Favourite op amp for driving ADCs?

So I'm building this board that has a very nice shiny ADS8319 ADC (16-bit, 500 ks/s, good specs). The input is capacitive, of course, so it needs some fairly beefy driver. For slower things I usually use the AD8605 series, which is an RRIO with a closed-loop output stage, but it's on the slow side for this, and its settling time is a bit on the slow side (typ 1 us to 0.01%).

Any faves you could suggest that won't break the bank?

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs
Loading thread data ...

I've used the OPA365 to drive the ADS8319 ADC.

A word of caution with this ADC. We've experienced some Manufacturing fallo ut with this ADC failing settling time. The ADC seems to have a falling edg e settling issue that older date codes do not exhibit. The manufacturer wil l not confirm the issue and claims the device works fine, however they did confirm their wafer process did change for this device (hmmm). Anyway, we a re in the process of designing this ADC out and replacing it it the ADS8862 .

Reply to
gravpoet

Good info, thanks. How bad is the problem?

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

If you only need settling accuracy to 2LSB in 16-bits then the part may be fine to use.

The issue I experienced with the ADS8319 is the time it take the device to transition from 2LSB to 1LSB settling accuracy, which can be as long as 15uS. Similar testing with the AD7988-5 and ADS8862 did not produce this issue.

Reply to
gravpoet

LMH6702 is a jellybean part around my house. Fast, quiet, well-behaved, if you don't mind using a current-feedback part.

-- john, KE5FX

Reply to
John Miles, KE5FX

Thanks. My prejudice is that CFAs have really crappy settling behaviour, though--the ones I've used were great to 10 bits but junk at higher accuracy. Are these ones better than that?

Thanks again

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

The data sheet claims 13.4 ns to 0.1%, so that's about 10 bits, I guess. For a 500 ks/s ADC you'd have some time to spare.

They actually include a graph of settling time versus load C and series R, so somebody at least cared enough about that behavior to characterize it.

-- john, KE5FX

Reply to
John Miles, KE5FX

LTC6362:

1mA Supply Current n Single 2.8V to 5.25V supply n Fully Differential Input and Output n 200?V Max Offset Voltage n 260nA Max Input Bias Current n Fast Settling: 550ns to 18-Bit, 8VP-P Output n Low Distortion: ?116dBc at 1kHz, 8VP-P n Rail-to-Rail Inputs and Outputs n 3.9nV/?Hz Input-Referred Noise n 180MHz Gain-Bandwidth Product n 34MHz ?3dB Bandwidth n Low Power Shutdown: 70?A

I don't work for LTC but I like their bits !

I'm using it with an LTC2378 - design in early stages - board due back soon so my affection for this part is entirely theoretical :-)

Michael Kellett

Reply to
MK

Is AN71 useful? (Get all bits you paid for)

Reply to
haiticare2011

That's interesting. Can you walk me through how you prototype with this mso p chip?

Reply to
haiticare2011

As an open question to ADC develpers here, how do you plan to transfer the ~1 mByte/sec data over to a PC? (ie the ADC output?)

Reply to
haiticare2011

That's familiar stuff, thanks. The main thing is to find something closer in cost to the AD8605 ($1.50 in qty 25) but that can run off at least a 5.5V supply. (I'm running it from +5 and -0.3 [i.e. one Schottky diode drop] to keep it well behaved near zero volts.)

The OPA365 is recommended in the ADC datasheet, but would run too close to its limits there (abs max VDD is 5.5V). The reference voltage is

4.096, so in a pinch I could put another diode in series with VDD, I suppose.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

In my case, it won't be running continuously at top speed. It's for the MCU/power supply board that my pair of hunchbacks are building, for an advanced laser noise canceller. It'll mostly be used for tweaking stuff--the analogue board needs eleven, count 'em, eleven tweaks to work at its best:

Vos Ios R_ee' (log conformance) Prebias current balancing (2x)

three jwRCs (detector capacitance) three -omega**2 LCs (wiring inductance and reactive input impedances)

Target specs are DC-10 MHz BW on both linear and log outputs, and 70 dB of noise cancellation throughout the bandwidth. That requires some pretty stiff specs on the photodiodes, e.g. centre-to-edge propagation delay variations of less than 30 picoseconds, which is tough to do unless the PD is either (a) really small or (b) a Schottky type.

It depends sufficiently critically on the properties of three or four boutiquey devices that I'm going to make a lifetime buy when I find out roughly how many might sell in that lifetime. I also want a few boards to use myself!

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Please don't hijack threads. If you want to know how to prototype MSOP (0.65mm spacing) then start your own thread. This one's about buffer amps, not prototyping.

Clifford Heath.

Reply to
Clifford Heath

You do know that this is the Usenet, right?

Reply to
krw

Worse than that, this is s.e.d. It's a bit like asking Somalian pirates if they know they are in the Arabian Sea.

--

Rick
Reply to
rickman

SED is not unique.

Reply to
krw

Breakout boards, e.g. the Bellin Dynamic Systems ones or else Schmartboards.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

No bother - I hardly ever breadboard anything now - I can get 4 layer boards cheap in 48 hours (no solder resist or silk screen) or full spec and cheap in 7 days. These modern fast fancy bits don't work except on boards anyway - the ADC clocks out the data at 70MHz max (64MHz on my board) and the FPGA it talks to is in a 0.5mm pitch BGA. So - to prototyping:

If it has leads and .4mm pitch or more I can hand solder under a microscope, currently use an ERSA iron with a hollow tip (ie a sort of cavity in the tip.) Not really necessary, at a pinch I can use pretty much any iron with the right size tip - the trick is to use the right flux (I use a very runny (like water) flux from Warton Metals called Future 315). The knack is to run a blob of solder across the pins while the surface tension keeps the blob on the iron and the pins end up not shorted. If the last pair of arow are shorted you used abit too much solder to you can take the surplus off with solder braid or do the next row and then use the iron to clear a short.

For chips with no leads (more and more common) I dot solder paste on

of crappy and now I see exactly how it works I think i could make one for the same money (air pressure regulator, solenoid valve, 0.01 to 0.5

collection of nozzles and needles to squirt the solder paste (they are very cheap if you buy bags of 50 but a bit pricey in ones so I bought a lifetime supply).

If there are several chips with no leads you will need a solder stencil

found a good cheap one).

Put the chips on the paste and fry up on a lump of 8mm thick Al on a bog standard electric cooking hotplate (while checking the temperature and watching the paste). Don't heat up too faste because if you do the voltaile stuff will still be boiling out of the paste when it melts and you'll get voids.

BGA is a new venture and I've bought a cheapo Chinese IR rework machine with top and bottom heat to do them The plan is to use flux and no paste

- I'm told it works - I'll try to let you know.

Michael Kellett.

Reply to
MK

Why are you so cost conscious? Is this a necessity or just a challenge for you? (I recall your paper - something like fast photodiode amplification for

Reply to
haiticare2011

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.