EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: thereza on April 05, 2014, 02:01:45 am
-
I'm struggling trying to find an instrumentation amplifier that can give me an undistorted output of a sine wave at 1Mhz that can operate at 3V. I need something with a gain in the range of 1-5. The closest ones I've found are the INA826 and the INA331, but both distort the waveform when it gets close to 1Mhz (contrary to what the datasheet implies).
I can use a boost circuit if I must get higher voltages, as well as using multiple IC, but I want to optimize the design on size, power consumption, and cost (in that order).
Any suggestions?
Thanks!
Reza
-
the best option i've found so far is to use a couple opa2835's to build an instrumentation amp.
-
Neither datasheet implies you will get an undistorted waveform at 1 MHz.
INA826 claims a bandwidth of 1 MHz. That's the half-power point. INA331 claims a 2 MHz bandwidth, and if that's the half-power point, you're already going to be losing amplitude at 1 MHz. Also, the slew rate is 5V/µs. The maximum slew rate of a waveform of frequency f and amplitude A is 2(pi)Af, so if you want 1Vp-p you need a slew rate of 2(3.14)(0.5)(1e6) = 3.14 V/µs, which isn't a hell of a lot of margin under 5 V/µs.
I'm not seeing a lot of options. What's the application?
-
Any suggestions?
Those fellows generally aren't fast - actually they are fairly slow by today's standards.
You may want to think about what exactly you are looking for and construct your own using opamps that meet / exceed that specification.
-
An instrumentation amp implies high CMRR, something difficult as the frequency rises. All stray capacitance has to be balanced, which is why a differential scope plug-in will have lots of trimmer capacitors so CMRR can be optimized at high frequencies. The ready-made instrumentation amps tend to be LF, so building your own seems like the best bet.
-
Neither datasheet implies you will get an undistorted waveform at 1 MHz.
INA826 claims a bandwidth of 1 MHz. That's the half-power point. INA331 claims a 2 MHz bandwidth, and if that's the half-power point, you're already going to be losing amplitude at 1 MHz. Also, the slew rate is 5V/µs. The maximum slew rate of a waveform of frequency f and amplitude A is 2(pi)Af, so if you want 1Vp-p you need a slew rate of 2(3.14)(0.5)(1e6) = 3.14 V/µs, which isn't a hell of a lot of margin under 5 V/µs.
I'm not seeing a lot of options. What's the application?
The INA826 is def. slow. The INA331's bode plot looks flat past 1Mhz, but the problem isn't a loss of amplitude, rather, the waveform gets deformed (see attachment - blue input white output - waveforms averaged).
I'm comparing two waveforms to measure the ratio of amplitudes and phases - both fed through the amp. As as long as the ratio of amplitude loss between them at high frequencies is consistent (i.e. they both experience the same percentage loss at any given frequency), then that should work, but the deformation in the waveform is messing up my phase measurements.
I think I'll build my own using the opa2835 -- it also has the best datasheet i've seen with instructions on how to build your own instrumentation amp (use 2 as buffers and the 3rd as a difference amp).
-
Another option that looks good is one OPA2832 as the input stage followed by the opa835 for the differential stage. Any thoughts on that configuration?
... actually, the quiescent current for the opa2832 ~10mA, or 40x the opa835. I think i'll stick with those.
-
I think I'll build my own using the opa2835
Whenever you have trouble finding some super-duper parts, you should step back and ask yourself "why is that such parts are difficult to find?"
Could it be that you have approached the problem in such a way that you have to rely on such super-duper parts to dig out a design mistake you made earlier?
My experience is that it is a lot easier to ask "Here is a problem I am trying to solve. Am I on the right track solving it this way?" than "Where can I find that super-duper part?".
The 1st question gets you all the experience of the people here; and the 2nd question gets you some free keystrokes on digikey.com.
-
Taking another look at the INA331 datasheet: also note this. While the Bode plot does imply that gain is still flat to 1 MHz at AV=5, look to the right at the CMRR vs. frequency plot. It falls at over 40dB/decade past a couple kilohertz. By the time you get to even 100 kHz, it's barely an instrumentation amp anymore. At that point, you might as well use a few filters to feed high frequency components around the INA331 to a separate high-frequency amplifier; you'll hardly sacrifice any of the practically-0dB CMRR...
Even your OPA835 only has 40dB CMRR at 1 MHz, which may be enough, or it may not. And that CMRR will be subject to other things in your circuit - you'll have to use very well-matched parts in a layout with well-matched parasitics.
I agree with dannyf that you're probably trying to paddle upstream with your hands here. And you'll have a tough time making one yourself and getting decent performance as well. What's the application?
-
MCP6N11?
-
Depend on application,
if you want simple and cheap solution, go for DC-DC converter (cheap,small, and simple)
like $5.77520
http://www.digikey.com/product-detail/en/RB-3.315D/RB-3.315D-ND/2303581 (http://www.digikey.com/product-detail/en/RB-3.315D/RB-3.315D-ND/2303581)
for the Instrumental amplifier maybe this one suitable
$35.41
http://www.digikey.com/product-detail/en/AD624ADZ/AD624ADZ-ND/1206698 (http://www.digikey.com/product-detail/en/AD624ADZ/AD624ADZ-ND/1206698)
in my case i make 1Mhz differential probe use 5V usb input using AD620 + DC-DC converter at <$10
-
@han - this is size-sensitive application, so those parts are way too large. I can implement a small boost converter if needed.
@matcat - bandwidth is only 500kHz
@c4757p/@dannyF - I don't think I'm backtracking. the design isn't simple and I needed to operate at up to 1Mhz. I was just shocked at how slow instrumentation amps were. I don't need very high CMRR, rather, I need high input impedance and a differential amplifier configuration. I believe there are parts that will work, but require higher supply voltages. This design only adds 1 additional IC to the original design (from 2x instrumentation to 3x dual op amp).
I'll order some amps to test the performance of the opa2835, and incorporate them into the circuit if they work out.
Thanks
-
I don't need very high CMRR, rather, I need high input impedance and a differential amplifier configuration.
Sounds like an instrumentation amp is the wrong medicine here.
-
I don't need very high CMRR, rather, I need high input impedance and a differential amplifier configuration.
???
CMRR is what makes it differential...
-
@matcat - bandwidth is only 500kHz
If it's just a sine wave @ 1MHz your bandwidth is going to be far below 500KHz....
-
@matcat - bandwidth is only 500kHz
If it's just a sine wave @ 1MHz your bandwidth is going to be far below 500KHz....
These things go down to DC, so a 500 kHz bandwidth implies it tops out at 500 kHz...
-
@matcat - bandwidth is only 500kHz
If it's just a sine wave @ 1MHz your bandwidth is going to be far below 500KHz....
A friend mentioned some sort of amplifier that works at AC and has higher bandwidths but I've not found 'em. I need to measures from 1Khz - 1Mhz -- are there other types of amplifier that are better suited? Else this is the topology that I'll be testing next.