Re: Interesting inductor

On 2024-03-13 12:04, John Larkin wrote:

On Wed, 13 Mar 2024 11:49:31 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:
 

On 2024-03-13 10:59, John Larkin wrote:

On Wed, 13 Mar 2024 12:49:04 -0000 (UTC), Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:
>

John Larkin <jl@997PotHill.com> wrote:

On Tue, 12 Mar 2024 23:17:57 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:
>

So I'm doing a new lab amp product.
Our existing one is 500 Hz -- 20 MHz, 1.1 nV/sqrt(Hz).
>
The new one is aiming to be 10 kHz -- 200 MHz, 0.25 nV/sqrt(Hz).  The
spherical cows love it, so we'll see when the test boards arrive later
this week.
>
As part of the design, I wanted to make an emitter follower with a
decent amount of inductance in series with its tail resistor, to avoid
the transistor turning off on fast negative edges and causing linearity
problems.
>
Searching on Digikey, I found this very interesting part:
<https://www.digikey.com/en/products/detail/epcos-tdk-electronics/B82498F1472J000/697521>.
>
4.7 uH 0805 wirewound, with a self-resonant frequency of _210 MHz_,
which is several times higher than many other parts of that description.
That corresponds to an effective parallel capacitance of 0.12 pF,
about that of a resistor of the same size, despite all the copper windings.
>
Pretty nifty, if true.  (Parts on order.)
>
Cheers
>
Phil Hobbs

>
Couldn't you have a high tail voltage and a big resistor, or maybe a
string of smaller inductors? Or something. We've made super wideband
inductors from a string of various values.

>
The first stage (paralleled pHEMTs with a BFU520A cascode and BFU520A
follower) has a gain of about 40 and flatband 1-Hz noise of 0.2 nV. That
means that the noise of the follower and the second stage is not
insignificant.
>
The second stage is a VCVS active lowpass using an OPA818 at a gain of 10,
and the output stage is an OPA695 CFA inverter, to make the overall circuit
noninverting and provide a gain adjustment. (TE now makes a low-inductance
pot that’s nearly as good as the old Murata PVA2 ones that you use. )
>
Keeping the supplies simple is important, and so is avoiding ground loops.
The box actually makes +7 and -5 by railsplitting a 24V wall wart, and then
using regulating cap multipliers. (The second and third stages’ supplies
are followers running off the quiet ones, to prevent unwanted feedback.)
>
Sooo, I want to run the follower on +7/0 if possible, which is where the
inductor comes in.  It doesn’t save any power, on account of the
railsplitter, so I can probably use the -5 rail instead.
>
There’s no overall feedback in this version, because it’s hard to do
without trashing the noise performance and/or stability.
>

>
I'm hassling with inductors now too, but at the other end of the speed
spectrum.
>
We want a programmable inductor, from maybe 1 mH to 500 mH or so,
maybe 100 mA. Sounds like an inductive DAC, a series string of
inductors with shorting relays. If the step inductance ratio were,
say, 1.8:1 we could have some hidden bits, more than the customer
sees, so we could get pretty close to his requested value.
>
We could test all 2^n steps, make a list, and select the closest to
his request.

>
We did something similar for choosing resistor taps in a low noise PGA.
Works okay, but is a bit of a pain.

>
We're simulating loads to an engine control computer, torque motors
and solenoids and steppers.

>
Fun.  Analog computers forever!
>

>
We are about to publicly announce the P940, our modular power system.
It would be tragic if I make my fortune selling power supplies and
dummy loads that work in the single digits of KHz.

>
If that happens, I'll commiserate appropriately. ;)

 Buy me a beer that I can cry in.
 

>
Making DACs with relays is humiliating.

>
Nah, relays are amazing.  There are low-power muxes that come close,
e.g. the TMUX1511 (5 ohms R_on, 2 pF C_off), but nothing that will take
any sort of power.
>
Of course you can do similar things with tubes. ;)

 If Ron * Coff is the figure of merit, in femtoseconds, no semi can
come within miles of a relay. We use a $1 DPDT telecom relay that is a
damned good 3 GHz 50 ohm switch.
 Tubes don't score well by that standard, except krytrons maybe.
 

Not identical things, just similar.  Dragging a grid up to +200V quickly and then leaving it there, with no turn-off charge injection and nearly no capacitive loading, is a job for a tube.  (I used an 811A for that BITD--it even had a B battery for the plate and a C battery for the grid bias.) :)
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510
http://electrooptical.net
http://hobbs-eo.com