Z9 vs Batteries

Is there a friendly way to open and re-close the battery case? I tried burnishing the label with my thumb to reveal a screw hole dimple but didn't find any.
I actually haven't opened a battery case yet, but I know there are no screws. According to some videos and comments, it's not hard to pry open. The difficulty seems to be finding a way to put it back together, that would allow opening it again.

All of my studies have relied on opening a window or small holes to access cell terminals, rather than taking the case apart.
 
I see that it's the "MIPI" battery communication protocol.
How did you confirm that?
I'm not sure any more. I can't find the datasheet at Weltrend. The sister part WT6615F does list MIPI so maybe that was the source of my alleged confirmation:
I decided to get set up so I could look at the serial communications with my scope, and I certainly learned a few things!

The Old

Starting with my old EN-EL4a batteries and MH-21 charger, I found the communications follows ordinary serial formatting such as used with RS-232. Voltage levels are approximately +3.2V and 0V, and logic '1' is high. Baud rate is nominal 2400 baud, or more exactly 2460 baud as implemented, 1 start bit, 8 data bits LSB first, no parity and 1 or 2 stop bits in battery messages. (Charger commands have bytes paced out at 10ms intervals). All messages include a checksum byte at the end, and start with two 'FF' sync bytes.

The charger is the master and the battery only transmits in response to a charger query. This happens once every 2 seconds while charging is in progress. There is also a single short exchange, different command/data content, at the start of charging. Communication stops when the battery reaches 100%.

During charging, the battery responds to charger queries with a message containing 30 data bytes. I found the byte containing the '% charge' data and I could watch it incrementing with my scope. I haven't identified the other data fields yet; there are about a dozen bytes that change during charging. It is also possible that other data messages are exchanged during charging at a much lower rate, but I haven't captured anything different yet.

The New

On to the EN-EL18d and MH-33 charger, a number of details have changed. It's still 8-bit serial data, same voltages and logic levels, but with Even parity, always 1 stop bit. No sync bytes at message start, and no checksum byte appended. Baud rate is nominal 9600 (9500 actual).

There are 6 different messages that are requested from the battery, in a constant rotation that takes 1.6 seconds to complete. Charger queries are 4 bytes. Battery messages are all different lengths, from 5 to 28 bytes. Again, I found the '% charge' reporting byte, and also found a message which contains individual cell voltages in millivolts, to 13-bit precision. I like that! Now I won't need to drill access holes into any of my EN-EL18 batteries to check the cells.

And there's a Z9 firmware update request for Nikon: Could you please add the cell voltages to the battery status menu page?

Next Steps

It should be straightforward to connect the serial line to a level shifter/inverter and run that into an RS-232 port to monitor the data exchanges. I might do that; it would certainly be easier, faster and more reliable than manually decoding the bits from the scope screen.

I will also try to instrument my camera, so I can see how it communicates with the battery.

Eventually, I may experiment with sending other command codes to the battery to see how it responds. Hopes are that there might be a factory-reset command code that can clear out adverse cell data. Some of my old EN-EL4a packs are in a condition that can't really get any worse.

--
Source credit: Prov 2:6
- Marianne
 
Last edited:
Wow Marianne! That's really fascinating. As a ham radio operator for many years and Z9 user for a year, I appreciate all that info.

--
Regards,
Ken - LR ACE
FAA Remote Pilot Certificate, ATP ASMEL
Mizzou PJ '66
www.kenseals.com
 
Last edited:
There’s a deep dive for sure. It seems like the difference in communication protocols explains which combinations of battery and body are compatible and which aren’t.

Right now I’m waiting on parts to make a supervisor that stops the charging at 80%. It will be noninvasive because the front ends of three optocouplers are already present on the MH-33.
 
It's interesting that your thought process and methodology came to the 20%-80% optimum charge range for maximizing usable cycles.

That's the same range that the Toyota engineers used for the first few generations of the Prius in the attempt to extend battery longevity.
 
Prolonging battery life

Lithium batteries are sensitive to high temperatures, and are also subject to voltage stress if kept near 100% charge for significant time. Both of these reduce battery capacity and shorten service life. Do Not store your camera with a fully-charged battery pack; run it down to 80% or less before putting it away (60% charge is better). Always protect batteries from sources of heat.

It also helps to avoid deep discharge. The Z9 will give a low-battery warning when charge drops to about 21%, and that is a good time to swap the battery. Occasionally using the full battery capacity is not damaging, but doing so routinely will shorten the battery's service life.
Researching an electric car ( or post purchase research ?), it becomes very clear how many of Marianne's suggestions here are identical for the battery tech in modern electric cars.

Electric cars have one advantage over our camera battery charger - they allow you to set a "maximum charge" target ( i.e., 70/80/90% ) - trying to live in the 20-80% battery range to keep the batteries healthy.
 
...

Nikon implements true charge (Coulomb) counting to track battery charge state

...
But how can this handle self discharge?
To clarify, that is not a complete description of the battery condition monitoring. It is only a statement that charge/discharge current plays a major role. The full algorithm must be more complex.

It is also not to say that Nikon's implementation is without issues. Leaving a battery to self-discharge for a very long period, so that one or more cell voltages fall below 3V, can result in a significant underestimation of the battery's health and loss of usable capacity due to adverse cell history that results.
 
Being shy about hacking into my Nikon MH-33 charger for the EN-EL18d batteries, and not feeling shy about hacking the Wasabi charger that came in a set with two of it's batteries, I got this ball rolling.

I tacked thin wires onto the very tiny surface mount charge status LEDs and brought them out of the case and over to analog inputs on an Arduino Uno. The circuit points I tapped are downstream of the limiting resistors, therefore a bit "distanced" from the Wasabi's charge controller chip providing a tiny bit of ESD safety. The voltage states are 0.0V and 2.0V for Off and On states respectively.

[If I had instead fed the wires over to digital inputs on the Arduino even the On condition falls below the digital logic transition of 2.5V; it would always be misinterpreted as "Off".]

The "is it on or off" assessment threshold of the analog signal is set at 1.0V.

The set of three LED state signals are sensed at 250 msec intervals. I keep that up for 8 readings, therefore two cycles of the flash pattern. The flashing pattern is a 50% duty cycle 1 Hz square wave just like the Nikon MH-33. After I parse the results into Highs and Lows I can tell if the LED is solidly Off, pulsing, or solidly On.

The pattern therefore reveals the charge level at the same coarse level that we're used to on the MH-33. When the "80" LED is solidly lit I pull the plug on the charger power.

3cf84d10a56d44cfa6cecc9804281833.jpg

Here's a sample of the running steam of debugging text coming out of the Arduino's serial port:

As soon as the Interpretation reaches the >=80% threshold, the power to the Wasabi charger will be cut out.
As soon as the Interpretation reaches the >=80% threshold, the power to the Wasabi charger will be cut out.

Obviously this is not a very travel friendly system, physically.

But a robust refinement is possible. It turns out that there is enough room inside the Wasabi case to accept a small custom PC board. An Arduino Pro Micro and a few MOSFet switches could run the show from entirely inside the Wasabi case! When the charge level reaches 80% the MOSFets could interrupt the battery charge line and the communication lines. Thank you, Wasabi, for exposing those runs in the form of wire leads.

Left to right: Old school linear power supply fed by AC mains; charge controller board; dismounted battery connector daughterboard
Left to right: Old school linear power supply fed by AC mains; charge controller board; dismounted battery connector daughterboard

The solder bumps on the surface mount LEDs are less than 1mm across, so use a tiny soldering iron tip.
The solder bumps on the surface mount LEDs are less than 1mm across, so use a tiny soldering iron tip.

[ATTACH alt="I'm shining that flashlight into an empty space with about 2" x 3" x 1" free clearance"]3307613[/ATTACH]
I'm shining that flashlight into an empty space with about 2" x 3" x 1" free clearance

The source code should transport with few modifications required. The missing elements at this point are a PCB which I would design in Eagle CAD, and a prototyping of the data interruption line. I'm presuming that a P-channel power MOSFet would be OK on the battery + lead. Failing either of those, then I could use relays.

The battery I'm charging right now will hit the 80% point in an hour or two. I'll report back with the Z9's power level readings of the four batteries I have. I hope to see a small spread.

--
Wag more; bark less.
 

Attachments

  • 30721e7acc0a4e16b5f448c08ef0a752.jpg
    30721e7acc0a4e16b5f448c08ef0a752.jpg
    1.5 MB · Views: 0
The set of three LED state signals are sensed at 250 msec intervals. I keep that up for 8 readings, therefore two cycles of the flash pattern. The flashing pattern is a 50% duty cycle 1 Hz square wave just like the Nikon MH-33. After I parse the results into Highs and Lows I can tell if the LED is solidly Off, pulsing, or solidly On.

The pattern therefore reveals the charge level at the same coarse level that we're used to on the MH-33. When the "80" LED is solidly lit I pull the plug on the charger power.
Or, you could just "pull the plug" immediately when the "100%" LED turns on, since it starts flashing at the same time that the "80%" LED goes solid.
 
Or, you could just "pull the plug" immediately when the "100%" LED turns on, since it starts flashing at the same time that the "80%" LED goes solid.
Yup, very simple, that. Maybe I was overly cautious about spikes.

But something was amiss on the initial run. In three successive checks, during only 15 seconds, the Wasabi's indicated battery state went from X<50 then 50<X<80 then 80<X<100. It pole vaulted over 80. That battery in the Z9 reported 95% charge. In the MH-33 afterwards it topped out in about 15 minutes so that seems consistent.
  • This could be a Wasabi quirk regarding the reported state, meaning that it's not fully in agreement with Nikon's intended system.
  • Or, and now I'm reaching here, this battery has been up and down by a moderate amount a handful of times without approaching either endpoint, up or down. It makes me wonder if the Calibrate function might end up with erroneous data accumulated in the battery's data record if it isn't given a chance to complete normally. Sometimes an algorithm might not always do it's housekeeping in a timely way.
I have the next run going in another battery at the moment.

***

Bonus surprise discovery: Before hacking the Wasabi I had tried optically detecting the MH-33's LEDs for a non invasive setup. My optical sensor was just a run of the mill LED; fun fact, many of them work both ways. Works a treat with a very bright flashlight glaring righr into it. But they are mediocre compared to real photodiodes and gave me too little from the MH-33 lights. Moving on...

The strongest signal I got was from the sensing LED simply being very near the charger, even with no optical path. It was rectifying RF energy from the charger as if it was a crystal radio picking up a strong AM station. That fake signal cycled, but independently of the LED flashes. I didn't time it but it was longer than 1 Hz.

Did you see any modulation of activities like that while you were probing current?
 
Or you could use a Z9 in the same manner you would a D5 and find out how long the pack would last in real world usage..and then plan accordingly.
 
But something was amiss on the initial run. In three successive checks, during only 15 seconds, the Wasabi's indicated battery state went from X<50 then 50<X<80 then 80<X<100. It pole vaulted over 80.
This is normal. There is no LED state for =80 (or =50).
That battery in the Z9 reported 95% charge. In the MH-33 afterwards it topped out in about 15 minutes so that seems consistent.
  • This could be a Wasabi quirk regarding the reported state, meaning that it's not fully in agreement with Nikon's intended system.
Yes, Wasabi Power batteries have plenty of quirks. Their charge-estimation algorithm is based primarily on battery voltage, and it doesn't even compensate for internal impedance effects (i.e., terminal voltage higher during charge and lower during discharge).

The reported charge state is an overestimation while charging, and an underestimation while discharging, especially when below 50%. It's simply a poor design, in my opinion.

Your charge supervisor should work well with Nikon batteries, but will give variable results with the Wasabi battery.
  • Or, and now I'm reaching here, this battery has been up and down by a moderate amount a handful of times without approaching either endpoint, up or down. It makes me wonder if the Calibrate function might end up with erroneous data accumulated in the battery's data record if it isn't given a chance to complete normally. Sometimes an algorithm might not always do it's housekeeping in a timely way.
Calibration actually has very limited authority, even for Nikon batteries. For example, it can't overcome adverse cell history that results from a battery being left to self-discharge for several years, even when Cal/recharge cycles prove the battery still has very good capacity.

With the Wasabi battery, I've found no evidence that Calibration has any effect at all - not surprising if it is relying mostly on voltage, which wouldn't require calibration.
***

Bonus surprise discovery: Before hacking the Wasabi I had tried optically detecting the MH-33's LEDs for a non invasive setup. My optical sensor was just a run of the mill LED; fun fact, many of them work both ways.
Any diode - or bipolar transistor - with a transparent package will respond to light. Some of the early transistor radios had transistors with clear encapsulation, and if you exposed the internals to lighting such as fluorescent light, it would cause all manner of noise.
The strongest signal I got was from the sensing LED simply being very near the charger, even with no optical path. It was rectifying RF energy from the charger as if it was a crystal radio picking up a strong AM station. That fake signal cycled, but independently of the LED flashes. I didn't time it but it was longer than 1 Hz.

Did you see any modulation of activities like that while you were probing current?
No, the current probe, by virtue of its closed magnetic circuit, is not sensitive to most noise sources. A scope voltage probe will certainly pick such things up, but I haven't explored that.
 
Or you could use a Z9 in the same manner you would a D5 and find out how long the pack would last in real world usage..and then plan accordingly.
Ultimately, that's what will happen. But December through February is off-season and I was anxious to have answers before the new season starts.

Very first competition of the year was last Saturday. It was a small event, and ran just a little over 4 hours. I took 9400 frames with the Z9, all with a single battery. Charge state at the beginning was 91%, and ended up at 28% when I was done.

This is a promising result, and does not differ greatly from battery performance when using the D5.
 
Or you could use a Z9 in the same manner you would a D5 and find out how long the pack would last in real world usage..and then plan accordingly.
Ultimately, that's what will happen. But December through February is off-season and I was anxious to have answers before the new season starts.

Very first competition of the year was last Saturday. It was a small event, and ran just a little over 4 hours. I took 9400 frames with the Z9, all with a single battery. Charge state at the beginning was 91%, and ended up at 28% when I was done.

This is a promising result, and does not differ greatly from battery performance when using the D5.
Promising indeed, but just to be clear: I assume you used an EN-EL18D in the Z9 and perhaps EN-EL18A or B, C in the D5? There is a difference between 2500mAh/27Wh for the A, B, and C vs. 3300mAh/36Wh for the D version.
 
Or you could use a Z9 in the same manner you would a D5 and find out how long the pack would last in real world usage..and then plan accordingly.
Ultimately, that's what will happen. But December through February is off-season and I was anxious to have answers before the new season starts.

Very first competition of the year was last Saturday. It was a small event, and ran just a little over 4 hours. I took 9400 frames with the Z9, all with a single battery. Charge state at the beginning was 91%, and ended up at 28% when I was done.

This is a promising result, and does not differ greatly from battery performance when using the D5.
That is a positive result. I’ve found from my heavy usage over the past year shooting birds, that when JUST shooting and using only the viewfinder, battery life is very good. Chimping and using the rear screen in general, however, really take a toll.

Thanks for your efforts in and sharing of your findings of your battery experiments.

Also, in light the imminent demise of DPR, thanks VERY much for all the excellent contributions you’ve made here on the forums over the years!

--
Alan Clark
https://arclark.smugmug.com/
 
Last edited:
Much good info above.

A clarification of what I meant to convey about the LED sequencing: the Wasabi states in trial 2 went from flashing 50 to flashing 100 without apparently visiting the state of flashing 80. If it did touch that state it couldn’t have lasted more than a few seconds. That was on a Wasabi battery. On the other hand, trial 3 was on a Nikon battery and that spent 4 hours in “flashing 80”. Thanks for your confirmation that I need to track battery brands separately. I don’t want to perceive more randomness than there is.

So far the most consistent observation is that the onset of flashing 100 means the battery is very close to 100%, not 80.

In one trial so far the onset of flashing 80 correlated to a 41% reading in the Z9. Maybe that’s fine for storage.

[EDIT]

The Wasabi LED indicators have an uncalibrated response sequence, it's as simple as that.

It was flashing the "80" which until my testing I had assumed meant "<80", but when I put the battery into the Z9 it reported 85%. I continued with the charging and as soon as the "80" went solid I put it in the Z9 who measured it as 98%. So if I want to make the Wasabi take me near 80% I'll need to have my supervisor gizmo stop watching LEDs earlier, then add some fixed time interval as a dead-reckoning extension. That's not hard. And if that misses 80 by 5% that's no hardship either.

--
Wag more; bark less.
 
Last edited:
does not differ greatly from battery performance when using the D5.
Thumbs up.

If you care to share them it would be nice to hear your AF experiences too.
 
Or you could use a Z9 in the same manner you would a D5 and find out how long the pack would last in real world usage..and then plan accordingly.
Ultimately, that's what will happen. But December through February is off-season and I was anxious to have answers before the new season starts.

Very first competition of the year was last Saturday. It was a small event, and ran just a little over 4 hours. I took 9400 frames with the Z9, all with a single battery. Charge state at the beginning was 91%, and ended up at 28% when I was done.

This is a promising result, and does not differ greatly from battery performance when using the D5.
Results very similar to mine. I photographed an airshow with my Z9 last September. 8000+ shots and 30% battery left.

--
Regards,
Ken - LR ACE
FAA Remote Pilot Certificate, ATP ASMEL
Mizzou PJ '66
www.kenseals.com
 
Last edited:
Very first competition of the year was last Saturday. It was a small event, and ran just a little over 4 hours. I took 9400 frames with the Z9, all with a single battery. Charge state at the beginning was 91%, and ended up at 28% when I was done.

This is a promising result, and does not differ greatly from battery performance when using the D5.
Promising indeed, but just to be clear: I assume you used an EN-EL18D in the Z9 and perhaps EN-EL18A or B, C in the D5? There is a difference between 2500mAh/27Wh for the A, B, and C vs. 3300mAh/36Wh for the D version.
Correct. All of the work I've done with the D5, used the EN-EL18a. That is my comparison basis, and I believe it's most relevant to other users experienced with the D5 as well. I'm fairly sure that across the entire user history for the D5, relatively few operating hours have been accumulated with an EN-EL18d battery.

But for completeness, I did show a plot for the D5 with EN-EL18d, in my post regarding run times and shots per charge, earlier in this thread.
 

Keyboard shortcuts

Back
Top