Is this right? 1125MHz 14-bit dual channel acquisition

mandag den 1. november 2021 kl. 22.25.51 UTC+1 skrev Fred Bloggs:

Last time I checked a 14-bit acquisition board costs a LOT more than that. I'm starting to suspect that's a Chinese "M" meaning x1000, western "k."

I'm maybe missing something important, but for me the specs page says

"Sample rate (real time): 40MS/s for 1410-40, 105MS/s for 1410-105,

125MS/s for 1410-125 Input impedance: 1MΩ||18pF Analog bandwidth: 1410-40: 20 MHz @ 3dB, 8 MHz @ 0.5dB, 4 MHz @ 0.1dB 1410-105: 70 MHz @ 3dB, 30MHz @ 0.5dB, 20 MHz @ 0.1 dB 1410-125: 70 MHz @ 3dB, 30MHz @ 0.5dB, 20 MHz @ 0.1 dB"

Cheers

Phil Hobbs

and the pictures show an AD9648, a dual 14bit, 125125 MSPS ADC

it is only the ADC and front end, you need one of these to use it:

On a sunny day (Mon, 1 Nov 2021 14:25:48 -0700 (PDT)) it happened Fred Bloggs snipped-for-privacy@gmail.com wrote in snipped-for-privacy@googlegroups.com:

Seems expensive for what you get, for double that you can get a real scope with probes etc. I mean 14 bits and 3dB error at 70 MHz, what is the point?

Ok, thanks. Those are still really good performance numbers for a 14-bit.

This is a National Instruments company, they know what they're doing. Apparently they see a growing demand for embedded measurement, and these kinds of products facilitate it. Here's a kind of nutty description of what they're trying to do:

On a sunny day (Tue, 2 Nov 2021 11:27:16 -0700 (PDT)) it happened Fred Bloggs snipped-for-privacy@gmail.com wrote in snipped-for-privacy@googlegroups.com:

Nothing new, I have an old Digilent board with what was it a Xilinx Spartan 2? Used it 15 years ago to brute force things (why I bought it) and do analog TV processing with it. Added an 8 bit FLASH ADC for video.. Card reader stuff (for the hack). Also had a simple frequency counter:

add-on board: think the black thing is the card reader... Pay TV hacking. Of course I stopped hacking when it became illegal, or rather some politician woman in the EU started screaming 'mouse' and called Tom the cat (but Jerry always wins) like in that Tom and Jerry comic strip... So now 14 bit, am sure even that was no problem back then. In 78 I build a video digitizer with normal logic... Digilent was expensive even then BTW. Let's Go Brandon!

All sorts of instruments can profit from that sort of performance.

Cheers

Phil Hobbs

On a sunny day (Wed, 3 Nov 2021 12:59:47 -0400) it happened Phil Hobbs snipped-for-privacy@electrooptical.net wrote in snipped-for-privacy@electrooptical.net:

When you start thinking in decibels 3 dB is down to about 70%

2^14 = 16384 70 % of 16384 = 11468.8 difference is 16384 - 11468.8 = 4915.2 That is close to 2^12 = 4096 So the thing screws up 12 bits or is basically worth 2 bits at that frequency for signals that fast. In other words 11111111111111 11X? ^ | 4096

Then you can better get a 100 MHz 8 bit flash ADC for peanuts. For DC that 14 bit thing should work.. Although there are plenty of good 16 bit audio ADCs I think for a fraction of that money.

So yes, probably some application exists for it but I aint buying for that money. My math is strange I but neural net beep

Straw man.

Everything has a rolloff. If you can't manage it, just use a narrower signal bandwidth.

The additive quantization noise of an 8-bit ADC is 36 dB worse. For ham radio things and oscilloscopes, they're fine, but high dynamic range measurements need fast, high resolution ADCs.

Cheers

Phil Hobbs

the same guys have a complete instrument for about the same as the motherboard needed for that adc module

On a sunny day (Wed, 3 Nov 2021 18:17:24 -0400) it happened Phil Hobbs snipped-for-privacy@electrooptical.net wrote in snipped-for-privacy@electrooptical.net:

Yes and no, could have put it better.

Many if not most analog video digitizing circuits used 8 bit flash ADCs The bandwidth needed was than < 10 MHz, maybe < 5 MHz, I was using a TDA8708A (video input processor with 8 bit ADC and 32 MHz sampling rate) in that FPGA project from around year 2009.

it is one the top board, sorry for bad picture, had hard time finding any picture from it. connectors right board are: video in, video out... Must still have that stuff somewhere... After digital video came here in 2006 things moved to 100% sofware...

The Vital Industries NTSC special effects system (SqueezeZoom) used a 12 bit ADC, in the mid '80s. It was a custom hybrid, that still sold for over $1400 in 1988. It wasn't a cheap system. Studios were waiting up to a year to sped $250,000 on each of them, as fast as they could be built. When used with our 1" Sony decks and RCA TK46A cameras we had the cleanest video in the Orlando market. The low video quality of a couple network stations was embarrassing. Digital TV was still a decade away, but they wouldn't spend money for anything that was beyond just staying on the air.

On a sunny day (Thu, 4 Nov 2021 04:51:57 -0700 (PDT)) it happened Michael Terrell snipped-for-privacy@gmail.com wrote in snipped-for-privacy@googlegroups.com:

I left broadcasting in 1976 it was I think, so have not worked with any of that stuff, we had Ampex VR2000 and AVR1 quadruplex, many of those. AVR1 had a sort of digital time base compensator but do not remember any ADC DAC details.. or did it not just switch analog delay lines? Ah, google knows: AMPEX AVR-1 2" Quad VTR, Accessorie & Part - BroadcastStore.com model_detail The servos would also allow playback of a tape without a control track. The timebase corrector used a series of glass delay lines for dropout compensation and ... Thing was called 'the buffer' IIRC.

Nice old tech! I once modified a Philips LDL1000 BW video tape recorder by adding 64 uS PAL glass delay lines so it could record and play color by sequentially recording lines in red, green and blue. And also did that by recording the down converted chroma as 'bias' on the FM video modulation, same system as Sony Umatic after they showed one in the studio ? 1973? Analog days! Not an ADC in sight!

Seems expensive for what you get, for double that you can get a real scope

So what? Video is a very low-SNR application with no serious interfering signals (once you get to the video signal, that is). Things like NMR are quite different--you're looking for a small signal in a lot of noise. Lots of optical measurements just have a huge dynamic range.

Using dither to get N more bits is exponentially slow--even granting a sufficiently clean clocking and dithering system, extending your 100 MHz

8-bit ADC to get six extra bits would slow it down to 400 kHz or thereabouts. (Using a high-resolution DAC to apply deterministic dither would help, but you still have to fight the 8-bit quantization noise.)

Cheers

Phil Hobbs

On a sunny day (Thu, 4 Nov 2021 10:36:04 -0400) it happened Phil Hobbs snipped-for-privacy@electrooptical.net wrote in snipped-for-privacy@electrooptical.net:

Yes, but why not use 14 fast comparators? I think you have lots of experience with low offset stuff?

What would you use them for? To make it a true flash converter, you need 2**14-1 comparators, not 14.

Cheers

Phil Hobbs