Thing is, the Johnson noise of the 1G pickoff resistor is forced into the follower output. That's only at low frequencies, and the corner freq can be pushed pretty low, at the cost of startup time. There's probably some kluge that would help the startup, at least.
That LT1012 is noisy, too.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
That's sort of what I do, but since mine are usually bootstraps, I care about the gain being really really close to 1, like 0.999 or so.
With the noninverting input connected directly to the source, the emitter of the current source follows the JFET's source. Thus the 500 ohm load appears right on the FET, leading to a gain of about
1-1/(g_M * 500ohms) ~ 0.95. Hanging a matching RC on the + input is one approach, and if the op amp's bias current isn't too high, it could be 100k and 1 uF. That would cause a gain error of only 0.0004, plus whatever the residual contribution of Q1's Early voltage is.It's also possible to switch the gate snooper circuit , so that you don't get the Johnson voltage noise during a measurement. With a servo as slow as this one, a bootstrapped relay is good for that.
You have to be careful with that sort of thing, though. I did a bootstrapped TIA a couple of years ago for a scanning surface potential probe used for semiconductor inspection. It used a DPDT latching relay to switch between 1G and 50G feedback resistors. The the relay coil had to be bootstrapped, so it was driven with a non-latching relay.
Interestingly, I had to short out the 1G resistor when in the 50G mode, because otherwise the Johnson noise current of the 1G would have gotten into the summing junction via the capacitance of the _open_relay_contact_ and blown the measurement out of the water.
Managing all the strays on that circuit was great fun. Once we got the layout strays ironed out, it worked great. The customer eventually decided just to go with a plain 20G model, so we stripped out all the relay stuff.
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
Sure, that could be bootstrapped. I was just interested in the servo part.
Good point.
That would be a powerup zero offset correction, which can be done downstream of the front end, especially if there's a uP around. What's left is temperature drift, which the uP can partially fix, too.
Yeah, switching the gate is tricky. Snooping the gate voltage is tricky. Expensive lunch.
That Johnson guy caused us no end of trouble.
A pair of perfectly matched fets can be use to run the follower at Idss. I think that's lower noise than a diffamp, but I'd have to verify that.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Sure, if you just ground the input and measure it, that works well. A lot of times I'm just doing the front end, and it's easier to make that function self-contained than spend time explaining to the firmware person what he should be doing, especially since the customer often doesn't want to pay for me to do that. "Here, put a 100 ms pulse on this wire before every measurement" is a lot harder to screw up.
DC-bootstrapping the drain makes the FET run at constant power, which helps the thermal problem a lot, as you know. One of these times I should instrument a bunch of BF862s to see if there's some regularity in the tempcos vs V_DS and I_D.
They're surprisingly linear devices--the usual parabolic thing doesn't seem to apply to BF862s. Their drain current just starts up from zilch someplace between -0.4 and -1V G-S, and increases with a nearly constant g_M of right around 40 mS. There's a bit of curvature near V_P of course, but all the variation in I_DSS seems to come from different pinchoff voltages, not different transconductance.
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
This fixes the problem that you noted
and takes the opamp noise out entirely at higher frequencies. The noise is now close the the jfet limit.
I actually have no clear use for this, so it's an interesting bag-o-tricks thing.
A uP doing a downstream offset/tempco correction would be better, if it can be calculated without temperature cycling every unit.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
On 06/08/2015 11:38 AM, John Larkin wrote:
Fun. With my invariable tendency to add parts, here's one with a 10x improvement in output impedance and different choices of I_DSS.
NB the NXP BF862 model overestimates its gate leakage by a factor of about 1000--it's really only about 2 pA with V_DS < 3V or so. Hey, Jim, how about fixing the model? I hacked it myself, but it's probably still wrong.
Cheers
Phil Hobbs
Version 4 SHEET 1 2272 916 WIRE 448 -160 400 -160 WIRE 496 -160 448 -160 WIRE 672 -160 496 -160 WIRE 784 -160 672 -160 WIRE 880 -160 784 -160 WIRE 496 -128 496 -160 WIRE 672 -96 672 -160 WIRE 784 -96 784 -160 WIRE 400 -80 400 -160 WIRE 432 -80 400 -80 WIRE -208 -32 -256 -32 WIRE -160 -32 -208 -32 WIRE 880 -16 880 -160 WIRE -160 0 -160 -32 WIRE 496 0 496 -32 WIRE 672 16 672 -16 WIRE 784 16 784 -32 WIRE 784 16 672 16 WIRE 672 48 672 16 WIRE 496 96 496 80 WIRE 528 96 496 96 WIRE 880 96 880 64 WIRE -160 112 -160 80 WIRE -160 160 -400 160 WIRE 384 160 -160 160 WIRE 448 160 384 160 WIRE 208 224 -48 224 WIRE 272 224 208 224 WIRE 496 224 496 192 WIRE 496 224 352 224 WIRE 672 224 672 144 WIRE 672 224 496 224 WIRE 736 224 672 224 WIRE 768 224 736 224 WIRE 208 240 208 224 WIRE -400 272 -400 160 WIRE -160 272 -160 160 WIRE 768 272 768 224 WIRE 208 320 208 304 WIRE 48 352 16 352 WIRE 80 352 48 352 WIRE 496 368 496 224 WIRE 16 384 16 352 WIRE 768 384 768 352 WIRE -400 400 -400 352 WIRE -48 400 -48 224 WIRE -16 400 -48 400 WIRE 128 416 48 416 WIRE 192 416 128 416 WIRE 320 416 272 416 WIRE 432 416 320 416 WIRE -160 432 -160 352 WIRE -16 432 -160 432 WIRE 128 448 128 416 WIRE 16 480 16 448 WIRE 48 480 16 480 WIRE 80 480 48 480 WIRE 320 528 320 416 WIRE -400 544 -400 480 WIRE -160 544 -160 432 WIRE -64 544 -160 544 WIRE 128 544 128 528 WIRE 128 544 0 544 WIRE 496 608 496 464 WIRE 160 752 128 752 WIRE 320 752 320 592 WIRE 320 752 160 752 WIRE 496 752 496 688 WIRE 496 752 320 752 FLAG 880 96 0 FLAG -400 544 0 FLAG -160 112 0 FLAG 736 224 SRC FLAG 384 160 GATE FLAG 448 -160 +5 FLAG -208 -32 -15 FLAG 48 352 -15 FLAG 48 480 +5 FLAG 160 752 -15 FLAG 208 320 0 FLAG 768 384 0 SYMBOL njf 448 96 R0 WINDOW 0 -55 6 Left 2 WINDOW 3 -111 -29 Left 2 SYMATTR InstName J1 SYMATTR Value {jfet} SYMBOL voltage 880 -32 R0 WINDOW 0 57 46 Left 2 WINDOW 3 62 80 Left 2 SYMATTR InstName V1 SYMATTR Value 5 SYMBOL res -416 256 R0 WINDOW 0 60 33 Left 2 WINDOW 3 59 67 Left 2 SYMATTR InstName R1 SYMATTR Value 50 SYMBOL voltage -160 -16 R0 WINDOW 0 57 46 Left 2 WINDOW 3 62 80 Left 2 SYMATTR InstName V2 SYMATTR Value -15 SYMBOL npn 432 368 R0 WINDOW 0 100 21 Left 2 WINDOW 3 77 57 Left 2 SYMATTR InstName Q1 SYMATTR Value BC847C SYMBOL res 480 592 R0 WINDOW 0 -75 34 Left 2 WINDOW 3 -78 71 Left 2 SYMATTR InstName R2 SYMATTR Value 300 SYMBOL Opamps\\LT1012A 16 480 M180 WINDOW 0 57 218 Left 2 WINDOW 3 22 183 Left 2 SYMATTR InstName U1 SYMBOL res -176 256 R0 WINDOW 0 -64 34 Left 2 WINDOW 3 -65 72 Left 2 SYMATTR InstName R3 SYMATTR Value 1G SYMBOL cap 0 528 R90 WINDOW 0 69 29 VBottom 2 WINDOW 3 78 28 VTop 2 SYMATTR InstName C1 SYMATTR Value 100p SYMBOL voltage -400 384 R0 WINDOW 0 57 46 Left 2 WINDOW 3 62 80 Left 2 WINDOW 123 62 108 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V3 SYMATTR Value PULSE(0 100m 10n 500p) SYMATTR Value2 AC 1 SYMBOL res 112 432 R0 WINDOW 0 57 52 Left 2 WINDOW 3 64 80 Left 2 SYMATTR InstName R4 SYMATTR Value 1 SYMBOL res 288 400 R90 WINDOW 0 -56 53 VBottom 2 WINDOW 3 -50 53 VTop 2 SYMATTR InstName R5 SYMATTR Value 10k SYMBOL cap 304 528 R0 WINDOW 0 -60 34 Left 2 WINDOW 3 -57 64 Left 2 SYMATTR InstName C2 SYMATTR Value 10u SYMBOL res 368 208 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R6 SYMATTR Value 1G SYMBOL cap 192 240 R0 SYMATTR InstName C3 SYMATTR Value 100p SYMBOL res 480 -16 R0 SYMATTR InstName R7 SYMATTR Value 70 SYMBOL pnp 608 144 M180 WINDOW 0 132 106 Left 2 WINDOW 3 -9 114 Left 2 SYMATTR InstName Q2 SYMATTR Value BFT92 SYMBOL res 656 -112 R0 SYMATTR InstName R8 SYMATTR Value 150 SYMBOL cap 768 -96 R0 SYMATTR InstName C4 SYMATTR Value 10m SYMBOL npn 432 -128 R0 SYMATTR InstName Q3 SYMATTR Value 2N3904 SYMBOL res 752 256 R0 SYMATTR InstName R9 SYMATTR Value {RL} SYMBOL res 624 80 R90 WINDOW 0 96 47 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R10 SYMATTR Value 22 TEXT 664 640 Left 2 !;op TEXT 664 680 Left 2 !;tran 20 uic TEXT 576 728 Left 2 !.tran 30n TEXT 648 464 Left 2 ;JFET Idss SERVO TEXT 616 520 Left 2 ;J Larkin / P Hobbs June 8, 2015 TEXT 840 656 Left 2 !.step dec param RL 100 100k 1 TEXT 984 144 Left 2 !.model BF862J NJF(Beta=47.800E-3 Betatce=-.5 Rd=.8 Rs=7.5000 Lambda=37.300E-3 Vto=-.57093\n+ Vtotc=-2.0000E-3 Is=424.60E-12 Isr=2.995p N=1 Nr=2 Xti=3 Alpha=-1.0000E-3\n+ Vk=59.97 Cgd=7.4002E-12 M=.6015 Pb=.5 Fc=.5 Cgs=8.2890E-12 Kf=87.5E-18\n+ Af=1)\n \n* Same with more realistic leakage\n.model BF862_1pA ako:BF862J Isr=6e-15 Is=6e-15;\n.model BF862_25mA ako:BF862J Beta=150.0E-3;\n.model BF862_10mA ako:BF862J Beta=37.0E-3; TEXT 1232 424 Left 2 !.MODEL BFT92 PNP( IS = 4.3756E-016 BF =
33.5815 NF = 1.0097 VAF = 23.3946\n+ IKF = 9.9538E-002 ISE = 8.7054E-014 NE = 1.94395 BR = 4.9472\n+ NR = 1.00254 VAR = 3.90385 IKR = 5.2816E-003 ISC = 3.5886E-014\n+ NC = 1.3933 RB = 5 IRB = 1E-006 RBM = 5 RE = 1 RC = 10 EG = 1.11\n+ XTI = 3 CJE = 7.4666E-013 VJE = .6 MJE = .35683 TF = 1.7492E-011\n+ XTF = 1.3546 VTF = .155654 ITF = 1E-003 PTF = 45 CJC = 9.371E-013\n+ VJC = .396455 MJC = .19995 XCJC = .106 TR = 8.422E-009\n+ VJS = .75 FC = .767856) TEXT 1784 288 Left 2 !.model 1 ako:bf862_25mA TEXT 1792 368 Left 2 !.model 3 ako: bf862_1pA TEXT 1792 328 Left 2 !.model 2 ako:bf862_10mA TEXT 1792 248 Left 2 !.step param jfet list 1 2 3-- 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
Does not compute. I got curious and tried to run this last netlist, but I got an error: Unknown schematic syntax. Rs=7.5000 Lambda=37.300E-3 Vto=-.57093\n+ Vtotc=-2.0000E-3 Is=424.60E-12
What might be wrong.
By the way LTSpice should let the users to copy error messages to the clipboard.
Probably line wrap.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Fix the line wraps. I'll try posting this again with the file attached.
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
Well, the one with the attachment appeared on Supernews and on Eternal September, but not on Google Gropes.
Any other Usenet providers that work with text attachments?
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
Agent is likes it.
Is it just your news client that "pre-wraps" the text? Is this wrapped:
"a long line of text a long line of text a long line of text a long line of text a long line of text a long line of text a long line of text"
If you sent me the file or put it up somewhere I can repost it
-- John Devereux
The long line wasn't wrapped as received, but it is now, because I hit ctl-R (Thunderbird). I think it's the OP's client, or maybe Google, because it isn't line wrapped on Supernews or Eternal September.
For anybody who's still reading, the file is at
Wrapping a bootstrap round the whole works is also a win.
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
It is amazing how well semiconductors perform that are from the days when we still had full hair on our heads. And cheap. Last year I did a TDR that works high in the gigeehoitzes and costs less than five bucks in materials. Analog is fun. The only thing more fun I could imagine is craft beer brewing. Which is what I'd go for if I were deciding my career again, keeping analog electronics as a hobby.
-- Regards, Joerg http://www.analogconsultants.com/
Well, I was born bald. I did have more hair in the middle there at some point. ;)
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
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.