Null Detector based on Conrad Hoffman design check

[hozone]

That's interesting, seems the difference is that your design has a + indicator as well as a - indicator.
Can you please share the LTspice file? I'll pack it with the detector files.
Later I'll share the KiCad source too, when all is done.
Find attached the .asm file for the 5V low monitor. As Kleinstein stated current consumption is pretty high, 4mA almost. Need to investigate an alternative way maybe.

[Kleinstein]

4 mA is indeed pretty high for battery operation. The alternative is a comparator / OP to compare the 2.5 V (half of 5 V) to some 45% of the input voltage. This could getaway with some 200-250 µA when using a MCP6001/2 and less if needed.

[iMo]

That's interesting, seems the difference is that your design has a + indicator as well as a - indicator.
Can you please share the LTspice file? I'll pack it with the detector files.
Later I'll share the KiCad source too, when all is done.
Find attached the .asm file for the 5V low monitor. As Kleinstein stated current consumption is pretty high, 4mA almost. Need to investigate an alternative way maybe.

The wiring is different, see yours and mine.
Attached the sim file. Mind it is a sim only, I would highly recommend to try it in HW whether it works fine. Also blinking the LEDs might be "dangerous" for your design - the weak virtual ground might cause issues with 1mA pos/neg pulses.

[hozone]

Thanks!

For the low voltage detector I'm moving to MX810, the one with 4.6V threshold, and I'll put it after the LDO. I've worked with MAX809 before (for other purpose), they work good. Do you think it will be ok even there?

I've notice the peak detection in your circuit changes regarding the led used, and op amp used. Find my asm attached. I've to check my design with LTspice too. For my purpose I think there will not be many blinking.

[iMo]

The behavior of my null indicator changes based on
a) the opamp is rail2rail or not,
b) the opamp can work at +/- 2.5V
c) the LED diode forward voltage - with only +/- 2.5V your LED's Vf has to fit somehow (based on the color the Vf is from 1.6V to 3.5V, afaik).

PS: I doubt your above indicator may work as there is none feedback manipulating the input threshold levels..

PPS: Recommended reading:
Jerald G. Graeme: Applications of Operational Amplifiers, Third-Generation Techniques, McGraw-Hill, 1973, p.112
 

[hozone]

I've to investigate about that specific implementation, but I can confirm you I've one built in another project that works. It's RC4558 based, +-15V rail and it peaks almost +-8.5V.
Note: posting here I've notice this +-2.1V implementation has +- rails inverted, I've to swap them.

[trobbins]

Thanks for the thread - it has inspired me to start preparing a null detector to drive the original galvanometer in a nice 1950's PYE wheatstone bridge I tweaked last year. 

I had been using the mV range on an Aneng AN8009 cheap DMM in lieu of the galvanometer for much better auto-ranging and discrimination, but I think a bit of interface circuitry based on an AZ opamp could well do a much better job, and could be easily powered by 5V from an external USB battery brick.  The application doesn't need accurate FS voltage ranging, and the load is likely a few tens of uA (although I haven't measured what the meter FS current actually is).  I'll try and keep the original initial/final keying of the galv, and retrofit the final key with a 'super final' press-button.

[Kleinstein]

A power bank could add quite some EMI issues. With some care the circuit can be relatively low power and well get away with direct battery supply with a linear regulator, as shown. The OP-amp can also work with less than 5 V so even a single LI cell can work.

[hozone]

In a couple of weeks I think I'll take this PCB to production, I'll keep you updated.

[hozone]

Forget to post here the peak-detector based on my design. Op amp is 741, and like the iMo design, changes in the Led and the Op amp makes changes in the R5 detector resistance.
One option could be reuse the 100k trimmer already used in the BOM in place of the R1,R5 or use one trimmer in place of R5 and a fixed R1.

[trobbins]

Further to my post #56, I ended up using an LTC2054 AZ opamp with 640x gain as a simple preamp to the galvanometer in a vintage PYE 7440/G wheatstone bridge.  This effectively gave a net gain of 10x to the null display, as I wanted to drive the combined meter and Ayrton-Mather damping network to maintain needle damping (and that imposes a 64x attenuation).  I can now easily discern sub 1 ohm step changes for a bridge measurement of 10kohm with 12Vdc bridge supply, whereas before it took about a 5-10 ohm step to see a discernable needle movement (unless I used a significantly higher than 12Vdc bridge supply).

It is a neat upgrade that just needed a 5Vdc input barrel socket added to the top panel, as I was able to use the front panel switch contact for an external galvanometer to power the preamp and connect it into the meter circuit.  I was able to use a 5Vdc USB power brick (a Sony CP-F2L) for powering the preamp - the brick does add 50mV transients each 240us (4kHz) on a 20mV base ramp, with some dither, but that was simply filtered with an RC and had no impact on opamp output.  The only other aspect of using a brick is the need for a load to keep it connected - which for this brick was 390R, or 65mW.

The LTC2054 output was nulled to <1mV in an un-energised bridge, and when the preamp was on it just noticeably deflected the meter needle, which can be easily reset to mid-point if needed by the top panel manual meter 'set zero' adjustment, so no ongoing need to adjust bias/offset for time and temperature. I'm now a convert!

[hozone]

Thanks! That's interesting.
I'm still evaluating the blinking circuit, unluckily I've little time to dedicate to this project.