New eevblog multimeter?

[Kleinstein]

Folks,

this weakness of the 121GW has nothing to do with Crest factor in first instance. This instrument  simply has no proper overload detection for its (internal) OpAmps, at about +/- 1.6V, or so. Therefore, signal clipping occurs, which falsifies of course the RMS calculation.

The crest factor is just a different way to express the peak voltage. With a 500 mV full scale range at the RMS converter, the 1.6 V limit of the amplifier would allow for a crest factor of about 3 when at the full scale. With more there will be clipping and thus a possibly large error, not just the slowly increasing error from the AD8536 data-sheet, that is valid with a smaller amplitude (e.g. 200 mV or 300mV with a higher supply voltage).

Specs for the peak value at something like 3 times nominal range may be less confusing. This would imply a limit for the crest factor of 3 at full scale, allowing more at lower voltage.

[EEVblog]

This overload detection requires a fast sampling of the amplified signal, sort of peak-window detection, which is not implemented. Even in ACV it's not implemented, what makes me wonder. The fast internal A/D could serve this purpose.
Better DMMs feature a separate fast window comparator for that.
So you have to live with this weakness and may anticipate manual upranging, when you expect 'difficult', non-sinusoidal signals.

Yes. Kane have had a lot of problems with the HY3131 chipset, and have specifically noted that there are issues with using both internal (fast and slow) ADC's at the same time and/or switching between them.
They have even sent engineers to Hycon several times to try and resolve this sort of stuff, but it seems there are just inherent limitations in the chip. Particularly so when AC and DC share the same switch position, which is the issue with the volts autoranging. That's not unique to Kane either, other brand meters that share AC/DC volts knob position can also have issues with detection of autoranging with DC offsets on AC signals. It's why the new design meter uses separate AC and DC switch positions.

[EEVblog]

Of course crest factor can't be an excuse if in the video an Aneng meter is showing correctly ... what are we talking about ? 
As I said in many many cases the current is in pulses , that's how a diode bridge rectifier + filter cap works . This is absolutly  basic

The waveform show is very different to a full or half wave bridge rectified signal. The one shown obviously has some form of zero volt crossover switching that makes it switch hard on, so gives a sharp turn on current edge. Why does everyone keep trying to equate it to a basic rectifier? 

[2N3055]

Of course crest factor can't be an excuse if in the video an Aneng meter is showing correctly ... what are we talking about ? 
As I said in many many cases the current is in pulses , that's how a diode bridge rectifier + filter cap works . This is absolutly  basic

The waveform show is very different to a full or half wave bridge rectified signal. The one shown obviously has some form of zero volt crossover switching that makes it switch hard on, so gives a sharp turn on current edge. Why does everyone keep trying to equate it to a basic rectifier? 

To me that looks like some SCR phase controlled switching waveform..

[Jay_Diddy_B]

Hi,

Here is a quick model of a 'wall wart' power supply, a very typical line frequency application:




The waveforms are:





The Crest Factor, peak / rms, for the input current is 2.7 This is a very typical application.

Regards,
Jay_Diddy_B

[rattnallen]

To me it looks like the current into the smooting capacitor, i.e. the current into a normal diode bridge rectifier with a smoothing capacitor after the bridge...
That is something almost every cheap gadget that you connect to the mains nowdays will show.

[Jay_Diddy_B]

Hi,

Here is a model for phase control with a resistive load, a typical Triac light dimmer:




The waveforms




And the measurements:




At 10^o The Crest factor is 7.4 but the peak current is low.
This is probably okay unless the DMM auto-ranges to a range that cannot handle the peaks.

This is probably the highest crest factor found in 'normal' applications.

Jay_Diddy_B

[Dr. Frank]

Hi,

To build on what Dr. Frank said.

I scanned the 121GW manual to see what it had to say about crest factor.

The manual echoes the Analog Devices specifications for the AD8436 as having CF of 10.

I am not sure if the crest factor is preserved in the design, because it would require a lot of headroom, without clipping, in the amplifier stage in front of the AD8436.


Crest factor is the peak / RMS

The Fluke 289 DMM has a crest factor of 3 at full scale and increases to 5 at mid-scale. This make sense because there is more headroom at mid-scale.

You can find tables on the internet that illustrate various crest factors.

A 5% duty-cycle square wave has a crest factor approximately equal to 4.5

A 1% duty-cycle square wave has a crest factor of 10

Most waveforms that you measure will be lower.

Regards,
Jay_Diddy_B

It's getting even more complicated, as inside the 121GW in ACI mode, the RMS converter is AC coupled only.

I experimented a bit with asymmetric current pulses, i.e. positive square wave pulses only, generated by a variable duty cycle pulse generator and a V/I converter, applied to the 121GW, BM869, KS34465A and a HP3458A in series, and often got about 4 different results, due to different coupling and RMS conversion methods. Of course the over range/clipping indication was also different.

That all was quite confusing, i.e. what would really be the meaning of the displayed RMS measurements, even on these other instruments which have proper (over-) ranging functionality. Let aside the 121GW, on all other DMMs it's always very difficult to find out from specification or the manual, how the AC signal is coupled internally and how the RMS conversion is really accomplished.

Frank

[Marco1971]

Hi,

To build on what Dr. Frank said.

I scanned the 121GW manual to see what it had to say about crest factor.

The manual echoes the Analog Devices specifications for the AD8436 as having CF of 10.

I am not sure if the crest factor is preserved in the design, because it would require a lot of headroom, without clipping, in the amplifier stage in front of the AD8436.


Crest factor is the peak / RMS

The Fluke 289 DMM has a crest factor of 3 at full scale and increases to 5 at mid-scale. This make sense because there is more headroom at mid-scale.

You can find tables on the internet that illustrate various crest factors.

A 5% duty-cycle square wave has a crest factor approximately equal to 4.5

A 1% duty-cycle square wave has a crest factor of 10

Most waveforms that you measure will be lower.

Regards,
Jay_Diddy_B

LTC1968 True RMS to DC converter used by Fluke 287/289

https://www.analog.com/media/en/technical-documentation/data-sheets/1968f.pdf

Marco1971

[Jay_Diddy_B]

Hi,
Analog Devices (LTC) provides this data for the LTC1968:




This is relative to 200mV output, which is mid-scale.

They use SCR waveforms, the same as modeled earlier.

This matches the Fluke 289 performance.

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

It's getting even more complicated, as inside the 121GW in ACI mode, the RMS converter is AC coupled only.

I experimented a bit with asymmetric current pulses, i.e. positive square wave pulses only, generated by a variable duty cycle pulse generator and a V/I converter, applied to the 121GW, BM869, KS34465A and a HP3458A in series, and often got about 4 different results, due to different coupling and RMS conversion methods. Of course the over range/clipping indication was also different.

Snip ...

Frank

1+

I don't think it is possible to AC couple an asymmetric waveform and measure the RMS .

Simple example:

10V dc with 2.8mV p-p sinewave ripple.

RMS is approximately 10V

If you ac couple you read 1mV RMS

Regards,
Jay_Diddy_B

[HKJ]

I don't think it is possible to AC couple an asymmetric waveform and measure the RMS .

That is exactly what good multimeters do. You get separate AC and DC values and you may also have a AC+DC that combines them.

[Kleinstein]

At least in theory the AC coupling does not cause much problem. The  total RMS including the DC part would be the square root of the DC reading squared plus the square of the AC coupled RMS reading. This general formula also applies to asymmetric waveforms. The AC coupling and limited bandwidth can be a problem with peak to peak readings, as this is sensitive to phase shifts between the different frequency components. Phase shifts can have quite some effect before the amplitude changes much.

The LTC1968 RMS converter is different from the classical ADxx36 ones in using a kind of SD ADC inside. This may lead to relatively hard internal clipping, while the analog nonlinear circuit is more like slowly increasing errors first. The more analog AD8436 is better behaved in this respect. However this does not help of the amplifier in front (or the absolute value part of the circuit) does hard clipping.
Chances are the GW121 will be much more tolerant to a high crest factor signal more in the center of the range.

A maximum crest factor of 1.5 like in the Fluke289 near the end of scale limits this to essentially sine only (already crest factor of 1.4). Compared to this the GW121 should be still good.

[Dr. Frank]

This overload detection requires a fast sampling of the amplified signal, sort of peak-window detection, which is not implemented. Even in ACV it's not implemented, what makes me wonder. The fast internal A/D could serve this purpose.
Better DMMs feature a separate fast window comparator for that.
So you have to live with this weakness and may anticipate manual upranging, when you expect 'difficult', non-sinusoidal signals.

Yes. Kane have had a lot of problems with the HY3131 chipset, and have specifically noted that there are issues with using both internal (fast and slow) ADC's at the same time and/or switching between them.
They have even sent engineers to Hycon several times to try and resolve this sort of stuff, but it seems there are just inherent limitations in the chip. Particularly so when AC and DC share the same switch position, which is the issue with the volts autoranging. That's not unique to Kane either, other brand meters that share AC/DC volts knob position can also have issues with detection of autoranging with DC offsets on AC signals. It's why the new design meter uses separate AC and DC switch positions.

Hi Dave,
I again dug deeper into the schematic and the HY3131 datasheet. Latter already has a programmable, fast window comparator on board, both lower and upper limits are programmable everywhere inside the V_DD, V_SS range, which is intended for monitoring the output of OP1 .. can't understand why this could not implemented for ACI, at least.

Should also work for this ACV + DCV problem.
Yes, the ACV signal is routed over OP1 again to the external RMS converter.
Therefore this comparator should be able to detect any overload conditions also.

Frank

[dcac]

This overload detection requires a fast sampling of the amplified signal, sort of peak-window detection, which is not implemented. Even in ACV it's not implemented, what makes me wonder. The fast internal A/D could serve this purpose.
Better DMMs feature a separate fast window comparator for that.
So you have to live with this weakness and may anticipate manual upranging, when you expect 'difficult', non-sinusoidal signals.

Yes. Kane have had a lot of problems with the HY3131 chipset, and have specifically noted that there are issues with using both internal (fast and slow) ADC's at the same time and/or switching between them.
They have even sent engineers to Hycon several times to try and resolve this sort of stuff, but it seems there are just inherent limitations in the chip. Particularly so when AC and DC share the same switch position, which is the issue with the volts autoranging. That's not unique to Kane either, other brand meters that share AC/DC volts knob position can also have issues with detection of autoranging with DC offsets on AC signals. It's why the new design meter uses separate AC and DC switch positions.

Hi Dave,
I again dug deeper into the schematic and the HY3131 datasheet. Latter already has a programmable, fast window comparator on board, both lower and upper limits are programmable everywhere inside the V_DD, V_SS range, which is intended for monitoring the output of OP1 .. can't understand why this could not implemented for ACI, at least.

Should also work for this ACV + DCV problem.
Yes, the ACV signal is routed over OP1 again to the external RMS converter.
Therefore this comparator should be able to detect any overload conditions also.

Frank

In AC modes the comparator is already used for the frequency measurement. Though ACI does (currently) not seem to show frequency so here it should be possible to free the comparator and, as you describe, use it for an out or range detector. Upper and Lower reference can conveniently be set at 0.1V and 3.5V - so if OP1 output exceeds this range it’s time to change range upwards or to show OFL.

Correction: ACI does actually show frequency.

And there does not seem to be any other route to the hy3131 counter when measuring AC. The CNT pin seems only connected in Hz-Duty modes.

[CDaniel]

AC current measurement is identical to AC millivolt or at least the chipset producers intended to be made in this way  , so should't be any issues detecting over range inside HY3131 if the firmware is working correctly , even if the input waveform is a pulse , has DC offset and so on . Or we have an issue with voltage too ? 
The external x10 amplifier could clip the signal at some point , but the firmware should be able to switch ranges by then .

[dcac]

Yes possibly a problem in ACV and ACmV too. Also here OP1 is used as 1x and 10x buffer - so it can be clipping the waveform before is reaches RMS chip.

Dr. Frank is correct the comparator presents a possible firmware solution - but as it is now it is already used for the frequency measurement in AC modes.

But to detect AC+DC voltages in DC mode and select the appropriate range - which it failed to do in Joe's test - should be doable without losing other functionality.

[Dr. Frank]

Yes possibly a problem in ACV and ACmV too. Also here OP1 is used as 1x and 10x buffer - so it can be clipping the waveform before is reaches RMS chip.

Dr. Frank is correct the comparator presents a possible firmware solution - but as it is now it is already used for the frequency measurement in AC modes.

But to detect AC+DC voltages in DC mode and select the appropriate range - which it failed to do in Joe's test - should be doable without losing other functionality.

Last point requires ultimately an overrange detection, that's not safely possible with DC mode conversion, I think .. last chance would be use of the fast A/D, but I'm not sure about its input range limitations and its capability to detect high count / low count conversions.
Otherwise, better sacrifice the 2nd display for frequency to have a safe and properly working AC functionality.

Frank

[dcac]

Yes possibly a problem in ACV and ACmV too. Also here OP1 is used as 1x and 10x buffer - so it can be clipping the waveform before is reaches RMS chip.

Dr. Frank is correct the comparator presents a possible firmware solution - but as it is now it is already used for the frequency measurement in AC modes.

But to detect AC+DC voltages in DC mode and select the appropriate range - which it failed to do in Joe's test - should be doable without losing other functionality.

Last point requires ultimately an overrange detection, that's not safely possible with DC mode conversion, I think .. last chance would be use of the fast A/D, but I'm not sure about its input range limitations and its capability to detect high count / low count conversions.
Otherwise, better sacrifice the 2nd display for frequency to have a safe and properly working AC functionality.

Frank

I measured the lower frequency limit in AC mode to be about 2Hz. So perhaps the comparator can alternately be used both for clipping detection and frequency - if you switch between the two at i.e. 1 sec intervals. One drawback would be a less responsive frequency display - if frequency changes it will take a moment to show.

But 1 sec period should enough time to 'sense' the input signal for potential clipping. The comparator has a latching output and 2Mhz BW at 40mV P-P, which seems more than fast enough for what we dealing with here.

Implementing something like this is however a more substantial rewrite of the firmware - more than I think we seen in previous updates.

[CDaniel]

I checked in mV range and surprise or not  for simple square wave 500mVpp 100Hz , no DC offset like Dave said it is the cause , it is not autoranging right if the duty cycle is below 9% . It is showing something ( clipped ) in the 50mV range and  is showing correctly in the 500mV range only if you manually switch it .

I used a Fluke 187 for comparison , and it is  fooled in the same way just below ~2% duty cycle .

Autoranging detection  from around 10% duty cycle is pretty bad and probably the cause for not measuring the current right . For currents it is much more visible as the pulses have low duty cycle ( of course  ) ...

I would speculate that the reason is the slow / inefficient firmware and not hardware

[Jay_Diddy_B]

CDaniel and the group,

If the signal is DC coupled the RMS value

 = pk  x ( duty_cycle)^0.5

for a 10% duty-cycle

peak 500mV

The correct answer = 158mV

If it is AC coupled

You calculate

 pos = +PK x (DUTY_CYCLE)^0.5

neg = -PK x (1 - duty_cycle) ^0.5

Total = ( pos² + neg²)^0.5

= 150.02 mV

The Crest Factor  = pk /rms

CF (dc coupled) = 0.5 / 0.158 = 3.16

CF (ac coupled) = 0.45 / 0.15 = 3.0

 I am not sure that you should be measuring a unipolar square wave on the AC range.

If you measure on the DC range

you should measure the pk x duty_cycle = average voltage = 50mV

But you may also run into clipping if the auto range is on.

Regards,
Jay_Diddy_B

[CDaniel]

Why you assume it's unipolar ?... I used alternating square wave to be clear , centered on 0V , that is what I meant when I said no offset .
And yes , the correct value is around 150mV but 121GW is showing about 52mV in 50mV range at 9% duty cycle  . With 1% duty cycle more added ( 10%)  will over range and measure correctly in 500mV range 

[Jay_Diddy_B]

Hi,

If you are dc coupled and +/- 250mV, symmetrical about ground.

The correct answer is 250mV

If you are ac coupled the and the signal is +/-250mV wrt ground the

The correct answer is 150.02mV

This is because the ac coupling removes the dc offset.

Jay_Diddy_B

[Jay_Diddy_B]

Hi,

I set my HP8116A to generate 100Hz 10% duty-cycle  High level 500mV Low Level 0V

I connected a 50 precision termination (0.1%) and measured the voltage with a few of the meters that I have in the lab.

Here are the results:

Datron 1281




I don't have an HP3458A. I have to make do with a Datron 1281 

Fluke 85 III







Fluke 289




HP 3457A on the AC range



HP3457A on the ACDC range





The HP8116A isn't super accurate or stable.

But all of these readings are close to what they should be.

Regards,
Jay_Diddy_B

[CDaniel]

Very well , this is the result to be expected , but 121GW will show something about 3 times lower in the 50mV range as long as the duty cycle is below 9% , then will jump and show ~150mV.

As I said a Fluke 187 will autorange correctly from 2% duty cycle . This meter is similar 50,000 counts and has  50mV and 500mV ranges

This makes all the difference when measuring current from a diode rectifier + filter cap wich has low duty cycle