From: jl@glen--canyon.com   
      
   On Thu, 29 Jan 2026 17:22:39 -0500, Phil Hobbs   
    wrote:   
      
   >On 2026-01-29 16:40, bitrex wrote:   
   >> On 1/29/2026 4:24 PM, Phil Hobbs wrote:   
   >>> On 2026-01-29 15:32, Liz Tuddenham wrote:   
   >>>> Phil Hobbs  wrote:   
   >>>>   
   >>>>> Hi, all,   
   >>>>>   
   >>>>> I'm doing a high-accuracy version of the laser noise canceller   
   >>>>>  .   
   >>>>>   
   >>>>> In particular, to get better cancellation accuracy, I want to   
   >>>>> get rid of the input offset voltages of a couple of op amps.   
   >>>>>   
   >>>>> One approach to this is to use a chopamp integrator to snoop   
   >>>>> the summing junction, and dork the noninverting input to force   
   >>>>> the summing junction to average 0.00000V.   
   >>>>>   
   >>>>> This is nice conceptually, but there are a couple of worries:   
   >>>>>   
   >>>>> 1. Chopamps kick out nasty switching spikes, which will have to   
   >>>>> be decoupled sufficiently well.   
   >>>>>   
   >>>>> 2. Weird-ass composite amplifiers always have weird settling   
   >>>>> behavior.   
   >>>>>   
   >>>>> I haven't done this lately, but I'm thinking of a TLV2333.   
   >>>>>   
   >>>>> Any wisdom?   
   >>>>   
   >>>> At that level of accuracy, beware thermocouple effects.   
   >>>   
   >>> It's all on one board, and the power level is low, so that   
   >>> shouldn't be a huge issue, I don't think.  Gradients on the board   
   >>> should be way under 1K in the quarter-inch or so separating the two   
   >>> amps.  I'll certainly put the power buffer some distance away.   
   >>>   
   >>>> If you are compensating a slow drift in offset, chop slowly and   
   >>>> sinusoidally, then the 'spikes' will matter less.   
   >>>   
   >>> I'm not the one doing the chopping--the spikes come from the CMOS   
   >>> switches inside the chopamp.   
   >>>   
   >>> The noise canceller works by splitting a larger photocurrent using   
   >>> a BJT diff pair, and adjusting the split ratio until the current in   
   >>> one arm exactly cancels a smaller photocurrent derived from the   
   >>> same laser.   
   >>>   
   >>> There are various fine points, but because the diff pair is a   
   >>> highly linear current splitter, the fluctuations split the same as   
   >>> the DC, so by adjusting the DC to zero, one in principle obtains   
   >>> cancellation of the fluctuations at all frequencies.  A slow servo   
   >>> loop lets you do AC- coupled measurements down at the shot noise   
   >>> even with noisy lasers.   
   >>>   
   >>> With a bit of math, you can use the delta V_BE of the diff pair to   
   >>> do the same thing inside the feedback loop bandwidth.   
   >>>   
   >>> An offset voltage in either the TIA or the integrating servo amp   
   >>> causes the cancellation to be in error by   
   >>>   
   >>> delta I = V_os / R_F.   
   >>>   
   >>> With a 5k ohm R_F, a millivolt of offset makes 200 nA of current   
   >>> imbalance.  With a 100-uA photocurrent, that limits the   
   >>> cancellation performance to   
   >>>   
   >>> Amax = 20*log(100uA/500nA) = 54 dB.   
   >>>   
   >>> It's better than that at higher photocurrent, but I'm chasing an   
   >>> honest 70 dB with this box, so the offsets have to be down in the   
   >>> tens of microvolts at most.   
   >   
   >>   
   >> So is the idea to LPF the crap out of the summing junction voltage,   
   >> send to a chopper amp used as an integrator, and then LPF the crap of   
   >> the chopper amp output sent to the TIA amp non-inverting input?   
   >   
   >Right, except that the output doesn't need filtering, just a voltage   
   >divider.   
   >   
   >> Is it spikes going forward to the TIA non-inverting input or going   
   >> backwards to the summing junction itself that's the most concern?   
   >   
   >The output is just ordinary noisy--53 nV in 1 Hz.  Chopamp inputs kick   
   >out evil microamp-level spikes of low duty cycle--the 70 pA bias current   
   >spec is basically the bits of the spikes that don't average to zero.   
   >>   
   >> The non-inverting input is the devil and I don't really like it   
   >> anywhere but bolted to ground in precision applications but I guess   
   >> there aren't a lot of other places to inject a correction that isn't   
   >> going to disturb the summing junction worse   
   >>   
   >   
   >Well, it's got a 10k:50R voltage divider to help keep it still.  The   
   >total adjustment range is thus +-12 mV or so, comfortably larger than   
   >the +-5 mV max offset over temperature.   
   >   
   >The voltage divider reduces the loop bandwidth by the same factor of 200   
   >for a given time constant, so to get a 5-Hz snoop loop bandwidth, it   
   >needs a time constant of   
   >   
   >tau = 1/200 / (2 pi * 5 Hz) = 160 us   
   >   
   >so the integrator has 200k * 820 pF.   
   >   
   >Since the 820 pF is connected between the inverting input and the   
   >(low-Z) output, it'll suck in most of the spikies, but just in case, I'm   
   >splitting the 200k in half and bypassing the midpoint with 1 nF to ground.   
   >   
   >Cheers   
   >   
   >Phil Hobbs   
      
   Couldn't the offset servo loop be mega-slow? I think it's correcting   
   thermals.   
      
      
      
      
   John Larkin   
   Highland Tech Glen Canyon Design Center   
   Lunatic Fringe Electronics   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)