Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Rohde and Schwarz AMIQ Modulation Generator Teardown

tomverbeure.github.io

Post reply on HN
Print view View original

21–27 of 27 posts

Re: Rohde and Schwarz AMIQ Modulation Generator Teardown

#21

Post by aidenn0 »

It would be interesting to benchmark this against consumer USB DACs that are marketed for sound. You can get 32-bit 192kHz bandwidth (384kHz sampling rate) DACs now for under $200. Obviously this is much lower bandwidth than the R&S, but I'd be interested to see the dynamic range of a modern DAC when approaching Nyquist; I don't have a strong belief one way or the other over which would win, given that 14-bits of dynamic range is about where the integrity of your analog path starts to matter a whole lot.

[edit]

Sound Blaster claims "up to" 130dB of dynamic range on theirs. Putting aside how much work the "up to" id doing, this would make the last 9 or 10 bits completely useless, even for undithered signals, but I was thinking of something cheaper, by one of those fly-by-night companies with names made of random letters.

Re: Rohde and Schwarz AMIQ Modulation Generator Teardown

#22

Post by CamperBob2 »

teleforce wrote: post #16

Wonderful article and I don't even know that this device existed perhaps due to Agilent/Keysight bias laboratory/shop. R & S AMIQ Modulation Generator general specs is available here [1]. Fun facts, the venerable Silicon Valley was started by Hewlett-Packard (HP, then Agilent, now Keysight) that mainly supplying function/signal/waveform generator for the then booming radar and electronics industry. HP Garage is now a…

Their first customer was actually Disney, who needed a wide-range audio oscillator to test sound hardware that would be used in theaters showing 'Fantasia.'

The oscillator's model number was 200A, because even back then they understood the SV maxim "Nobody wants to buy the first version of anything."

Re: Rohde and Schwarz AMIQ Modulation Generator Teardown

#23

Post by tverbeure »

dusted wrote: post #17

I'm trying really hard not to want to turn this into a dac for my stereo system. They build really nice boxes.. I have a PSL-3 industrial controller (well, two, but one is broken) on the shelf behind me that I'm using for the noble purpose of playing Theme Hospital in Windows 98.

It's not impossible to hack the hardware to do this: the output of the SDRAMs go to some 74-series chips which drive both the DAC and the connector for the 16-bit output.

You could remove the SDRAM path altogether, reverse the signal path from the connector with some PCB surgery and drive the DAC from the connector with your own digital hardware. That's one way...

There's also a hardware path from the ISA bus straight to the DAC, bypassing the SDRAM, for diagnostic purposes probably. So there might by options there too, but though it would require a ton of reverse engineering or even reprogramming the FPGAs (which are so old that the software isn't available anymore on the Altera website.)

Whether or not all of that is worth doing for a 14-bit DAC output is a matter of personal opinion.

Re: Rohde and Schwarz AMIQ Modulation Generator Teardown

#24

Post by tverbeure »

myself248 wrote: post #19
lnsru wrote: post #18

Earlier quoted context omitted.

Many modern measurement hardware is built this way - motherboard with custom PCIe boards. It dramatically simplifies GUI design using standard tools and keeps the complexity separated on the external boards. The PCIe boards also can be designed for one/two measurement channels making the equipment rather modular and customizable.

Yeah, but PCIe is a bit harder to DIY an interface for! ;)

I haven't done it myself... yet... but these days it's not that horrible to DIY PCIe with an FPGA. There's even a fully open source PCIe core, and there are example designs for it: https://github.com/enjoy-digital/litepcie.

Re: Rohde and Schwarz AMIQ Modulation Generator Teardown

#25

Post by jacquesm »

HeyLaughingBoy wrote: post #20
lnsru wrote: post #18

Earlier quoted context omitted.

Many modern measurement hardware is built this way - motherboard with custom PCIe boards. It dramatically simplifies GUI design using standard tools and keeps the complexity separated on the external boards. The PCIe boards also can be designed for one/two measurement channels making the equipment rather modular and customizable.

A lot of hardware is built like this. Separating the GUI from the real time or safety-critical aspects of the system is a common enough need that there are many System-on-Module (SoM) boards that can run Linux or Windows and also include an ARM processor for real time behavior on the same die. Your UI can be built in Python/Qt/Tk while the safety critical stuff is programmed in C running on an RTOS.

I think you meant 'time critical'.

Re: Rohde and Schwarz AMIQ Modulation Generator Teardown

#26

Post by HeyLaughingBoy »

jacquesm wrote: post #25
HeyLaughingBoy wrote: post #20

Earlier quoted context omitted.

A lot of hardware is built like this. Separating the GUI from the real time or safety-critical aspects of the system is a common enough need that there are many System-on-Module (SoM) boards that can run Linux or Windows and also include an ARM processor for real time behavior on the same die. Your UI can be built in Python/Qt/Tk while the safety critical stuff is programmed in C running on an RTOS.

I think you meant 'time critical'.

I meant safety critical, although a similar design decision is often made for time critical behavior.

In some cases, e.g., for IEC-62304 compliance, software is designated as various safety classes based on likelihood of causing Harm. If you can extract the more safety critical software from the rest of the system and prove that it is sufficiently segregated (e.g., by putting it on a separate processor), you can substantially reduce your Verification and documentation burden.

Re: Rohde and Schwarz AMIQ Modulation Generator Teardown

#27

Post by jdsully »

tverbeure wrote: post #23
dusted wrote: post #17

I'm trying really hard not to want to turn this into a dac for my stereo system. They build really nice boxes.. I have a PSL-3 industrial controller (well, two, but one is broken) on the shelf behind me that I'm using for the noble purpose of playing Theme Hospital in Windows 98.

It's not impossible to hack the hardware to do this: the output of the SDRAMs go to some 74-series chips which drive both the DAC and the connector for the 16-bit output. You could remove the SDRAM path altogether, reverse the signal path from the connector with some PCB surgery and drive the DAC from the connector with your own digital hardware. That's one way... There's also a hardware path from the ISA bus straigh…

A 14-bit dac at 105msamples/s with delta sigma modulation means you can get an effective bit depth of ~50 bits @20khz which is a ridiculous amount of dynamic range. Far more than what is implied by "just" 14-bits.
Post reply on HN