Noise measurement, analog mult...

G

George Herold

Guest
So the last time I measured noise I used an analog multiplier
(AD734, I was going to say ~$20, but now $35!
https://www.digikey.com/product-detail/en/analog-devices-inc/AD734ANZ/AD734ANZ-ND/751025


So I'm thinking about methods to measure single channel noise
(V^2, V**2) and two chan. correlations V_a(t)*V_b(t)

0.) Digitize everything and do it in software.
This is not my strength, so someone else's box would
be my starting place. I must admit a fast (10 MHz) two
channel digitizer sounds like everyone's cheap DSO.*

But I'm looking for ideas.. one problem with cheap scopes
is the record length. You can save so much and then have to
down load it.

1.) Some switched gain thing like cheap lockin's
But I'd be doing Va(t)*sgn(Vb(t))
(where sgn is the sign of the voltage wrt gnd.)
or Vb*sgn(Va). Well it's not all that clear how
this relates to the correlations.. you get some sort of measure
if Vb and Va are the same size.
(I'll just note that Va*sgn(Vb) = Va*Vb/|Vb|)

or some other silly ideas...diode mixer?

I'm going to order a spendy AD734 as a reference.

George H.
*Does my cheap rigol do A*B? hang on... Yeah it does..
I really need the average of A*B
 
On 2020-02-18 08:47, George Herold wrote:
So the last time I measured noise I used an analog multiplier
(AD734, I was going to say ~$20, but now $35!
https://www.digikey.com/product-detail/en/analog-devices-inc/AD734ANZ/AD734ANZ-ND/751025


So I'm thinking about methods to measure single channel noise
(V^2, V**2) and two chan. correlations V_a(t)*V_b(t)

0.) Digitize everything and do it in software.
This is not my strength, so someone else's box would
be my starting place. I must admit a fast (10 MHz) two
channel digitizer sounds like everyone's cheap DSO.*

But I'm looking for ideas.. one problem with cheap scopes
is the record length. You can save so much and then have to
down load it.

1.) Some switched gain thing like cheap lockin's
But I'd be doing Va(t)*sgn(Vb(t))
(where sgn is the sign of the voltage wrt gnd.)
or Vb*sgn(Va). Well it's not all that clear how
this relates to the correlations.. you get some sort of measure
if Vb and Va are the same size.
(I'll just note that Va*sgn(Vb) = Va*Vb/|Vb|)

or some other silly ideas...diode mixer?

I'm going to order a spendy AD734 as a reference.

George H.
*Does my cheap rigol do A*B? hang on... Yeah it does..
I really need the average of A*B
Click to expand...

What bandwidth do you need? AD makes true-RMS chips for $5 or so.

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
 
On Tuesday, February 18, 2020 at 12:52:13 PM UTC-5, George Herold wrote:
On Tuesday, February 18, 2020 at 11:37:42 AM UTC-5, Phil Hobbs wrote:
On 2020-02-18 08:47, George Herold wrote:
So the last time I measured noise I used an analog multiplier
(AD734, I was going to say ~$20, but now $35!
https://www.digikey.com/product-detail/en/analog-devices-inc/AD734ANZ/AD734ANZ-ND/751025


So I'm thinking about methods to measure single channel noise
(V^2, V**2) and two chan. correlations V_a(t)*V_b(t)

0.) Digitize everything and do it in software.
This is not my strength, so someone else's box would
be my starting place. I must admit a fast (10 MHz) two
channel digitizer sounds like everyone's cheap DSO.*

But I'm looking for ideas.. one problem with cheap scopes
is the record length. You can save so much and then have to
down load it.

1.) Some switched gain thing like cheap lockin's
But I'd be doing Va(t)*sgn(Vb(t))
(where sgn is the sign of the voltage wrt gnd.)
or Vb*sgn(Va). Well it's not all that clear how
this relates to the correlations.. you get some sort of measure
if Vb and Va are the same size.
(I'll just note that Va*sgn(Vb) = Va*Vb/|Vb|)

or some other silly ideas...diode mixer?

I'm going to order a spendy AD734 as a reference.

George H.
*Does my cheap rigol do A*B? hang on... Yeah it does..
I really need the average of A*B


What bandwidth do you need? AD makes true-RMS chips for $5 or so.
Huh, BW... well as much as possible as always. :^)
I think my DMM goes to ~100kHz or so... I should check.
There's the AD8436
https://www.analog.com/media/en/technical-documentation/data-sheets/AD8436.pdf

That looks like a nice chip.
For the part with leads and in onsies, $10.
And no way to do correlations.

The bad thing about any switched gain thing is switching
spikes and slew rate....
the two bad things about switched gain are..

Well just searching other analog multipliers.
For ~$10 you can get the AD633, 1MHz small sig BW.
(small sig = 100 mV rms... Given the DC offsets
in analog mults, I mostly think about running 'em
balls to the walls.. (just the wee-ist bit of clipping
of the sig at the power rails.)* So I mostly want to
know the full power bandwidth. How much longer will the
AD734's be in production? Or better maybe is; What is the
market for these things?

George H.
* with noise if you wait long enough you'll always see some clipping..
(assuming you are above the 'small signal' level. :^)

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
Click to expand...

I just wanted to add, that someone's 'scope. keysight, rigol ?
Seems like it might be the right answer.
I need it to filter the math function AxB and output a number,
as a function of time..(or trigger signal?)
Maybe this is just my excuse to go buy a keysight scope?
GH
 
On Tuesday, February 18, 2020 at 11:37:42 AM UTC-5, Phil Hobbs wrote:
On 2020-02-18 08:47, George Herold wrote:
So the last time I measured noise I used an analog multiplier
(AD734, I was going to say ~$20, but now $35!
https://www.digikey.com/product-detail/en/analog-devices-inc/AD734ANZ/AD734ANZ-ND/751025


So I'm thinking about methods to measure single channel noise
(V^2, V**2) and two chan. correlations V_a(t)*V_b(t)

0.) Digitize everything and do it in software.
This is not my strength, so someone else's box would
be my starting place. I must admit a fast (10 MHz) two
channel digitizer sounds like everyone's cheap DSO.*

But I'm looking for ideas.. one problem with cheap scopes
is the record length. You can save so much and then have to
down load it.

1.) Some switched gain thing like cheap lockin's
But I'd be doing Va(t)*sgn(Vb(t))
(where sgn is the sign of the voltage wrt gnd.)
or Vb*sgn(Va). Well it's not all that clear how
this relates to the correlations.. you get some sort of measure
if Vb and Va are the same size.
(I'll just note that Va*sgn(Vb) = Va*Vb/|Vb|)

or some other silly ideas...diode mixer?

I'm going to order a spendy AD734 as a reference.

George H.
*Does my cheap rigol do A*B? hang on... Yeah it does..
I really need the average of A*B


What bandwidth do you need? AD makes true-RMS chips for $5 or so.
Click to expand...
Huh, BW... well as much as possible as always. :^)
I think my DMM goes to ~100kHz or so... I should check.
There's the AD8436
https://www.analog.com/media/en/technical-documentation/data-sheets/AD8436.pdf

That looks like a nice chip.
For the part with leads and in onsies, $10.
And no way to do correlations.

The bad thing about any switched gain thing is switching
spikes and slew rate....
the two bad things about switched gain are..

Well just searching other analog multipliers.
For ~$10 you can get the AD633, 1MHz small sig BW.
(small sig = 100 mV rms... Given the DC offsets
in analog mults, I mostly think about running 'em
balls to the walls.. (just the wee-ist bit of clipping
of the sig at the power rails.)* So I mostly want to
know the full power bandwidth. How much longer will the
AD734's be in production? Or better maybe is; What is the
market for these things?

George H.
* with noise if you wait long enough you'll always see some clipping..
(assuming you are above the 'small signal' level. :^)
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
Click to expand...
 
On Wednesday, February 19, 2020 at 4:52:13 AM UTC+11, George Herold wrote:
On Tuesday, February 18, 2020 at 11:37:42 AM UTC-5, Phil Hobbs wrote:
On 2020-02-18 08:47, George Herold wrote:
So the last time I measured noise I used an analog multiplier
(AD734, I was going to say ~$20, but now $35!
https://www.digikey.com/product-detail/en/analog-devices-inc/AD734ANZ/AD734ANZ-ND/751025


So I'm thinking about methods to measure single channel noise
(V^2, V**2) and two chan. correlations V_a(t)*V_b(t)

0.) Digitize everything and do it in software.
This is not my strength, so someone else's box would
be my starting place. I must admit a fast (10 MHz) two
channel digitizer sounds like everyone's cheap DSO.*

But I'm looking for ideas.. one problem with cheap scopes
is the record length. You can save so much and then have to
down load it.

1.) Some switched gain thing like cheap lockin's
But I'd be doing Va(t)*sgn(Vb(t))
(where sgn is the sign of the voltage wrt gnd.)
or Vb*sgn(Va). Well it's not all that clear how
this relates to the correlations.. you get some sort of measure
if Vb and Va are the same size.
(I'll just note that Va*sgn(Vb) = Va*Vb/|Vb|)

or some other silly ideas...diode mixer?

I'm going to order a spendy AD734 as a reference.

George H.
*Does my cheap rigol do A*B? hang on... Yeah it does..
I really need the average of A*B


What bandwidth do you need? AD makes true-RMS chips for $5 or so.
Huh, BW... well as much as possible as always. :^)
I think my DMM goes to ~100kHz or so... I should check.
There's the AD8436
https://www.analog.com/media/en/technical-documentation/data-sheets/AD8436.pdf

That looks like a nice chip.
For the part with leads and in onsies, $10.
And no way to do correlations.

The bad thing about any switched gain thing is switching
spikes and slew rate....
the two bad things about switched gain are..

Well just searching other analog multipliers.
For ~$10 you can get the AD633, 1MHz small sig BW.
(small sig = 100 mV rms... Given the DC offsets
in analog mults, I mostly think about running 'em
balls to the walls.. (just the wee-ist bit of clipping
of the sig at the power rails.)* So I mostly want to
know the full power bandwidth. How much longer will the
AD734's be in production? Or better maybe is; What is the
market for these things?
Click to expand...

Probably forever. If the 741 and the 555 are still around, legacy designs and legacy designers can clearly keep totally hopeless crap in production.

The 734 is actually a pretty neat part, and if it is still selling at the current price it's likely to last forever.

--
Bill Sloman, Sydney
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top