What changes when a Rigol DHO800/900 or MHO900 runs NovaOS instead of Rigol's firmware, line by line: where it is better, what is new, where it matches, and where stock is still ahead. Every line was measured on our own DHO924S or MHO934, the same way on both systems.
32 better or new3 the same4 stock ahead4 not measured yet
better than stockMHO900 MHO934: with the scope running, NovaOS has about 257 MiB of RAM in use against stock's 535-542 MiB on the same unit (MemTotal - MemAvailable, 60 s after the screen, three boots each).
for informationDHO + MHO NovaOS replaces the scope's Android 7.1 / Linux 4.4 system with Linux 6.12, systemd and a scope engine and UI written in Rust; the same image runs on the DHO800/DHO900 and the MHO900.
Boot
better than stockDHO800/900 DHO924S: a reboot reaches the scope screen about 12.5 s after the kernel starts; Rigol's firmware on the same unit takes 44-50 s (ten logged stock boots).
better than stockDHO800/900 DHO: the acquisition FPGA's bring-up takes about 2.5 s, where stock waits through fixed delays of about 16.5 s.
better than stockMHO900 MHO934: a reboot reaches the scope screen about 10.4 s after the kernel starts; Rigol's firmware on the same unit takes about 48 s (median of ten timed stock boots, 47.6-52.7 s).
not measured yetDHO800/900 The first boot after a NovaOS update takes longer (DHO r44: 29.5 s) because the new system is verified byte for byte before it runs; later boots skip the check. Stock's own time after a firmware update is not measured yet.
Acquisition and display
only in NovaOSMHO900 Remote view: about 76 ms from a change on the scope to its picture leaving the scope, hardware H.264.
matches stockDHO800/900 Trigger rate at slow timebases matches stock (25 a second at 1 ms/div on a 50 Hz signal).
matches stockDHO800/900 Segmented recording holds as many frames as stock: 120, 960 and 7,680 at 1 Mpts, 100 kpts and 10 kpts.
stock is aheadDHO800/900 Rigol advertises up to 1,000,000 waveforms/s in its UltraAcquire mode; NovaOS's fastest mode measured 34,498 waveforms/s on the DHO924S. We have not yet measured stock's rate on our unit, and we expect to trail it here.
not measured yetDHO800/900 DHO: 31,509 waveforms a second at 20 ns/div into the density display; stock's rate at this setting is not measured yet, because it has no rate readout.
not measured yetDHO800/900 DHO: 10,702 waveforms a second at 2 µs/div into the density display; stock's rate at this setting is not measured yet.
not measured yetDHO800/900 DHO: 3,575 waveforms a second with all four channels in the density display at 2 µs/div; stock's rate is not measured yet.
Signal quality
better than stockDHO800/900 DHO: the trigger point sits within one sample in the raw record; on stock it comes up to 40 ns early.
better than stockMHO900 MHO: even with no stored calibration, NovaOS's start-up core equalisation brings the fs/4 interleave spur to 0.65-1.5 mVpk, against 4.0-8.1 mVpk on stock across boots.
better than stockMHO900 MHO934: with NovaOS's own stored self-calibration the 2 GSa/s interleave spurs measured 0.13-1.22 mVpk across restarts, against 2.3-8.1 mVpk on stock across boots.
better than stockMHO900 MHO at 500 MSa/s: the line where its two ADCs meet (fs/2) measures 0.23-1.02 mVpk, against 0.9-5.3 mVpk on stock from boot to boot.
better than stockMHO900 MHO: NovaOS keeps its own self-calibration result, written atomically and kept across updates, and never writes Rigol's calibration partition.
matches stockMHO900 Without any calibration, NovaOS's MHO interleave spurs sit inside stock's own boot-to-boot spread (stock: 2.3-7.1 mVpk at fs/8, 4.0-8.1 mVpk at fs/4 over four boots).
stock is aheadMHO900 MHO: the open-input DC offset at 1 V/div is larger than stock's today (-17.5 mV against -8.7 mV at 1 MΩ; -13.1 against -11.7 mV at 50 Ω).
Features stock lacks or sells separately
only in NovaOSDHO + MHO No option tiers: NovaOS is its own engine that drives the FPGA directly, so every feature the hardware has is available on every unit; it never reads or alters Rigol's option keys or licence files.
only in NovaOSDHO + MHO 15 bus decoders: stock's nine plus DMX512 with RDM, Modbus RTU and ASCII, 1-Wire, SWD, JTAG and USB-PD.
only in NovaOSDHO + MHO Eye diagram and jitter analysis (TIE, RJ/DJ/PJ, bathtub, mask templates) on every model; on stock they are paid options.
only in NovaOSDHO + MHO Identify names the signal on a channel (DC, sine with its THD, square, PWM, pulse train, ramp, noise, burst, AM, FM, serial data) and offers the matching trigger, decoder or eye; stock has no equivalent.
only in NovaOSDHO + MHO Remote view and touch control from a paired display, about 0.1 s behind the scope; stock offers a browser page only.
only in NovaOSDHO + MHO NovaOS comes with an AI agent interface (an MCP server on the PC) that can describe, measure, capture, decode and screenshot the scope through its normal SCPI port, with calibration, firmware and other risky commands refused; Rigol's firmware has no such interface.
only in NovaOSMHO900 MHO: both built-in generator channels work on every MHO900, with Bode on either, up to 20 Vpp; on stock they depend on an installed option.
better than stockDHO + MHO NovaOS updates are signed and installed over the network or from a USB stick; a release that fails to start is rolled back automatically (tested on our DHO924S: update in 3 min, automatic fallback in 4.5 min).
stock is aheadDHO + MHO Screen languages: 10 of stock's 13 today, all our own translations; Simplified Chinese is being added.
Stock defects NovaOS fixes or avoids
better than stockDHO + MHO SCPI queries return real values where stock answers 1E+24 placeholders, errors on settings it accepts, or drops the connection.
better than stockDHO800/900 DHO: segmented-record frame times are reported correctly; stock reads them 3.2 times too short (it counts 250 ps ticks on an 800 ps counter).
better than stockDHO800/900 DHO: an exported FFT waveform carries its x step and origin in Hz; stock's export mixes Hz and seconds.
better than stockDHO800/900 DHO: Per.Vrms is the RMS over one period; stock returns the top level instead (1.92 V on a 1 kHz square whose period RMS is 1.55 V).
better than stockDHO800/900 DHO: the board check no longer shows a -273 °C FPGA temperature (the FPGA image has no such sensor reading); NovaOS watches the SoC and GPU temperatures from the first check.
better than stockDHO800/900 DHO: a network time update no longer restarts a healthy scope; the software is supervised on a monotonic clock.
better than stockDHO800/900 DHO: the power key shuts the scope down cleanly; on stock a press could crash its launcher.
better than stockMHO900 MHO: no false 'ADC Clk Freq: Fail' at every start; stock compares the MHO's ADC clock count against the DHO's.
better than stockMHO900 MHO: every start checks that the FPGA configured, that the sample clock locked and that the board's own clock table loaded; stock's loader reports success even when the FPGA did not configure.
Robustness and maintenance
better than stockDHO + MHO Rigol's firmware never shuts down cleanly (the power is cut), so every stock boot we captured replays two file-system journals; NovaOS shuts down cleanly and leaves the stock partitions clean.
better than stockDHO + MHO Rigol's firmware rewrites a factory calibration file at every start; NovaOS mounts the calibration partition read-only and never writes it.
better than stockDHO + MHO Stock exposes an unauthenticated root debug port (adb on TCP 55555) on the network; NovaOS ships with SSH off, keys only, and HTTPS for the web pages.
better than stockDHO800/900 DHO: no per-boot log copies; stock writes 0.5 to 5 MB of logs at every boot.
better than stockDHO800/900 DHO: if one part of the scope software fails, only that part restarts, from about 1 s; stock relaunches its whole app after 17 to 26 s and the USB instrument drops off the bus meanwhile.
stock is aheadDHO800/900 Switching from NovaOS back to Rigol's firmware passed 16 clean cycles on the MHO; the same soak test on the DHO is still to run.
This page is generated from our comparison ledger, which keeps both sides' values, the method and the evidence for each line. When a measurement changes, the page changes with it. Questions about any line: [email protected].