Forums

how to optimize for multistage gain selection

Started by Unknown April 17, 2019
On Fri, 19 Apr 2019 10:01:38 -0700 (PDT), George Herold
<gherold@teachspin.com> wrote:

>On Friday, April 19, 2019 at 11:57:53 AM UTC-4, upsid...@downunder.com wrote: >> On Fri, 19 Apr 2019 08:10:29 -0700 (PDT), George Herold >> <gherold@teachspin.com> wrote: >> >> >On Friday, April 19, 2019 at 10:18:22 AM UTC-4, gnuarm.de...@gmail.com wrote: >> >> On Friday, April 19, 2019 at 9:35:10 AM UTC-4, George Herold wrote: >> >> > On Thursday, April 18, 2019 at 9:18:47 PM UTC-4, gnuarm.de...@gmail.com wrote: >> >> > > On Thursday, April 18, 2019 at 7:18:04 PM UTC-4, George Herold wrote: >> >> > > > On Wednesday, April 17, 2019 at 8:45:52 PM UTC-4, Clifford Heath wrote: >> >> > > > > On 18/4/19 8:24 am, al.basili@gmail.com wrote: >> >> > > > > > I'm trying to design a two stage amplifier for a sensor readout and I would like those stages to be configurable, but I also need to have the overall gain vs. configuration curve to be such that every new step is ~5% increase w.r.t. the previous. >> >> > > > > > >> >> > > > > > So let's say I have 3 control bits on the first amplifier and 4 for the second, how would I design the set of values those gains need to be at? Is there an algorithm to do that? >> >> > > > > > >> >> > > > > > I'm currently guesstimating it, but I doubt there's no formal method to do so. >> >> > > > > > Any pointer/comment is appreciated. >> >> > > > > > >> >> > > > > >> >> > > > > Gain is easy; stepped attenuation is easy. Stepped gain is usually a bad >> >> > > > > idea because it either changes the input impedance or lengthens and adds >> >> > > > > reactance to the feedback path. >> >> > > > > >> >> > > > > Use fixed gain and stepped attenuation. >> >> > > > >> >> > > > That's what I did when I wanted to keep the BW >> >> > > > approx. constant. Like a 'scope, at least analog ones. >> >> > > >> >> > > How do you deal with overloading the gain stage? Is the attenuator ahead of the first stage? >> >> > There was a lot of gain. You can switch in x10 gain stages up front, >> >> > The attenuator in the ~middle and then fixed gain at the end. >> >> >> >> Hmmm... I'm not getting it. The idea is to have high gain in the amps at all settings so as to minimize the bandwidth??? Sounds counter productive. >> >Ya know Rick, if you were inquisitive and polite, rather than argumentative >> >and insulting, I would be happy to share details. It's not anything >> >amazing. But we sell them without complaints. Gain from 10 to 10^4 >> >and BW from DC to 1 MHz. (opamps are amazing.) >> >> So the gain range would be 20 to 80 dB and the maximum gain would be >> about 35 and 45 dB/stage, this at 1 MHz. For proper feedback action, >> the open loop gain would have to be, say, 20 dB better, so with >> standard compensation, the open loop gain would drop to unity around 1 >> GHz, so an amazing op-amp is needed. With more aggressive >> compensation, a lower open loop limit would suffice, but since this is >> a variable gain stage, the stability at lower gain settings must be >> analyzed. > >Here's a schematic. >https://www.dropbox.com/s/oeceem6by903xjb/gain.png?dl=0
Thanks, this clarifies some things. Hopefully the signal is AC, since switching in an inverting stage will reverse the polarity. I calculated 1.5 bits for the two first stages and 3 bits for the attenuator. From previous posts I assumed that there would be 3+4 control bits. Anyway, all those op-amps are really marginal for 1 MHz bandwidth. The full voltage swing stops at about 200 kHz and at 1 MHz, the swing is only a few volts. The open and close loop gain are marginal, some of which drops a few dB at 1 MHz. In order to have, say the -3 dB full chain response at 1 MHz, the gain of each individual stage should drop by less than -1 dB, Better op-amps may help.
> >George H.
On Friday, April 19, 2019 at 4:52:15 PM UTC-4, upsid...@downunder.com wrote:
> On Fri, 19 Apr 2019 10:01:38 -0700 (PDT), George Herold > <gherold@teachspin.com> wrote: > > >On Friday, April 19, 2019 at 11:57:53 AM UTC-4, upsid...@downunder.com wrote: > >> On Fri, 19 Apr 2019 08:10:29 -0700 (PDT), George Herold > >> <gherold@teachspin.com> wrote: > >> > >> >On Friday, April 19, 2019 at 10:18:22 AM UTC-4, gnuarm.de...@gmail.com wrote: > >> >> On Friday, April 19, 2019 at 9:35:10 AM UTC-4, George Herold wrote: > >> >> > On Thursday, April 18, 2019 at 9:18:47 PM UTC-4, gnuarm.de...@gmail.com wrote: > >> >> > > On Thursday, April 18, 2019 at 7:18:04 PM UTC-4, George Herold wrote: > >> >> > > > On Wednesday, April 17, 2019 at 8:45:52 PM UTC-4, Clifford Heath wrote: > >> >> > > > > On 18/4/19 8:24 am, al.basili@gmail.com wrote: > >> >> > > > > > I'm trying to design a two stage amplifier for a sensor readout and I would like those stages to be configurable, but I also need to have the overall gain vs. configuration curve to be such that every new step is ~5% increase w.r.t. the previous. > >> >> > > > > > > >> >> > > > > > So let's say I have 3 control bits on the first amplifier and 4 for the second, how would I design the set of values those gains need to be at? Is there an algorithm to do that? > >> >> > > > > > > >> >> > > > > > I'm currently guesstimating it, but I doubt there's no formal method to do so. > >> >> > > > > > Any pointer/comment is appreciated. > >> >> > > > > > > >> >> > > > > > >> >> > > > > Gain is easy; stepped attenuation is easy. Stepped gain is usually a bad > >> >> > > > > idea because it either changes the input impedance or lengthens and adds > >> >> > > > > reactance to the feedback path. > >> >> > > > > > >> >> > > > > Use fixed gain and stepped attenuation. > >> >> > > > > >> >> > > > That's what I did when I wanted to keep the BW > >> >> > > > approx. constant. Like a 'scope, at least analog ones. > >> >> > > > >> >> > > How do you deal with overloading the gain stage? Is the attenuator ahead of the first stage? > >> >> > There was a lot of gain. You can switch in x10 gain stages up front, > >> >> > The attenuator in the ~middle and then fixed gain at the end. > >> >> > >> >> Hmmm... I'm not getting it. The idea is to have high gain in the amps at all settings so as to minimize the bandwidth??? Sounds counter productive. > >> >Ya know Rick, if you were inquisitive and polite, rather than argumentative > >> >and insulting, I would be happy to share details. It's not anything > >> >amazing. But we sell them without complaints. Gain from 10 to 10^4 > >> >and BW from DC to 1 MHz. (opamps are amazing.) > >> > >> So the gain range would be 20 to 80 dB and the maximum gain would be > >> about 35 and 45 dB/stage, this at 1 MHz. For proper feedback action, > >> the open loop gain would have to be, say, 20 dB better, so with > >> standard compensation, the open loop gain would drop to unity around 1 > >> GHz, so an amazing op-amp is needed. With more aggressive > >> compensation, a lower open loop limit would suffice, but since this is > >> a variable gain stage, the stability at lower gain settings must be > >> analyzed. > > > >Here's a schematic. > >https://www.dropbox.com/s/oeceem6by903xjb/gain.png?dl=0 > > Thanks, this clarifies some things. > > Hopefully the signal is AC, since switching in an inverting stage will > reverse the polarity.
Yeah, noise amplifier.
> > I calculated 1.5 bits for the two first stages and 3 bits for the > attenuator. From previous posts I assumed that there would be 3+4 > control bits.
Oh I'm not the OP. I was just posting a schematic of a gain with switched attenutators after..
> > Anyway, all those op-amps are really marginal for 1 MHz bandwidth. The > full voltage swing stops at about 200 kHz and at 1 MHz, the swing is > only a few volts. > > The open and close loop gain are marginal, some of which drops a few > dB at 1 MHz. In order to have, say the -3 dB full chain response at 1 > MHz, the gain of each individual stage should drop by less than -1 dB, > > Better op-amps may help.
Oh yes! This circuit is from ~10 years ago. Before putting new opamps in the gain stage, I could use faster opamps in the 100kHz filter stage. (not shown.) In this circuit, amplifying noise, you start to get clipping of the voltage before hitting the slew rate limit of the output opamps. Vrms <=~ 1V. I measured a ~2 MHz BW of this gain stage, sometime in the distant past, under those conditions. Which was good enough at the time. George H.
> > > > >George H.