• Open source OS for Rigol scopes

    From Phil Hobbs@3:633/10 to All on Tuesday, November 17, 2026 12:38:48

    https://hackaday.com/2026/11/17/newer-rigol-scopes-get-a-custom-linux-firmware/

    Looks pretty cool, and runs off the built-in SD card, so you can switch
    back to the factory software anytime.

    Cheers

    Phil Hobbs

    --
    Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Jan Panteltje@3:633/10 to All on Tuesday, November 17, 2026 15:01:03
    Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net>wrote:
    https://hackaday.com/2026/11/17/newer-rigol-scopes-get-a-custom-linux-firmware/

    Looks pretty cool, and runs off the built-in SD card, so you can switch
    back to the factory software anytime.

    Cheers

    Phil Hobbs

    Impressive!


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Phil Hobbs@3:633/10 to All on Tuesday, November 17, 2026 17:04:44
    Jan Panteltje <alien@comet.invalid> wrote:
    Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net>wrote:

    https://hackaday.com/2026/11/17/newer-rigol-scopes-get-a-custom-linux-firmware/

    Looks pretty cool, and runs off the built-in SD card, so you can switch
    back to the factory software anytime.

    Cheers

    Phil Hobbs

    Impressive!



    It?s clearly not ready for prime time yet, but if the guy keeps cranking on
    it, it might be pretty useful.

    Simon has been using open-weight LLMs to build a bunch of lab automation
    stuff, which is promising.

    Cheers

    Phil Hobbs

    --
    Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From john larkin@3:633/10 to All on Tuesday, November 17, 2026 12:24:04
    On Tue, 17 Nov 2026 17:04:44 -0000 (UTC), Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:

    Jan Panteltje <alien@comet.invalid> wrote:
    Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net>wrote:

    https://hackaday.com/2026/11/17/newer-rigol-scopes-get-a-custom-linux-firmware/

    Looks pretty cool, and runs off the built-in SD card, so you can switch
    back to the factory software anytime.

    Cheers

    Phil Hobbs

    Impressive!



    It?s clearly not ready for prime time yet, but if the guy keeps cranking on >it, it might be pretty useful.

    Simon has been using open-weight LLMs to build a bunch of lab automation >stuff, which is promising.

    Cheers

    Phil Hobbs

    My kids are doing astounding stuff with Claude.

    I designed a high voltage pulse generator, with an LTC3803 booster
    feeding a voltage multiplier in the 0-700 volt supply. A filtered PWM
    from an FPGA boogers the feedback pin to set the HV level.

    One of my guys pushed the schematic and Spice model into Claude and
    asked it to compute his cal factors. Then he ran the board and asked
    for 300 volts. He got 301.

    That's terrifying. AI is unemploying coders and potentially circuit
    desigers too.


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Lasse Langwadt@3:633/10 to All on Wednesday, November 18, 2026 22:42:47
    On 11/17/26 21:24, john larkin wrote:
    On Tue, 17 Nov 2026 17:04:44 -0000 (UTC), Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:

    Jan Panteltje <alien@comet.invalid> wrote:
    Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net>wrote:

    https://hackaday.com/2026/11/17/newer-rigol-scopes-get-a-custom-linux-firmware/

    Looks pretty cool, and runs off the built-in SD card, so you can switch >>>> back to the factory software anytime.

    Cheers

    Phil Hobbs

    Impressive!



    It?s clearly not ready for prime time yet, but if the guy keeps cranking on >> it, it might be pretty useful.

    Simon has been using open-weight LLMs to build a bunch of lab automation
    stuff, which is promising.

    Cheers

    Phil Hobbs

    My kids are doing astounding stuff with Claude.

    I designed a high voltage pulse generator, with an LTC3803 booster
    feeding a voltage multiplier in the 0-700 volt supply. A filtered PWM
    from an FPGA boogers the feedback pin to set the HV level.

    One of my guys pushed the schematic and Spice model into Claude and
    asked it to compute his cal factors. Then he ran the board and asked
    for 300 volts. He got 301.

    That's terrifying. AI is unemploying coders and potentially circuit
    desigers too.


    they become "managers" of all those AI minions


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From someone@3:633/10 to All on Thursday, November 19, 2026 18:15:02
    The de facto leader in instrumented automation includes Rigol as a minuscule subset of their capabilities.

    National Instruments (NI), now operating as part of Emerson, has integrated artificial intelligence directly into its test and measurement automation portfolio through two primary initiatives: NI Nigel? AI and NI AI Extensions (MCP Servers).
    These additions move AI beyond simple chat assistance into active code authoring, sequence generation, and direct hardware control across more than 400 NI products.

    1. NI Nigel? AI (Built-In Test Assistant & Author)

    Originally introduced as an advisor, NI Nigel? AI is a test-optimized large language model embedded directly into the NI LabVIEW+ Suite. It is trained on NI?s proprietary test and measurement domain expertise to operate within engineering constraints.
    ? Prompt-Based Code & Sequence Generation. Engineers can use natural language prompts inside NI LabVIEW to generate graphical code snippets and automatically build test sequences inside NI TestStand.
    ? Cross-Platform Integration. Nigel AI is embedded across core daily engineering tools, including NI InstrumentStudio, NI FlexLogger, NI VeriStand, and NI SystemLink?.
    ? Automated Analysis & Reporting. Beyond writing code, Nigel analyzes test runs to pinpoint Device Under Test (DUT) failure points, verify measurement integrity, recommend compatible hardware modules, and generate product reports.

    2. NI AI Extensions (Model Context Protocol Servers)

    Showcased at electronica 2026, NI introduced open-standard Model Context Protocol (MCP) servers via NI AI Extensions. This allows engineers to connect external, general-purpose AI coding assistants?such as Claude Code, GitHub Copilot, Cursor, OpenAI Codex, and Google Gemini?directly to NI's hardware and software ecosystem.
    ? NI Device Drivers MCP Server. A local server that provides third-party AI tools with deep, accurate knowledge of NI hardware drivers across more than 400 DAQ and PXI instruments. Going beyond basic SCPI commands, it enables external AI agents to write accurate driver-level code, synchronize measurements, and debug hardware behavior.
    ? NI Software MCP Servers. Connects external AI assistants to the LabVIEW+ Suite and SystemLink? environments, giving AI agents the context needed to edit projects, automate software functions, and troubleshoot system architectures.

    November 19, 2026 ? The NI Software MCP servers connect compatible AI applications with context and supported capabilities for NI LabVIEW+ Suite produ...
    Show all

    Yes, you can program RIGOL oscilloscopes using National Instruments software (such as LabVIEW and NI-VISA), though native driver and AI support varies significantly depending on the oscilloscope model and whether you are using standard NI drivers or NI's new AI tools.

    1. Standard NI Software & Driver Support

    NI provides a few ways to connect and automate RIGOL instruments:
    ? Certified LabVIEW Plug and Play Drivers. NI hosts official, certified third-party drivers on the NI Instrument Driver Network (IDNet) for several legacy and mid-range RIGOL lines, including the DS1000, DS2000, DS4000, and DS6000 series. These give you pre-built Virtual Instruments (VIs) for quick drag-and-drop programming in LabVIEW.
    ? RIGOL-Supplied LabVIEW Examples. For models without an NI-certified IDNet driver (such as the DS1000Z series), RIGOL's Software Download Portal provides standalone LabVIEW demo code and VIs that communicate over USB or LAN.
    ? Direct SCPI Commands via NI-VISA. For newer series like the MSO5000, MSO8000, or DHO lines where dedicated LabVIEW drivers are often unavailable, you can still automate them in LabVIEW, Python, or C/C++ by installing NI-VISA and sending standard text-based SCPI commands directly to the scope.

    2. How NI's AI Tools Handle RIGOL Scopes

    ? NI Device Drivers MCP Server (NI Hardware Only). NI's dedicated hardware AI server is strictly trained on NI?s proprietary driver APIs (such as NI-SCOPE, NI-DAQmx, and NI-DMM) across its own PXI and DAQ hardware. It does not natively include third-party RIGOL drivers.
    ? NI Nigel? AI & SCPI Assistance. While Nigel AI cannot automatically configure a RIGOL scope inside NI InstrumentStudio (which is reserved for compatible NI hardware), it can still assist you in LabVIEW and TestStand. You can use Nigel to structure VISA read/write loops, parse waveform arrays, or help format standard SCPI command strings if you supply the RIGOL programming guide syntax.
    If you're looking to set up automation for a specific bench setup, let me know: ? Which RIGOL oscilloscope model you are using (e.g., DS1054Z, MSO5000, DHO800/900)?
    ? Your preferred programming environment (LabVIEW, TestStand, or Python)?
    I can point you to the exact driver package or help write a starter SCPI script for capturing waveforms.
    Try without personalization
    +5
    SponsoredIf you're looking to set up your programming environment, you can explore these software and driver resources from National Instruments and RIGOL.
    NI
    National Instruments? LabVIEW - Graphical Programming Software - More Time-Saving
    Software for applications that require test with rapid access to hardware & data insights.
    NI Test Hardware
    NI LabVIEW Trial Download
    RIGOL Technologies USA
    RIGOL Firmware and Drivers - RIGOL Software Download Area - Buy Direct Online Software and Firmware Download from RIGOL test and measurement instruments
    ?
    RIGOL Technologies
    Software and Firmware Download | RIGOL
    This example program retrieves trace data from Rigol Spectrum Analyzers (DSA1000, 1000A, and 800 series). Requires LabVIEW 2010 or...
    ?

    --
    For full context, visit https://www.electrondepot.com/electrodesign/open-source-os-for-rigol-scopes-4424175-.htm


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From someone@3:633/10 to All on Thursday, November 19, 2026 18:15:02
    What is " boogers the feedback pin"?

    --
    For full context, visit https://www.electrondepot.com/electrodesign/open-source-os-for-rigol-scopes-4424175-.htm


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From someone@3:633/10 to All on Thursday, November 19, 2026 18:15:02
    "That's terrifying. AI is unemploying coders and potentially circuit desigers too."

    That would be anything that doesn't require creativity or more than one (1) dimensional consideration beyond the task at hand.

    The few remaining experienced practitioners with a track record of accomplishment will be used to mentor the corporate AI instead of new hires.

    --
    For full context, visit https://www.electrondepot.com/electrodesign/open-source-os-for-rigol-scopes-4424175-.htm


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Phil Hobbs@3:633/10 to All on Thursday, November 19, 2026 23:50:18
    someone <2a59d59e3809f827ce709d3815e3950eef4a6a93af5557a93a7fdfba71460843@example.com> wrote:
    What is " boogers the feedback pin"?


    Your AI couldn?t tell you that?

    Cheers

    Phil Hobbs

    --
    Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Jan Panteltje@3:633/10 to All on Friday, November 20, 2026 05:15:24
    someone <2a59d59e3809f827ce709d3815e3950eef4a6a93af5557a93a7fdfba71460843@example.com>wrote:
    The de facto leader in instrumented automation includes Rigol as a minuscule subset of their capabilities.

    Maybe you should use AI to figure out how to limit your line length to about 80 in your
    postings on Usenet

    Not reading your stuff if you do not go normal.




    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Bill Sloman@3:633/10 to All on Friday, November 20, 2026 17:07:28
    On 18/11/2026 7:24 am, john larkin wrote:
    On Tue, 17 Nov 2026 17:04:44 -0000 (UTC), Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:

    Jan Panteltje <alien@comet.invalid> wrote:
    Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net>wrote:

    https://hackaday.com/2026/11/17/newer-rigol-scopes-get-a-custom-linux-firmware/

    Looks pretty cool, and runs off the built-in SD card, so you can switch >>>> back to the factory software anytime.

    Cheers

    Phil Hobbs

    Impressive!



    It?s clearly not ready for prime time yet, but if the guy keeps cranking on >> it, it might be pretty useful.

    Simon has been using open-weight LLMs to build a bunch of lab automation
    stuff, which is promising.

    Cheers

    Phil Hobbs

    My kids are doing astounding stuff with Claude.

    I designed a high voltage pulse generator, with an LTC3803 booster
    feeding a voltage multiplier in the 0-700 volt supply. A filtered PWM
    from an FPGA boogers the feedback pin to set the HV level.

    One of my guys pushed the schematic and Spice model into Claude and
    asked it to compute his cal factors. Then he ran the board and asked
    for 300 volts. He got 301.

    He should have asked for 300.00V and got it.

    That's terrifying. AI is unemploying coders and potentially circuit
    designers too.

    Somebody who asks for 300V and is happy to get 301V isn't any kind of potential circuit designer. It's terrifying that you didn't notice.

    --
    Bill Sloman, Sydney



    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From john larkin@3:633/10 to All on Friday, November 20, 2026 08:14:44
    On Fri, 20 Nov 2026 17:07:28 +1100, Bill Sloman <bill.sloman@ieee.org>
    wrote:

    On 18/11/2026 7:24 am, john larkin wrote:
    On Tue, 17 Nov 2026 17:04:44 -0000 (UTC), Phil Hobbs
    <pcdhSpamMeSenseless@electrooptical.net> wrote:

    Jan Panteltje <alien@comet.invalid> wrote:
    Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net>wrote:

    https://hackaday.com/2026/11/17/newer-rigol-scopes-get-a-custom-linux-firmware/

    Looks pretty cool, and runs off the built-in SD card, so you can switch >>>>> back to the factory software anytime.

    Cheers

    Phil Hobbs

    Impressive!



    It?s clearly not ready for prime time yet, but if the guy keeps cranking on >>> it, it might be pretty useful.

    Simon has been using open-weight LLMs to build a bunch of lab automation >>> stuff, which is promising.

    Cheers

    Phil Hobbs

    My kids are doing astounding stuff with Claude.

    I designed a high voltage pulse generator, with an LTC3803 booster
    feeding a voltage multiplier in the 0-700 volt supply. A filtered PWM
    from an FPGA boogers the feedback pin to set the HV level.

    One of my guys pushed the schematic and Spice model into Claude and
    asked it to compute his cal factors. Then he ran the board and asked
    for 300 volts. He got 301.

    He should have asked for 300.00V and got it.

    That's terrifying. AI is unemploying coders and potentially circuit
    designers too.

    Somebody who asks for 300V and is happy to get 301V isn't any kind of >potential circuit designer. It's terrifying that you didn't notice.

    Hitting that close was mostly luck. The switcher reference and the
    many resistors all have about 1% tolerances, so it might have been 10
    volts off.

    What astounded me is that it understood a pretty complex boost
    switcher+voltage multiplier+unusual control loop and generated good
    cal factors. We'll of course tweak the cal factors when we test
    production units.

    I fiddled the design in Spice to get the resistor values and tune the
    loop dynamics, which was easier than cranking out the math.

    Next thing like this, I'll see if Claude can do the whole design.


    John Larkin
    Highland Tech Glen Canyon Design Center
    Lunatic Fringe Electronics

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Bill Sloman@3:633/10 to All on Saturday, November 21, 2026 03:57:15
    On 21/11/2026 3:14 am, john larkin wrote:
    On Fri, 20 Nov 2026 17:07:28 +1100, Bill Sloman <bill.sloman@ieee.org>
    wrote:

    On 18/11/2026 7:24 am, john larkin wrote:
    On Tue, 17 Nov 2026 17:04:44 -0000 (UTC), Phil Hobbs
    <pcdhSpamMeSenseless@electrooptical.net> wrote:

    Jan Panteltje <alien@comet.invalid> wrote:
    Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net>wrote:

    https://hackaday.com/2026/11/17/newer-rigol-scopes-get-a-custom-linux-firmware/

    Looks pretty cool, and runs off the built-in SD card, so you can switch >>>>>> back to the factory software anytime.

    Cheers

    Phil Hobbs

    Impressive!



    It?s clearly not ready for prime time yet, but if the guy keeps cranking on
    it, it might be pretty useful.

    Simon has been using open-weight LLMs to build a bunch of lab automation >>>> stuff, which is promising.

    Cheers

    Phil Hobbs

    My kids are doing astounding stuff with Claude.

    I designed a high voltage pulse generator, with an LTC3803 booster
    feeding a voltage multiplier in the 0-700 volt supply. A filtered PWM
    from an FPGA boogers the feedback pin to set the HV level.

    One of my guys pushed the schematic and Spice model into Claude and
    asked it to compute his cal factors. Then he ran the board and asked
    for 300 volts. He got 301.

    He should have asked for 300.00V and got it.

    That's terrifying. AI is un-employing coders and potentially circuit
    designers too.

    Somebody who asks for 300V and is happy to get 301V isn't any kind of
    potential circuit designer. It's terrifying that you didn't notice.

    Hitting that close was mostly luck. The switcher reference and the
    many resistors all have about 1% tolerances, so it might have been 10
    volts off.

    1% of 300V is three volts. Resistor tolerances can be positive or
    negative so if there were lots of resistors in the train the net effect
    starts averaging out. Tolerancing a circuit is a specific task - at
    least in places that take circuit design seriously.

    What astounded me is that it understood a pretty complex boost switcher+voltage multiplier+unusual control loop and generated good
    cal factors. We'll of course tweak the cal factors when we test
    production units.

    I fiddled the design in Spice to get the resistor values and tune the
    loop dynamics, which was easier than cranking out the math.

    Next thing like this, I'll see if Claude can do the whole design.

    You'd have to tell it what you wanted in unambiguous terms, which turns
    out to be skill that improves with practice.

    --
    Bill Sloman, Sydney


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From john larkin@3:633/10 to All on Friday, November 20, 2026 09:28:05
    On Sat, 21 Nov 2026 03:57:15 +1100, Bill Sloman <bill.sloman@ieee.org>
    wrote:

    On 21/11/2026 3:14 am, john larkin wrote:
    On Fri, 20 Nov 2026 17:07:28 +1100, Bill Sloman <bill.sloman@ieee.org>
    wrote:

    On 18/11/2026 7:24 am, john larkin wrote:
    On Tue, 17 Nov 2026 17:04:44 -0000 (UTC), Phil Hobbs
    <pcdhSpamMeSenseless@electrooptical.net> wrote:

    Jan Panteltje <alien@comet.invalid> wrote:
    Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net>wrote:

    https://hackaday.com/2026/11/17/newer-rigol-scopes-get-a-custom-linux-firmware/

    Looks pretty cool, and runs off the built-in SD card, so you can switch >>>>>>> back to the factory software anytime.

    Cheers

    Phil Hobbs

    Impressive!



    It?s clearly not ready for prime time yet, but if the guy keeps cranking on
    it, it might be pretty useful.

    Simon has been using open-weight LLMs to build a bunch of lab automation >>>>> stuff, which is promising.

    Cheers

    Phil Hobbs

    My kids are doing astounding stuff with Claude.

    I designed a high voltage pulse generator, with an LTC3803 booster
    feeding a voltage multiplier in the 0-700 volt supply. A filtered PWM
    from an FPGA boogers the feedback pin to set the HV level.

    One of my guys pushed the schematic and Spice model into Claude and
    asked it to compute his cal factors. Then he ran the board and asked
    for 300 volts. He got 301.

    He should have asked for 300.00V and got it.

    That's terrifying. AI is un-employing coders and potentially circuit
    designers too.

    Somebody who asks for 300V and is happy to get 301V isn't any kind of
    potential circuit designer. It's terrifying that you didn't notice.

    Hitting that close was mostly luck. The switcher reference and the
    many resistors all have about 1% tolerances, so it might have been 10
    volts off.

    1% of 300V is three volts. Resistor tolerances can be positive or
    negative so if there were lots of resistors in the train the net effect >starts averaging out.

    Sensible engineering doesn't assume that component tolerances will
    average out. Thinking like that kills people.




    John Larkin
    Highland Tech Glen Canyon Design Center
    Lunatic Fringe Electronics

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Bill Sloman@3:633/10 to All on Saturday, November 21, 2026 05:15:55
    On 21/11/2026 4:28 am, john larkin wrote:
    On Sat, 21 Nov 2026 03:57:15 +1100, Bill Sloman <bill.sloman@ieee.org>
    wrote:

    On 21/11/2026 3:14 am, john larkin wrote:
    On Fri, 20 Nov 2026 17:07:28 +1100, Bill Sloman <bill.sloman@ieee.org>
    wrote:

    On 18/11/2026 7:24 am, john larkin wrote:
    On Tue, 17 Nov 2026 17:04:44 -0000 (UTC), Phil Hobbs
    <pcdhSpamMeSenseless@electrooptical.net> wrote:

    Jan Panteltje <alien@comet.invalid> wrote:
    Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net>wrote:

    https://hackaday.com/2026/11/17/newer-rigol-scopes-get-a-custom-linux-firmware/

    Looks pretty cool, and runs off the built-in SD card, so you can switch
    back to the factory software anytime.

    Cheers

    Phil Hobbs

    Impressive!



    It?s clearly not ready for prime time yet, but if the guy keeps cranking on
    it, it might be pretty useful.

    Simon has been using open-weight LLMs to build a bunch of lab automation >>>>>> stuff, which is promising.

    Cheers

    Phil Hobbs

    My kids are doing astounding stuff with Claude.

    I designed a high voltage pulse generator, with an LTC3803 booster
    feeding a voltage multiplier in the 0-700 volt supply. A filtered PWM >>>>> from an FPGA boogers the feedback pin to set the HV level.

    One of my guys pushed the schematic and Spice model into Claude and
    asked it to compute his cal factors. Then he ran the board and asked >>>>> for 300 volts. He got 301.

    He should have asked for 300.00V and got it.

    That's terrifying. AI is un-employing coders and potentially circuit >>>>> designers too.

    Somebody who asks for 300V and is happy to get 301V isn't any kind of
    potential circuit designer. It's terrifying that you didn't notice.

    Hitting that close was mostly luck. The switcher reference and the
    many resistors all have about 1% tolerances, so it might have been 10
    volts off.

    1% of 300V is three volts. Resistor tolerances can be positive or
    negative so if there were lots of resistors in the train the net effect
    starts averaging out.

    Sensible engineering doesn't assume that component tolerances will
    average out.
    It certainly doesn't. But in any individual case it probably will.

    We were talking about a specific example of a circuit. Tolerancing adds
    up all the worst case deviations to get the worst cases possible - two
    very improbable worst cases, but if you can deal with them you can deal
    with anything in between.

    After you have put a circuit into production you have the luxury of
    seeing what random variation actually does, but if your 5% resistors
    came from a batch which had been automatically sorted into +/-1% and
    +/-2% bins and the +/-5% bins got the parts that didn't make into the
    +/-1% and +/-2% bins, their distribution is gong to be a bit odd.

    Thinking like that kills people.

    You don't do a lot of thinking, and when you try to do it you don't do
    it all that well.

    --
    Bill Sloman, Sydney

    --
    Bill Sloman, Sydney


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From john larkin@3:633/10 to All on Friday, November 20, 2026 11:22:23
    On Thu, 19 Nov 2026 18:15:02 +0000, someone <2a59d59e3809f827ce709d3815e3950eef4a6a93af5557a93a7fdfba71460843@example.com> wrote:

    What is " boogers the feedback pin"?


    https://www.dropbox.com/scl/fi/w190wul8nctd41r51e9ag/R250_HV_Supply.jpg?rlkey=vf967i4lf5zcsxw8salog4okc&dl=0

    The control input is a 3.3 volt PWM from an FPGA. It sort of works
    backwards.

    This can probably make 1000 volts, but we can't find GaN fets that go
    that high. There seems to be some physics stopping them all at 650.


    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)