BM235 LoZ Mode

[mbarszcz]

I use my multimeters for a combination of electronics and electrical work, and the LoZ mode of the BM235 was something that appealed to me, but the implementation of the mode seems to be quite different (worse) than the behavior on my Klein CL900 clamp meter (with LoZ).

On the Klein, the LoZ mode behaves much like the regular AC voltage measurement mode, except with a Low Impedance.  If I measure between the floating ground on my isolation transformer and the line/neutral, I get ~56VAC on the standard mode.  If I measure it with the LoZ setting, I get 0.0VAC on the Klein, exactly what I would expect a LoZ mode to do and be used for.

The manual of the BM235 seems to suggest that this is not simply a fixed low impedance mode (like what the Klein seems to do), but some sort of variable LoZ functionality, depending on what it is measuring.  It doesn't really get into how it behaves, and at what voltages, but just that it "ramps up".  The way it is worded also sounds like the LoZ really isn't that Lo when measuring higher voltages.

The manual states:

*Ghost-voltage Buster: Ghost-voltages are unwanted stray signals coupled from
adjacent hard signals, which confuse common multimeter voltage measurements. The
AutoV mode provides low (ramp-up) input impedance (approx. 2.1kΩ at low voltage) to
drain ghost voltages leaving mainly hard signal values on meter readings. It is an
invaluable feature for precise indication of hard signals, such as distinguishing between
hot and open wires (to ground) in electrical installation applications.

If I measure my floating ground again with the BM235, on regular AC volts mode it reads ~56VAC much like the Klein.  However on LoZ mode, it reads ~7VAC.  Not 0V, but always 7V.  The LoZ is obviously doing something, but is it this variable ramp up nonsense that is causing it to read 7V instead of just bleeding off the stray voltage?  I wish Brymen wouldn't have made it with the annoying Auto AC/DC/Ranging setting.
 Why couldn't we just have the AC or DC modes with a Low Impedance?  It looks like the Fluke 117 has this strange auto ranging LoZ mode as well, maybe that's where Brymen copied it from.  Either way, dumb.

Anyway, does anyone have any more information around how the LoZ mode works on the BM235 better than the manual explains it?  Why does it not bleed off all the stray voltage like the Klein?  User error?  Strange design decisions?  Faulty?  Is it only a low impedance up to a certain voltage?

[Kean]

The purpose of LoZ mode is not to give accurate readings at low voltage, but to indicate if a voltage reading is low impedance (potentially hazardous) or a high impedance (ghost) voltage.

If you read 56 VAC in normal AC measurement but something like 7 VAC in LoZ, then it indicates high impedance.
If you read 56 VAC in both modes, then it is low impedance and likely hazardous.

It is not a fixed input resistance in LoZ mode.  It is a combination of a fixed resistor in series with a PTC.  The PTC will start to heat up when a higher voltage is applied, and will increase in resistance the more it heats up.

This generally won't affect accuracy of low impedance readings too much, but if the input is left connected for a while to a low impedance voltage then the PTC can get quite hot and dramatically increase in resistance value.  It can take quite a few minutes for it to cool down and operate in the normal LoZ mode again, and thus high impedance voltage sources might read higher than expected.

[bdunham7]

The manual of the BM235 seems to suggest that this is not simply a fixed low impedance mode (like what the Klein seems to do), but some sort of variable LoZ functionality, depending on what it is measuring.  It doesn't really get into how it behaves, and at what voltages, but just that it "ramps up".  The way it is worded also sounds like the LoZ really isn't that Lo when measuring higher voltages.

Anyway, does anyone have any more information around how the LoZ mode works on the BM235 better than the manual explains it?  Why does it not bleed off all the stray voltage like the Klein?  User error?  Strange design decisions?  Faulty?  Is it only a low impedance up to a certain voltage?

All of the Lo-Z function meters use a few kilohms in series with a PTC for the load.  Your Klein is no different, its manual states that the Lo-Z impedance is ">3kΩ" and it will go much higher as you apply more voltage.  If it didn't and you applied 480V at 3k, you'd be dissipating about 75 watts which wouldn't work out very well.  If you have a small leakage current then most Lo-Z meters will indicate near zero and the impedance will indeed be low.  The BM235 apparently has a design quirk feature (not the variable impedance thing, something else) that results in a minimum voltage drop of about 8V in the Lo-Z load circuit.  IDK what is in there that does that, but it is mentioned in the manual.  It really isn't important for the most commonly intended use of Lo-Z, but it is a bit weird I suppose.  The Fluke 117 should behave just like your Klein meter.

From the manual:

Dave's Note: You can use the AutoV LowZ mode to discharge capacitors to a safe
level! (Can be slow on big caps) You'll even get a handy voltage reading to know
when it's safely discharged. Note that the LowZ mode does not work under 8VDC,
at which point it will be a 10M discharge.

[J-R]

If you read a bit further in the EEVblog BM235 manual, you'll find your answer to this behavior on page 26:

"Not specified at <8VDC    [I think they mean 8VAC]
Threshold: > 1VAC nominal
Input Impedance:
Initially approx. 2.1k, 164pF nominal; Impedance increases abruptly within a fraction
of a second as display voltage is above 8V (typical).
Impedances vs display voltages typically are:
12k @100V
100k @300V
240k @600V
580k @1000V
NOTE: The meter will NOT be low impedance for voltages under approximately 8V
DC. It will revert to the standard 10M"   [Who knows exactly what they mean here, see below.]


There is a more accurate explanation in the current version of the Brymen BM235 manual on page 19: http://www.brymen.com/images/ProductsList/BM230_List/BM230-print1-woCATII.pdf

"Not specified at <1VAC
Threshold: > 1VAC nominal
Approximate input impedance (//164pF) for reference:
At direct input </=50Vac (typical) from quiescence:
>8M@ < 5.6Vac
22k@ 7Vac
12k@ 8Vac
2.6k @ 50Vac
At direct input >>50V (typical) from quiescence:
Initial impedance is approximately 2.1k. Impedance increases abruptly within a
fraction of a second as display voltage (hard signal) is much higher than 50V (typical).
End-up impedances vs display voltages typically are:
12k@100V
100k@300V
240k@600V"

SO, that is the reason for the 7V AC reading.  This specific ghost voltage and the input impedance behavior of the BM235 are at equilibrium at around 7V AC.

I could make the case that the BM235 LoZ mode is at minimum misleading.  If it's not LoZ below 7V, then the label is essentially incorrect.

The BM789 LoZ mode behaves more like what we would expect.

[bdunham7]

SO, that is the reason for the 7V AC reading.  This specific ghost voltage and the input impedance behavior of the BM235 are at equilibrium at around 7V AC.

I could make the case that the BM235 LoZ mode is at minimum misleading.  If it's not LoZ below 7V, then the label is essentially incorrect.

I don't have one to test or disassemble, but I'm imagining there is something in the Lo-Z PTC circuit that needs to 'break down' to allow conductance, like a TVS or zener pair or something.  Thus it isn't an "equilibrium" but rather a threshold so that almost no 'ghost voltage' will ever read less than about 7V.  I agree it is misleading--not a problem for actual household electrical work but definitely an issue in other areas where you might be counting on a 2-3k impedance.

[floobydust]

LowZ is typically a PTC+surge resistor 1kΩ, or a couple PTC's in series, and the mode switch. Nothing else. You can look at BM235 teardown pics.
There are very few PTC's rated for 600V or 1,000V with regulatory approvals and high (cold) ohmic value in the many kΩ.
They are semiconductors after all, no specs at low voltage because the datasheet cares about the other end. 
I would think it's a semiconducting property for the "7V" spec? But they have different compositions as well.
Vishay PTCCL Series looks linear down to ~1V but the graph ends there.

[bdunham7]

They are semiconductors after all, no specs at low voltage because the datasheet cares about the other end. 
I would think it's a semiconducting property for the "7V" spec? But they have different compositions as well.
Vishay PTCCL Series looks linear down to ~1V but the graph ends there.

PTCs may be semiconductors but they don't have polar junctions and I've never seen one behave like anything other than a temperature sensitive resistor at any signal level. The I/V graph you show is logarithmic so it has to stop somewhere before zero, but you can see it is a straight line to that point.  IDK, the whole thing is pretty weird and apparently quite consistent.  I just tested the PTCs in my Fluke 289 (2 huge YS4020 PTCs in series with a 1k surge resistor for about 3.2k cold) and they are linear down to about 1.6mV or about 500µV per PTC.  If Brymen is using some part that behaves this way I'd like to know what the heck it is.  Furthermore, since they typically also use the PTC in the resistance circuit protection scheme, IDK how they would deal with that using a PTC with such behavior.

My first thought was that they were somehow routing the Lo-Z signal through the shunt protection bridge for some reason, perhaps a layout or switching efficiency thing, but I can't really see how that pencils out.  I'm almost curious enough to buy one to see what is going on--but not quite.  Perhaps Dave Jones could answer this question.

[floobydust]

In the 121GW, for LowZ two series PTC's are used but I don't know their p/n. Pic shows BM235 smaller one PTC1 is for LowZ and again no idea of p/n.

LowZ commonly uses different PTC's than for the Ohms-source protection where the YS4020 is popular for that. If you think about it, it can't have a "deadband" at all or you would lose compliance voltage. The part is expensive and basically huge, no coating on it either.

Thermistors have P or N-type materials, oxides, doping and valence bands doing their stuff.
Best part is we don't really know how they work. Cooper-pairs seems to be the best guess.

I'm suspecting zener-like behaviour in this LowZ PTC. I could be out to lunch but seeing multimeter schematics there is no other excuse I can find.
Working on 5V automotive, I'd expect LowZ to work, so moisture does not give misleading readings.

[bdunham7]

I'm suspecting zener-like behaviour in this LowZ PTC. I could be out to lunch but seeing multimeter schematics there is no other excuse I can find.

I'm not convinced, but I have no plausible alternative to offer.  I'll bug Dave, this is one of his meters and he did write that note in the manual.

[EEVblog]

I'm not convinced, but I have no plausible alternative to offer.  I'll bug Dave, this is one of his meters and he did write that note in the manual.

As others have said, LowZ meters are a series resistor plus non-linear PTC.
As such you can expect different behaviour based on the exact type of PTC used.
The spec exists for a reason, don't expect it to perform beyond that spec.
I do not have a schematic or a BOM.

As for the note, I don't recall my exact testing of this, but basically I'm saying that if the PTC doesn't conduct then you are going to get the normal 10M input impedance in parallel.

EDIT: I just measured LowZ input using another meter and it does indeed read 10Mohms, so the PTC is not conducting at all.
I also just measured the new BM2257 and it reads 2.2k using another meter vs 10M on the BM235, so it's obviously using a different PTC.

[bdunham7]

As such you can expect different behaviour based on the exact type of PTC used.

The part about that seems odd to me is that there are plenty of PTCs that don't behave that way and I've never seen or heard of one that does. Of course there's a lot I haven't seen or heard of so that might not mean much.  Still, given the wide availability of PTCs without this "feature" and no good reason that I can think of to deliberately include it, using such a part would seem an odd choice.

If anyone has a dead BM235 they want to send me, I'd be happy to dissect it and see what's going on.

[bdunham7]

OK, I think I've figured it out from an old video posted by joeqsmith.  Basically they use the same PTC for the LoZ, ohms and diode ranges, which isn't uncommon, but to minimize parts count and selector switch contacts they don't shunt the other side of the PTC to ground but rather send it through the same two-transistor protection clamp that serves the ohm/diode ranges.  That two-transistor Q8/Q9 clamp activates at a voltage of the BE-zener voltage plus one diode drop or about 7-8V. 

In the attached picture, you can see PTC2 (large) that is the one the LoZ/ohms/diode ranges use.  That goes to the aforementioned Q8/Q9 clamp and I presume that in LoZ mode the stuff to the right is just switched open by the selector switch.  The voltage reading is then taken through the upper circuit R7/PTC1/etc just like any other.  This accounts entirely for the observed behavior. 

[Kean]

Thanks, that explains the behaviour.
A model of clamp for simulation is at https://www.eevblog.com/forum/blog/eevblog-1157-transistor-zener-clamp-circuit/msg2040238/#msg2040238

This has saved me from having to pull one of my BM235s apart tomorrow to trace the circuit. 

[mbarszcz]

Thanks everyone for the good discussion and information.

The two transistor clamp protection clicked for me today after watching Dave's video:

Strange choice to send the LoZ mode on the BM235 through that.  It wouldn't matter in a lot of other applications, but in a multimeter where you are literally trying to read the voltage, it seems like a poor design decision on Brymen's part.

[EEVblog]

EDIT: I just measured LowZ input using another meter and it does indeed read 10Mohms, so the PTC is not conducting at all.
I also just measured the new BM2257 and it reads 2.2k using another meter vs 10M on the BM235, so it's obviously using a different PTC.

FYI, the new BM2257 will autorange down to 6V range in LowZ mode, so you get the full 1mV resolution all the way down to 0V.
Completely changed from the BM235