...

[Martin.M]

...



 

[bd139]

STM32 has USB integrated you know 

Yeah it has that, the ICs are a lot cheaper and LCSC will place them. However I don't know STM32. Yet. And I've written an event driven CMT kernel on AVR I don't want to have to port.

I actually have an STM32 discovery board somewhere!

Ahh fuck it. If I blow a couple of days on that, so be it 

[worsthorse]

I'll crack open the 114 as soon as I get it and we'll compare. Supposedly it's fully functional but if I need some GE transistors I'll be PM'ing you. 

Excellent. I'd like to see if your unit has some of the same mods as mine. I am on the road until Saturday, so I won't be able to pick up on it for a couple of days. I will also look at my crate of Ge transistors this weekend and see if I have any of the tek part substitutes, just in case.

[bd139]

So there's STM32 out the window. I have just spent an hour trying to work out what device I need. Turns out their low end devices with USB are expensive and their mid range devices with USB are cheaper but complete overkill.  1000pcs quote:

STM32 $1.43 + headache + learning curve

ATmega16 + CH340 $1.32 + no headache + no learning curve

That's $110 difference over 1000pcs or ~ 2 hours cost. Screw it. Time-value proposition doesn't work.

If I could get one with a decent enough ADC and DAC like PCM2900C which is $4.5 a go that'd make a cost difference. Hmm.... thinking.

[med6753]

So there's STM32 out the window. I have just spent an hour trying to work out what device I need. Turns out their low end devices with USB are expensive and their mid range devices with USB are cheaper but complete overkill.  1000pcs quote:

STM32 $1.43 + headache + learning curve

ATmega16 + CH340 $1.32 + no headache + no learning curve

That's $110 difference over 1000pcs or ~ 2 hours cost. Screw it. Time-value proposition doesn't work.

If I could get one with a decent enough ADC and DAC like PCM2900C which is $4.5 a go that'd make a cost difference. Hmm.... thinking.

What did I miss? Are you attempting to make and sell a product? If so, what? 

[mnementh]

If you keep doing the math this way instead of considering the time invested in learning STM32 as an investment in professional development, you'll never ever spend the time.

We RC hobbyists abandoned Atmel ages ago for driving ESCs... not fast enough to keep up with ever-escalating commutation timings. STM32F1 became the bare minimum for flight controllers somewhere around 2012 or so; current tech is pushing STM32F7 family to its limits, and dual-processor designs are commonplace now.

This is 100% hobbyist tech too... not even legitimately anything approaching commercial product (even though there's millions of dollars in sales annually); Atmega is still viable for frontend control, but the actual work gets done with STM32s where larger horsepower is needed and SiLabs EFM8 BusyBees where smaller CPU load but higher speeds (and hardware PWM) are needed.

mnem
Come over to the dork side; we have cookies. Also integrated USB with signed drivers. 

[BU508A]

BU508 you are restorating a R+S DVM there?

Nope. Guess again.

I have two R&S DVM, a UDG 51 and a UWGD BN 1110 but none of them has the Z566M or Z567M in them.
Here are some pictures of the (my) UGD 51:
https://www.eevblog.com/forum/reviews/lets-see-your-nixie-tube-equipment/msg857874/#msg857874



This is the UWGD:


I wasn't aware, that there is R&S equipement around with Z566M / Z567M nixie tubes in it.
Can you provide an example?

And, just to be clear: I didn't have ordered those Nixie tubes. They come in one device. 

[bd139]

If you keep doing the math this way instead of considering the time invested in learning STM32 as an investment in professional development, you'll never ever spend the time.

We RC hobbyists abandoned Atmel ages ago for driving ESCs... not fast enough to keep up with ever-escalating commutation timings. STM32F1 became the bare minimum for flight controllers somewhere around 2012 or so; current tech is pushing STM32F7 family to its limits, and dual-processor designs are commonplace now.

This is 100% hobbyist tech too... not even legitimately anything approaching commercial product (even though there's millions of dollars in sales annually); Atmega is still viable for frontend control, but the actual work gets done with STM32s where larger horsepower is needed and SiLabs EFM8 BusyBees where smaller CPU load but higher speeds (and hardware PWM) are needed.

mnem
Come over to the dork side; we have cookies. Also integrated USB with signed drivers. 

Investors invest wisely. This is a commercial project. Therefore time = money. And for all of us time = finite. Priorities need to be well defined and if I can get an MVP on the market without having to learn again, reinvent several wheels and pay slightly more for the privilege then it makes no sense at all to even nudge anything in that direction. This isn't even the most important bit of the product. Thus I'm reluctant to even bother.

I actually have got to the point I might even scrap what I have and use AmForth for this as I can just stream FORTH primitives to the device over serial from the PC as it has excellent SPI primitive support. That means there is literally no code to write on the device. Everything is software defined on the client.

Also CH340 drivers are signed if you get them from the right place, and I own a DigiCert EV code signing key so  (someone wanted this so i can sign a windows MSI package for something and now pays me to maintain it yearly)

Edit: Check this. Out of box. Can issue these commands in immediate mode straight over a UART plugged into a CH340:

[GreyWoolfe]

I just hit Ctrl+S and it disappears! Windows build is bog standard.

Cheapduino here - CH340 chipset. China's finest 

I usually use a USBasp programmer, bare AVR and AVR-gcc / AVR-libc / avrdude / make because that's closest to home (C + linux toolchain) for me.

And fuck STM32. I can't even work out where to start. I look into it and the "can't be arsed" switch goes off.

Then I think you need to take The Blue Pill.  It's made for guys like us.

mnem

Combined IQ of all involved of about 150. With the chimp scoring 70. What a bunch of morons. 

Kind of reminds me of upper manglement cutting newbies loose with minimal training.  Oops, did I say that out loud?

[beanflying]

So there's STM32 out the window. I have just spent an hour trying to work out what device I need. Turns out their low end devices with USB are expensive and their mid range devices with USB are cheaper but complete overkill.  1000pcs quote:

STM32 $1.43 + headache + learning curve

ATmega16 + CH340 $1.32 + no headache + no learning curve

That's $110 difference over 1000pcs or ~ 2 hours cost. Screw it. Time-value proposition doesn't work.

If I could get one with a decent enough ADC and DAC like PCM2900C which is $4.5 a go that'd make a cost difference. Hmm.... thinking.

32U4 and dump the CH340/FTDI. The little pro micro boards with them are not as cheap as some but it makes for a low parts count.

[mnementh]

If you keep doing the math this way instead of considering the time invested in learning STM32 as an investment in professional development, you'll never ever spend the time.

We RC hobbyists abandoned Atmel ages ago for driving ESCs... not fast enough to keep up with ever-escalating commutation timings. STM32F1 became the bare minimum for flight controllers somewhere around 2012 or so; current tech is pushing STM32F7 family to its limits, and dual-processor designs are commonplace now.

This is 100% hobbyist tech too... not even legitimately anything approaching commercial product (even though there's millions of dollars in sales annually); Atmega is still viable for frontend control, but the actual work gets done with STM32s where larger horsepower is needed and SiLabs EFM8 BusyBees where smaller CPU load but higher speeds (and hardware PWM) are needed.

mnem
Come over to the dork side; we have cookies. Also integrated USB with signed drivers. 

Investors invest wisely. This is a commercial project. Therefore time = money. And for all of us time = finite. Priorities need to be well defined and if I can get an MVP on the market without having to learn again, reinvent several wheels and pay slightly more for the privilege then it makes no sense at all to even nudge anything in that direction. This isn't even the most important bit of the product. Thus I'm reluctant to even bother.

I actually have got to the point I might even scrap what I have and use AmForth for this as I can just stream FORTH primitives to the device over serial from the PC as it has excellent SPI primitive support. That means there is literally no code to write on the device. Everything is software defined on the client.

Also CH340 drivers are signed if you get them from the right place, and I own a DigiCert EV code signing key so  (someone wanted this so i can sign a windows MSI package for something and now pays me to maintain it yearly)

   Edit: Check this. Out of box. Can issue these commands in immediate mode straight over a UART plugged into a CH340:

I dunno... seems to me that for someone in the trade, even just dabbling, that nowadays not being able/willing to code for STM32 is a bit like not being able/willing to script in PowerShell. I know my weak Kung-Fu (and an overpowering allergy to the dissolving fluidity of corporate command structure) is what keeps me from being on the other end of the shit-shoveler/remote sysadmin equation.

If you can work it out in FORTH, can't you as easily go STM32? There have to be well-developed libraries by now, don't there? Unless their standard 12-bit ADCs are a limitation, speed or resolution-wise...

mnem
*back into hell*

[bd139]

I know someone who delivers a project with a Parallax Basic stamp. What matters really is the delivery if you’re in control of the whole product cycle. If I was selling myself I’d be on STM32 a long time ago. I don’t use ADC. I’ve got a USB audio device doing that.

Let’s not go down the PowerShell hole. I don’t.

ATmega32U4 Is $2.25 more expensive per unit than a cheap AVR and CH340. Also a point to note, it’s a lot harder debugging an integrated USB peripheral than a UART with RX and TX test points

[Kosmic]

The best tools are often the one you know and master. So if for you it's AVR and FORTH, go for it 

[med6753]

I think I need to make more popcorn. 

 

[bd139]

I’m done now. Gone to bed 

Back on TEA, apparently keysight fucked up my 34401A display order. Took details but didn’t bill my card. Grr will call them tomorrow.

[VK5RC]

Re Keysight, a year ago I had a DMM calibrated, they were supposed to email, I would pay the bill, then they would send it back, I was surprised when I got the unit back - I suppose for them I am 'loose change'. 😊
PS re micros - I am in the luddite camp, while I am moderately OCD, getting me to do some programming is the quickest way to get something hit! 10 mins to find a space that shouldn't have been there.