ADC "stacking"

Jim Thompson skrev:

snip

I believe it is based on paracetamol?, I don't know if double is enough but the limit between enough and a too much is slim. and afaik if you get too much you won't notice untill after a few days and then it is probably too late do anything about it.....

Here only pharmacies are allowed to sell you more than 10 a day

-Lasse

Hello Fred,

Respectfully disagree. Made it work, several times. Not just on the lab bench but also for production runs of many years. The trick is to not randomize but auto-adjust. You are right, all three parameters you mentioned (gain, offset, aperture uncertainty) matter. But they can be adjusted. The only requirement is that there be a time slot, preferably every so many hundred msec, where that adjustment can automatically take place. In a video system for example that is no big deal since there is a whole lot of idle time in the frame reset window or V-sync. The same goes for sonar, ultrasound, radar, you name it. Adjusting only once at system start could be iffy.

I found that the quantization errors were not much of an issue and most of the time I did use Analog Devices converters. Some day I want to try that with converters like the Texas ADS5500. At least when it comes off its $50 plinth.

Regards, Joerg

I've been assuming that the ESD protectors were somehow tuned to pass the test for ESD protectors being there but that they really don't protect.

I'll have to try touching one with my finger after I get out of my chair and see if it still works. This test will have to wait for the next bit of free time I get.

In article , Joerg wrote: [...]

Back in the mists of time I did something very like that. The "subtracting node" is a real trouble maker. You need a fairly exact ratio, and quite a good bandwidth to make it work. The linearity of the DAC can also be a problem.

The ratio and linearity issues I handled by applying a reference signal one out of every N converts. The reference was a very smooth ramp. Inside the DSP, I kept a table indexed by the values going to the DAC and holding the amount to add to the second ADC's output. This greatly relaxes the DAC and subtract requirements but requires a few minutes before the first measurement is taken.

Don't do that. ESD diodes are there for SHIPPING and controlled handling, and can be as low as 200V on low capacitance pins. Once in a board MOST IC's don't need ESD protection.

...Jim Thompson

Why not just order a sample? I just got an ADS5444 that is a 13-bit 250MSps. You can only get 1 at a time but might be worth spending some time gathering a couple to try it? Would be very nice to get atleast 1Gsps by interleaving a couple of these as it should give me a 500mhz bandwidth which would be awesome.

Jon

Hello Ken,

It's been a long time ago since I designed in the last SD5400 but I was always under the impression that they didn't have any ESD coverage to speak of.

I had a tech who almost threw his coffee mug at me after several that he soldered in were dead. He was sure my circuit didn't work. Then I soldered one in myself and it worked, and explained to him how to do a

I wouldn't do that test. You can damage them completely or just a little. Then they still work but not to spec anymore.

Regards, Joerg

Hello Jon,

That is an option. Or I could just buy some. But I tend to only do things when there is a purpose, like a product that this solution could go into.

Just imagine 10 of those in a stash ;-)

Regards, Joerg

In article , Jim Thompson wrote: [....]

Spoil sport. I wasn't even going to use the 3KV 100mA supply for it.

And more reason might have led him to a solution after 10. Edison was notoriously math and science-phobic. He actually seems to not have had a good feeling for Ohm's Law, and others had a hard time convincing him that low-voltage generation and distribution wouldn't be economical at large scales.

But he gets an "E" for energy.

John