Re: Wideband ammeter

jlArbor.com wrote:

On Sun, 23 Mar 2025 03:59:46 -0000 (UTC), Sergey Kubushyn
<ksi@koi8.net> wrote:
 

Bill Sloman <bill.sloman@ieee.org> wrote:

On 23/03/2025 12:49 pm, john larkin wrote:

 
How about thermal imaging a surface-mount resistor?

 
Why bother? Measuring the voltage drop across the same device is easier,
and just as fast, if not faster.

>
That depends on what you actually want to measure. And "wideband" makes it
even more difficult.
>
Wideband current shunts made for AC-DC transfer are all of very special
costruction and cost arm and leg. If you want to measure the voltage drop
over those resistors, without making AC-DC transfer, you're up to another
challenge, measuring the AC voltage. Should start from the definition, what
IS the AC voltage? What the actual number your measurement shows means and
so on.
>
Look at e.g. not all that precise but much better than most LT1088 chip,
long obsolete. There is another one, proprietary and much better precision
inside e.g. Fluke 5790A Standard (which is a misnomer -- it is actually an
AC and DC voltmeter, 10x more precise that the venerable HP/Agilent/Keysight
3458A).
>
---
******************************************************************
*  KSI@home    KOI8 Net  < >  The impossible we do immediately.  *
*  Las Vegas   NV, USA   < >  Miracles require 24-hour notice.   *
******************************************************************

 
Many otherwise great DVMs have an AC bandwidth that doesn't even
handle the audio range.
 
The resistor has great common-mode rejection too.

Fluke 5790A is good to 1MHz on itself, without the wideband option. The
option makes it to 30 MHz but with limited voltage. The 5790A measures up to
1kV at 50 kHz.

However, with its accuracy and resolution it is not all that trivial to do
precise measurements at higher frequencies. You wouldn't notice it at all on
8-1/2 digit (on some ranges :)) 3458A. However, on the 5790A even a 12" coax
is NOTICEABLE starting from 50kHz and shows SIGNIFICANT drop at higher
frequency. 12" N-type to BNC cable like this:

https://www.amazon.com/dp/B07SJZMKRV

and BNC to dual banana like this:

https://www.amazon.com/dp/B0BTYKYS3Z

show significant losses at higher frequencies. Actually Fluke tells that 12"
is the MAXIMUM cable length for everything over 100kHz but they don't tell
WHAT cable it is :)

The N-type side is connected to 5790A, the banana part goes into my
Wavetek/Datron 4808 calibrator. 4808 stands right on the 5790A and the
shortest cable that I can use is a custom made 8" thick LMR400 with
right-angle N-type connector on the 5790A side. This cable is not
noticeable.

Hoping to measure wideband AC current by measuring a voltage drop over
regular resistor is simply insane.

This is what's used to measure AC current ACCURATELY:

https://www.ebay.com/itm/126975629050

It is not that it is the only option and it is very old (but still good and
usable) up to 50-100kHz range. They are guaranteed to be 1:1 for DC and AC
within that range and one should do a TRANSFER, not direct voltage
measurement. That is why the ABSOLUTE ohmic value is not all that important.
One applies that unknown AC current and gets a reading on the voltmeter.
Then, some KNOWN and characterized DC current, close to the expected unknown
AC current is applied and DC reading is taken. The shunts are guaranteed to
be 1:1 AC:DC so knowing the DC current and the ratio between known DC and
unknown AC readings allows to find the AC current.

That is if one wants good precise measurement. If it is to, say, plus/minus
couple per cents, all that complication is not needed.

And that is up to something like 50kHz, maybe up to 100kHz tops. Higher
frequencies require more sophisticated and significantly more expensive
measures.