current-mode opamp noise

I need a fast, high slew, low noise amplifier with a precise gain of

+3. The fastest opamps around are current-mode parts. They tend to have impressive voltage noise specs, but they have terrible DC specs and tons of current noise. Low open-loop gains, too.

They usually have a required/optimum value of feedback resistor, so the inverting input current noise dumps into Rf and makes output noise, which is especially bad in low gain situations like mine. Rf has its own Johnson noise too.

THS3201 is actually pretty good, a mere 20 pA/rthz typ inverting noise current. Some parts have 60 pA/rthz current noise on the inverting input.

In the sim, I made the inverting and non-inverting noise currents explicit, 20 and 13.4 pA per root Hz in this case. The value of Rf is as suggested on the data sheet.

I'd do better using a voltage-mode part, like ADA4817, which has more voltage noise but essentially no current noise. I can make the feedback divider resistances as low as I like, until I run out of current. It is a tad slow slewing, as voltage amps usually are.

Version 4 SHEET 1 880 692 WIRE -624 -112 -672 -112 WIRE -592 -112 -624 -112 WIRE -448 -112 -496 -112 WIRE -416 -112 -448 -112 WIRE -672 -80 -672 -112 WIRE -496 -80 -496 -112 WIRE -672 48 -672 0 WIRE -496 48 -496 0 WIRE 240 64 208 64 WIRE 256 64 240 64 WIRE 208 112 208 64 WIRE -384 128 -672 128 WIRE 16 128 -384 128 WIRE 176 128 16 128 WIRE 320 144 240 144 WIRE 368 144 320 144 WIRE 400 144 368 144 WIRE 176 160 96 160 WIRE -384 176 -384 128 WIRE 208 224 208 176 WIRE 240 224 208 224 WIRE 256 224 240 224 WIRE -736 272 -832 272 WIRE -112 288 -208 288 WIRE 16 288 -48 288 WIRE 96 288 96 160 WIRE 96 288 16 288 WIRE 160 288 96 288 WIRE 320 288 320 144 WIRE 320 288 240 288 WIRE -672 304 -672 128 WIRE -736 320 -736 272 WIRE -720 320 -736 320 WIRE -384 320 -384 256 WIRE -48 320 -48 288 WIRE -832 336 -832 272 WIRE -112 336 -112 288 WIRE -96 336 -112 336 WIRE 96 336 96 288 WIRE -208 352 -208 288 WIRE -720 368 -736 368 WIRE -96 384 -112 384 WIRE -736 416 -736 368 WIRE -672 416 -672 384 WIRE -672 416 -736 416 WIRE -112 432 -112 384 WIRE -48 432 -48 400 WIRE -48 432 -112 432 WIRE -832 464 -832 416 WIRE -672 464 -672 416 WIRE -384 464 -384 400 WIRE -208 464 -208 432 WIRE -48 464 -48 432 WIRE 96 464 96 416 FLAG -384 464 0 FLAG 96 464 0 FLAG -48 464 0 FLAG -208 464 0 FLAG -672 464 0 FLAG -832 464 0 FLAG -672 48 0 FLAG -496 48 0 FLAG -624 -112 +5 FLAG -448 -112 -5 FLAG 240 224 +5 FLAG 240 64 -5 FLAG 368 144 OUT FLAG 16 128 IN+ FLAG 16 288 IN- SYMBOL Opamps\\UniversalOpamp2 208 144 M180 WINDOW 0 14 173 Left 2 SYMATTR InstName U1 SYMATTR Value2 Avol=1Meg GBW=1g Slew=1g SYMATTR SpiceLine ilimit=100m rail=0 Vos=0 phimargin=45 SYMATTR SpiceLine2 en=1.65n enk=10K in=0 ink=0 Rin=500Meg SYMBOL res -368 272 R180 WINDOW 0 -46 68 Left 2 WINDOW 3 -45 30 Left 2 SYMATTR InstName R1 SYMATTR Value 50 SYMBOL res 256 272 R90 WINDOW 0 68 32 VBottom 2 WINDOW 3 74 29 VTop 2 SYMATTR InstName Rf SYMATTR Value 768 SYMBOL res 80 320 R0 WINDOW 0 56 56 Left 2 WINDOW 3 52 92 Left 2 SYMATTR InstName Rg SYMATTR Value 384 SYMBOL g -48 304 R0 WINDOW 0 55 46 Left 2 WINDOW 3 47 78 Left 2 SYMATTR InstName Gn SYMATTR Value 20m SYMBOL res -224 336 R0 WINDOW 0 -53 21 Left 2 WINDOW 3 -51 52 Left 2 SYMATTR InstName Rn SYMATTR Value 60 SYMBOL voltage -496 -96 R0 WINDOW 0 59 47 Left 2 WINDOW 3 61 82 Left 2 SYMATTR InstName V- SYMATTR Value -5 SYMBOL voltage -672 -96 R0 WINDOW 0 -81 44 Left 2 WINDOW 3 -71 78 Left 2 SYMATTR InstName V+ SYMATTR Value 5 SYMBOL voltage -384 304 R0 WINDOW 0 -98 39 Left 2 WINDOW 3 -181 111 Left 2 WINDOW 123 -111 71 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName Vin SYMATTR Value SINE(0 0.1 1e8) SYMATTR Value2 AC 1 SYMBOL g -672 288 R0 WINDOW 0 57 39 Left 2 WINDOW 3 54 73 Left 2 SYMATTR InstName Gp SYMATTR Value 13.4n SYMBOL res -848 320 R0 WINDOW 0 -53 21 Left 2 WINDOW 3 -51 52 Left 2 SYMATTR InstName Rp SYMATTR Value 60 TEXT -168 -24 Left 2 !;tran 100n TEXT -288 16 Left 2 !.noise V(out) Vin dec 20 1K 1G TEXT -288 -112 Left 2 ;Noise Analysis Current-Mode Opamp TEXT -192 -64 Left 2 ;JL Mar 31, 2017 TEXT 184 8 Left 2 ;THS3201 TEXT -968 408 Left 2 ;1 nV/rthz TEXT 240 408 Left 2 ;G=3

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics
Reply to
John Larkin
Loading thread data ...

Voltage mode op amp

formatting link

is 6,600V/us around 7.5 times faster slew then 4817, 2nv/sqrthz noise

-- Kevin Aylward

formatting link
- SuperSpice
formatting link

Reply to
Kevin Aylward

That's impressive, but I need more output voltage swing. What would be great would be a more conventional opamp with that guy's front-end transistors.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics
Reply to
John Larkin

That's fast enough, but it still has a lot of current noise.

In your case, it sounds like the forward voltage gain will be Rf/Rg, where Rg is 1/Gm of the jfet and Rf is about 1K maybe. Voltage gain ballpark 20? The current noise of the opamp inv input is 17 pA/rthz, which dumps into Rf, which makes 17 nV/rthz, isn't bad as input-referred noise, with a gain of 20.

I need a gain of 3, which isn't as nice. Lower gain, more input-referred noise! In a CFB amp, Rf is constrained by dynamics.

I need gain accuracy, too, 0.5% for some reason, and fet Gm won't be well controlled.

It is an interesting/outrageous idea!

Again, I have the silly requirements for analog gain precision, low noise, high slew, and very low DC offset, way beyond the reasonable limits of the application. I have decided that, rather than fighting (which isn't working) I'll accept it as a design challenge and, eventually, an expensive product.

I can fix the DC offset with an autozero servo.

There seems to be some basic physics about bipolar wideband amps: you need more input transistor current for speed, and that makes more noise. Jfets fix that but add capacitance. There are MMICS that have outrageous speed/noise ratios, but that technology doesn't seem to appear in opamps. Maybe there's not a lot of demand.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics
Reply to
John Larkin

The current noise gets dumped into the bootstrap output (Z0~1 ohm, with local feedback). The result is a quiet, fast slewing voltage feedback amp. I'm using a megohm resistor for overall feedback to the gate of the FET, so the current noise is a couple of hundred femtoamps. It's a TIA, so the closed-loop voltage gain is 1. Noise gain is about 3 due to correcting for the parallel capacitance of the feedback resistor.

I was using the THS3091 to drive the 50 ohm output, so since the ADA4898 was too slow, I tried that on a lark, and it worked fine. The LM6171A is cheaper and doesn't load down the bootstrap, so it worked a bit better for my situation. Cheers

Phil Hobbs

Reply to
pcdhobbs

OK, you are closing an outer loop. Do you even have a feedback resistor locally on the current-mode amp? Probably not.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics
Reply to
John Larkin

Right--no inner loop at all.

In this application the PD leakage dominates the tempco, so dV_gs/dT isn't such a worry.

Cheers

Phil Hobbs

Reply to
pcdhobbs

One thing I should add--I had antiparallel diodes across the amp inputs to prevent a latch state on power-up.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

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.