On Sun, Nov 07, 2021 at 12:05:23PM -0500, Harvey White wrote:

Interleaved responses.

Both arduino NANO and the Nucleo-32 boards have the same pinouts and form
factors, so you can buy a breakout board made for a NANO and plug the Nucleo
board in.? Be careful of the voltages, as ARM processors are 3.3 volts
rather than 5.0 volts.

The Nucleo-32 could be a good solution, but I think I'd take issue with
the NANO. With only 2K of SRAM for stack and variable space it seems a
bit small for doing any significant graphics. Heck, that's not even enough
for a frame buffer. Maybe a sophisticated programmer could do something
with it, but that's still a stretch.

I have a project using a NANO now and I've had to painfully optimize every
byte of storage to get it to fit in. If you go over 90% SRAM utilization,
you're very likely to overwrite the stack during runtime. And of course
the only way it tells you that is by going into either a reset loop or
OTL. Not really good for starting out.

I'd move to the Nucleo, but NANOs are just too darn cheap. I'd be happy
with a similar form factor for the ATMEGA644p. Still a bit small for
this, I think.

Paul

--
Paul Amaranth, GCIH | Manchester MI, USA
Aurora Group of Michigan, LLC | Security, Systems & Software
paul@... | Unix/Linux - We don't do windows

The ST tools (CubeMX and CubeMX_IDE) allow you to do this through a graphics interface, and assign and remap only legal pins to legal functions.? You do not need to wade through 1200 pages of documentation.? The setup code is then inserted into your project.

Harvey

On 11/7/21 09:39, peter bunge wrote:

I am not a programmer and like to keep things simple.
The reason I like the PIC over Arduino and RaspberryPi is that I can make custom PCBs and program the pins on the PIC to be convenient.

micropython on a STM32F4 or others might interest you. It's still a learning curve. STM32's have lots of GPIOs ports and such that can be remapped in various alternate ways, but not pin by pin. The power and grounds and XTAL in/out don't remap, but some redundant ones can be left off.

On 11/7/21 12:28 PM, Bruce wrote:

I'd like you to expand on the "toxic ecosystem" of PIC.
I use them occasionally and find the IDE reasonably - what is the complaint ??

Well, you said part of it right there: "the IDE". "THE". Microcontrollers and toolchains should never be treated as a single-source unit, Microchip tries hard to force people into their proprietary tools, and people who don't know any better just "go with it". "PIC" is a microcontroller architecture...when you choose a microcontroller architecture, you should be doing just that, choosing a microcontroller architecture. When you choose PIC, you are immediately shoe-horned into one set of tools, from the chip vendor, who supports what they want to support and de-supports what they no longer want to sell. It is a completely vendor-controlled ecosystem.

Projects and companies have failed due to situations like this.

Yes, there are tools available from other vendors, and there are some free PIC tools. Some of them are decent...CCS' C compiler comes to mind; it's very good. But all of these tools force you to write tool-specific code with almost no reusability if you ever dare step outside the PIC architecture. That's reusability of either code from a project or knowledge gained in working with the tools. You end up starting from scratch for both!

And the tools themselves (with some exceptions, like CCS C) are generally pretty crappy by modern professional standards. You can get work done, but it's not easy, not standards-based, and leads straight to one of the most dangerous kinds of vendor lock-in. It's ok for hobbyist-level playing around and doing occasional small, simple projects. But trying to do anything complex, or heaven forbid trying put dinner on the table with it, is a joke.

Microchip themselves have gotten better with the introduction of the MIPS-based PIC32 architecture (for which there are great tools) and their acquisition of Atmel a few years ago, bringing the fairly nice AVR architecture into the fold, which also has great tools...So it's a PIC problem, not a Microchip problem.

The only good thing the Microchip IDE itself has going for it is that it's built on Eclipse, which is widely supported and extensible, so the user actually has some control over it.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

I'd like you to expand on the "toxic ecosystem" of PIC.

I use them occasionally and find the IDE reasonably - what is the complaint ??

Cheers!

Bruce


Quoting Dave McGuire <mcguire@...>:

开云体育

Interleaved responses.

I find all that discouraging and confusing. I am not a programmer and like to keep things simple.

If you don't do any programming at all, then you are stuck buying already built solutions.

If you do hardware design, then you need to have a programmer as a VERY good friend, or you need to start learning until you feel confident.

The reason I like the PIC over Arduino and RaspberryPi is that I can make custom PCBs and program the pins on the PIC to be convenient. I can use 8 ADC inputs or make the same pins I/O.

You can do that on arduino, or ARM.?

Almost all microprocessors do that kind of thing (pin sets, adc).? The task in each processor environment is to figure out how *this* microprocessor does *that*.

Both arduino NANO and the Nucleo-32 boards have the same pinouts and form factors, so you can buy a breakout board made for a NANO and plug the Nucleo board in.? Be careful of the voltages, as ARM processors are 3.3 volts rather than 5.0 volts.

I can use a 14 pin PIC or a 40 pin PIC. I can use DIP packages that I can see to solder.

That's one reason why you can (and often want to) buy a prebuilt board.? Some people can do SMT work, some can't.?

I don't think the PIC is a good choice to drive a graphics LCD but it can be done. I need something more powerful and friendlier. Back to the point of this thread.

I agree, but it's just because I don't like the PIC architecture for general purpose work (stopped following them years ago, so very possible they might have changed).

All depends on what you want to do, and how you want to do it.? You can do an arduino solution with plugin hardware with a compatible display, and use the arduino's inbuilt ADC to digitize a scaled copy of the screen voltages.?

Harvey

Thanks Harvey & Dave & others;

There are actually several projects being tossed around so I will try to define them.

?

Project 1: Tektronix 576 Curve Tracer. This would require a very large amount of signal scaling from differential amplifiers, some transistor but maybe some tube circuits. All those switch positions that wind up in the light pipe digital displays could be a problem. The LCD would have to be physically mounted in place of the CRT. The sort of thing I can do except for the LCD software which I could probably manage with a lot of help. It is not a project I have time for.

?

Project 2: Replacing ‘scope CRTs. As mentioned it is better to buy one of many cheap DSOs like I did. I bought a Siglent.

?

Project 3: Adding a Graphic display to the HP4192A LF Impedance Analyzer and the HP4193A Vector Impedance Meter. It is more than just adding a display that reads a couple of ADCs. They change range automatically and you have to set them up with start/stop frequencies, etc. I am doing this with a Win7 laptop running EZGPIB and Prologix but there is no display until I massage the data and plot it. The 7470A emulator will not work with these. They are designed to drive an analog pen plotter. The solution is not simple. Will the F469I-Disco board control via GPIB? Is there a VESA library it can access? There needs to be an interactive input and digital capture which is why I suggested the RaspberryPi. It has an LCD display and keyboard, I’m sure I could grab a gif or jpg of the graph, it has a C compiler. Is there a VESA library or some way to communicate through the Prologix. Is the software beyond my capabilities? Must I use Python? Another option is a Win10 laptop and Prologix but I need to commit to a software package and learn it. This is something I would tackle preferably with help from any interested parties. Please contact me privately if you are interested. Adding Sonar to my ROV is my first priority so this project cannot take over my life but I will be working on C and Graphics starting now.

?

Project 4: I would like to be able to write programs on a Win10 computer in C, and do graphics on it. I will have another go at MS Visual C++. It ties in with everything else.

Project 5: I would like to control my Prologix with a C program on a Win7/10 computer. This ties in with Project 4.

?

Project 6: I would like to handle graphics on an LCD so it may be worth buying an F469I-Disco board to play with, if it can be done in C and I don’t have to learn a new microprocessor system. I have been using the PIC for 20 years and my ROV is full of them (9 if I include the robotic arm). I already do this with low res LCDs of 128x64 using PICs and a Digole adapter.

?

Project 7: A friend has a problem with his system that is going obsolete. It is driven by Laview on a Win7 computer with no spare. It controls an oven up to 1400 deg C and reads a thermocouple, it drives a linear actuator to move the sample. I cannot repair the actuator controller because there are no schematics and it has a multilayer PCB. Labview also controls an HP8753ES and takes the readings but it has to be fast before the heated sample cools. His software guy is no longer available. I have no desire to learn Labview but I do maintain his VNAs. I don’t think I can help him very much without a huge commitment. If I was already programming in C on a Win10 and knew how to control the VNA via Prologix I could help. These are skills I want to learn but it may be too late if he has a critical failure which could shut down his business.

These projects are driving my desire to learn C along with Graphics on Windows.

?

Project 8: Filling in our swimming pool. Finished now but the reason I did?not get?much done on?the other projects this?summer.

I rented a jackhammer for a week and broke up the concrete then used a skill saw with cut off blade to slit the steel liner. I hired a contractor to lift the steel out, cart the rubble away, and truck in fill and pack it down. Then another firm with a broadcaster to spread topsoil. Now I have the time to learn the skills I need for new projects. Photos below taken in June and September. I have cut the grass twice since then. We used it for 35 years and are simplifying our lives. I am 81, past my best-before date but not yet at my expiry date. Peter

??

On Sun, Nov 7, 2021 at 10:17 AM Dave McGuire <mcguire@...> wrote:

? ?Keil...well, there's a lot to know about that.? Here goes.

? ?If a person works hard enough, they can always find a way to pay for
something that the rest of the world gets for free.? Keil's lumbering
IDE, like nearly all IDEs, is a wrapper around existing command-line
compilers, and the compiler underlying Keil's IDE is, depending on which
version you're running, either GCC or LLVM.? Both of these compilers are
free.

? ?GCC is ARM Ltd.'s reference compiler for the ARM architecture.? They
maintain the back-ends and provide reference builds.? While GCC and LLVM
come from the UNIX world, Keil's IDE only runs under Windows, last I
checked.? And to make the picture even more interesting, ARM Ltd bought
Keil some years ago, wanting to control the whole eocsystem.

? ?Keil's system does give you a value-add atop GCC: lots of device
libraries to do various things.? But as you observed, you pay a lot for
it, and all of that stuff is available elsewhere for free.? And Keil
ties you to Windows...a high price on top of an already high price.

? ?There's one real reason that Keil's IDEs have some market share: Keil
goes back a very long time, to the early days of mcs51.? I used their C
compiler (a DOS product back then) for 8051 development in the late
1980s, at a large defense-related organization that did everything in a
vacuum.? Though the world of free software was well-established at that
time, embedded development was less mainstream than it is now, so there
was very little free software out there in that space.? When a company
like that starts looking at a new microcontroller, they ask the vendor
of their current development software if they support it, and if they
do, a new generation of sales is born.

? ?And since ARM Ltd owns Keil, and lots of companies are run by suits
who don't know any better, when a suit asks "where do I BUY stuff to
develop for ARM?", ARM Ltd is happy to sell them the Keil IDE and not go
out of their way to tell them "Oh by the way, this stuff is actually free.."

? ?For those who need the hand-holding of an IDE, Eclipse is free, and
its installed base dwarfs Keil's.? Many organizations do ARM development
with GCC or LLVM running under Eclipse.

? ?It's worth knowing what the industry's best practices are.

? ? ? ? ? ? ? ?-Dave

On 11/7/21 12:31 AM, David C. Partridge wrote:
> The main ARM development platform I'm aware of is Keil MDK and for sure that isn't even close to free (nor is it forgiving).
>
> -----Original Message-----
> From: [email protected] <[email protected]> On Behalf Of Dave McGuire
> Sent: 06 November 2021 23:42
> To: [email protected]
> Subject: Re: [HP-Agilent-Keysight-equipment] Question: XY (XYZ or Vector) instrument CRT, replacement with an LCD ?
>
> :
>
>? ? ?They are complicated, but they're very forgiving, and they don't have
> the toxic ecosystem of the PIC family.? Once you get into small ARM
> microcontrollers (and now, RISC-V) you will never look back.
>
>
>
>
>
>


--
Dave McGuire, AK4HZ
New Kensington, PA

On Sun, Nov 7, 2021 at 02:44 AM, Richard Merifield wrote:

> ...
> I measured the voltage at the R40/R31/R30 junction at 132V DC and -5.8V and -5.3V on the other side of R30 and R31 respectively
> with respect to chassis ground. Why would R30 get twice as hot as R31 given those measurements?
> Don't they effectively have almost the same current flowing across them ? I desoldered R30 and it measure 3.1kOhm but
> perhaps something else is wrong with it ?
----
There could be an AC voltage difference, RMS voltage on the resistors could be different. If you look at the voltage at two terminals of C12 with respect to ground with an oscilloscope (using a probe that can work at those voltage) what do you observe? When the circuit is perfectly balanced and C12 is doing its job AC voltage at two ends of C12 (on R30/R31) should be small. Ideally zero volts if everything was perfectly balanced but there will be some AC because of mismatches.

Ozan

Keil...well, there's a lot to know about that. Here goes.

If a person works hard enough, they can always find a way to pay for something that the rest of the world gets for free. Keil's lumbering IDE, like nearly all IDEs, is a wrapper around existing command-line compilers, and the compiler underlying Keil's IDE is, depending on which version you're running, either GCC or LLVM. Both of these compilers are free.

GCC is ARM Ltd.'s reference compiler for the ARM architecture. They maintain the back-ends and provide reference builds. While GCC and LLVM come from the UNIX world, Keil's IDE only runs under Windows, last I checked. And to make the picture even more interesting, ARM Ltd bought Keil some years ago, wanting to control the whole eocsystem.

Keil's system does give you a value-add atop GCC: lots of device libraries to do various things. But as you observed, you pay a lot for it, and all of that stuff is available elsewhere for free. And Keil ties you to Windows...a high price on top of an already high price.

There's one real reason that Keil's IDEs have some market share: Keil goes back a very long time, to the early days of mcs51. I used their C compiler (a DOS product back then) for 8051 development in the late 1980s, at a large defense-related organization that did everything in a vacuum. Though the world of free software was well-established at that time, embedded development was less mainstream than it is now, so there was very little free software out there in that space. When a company like that starts looking at a new microcontroller, they ask the vendor of their current development software if they support it, and if they do, a new generation of sales is born.

And since ARM Ltd owns Keil, and lots of companies are run by suits who don't know any better, when a suit asks "where do I BUY stuff to develop for ARM?", ARM Ltd is happy to sell them the Keil IDE and not go out of their way to tell them "Oh by the way, this stuff is actually free.."

For those who need the hand-holding of an IDE, Eclipse is free, and its installed base dwarfs Keil's. Many organizations do ARM development with GCC or LLVM running under Eclipse.

It's worth knowing what the industry's best practices are.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

Keil is nowhere near free, except a politician's definition.

However, ST Micro has free tools.

I think that Freescale has as well (although it may be limited).

I think NXP's tool is free.

TI's tool is free for certain product lines.

Even Xilinx has a free tool for some of their FPGA/CPLD; and Atmel (oops... microchip) has or had a free tool for the ATMEGA and XMEGA series.? Haven't used it in years, though.

I've used Keil, it has some nicer features than the ST Micro tools (mostly in figuring out what's doing what), but then again, I think that ST may have them, somewhere.

For the manufacturer's tools, the software is locked to their products.? That may not be all that bad.

PlatformIO seems to be a generic style compiler (will do ESP8266, ATMEL, and so on.? Even the Arduino IDE will do some other chips, but the lack of debugging is a show stopper... unless they added it).

If I were a company, and I had big dollars, I'd buy a commercial package for support issues.? Such commercial packages may have traceability back to industry standards demanded by customers....

I'm not a business, and don't have the deep pockets.

Free is good.? Quality Free is better.

For 13 to 30 dollars (or so) for basic ST development boards of various form factors and capabilities, and considering the lack of chip microprocessors, ST isn't bad.

But, the free tools are out there.

Harvey

I repaired L2 as best I could and it measured ok so thought I would give it a try. I suspect on 6AU6 may be part of my problem. I swapped them around and I now have a perfectly functional unit as far as I can tell, except R30 gets scolding hot still. I have 4 more 6AU6 coming.

I measured the voltage at the R40/R31/R30 junction at 132V DC and -5.8V and -5.3V on the other side of R30 and R31 respectively with respect to chassis ground. Why would R30 get twice as hot as R31 given those measurements? Don't they effectively have almost the same current flowing across them ? I desoldered R30 and it measure 3.1kOhm but perhaps something else is wrong with it ?

I think I am getting closer

Regards

Richard

On 2021-11-07 12:31 a.m., David C. Partridge wrote:

The main ARM development platform I'm aware of is Keil MDK and for sure that isn't even close to free (nor is it forgiving).

The free ones work fine.

-----Original Message-----
From: [email protected] <[email protected]> On Behalf Of Dave McGuire
Sent: 06 November 2021 23:42
To: [email protected]
Subject: Re: [HP-Agilent-Keysight-equipment] Question: XY (XYZ or Vector) instrument CRT, replacement with an LCD ?
:
They are complicated, but they're very forgiving, and they don't have
the toxic ecosystem of the PIC family. Once you get into small ARM
microcontrollers (and now, RISC-V) you will never look back.

The main ARM development platform I'm aware of is Keil MDK and for sure that isn't even close to free (nor is it forgiving).

开云体育

L2 provides plate feed to V1 so if it's open the DC conditions will be SEVERELY unbalanced. This is almost certainly why R30 is having a hard time, to say nothing of the various tubes. Hopefully they have not been permanently damaged. Running a pentode like a 6AU6 with full screen volts and no plate volts is a recipe for disaster. You can temporarily replace L2 with, say, a 100 ohm resistor or even a short circuit and if the tubes have survived then R30 should run cool and the unit should work OK with no clipping, at the lower frequencies anyway.

L2 and L3 are described in the parts list as RF reactors (chokes). If you can't repair L2 then get a couple of axial 2 mH RF chokes, those ones that look like resistors (or two 1 mH chokes in series) and install them. I wouldn't use ferrite cored power inductors intended for switchmode power supplies as they might not work properly at the frequencies of interest.

Morris

开云体育

In Tektronix (and I'd assume HP as well), some of the power supply decoupling networks were made of fine wire wound on a convenient resistor (say 10 ohms).? They're easy to repair, just match the wire gauge, and wind.? One or two extra turns will not make too much of a difference in a decoupling network.

Harvey

Rasberri Pis are fine little systems; they are just complete
Unix/Linux systems running on a card sized processor. You can
use any of the enormous number of programming tools, compilers,
graphics toolkits, etc available on linux. Throw away your
Microsoft world and move into the arena of decent software.
And you won't have to shell out any money, either.

I'm working on a project to build a radio/gps/music player for
my truck using a Pi as the controller. It's a good match for
that with open source components available for many functions
that I need. It's probably overkill for this application, but
ease of development and low cost can provide some justification.
Learning a tool chain and how to actually get code running on a
processor can take a lot of time to get working (and be frustrating).

Go with the solution that makes development easy and avoid
proprietary (and costly) solutions. I have an aversion
to Python but there are graphic scripting languages such
as Tk that would make this a fairly easy project. Qt is
a nice toolkit, but my only objection is that the versions
change often and it keeps getting bigger. Projects using
older versions of Qt sometimes get hard to build.

The choice of languages and toolkits will make a huge
difference in whether you finish the project.

Arduinos are way too small for this and Rasberri Pis are
arguably too big, but something in the middle will need
the appropriate tool chain, toolkits and libraries to
be useful. Pis are cheap enough to not worry about wasting
cycles. And you really don't want to get halfway through
a project and find your target is too small.

As an aside, I started my career on a Honeywell Multics
system back in the dark ages (Multics was a precursor
of Unix). The Rasberri Pi now can run a Multics emulator
that is faster than the last processor sold by Honeywell
(which cost in the millions). All for around $35.

Paul

--
Paul Amaranth, GCIH | Manchester MI, USA
Aurora Group of Michigan, LLC | Security, Systems & Software
paul@... | Unix/Linux - We don't do windows