NEW PDVS2mini DC Voltage Reference from Ian Johnston - Reviewed - NOW WRYTECH

[Gertjan]

Good test. I wonder if you could make improvements though.

1. If the tin is mostly sealed the air inside will stratify, I.e. temperature layering causing issues, especially since the PDVS2mini generates heat itself.

2. If going below ambient the tin may condensate causing temperature and humidity issues.

I think changing out the tin and using a different material to barrier from direct airflow would be better.

Ian

Hi Ian,

Good points.
I must admit that I used the cookie tin, because I use it for more shielded measurements. (The tin is connected to the guard of the 3458A)

It is always good to consider imperfections in ones measurement setup, especially when results are different from expected.

1 Re stratification: The LM35  ambient temperature sensor was at the same height (about the middle of the PDVS2mini box), and positioned that it would not be influenced too much by the PDVS2 temp, and also not too close to the cookie tin wall.

2 The temperature changes were very slow (taking 2 hours to go from 20,7°C to 14°C). So I do not think condensation occurred.
  Risk of condensation was the reason I did not want to go any colder. Maybe next time I should also put a humidity sensor in the box.

I already did a second measurement, with even slower temperature changes. I hope to post the results later today.

regards, Gertjan.

[Gertjan]

a diagram with internal temperature as x-axis would be interesting.
And also if there is a difference between ramp up vs. ramp down (to see whether there is any hysteresis).

Hi Andreas,

Indeed,a diagram with internal temperature as x-axis would is interesting, and I make them of every tempco measurement.

I choose not to present them this time because the measurement sweep did not return to the starting temperature, and thus this plot is less informative.
Meanwhile I did a second measurement, with a full sweep, and I will show these diagrams. Hopefully later today.

regards, Gertjan.

[Gertjan]

2nd PDVS2mini v2 Tempco measurement

I did a second tempco measurement, using the experience gained from the first. (But unfortunately before Ians ideas for improvement)

Now the PDVS2mini v2 was powered externally, so no battery problems. The PDVS2mini was switched on the evening before, so the batteries were fully charged, and extra warmth of the charging was gone long before the measurement started.

This time the temperature changes were even slower, so the PDVS2mini internal temperature could track the ambient better. And at the top and bottom temperatures I let the temperature dwell extra long, until PDVS2mini int. temp and output were stable too.


Full size graph: PDVS2mini-v2-tempco-ext-ps-29-09-2025-1200pix.png

Blue trace is the ambient temperature. After letting everything stable for 30 minutes, the temperature goes slowly down in 2hours to 14.5°C. There it dwells for 1:20hour, until PDVS2mini internal temp and output are stable too.
Then the temperature rises slowly, in 4hours to the maximum temp of 31.7°C. There it dwells again for 1:30hours, again until PDVS2mini internal temp and output are stable too.
Then in 2hours back to the starting temperature. Unfortunately there was some overshoot. So, with power from the Peltier removed, I let it dwell towards an end temperature, with everything stable.

Green is the PDVS2mini v2 internal temperature. It is lagging the ambient temp. One can see where the ambient temp is constant, that the internal temp needs about an hour to really catch up.

And red is the output voltage of the PDVS2mini v2. As measured with the 3458A, and corrected with the calibration offset. Again one vertical division is 10uV = 1ppm.


The same measurement in Excel:


Full size graph: PDVS2mini-v2-tempco-ext-ps-29-09-2025-Excel-1500pix.png

I used the Excel table to calculate the boxed method tempco. Points were chosen near the end of the dwell periods, where all temps and output were stable.
The result was a  box method tempco  of -0,83ppm/°C related to the ambient temperature. Related to the PDVS2mini internal temperature, I got a tempco of -0,94ppm/°C
These values are close to the ones found in the first measurement.

Also in the Excel graph are the additional traces with the tempco calculated out
The light-blue trace is where the Ambient tempco was calculated out of the output voltage. Best result I got with a nominal temp of 22°C, and a tempco of -0,8ppm/°C

The light-green trace is where the tempco referenced to the PDVS2mini v2 internal temperature was calculated out. Here the best result I got with a nominal temp of 29°C, and a tempco of -0,9ppm/°C. This got me a very flat line (<±1ppm deviation, and very close to the nominal 10V output.
These tempco values are close to the boxed method values.


Another look at the tempco is plotting the PDVS2mini output against temperature. This is for the tempco referenced to the ambient temperature:


Full size graph: PDVS2mini-v2-tempco-ext-ps-29-09-2025-Ambient-1500pix.png

Blue is the  PDVS2 mini v2 output plotted against ambient temperature. Ideal would be a straight line. If tempco would be zero, a horizontal line. If the tempco would be non-linear, the line would not be straight (but a curve, or even more complex)
Here we see clearly that the output voltage is lagging the ambient temperature: it is shown by the distance between the the ascending and descending lines in the graph.
A good thing is that the beginning and ending are meeting: there is no long term hysteresis (a more permanent shift in output after the temperature changes)

Light-blue is the same plot, but with the tempco calculated out of it. This should be as horizontal as possible. I use this to judge if my tempco calculation was correct.


Here is the same graph, but this time the tempco referenced to the PDVS2mini internal temperature:


Full size graph: PDVS2mini-v2-tempco-ext-ps-29-09-2025-Internal-1500pix.png

Immediately we see that the ascending and descending lines are now much closer: Less hysteresis, the coupling between the internal temp and output voltage is much better.
The line is straight: only first order tempco. Good!
Beginning and end of the line are close together: no long term term hysteresis. Also good!

The light-green line, with the tempco related to the internal temp calculated out of it, is more like a horizontal line. Showing that the tempco calculation was correct.

Conclusions
Without the pressure of battery time, I was able to do  a slower temp sweep. With longer dwell times resulting in a more reliable boxed method tempco calculation.
Starting and stopping at the same temp made (the absence of) long term hysteresis visible.
The found tempco figures maching pretty good with the results of the first measurement, strengthening confidence.

But the last graph, showing the tempco referenced to internal temperature is in my view the most important one.
It shows clearly that, although the existing tempco is higher then expected, it is linear (first order), and good coupling between temperature and output.

This means that the tempco could be calculated out of the output  by the PDVS2 mini v2 firmware. It would result in a big improvement: in a 15°C temperature range output deviation would decrease from about 12ppm to 1...2ppm. Meaning the last digit becomes a lot less temperature dependent, and thus more reliable.
The PDVS2mini v2 is a good and useful reference as it is. I am happy with it. But it seems there is room for further improvement. If I were the designer, I would be very tempted to further improve an already very good device...

regards, Gertjan.

[IanJ]

Certainly possible to adjust the cal map in the PDVS2mini based on temperature and see a more accurate output relative to tempco.......however, the issues would be:

This complicates calibration during production.
At a guess, each PDVS2mini wouldn't necessarily have the same swing in output for a given temperature change......so this would require in order to get it spot on for each and every PDVS2mini to calibrate during production at least 2 different temps, a min and a max and map from there.
The hope of doing this would have to be testing a bunch of PDVS2mini's in a chamber, and from completely different batches. If they differ then this whole approach falls apart.

Another issue could be the cal map changing as the temperature changes.....and if you were logging or doing something critical then you'd see a jump in output. The way around this could be to only switch maps on restart of the PDVS2mini......but I guess that kinda self defeats the whole thing.

At the end of the day, IMHO, I'd leave it as is and only try to improve tempco at a hardware level.

PS. PDVS2mini and PDVS2mini V2 differ. I don't have a V2 so can't comment (I've been asked).

Great idea though!

Ian.

[aronake]

I proposed investigating temperature dependent calmapping earlier in this thread.

How i would to it would be one set of parameter for sat 17 degrees, one for 23 degrees and one for 28 degrees. Then do cubic interpolation depending on actual temperature. And either have a separate temp sensor for ambient temp or use the internal and use delta temp to figure out ambient temp or something. Then there would not be jumps for different temperatures. And an option in the settings to just use the 23 degree cal map or this temp interpolated cal map.

There are quite some complexities, but also seems rather doable and may or may not work, and may be tricky to verify, but could also be a cheap way to get away with TC compared to tweaking components.

[Gertjan]

Certainly possible to adjust the cal map in the PDVS2mini based on temperature and see a more accurate output relative to tempco.......however, the issues would be:

This complicates calibration during production.
At a guess, each PDVS2mini wouldn't necessarily have the same swing in output for a given temperature change......so this would require in order to get it spot on for each and every PDVS2mini to calibrate during production at least 2 different temps, a min and a max and map from there.
The hope of doing this would have to be testing a bunch of PDVS2mini's in a chamber, and from completely different batches. If they differ then this whole approach falls apart.

Another issue could be the cal map changing as the temperature changes.....and if you were logging or doing something critical then you'd see a jump in output. The way around this could be to only switch maps on restart of the PDVS2mini......but I guess that kinda self defeats the whole thing.

At the end of the day, IMHO, I'd leave it as is and only try to improve tempco at a hardware level.

Hi Ian (and aronake),

Interesting ideas & comments on tempco correcting the PDVS2mini v2.
Extra interesting to hear the thoughts of the original developer.

Indeed I do not think that compensating each PDVSmini v2 for it’s optimum tempco would be a good idea. It is just too impractical in production.
But compensating the PDVS2mini v2 with an average tempco correction seems very well possible.
Measure a few, and compensate all with the average. Even if it only would reduce influence of temperature to half or a third, it would be very worthwhile.

I am thinking of a simple correction. The tempco is linear / first order. So no complicated corrections needed. In my Excel I used this simple formula:
Output + (nominal temp – internal temp) * tempco coefficient = corrected output.
You can download my original Excel sheet here if you want to play with it.

Why do you think this will complicate calibration?
I think you can calibrate with the correction in place. After all, if calibrating at the nominal internal temperature, the correction will do nothing.

But maybe I am missing things, and is my solution too simple. And of course, when tempo is reduced, other sources of errors will become dominant . I can not judge that.

I agree that improving at a hardware level is always the best idea.
Of course I measured only one sample. Wrytech’s findings for the v2’s tempco are better than mine. So it is also possible that there is a hardware difference between mine, and the one measured by Wrytech.
Maybe another brand resistors with worse tempco was used in production then for the measured prototype, etc.

In the end the PDVS2mini v2 is now a commercial product of Wrytech. It is up to them what to do with this info.
I just wanted to give them a nudge.

Regards, Gertjan.

[wrytech]

Nice to see the detailed measurements and effort put in by Gertjan, and the resulting discussion! Much appreciated, as any further testing adds to the standing of the units.

I did, of course, think about adding some active tempco compensation logic around the temperature sensor included in all wrytech units. The main problem there is the Differential Nonlinearity (DNL) of the DAC, making the adjustments done by a compensation logic not perfectly predictable and possibly leading to unwanted results. One solution would be to measure the exact change in output voltage above and below the calibration points by a couple of counts each. The actual change in voltage will differ in practice from the nominal ~10 µV, and with this information a predictable compensation of the tempco could be done — at the cost of making the calibration more complex due to the amount of additional information needed, and the approach only works for the exact calibration points. Additionally, this should be paired with an individual measurement of the tempco for each unit, further adding to the work to be done.

Of course I measured only one sample. Wrytech’s findings for the v2’s tempco are better than mine. So it is also possible that there is a hardware difference between mine, and the one measured by Wrytech.
Maybe another brand resistors with worse tempco was used in production then for the measured prototype, etc.

As described in the blog post, I used the measurements of the three version 1.x units to check whether my measurements matched those done with the prototype by xDevs (thus verifying my test setup to some extent), which was the case, so I went on with testing the prototypes of v2. The v2 design has gone through multiple iterations, with some measuring worse in the absolute tempco and/or in the distribution among prototypes tested. So I have seen some worse results as well with the temperature chamber setup and found that Ian’s initial design is already very close, and changes to the critical parts of the design might not lead to improvements in practice (no matter how good they look on paper). The three units tested for v1 and the final v2 are not a large sample size; however, the results are close for each version, so I went ahead here.

A point I would like to stress is that I built only three of the final v2 units initially, burned them in, and did the testing as shown in the blog post. No further selection for good or closely matching units was done for either v1 or v2. The same goes for the components used — the exact same ones were used for all v2 (and v1.6) units sold, with no changes made for worse-specced parts or similar. I am very cautious here, as not only the absolute tempco is important, but also the matching of the tempco, so without further measurements (that I really do not want to do) I would not make any changes.

I just wanted to give them a nudge.

Which, again, is much appreciated. I have been quite busy in the last weeks with the day job and private matters, and have been working on implementing SCPI-style commands as an additional possibility for remote control, also suggested by Gertjan. Once the firmware is tested, new units will ship with it directly, and it will also be made available on the website so users can update their existing v2 units themselves.

Not committing to a timeline, but I will also implement an optional, experimental active tempco compensation function in firmware, available for anyone who wants to measure the tempco and experiment with adjustments. As long as this stays in firmware, it doesn’t require much additional effort on my side, but it could provide some interesting insights for everybody.

Overall, I think — and testing for v2 has shown — that the reference is the limiting factor, and any further gains in hardware would come at significant cost and effort for testing.

So, a lot of interesting directions to explore in the coming winter days, I would say.

Best,
Johannes

[Gertjan]

Hi Johannes,

Good to see you back in this topic. I know you are extremely busy.
Thank you for sharing background information and your thoughts.

I think that individual measurement of tempco as part of calibration can not be expected at this price point.

Of course I measured only one sample. Wrytech’s findings for the v2’s tempco are better than mine. So it is also possible that there is a hardware difference between mine, and the one measured by Wrytech.

After reading this post by branadic, I was expecting that the circuits after the reference (DAC etc.) were diluting the ADR1399 tempco.
However, reading the ADR1399 datasheet again more carefully, the differences in our measurements could just be different ADR1399 samples. The headline spec is 0.2 ppm/℃. But the graphs are showing tempco’s between +0.44 ppm/℃.  and -0.66 ppm/℃.

one example:

ADR1399-datasheet-tempco-1000pix.png

(I do not know whether you are using  the TO-46 or the LCC version)

So the difference could just be a different ADR1399. With mine being a bit of an outlier.
And you could very well be right:

Overall, I think — and testing for v2 has shown — that the reference is the limiting factor, and any further gains in hardware would come at significant cost and effort for testing.

Re component choices:

I am very cautious here, as not only the absolute tempco is important, but also the matching of the tempco, so without further measurements (that I really do not want to do) I would not make any changes.

Good to hear that you are careful and cautious! Indeed it is easy to try to improve on one point, and find out later that something else got worse. Tempco is just one aspect, linearity and long term stability are important others.

Not committing to a timeline, but I will also implement an optional, experimental active tempco compensation function in firmware, available for anyone who wants to measure the tempco and experiment with adjustments. As long as this stays in firmware, it doesn’t require much additional effort on my side, but it could provide some interesting insights for everybody.

I think this is an excellent idea, to make tempco compensation optional in software!
This will not require more calibration effort, nor add BOM cost. So no increase of selling price for people who do not need the feature.
And those that like the extra accuracy gained by tempco compensation can do their own measurements and adjustment. (The increase in cost is now their own time )
I certainly will use it!
Commercially, it is a good way to add a selling point to your product.

have been working on implementing SCPI-style commands as an additional possibility for remote control

Nice to hear that you picked up this suggestion! I will update the TestController driver when the new firmware is published.

Best regards, Gertjan.

[KungFuJosh]

Does anybody know what the differences are between v1.3 and the new v2? Is there anything important besides the ADR1399?

Thanks,
Josh

[Cyclotron]

Does anybody know what the differences are between v1.3 and the new v2? Is there anything important besides the ADR1399?

Thanks,
Josh

Great question. I have a v1.6 with the ADR1399, and I'm unsure what the differences are between it and the 2.0 ADR1399.

[KungFuJosh]

Does anybody know what the differences are between v1.3 and the new v2? Is there anything important besides the ADR1399?

Thanks,
Josh

Great question. I have a v1.6 with the ADR1399, and I'm unsure what the differences are between it and the 2.0 ADR1399.

Hmm. I assumed the ADR1399 was part of v2, so I dunno if that's different either. It would be nice if somebody knows.

[bnz]

https://www.wrytech.eu/blogs/v2-refining-the-pdvs-2-mini/pt-2-analogue-improvements
https://www.wrytech.eu/blogs/v2-refining-the-pdvs-2-mini/pt-1-practical-refinements

[Gertjan]

Indeed, answers are easy to find on the Wrytech website .

The PDVS 2 Mini v2 is the first version with updates / redisign by Wrytech.

Wrytech also updated the command set for communication over the serial interface.
That is why my TestController driver for the PDVS 2 Mini v2 only works with the v2.

regards, Gertjan.

[KungFuJosh]

The communication and your TC driver is good info.

Unfortunately, their website doesn't mention v1.3.

Thanks,
Josh

[bnz]

The communication and your TC driver is good info.

Unfortunately, their website doesn't mention v1.3.

Thanks,
Josh

v1.3 is the version from Ian, https://www.wrytech.eu/collections/legacy-products describes the v1.6 and mentiones the differences to ian's version.

[KungFuJosh]

v1.3 is the version from Ian, https://www.wrytech.eu/collections/legacy-products describes the v1.6 and mentiones the differences to ian's version.

Thanks, but I prefer specific details. Their site mentions v1.0 and v1.4 (with noise comparisons to v1.6 and v2.0), but nothing directly comparing v1.3. They show noise for the others, but not that one.

It's not a big deal, I won't likely buy it, but vague marketing references comparing non-specific versions doesn't do anything for me. The other differences their site mentioned were related to manufactuing, which I have no interest in. It being easier for them to manufacture is not a comsumer benefit, especially when the end result is a much higher price. 

Thanks,
Josh

[Cyclotron]

The communication and your TC driver is good info.

Unfortunately, their website doesn't mention v1.3.

Thanks,
Josh

v1.3 is the version from Ian, https://www.wrytech.eu/collections/legacy-products describes the v1.6 and mentiones the differences to ian's version.

There are actually two variants of the 1.6. There's an LM399AH and an ADR1399 variant. I have the ADR1399 variant. The ADR1399 variant also underwent circuit changes and received gold contacts.

I don't see any data showing on the 1.6 ADR1399; the posts you linked to don't actually specify which variant was tested, though. It says "first v1.6 unit I built with the initial wrytech design changes".

[Cyclotron]

Indeed, answers are easy to find on the Wrytech website .

The PDVS 2 Mini v2 is the first version with updates / redisign by Wrytech.

Based on the website, v2 is not the first version with updates/redesign by Wrytech.

Wrytech also updated the command set for communication over the serial interface.
That is why my TestController driver for the PDVS 2 Mini v2 only works with the v2.

regards, Gertjan.

Serial communications over serial was available before 2.0, and it was updated to add the temperature sensor.

[IanJ]

Indeed, answers are easy to find on the Wrytech website .

The PDVS 2 Mini v2 is the first version with updates / redisign by Wrytech.

Based on the website, v2 is not the first version with updates/redesign by Wrytech.

Wrytech also updated the command set for communication over the serial interface.
That is why my TestController driver for the PDVS 2 Mini v2 only works with the v2.

regards, Gertjan.

Serial communications over serial was available before 2.0, and it was updated to add the temperature sensor.

My original helluva bespoke serial command set/protocol was updated by Wrytech.......and good on them too for doing that!
When you make a design decision way way back at the start and later on you wished you'd taken a bit more time to get it nice and clean......you sometimes can't backtrack without breaking something.
To be fair, I always felt the serial interface was an extra that the vast majority of folks wouldn't use and I was more interested in just using it for auto-calibration using WinGPIB.
However, more and more folks seem to be using it these days!

I've only ever used Test Controller very lightly, but I can't see any reason why it can't be adapted to the original PDVS2mini protocol. Maybe, if I get time.

Ian.

[SHF]

PDVS 2 Mini v2 is available again

https://www.wrytech.eu/products/pdvs-2-mini-v2

[Gertjan]

PDVS 2 Mini v2 is available again

https://www.wrytech.eu/products/pdvs-2-mini-v2

Good news! From the Wrytech newsletter:

"PDVS 2 Mini v2 units are now back in stock and ready for immediate shipment.
The IUSL Isolated USB-to-Serial Adapter and shorting links are also available.

In the meantime I have been working on a new firmware release that is planned to include the following additional optional capabilities:
•   Negative output range down to –10 mV (useful for testing ADC behaviour around the zero transition for example)
•   Active temperature-coefficient compensation using the integrated temperature sensor
These features are currently undergoing extended validation and testing. Provided the tests complete successfully, the update will be released as a free upgrade (requiring an UPDI programmer) for all existing and new PDVS 2 Mini v2 units."

[Cyclotron]

PDVS 2 Mini v2 is available again

https://www.wrytech.eu/products/pdvs-2-mini-v2

Good news! From the Wrytech newsletter:

"PDVS 2 Mini v2 units are now back in stock and ready for immediate shipment.
The IUSL Isolated USB-to-Serial Adapter and shorting links are also available.

In the meantime I have been working on a new firmware release that is planned to include the following additional optional capabilities:
•   Negative output range down to –10 mV (useful for testing ADC behaviour around the zero transition for example)
•   Active temperature-coefficient compensation using the integrated temperature sensor
These features are currently undergoing extended validation and testing. Provided the tests complete successfully, the update will be released as a free upgrade (requiring an UPDI programmer) for all existing and new PDVS 2 Mini v2 units."

I guess no updates will be made for anything prior to the v2 units.

[Gertjan]

iPowerUtra 800mAh Li-Polymer batteries

I bought there iPowerUtra 800mAh for use in my PDVS2mini v2.
These batteries have a specified output voltage of 8,4V...6,0V, so consist of two Li-Ion cells in a 6F22 package, just like the EBL ones.


IMG_4637__iPower_800mAh_batteries-2000pix.jpg

iPowerUS is a brand that is oriented at the professional Broadcast industry. Targeted for professional use means better quality and reliability. It also means that the batteries are fully specified with a datasheet.

I compared the runtime of the EBL 600mAh batteries with the iPowerUltra 800mAh. The PDVS2mini used as battery tester .
For easy comparison I kept the time axis the same for both graphs.


Full size graph: EBL-600mAh-Battery-test-1200pix.png


Full size graph: iPowerUltra-800mAh-Battery-test-1200pix.png

Most important line is the red one: the Battery Voltage. On the very left side, one can see the voltage dropping down after the external power was removed. To the right, the vertical lines are showing the auto-shut down of the PDVS2mini because of low battery voltage.

With the EBL 600mAh batteries, the PDVS2mini auto-shutdown after 7:55 hours, The iPowerUltra's lasted 13:01 hours before shutdown.

The batteries were pre-balanced (charged separately on their native chargers), and then in the PDVS2mini partly discharged, and charged again. The PDVS2mini was in it's default settings (Batt Low shutdown: 12.4V)


IMG_4644__PDVS2mini_with_iPowerUS_800mAh-2000pix.jpg

In the EBL 9V battery topic I compare the EBL and iPowerUS batteries, and show capacity test results: the iPowerUltra batteries are offering substantial more capacity then their EBL counterparts: ~680mAh versus ~450mAh.
But they are also substantial more expensive: iPowerUltra ~€ 23,- versus EBL ~€ 6,- (Professional product.....)

Personally, I like these iPowerUltra_800mAh batteries. They perform without any problems in my PDVS2mini v2. I am happy with the longer run-time, and think that is worth the higher price.

regards, Gertjan.

[Andreas]

Hello,

when looking for higher capacitive 9V batteries I found the Pallus 850 mAh devices.

https://www.amazon.de/dp/B0BFWJ4KF4

- price slightly higher than the EBL 600 (26 Eur for 4 batteries)
- capacity is higher than EBL 600
- downside: I measured a rather large stray (23%) from battery to battery -> I would select and pair them for usage in PVDS2mini

Measurement setup:
- charging freshly bought with EBL charger (the Pallus charger was not available when I bought the 9V blocks).
- charge termination voltage is about 8.4V (the Pallus cells might have their maximum capacity at a higher voltage since 8.7V is the switch off voltage for the 9V block).
- discharging with a total resistance of 194 Ohms (giving around 40 mA instead of the 60 mA of Gertian).
- as comparison I also tested some EBL 600 mAh batteries (already used some months occasionally for my low noise amplifier).
- DMM DT80000 (80000 Counts) for voltage measurement directly across the discharge resistors.
- switchoff by relay at 6.2 V discharge voltage

on the first run I did not take care to start discharging immediately after full charging.
So I repeated the measurement on the first 2 batteries.
Following I recharged each battery immediately before discharging.

Results:
EBL    447-477 mAh ( 7% stray) comparable to Gertians measurement regarding the 40 vs 60 mA load.
Pallus 604-741 mAH (23% stray) compared to the 680 mAh of the iPowerUltra.

with best regards

Andreas

[Gertjan]

When looking for higher capacitive 9V batteries I found the Pallus 850 mAh devices.

https://www.amazon.de/dp/B0BFWJ4KF4

- price slightly higher than the EBL 600 (26 Eur for 4 batteries)
- capacity is higher than EBL 600
- downside: I measured a rather large stray (23%) from battery to battery -> I would select and pair them for usage in PVDS2mini

Hi Andreas,

Thank you for bringing these Pallus batteries to our attention.

Most important take-away is that these batteries are offering a higher capacity alternative to the EBL's. Very affordable and easy to get (Amazon...)

Of course, as these are no name batteries (I could not find anything on the Pallus brand, let alone specifications), no one knows how long you can buy them. And tomorrow there could be different cells inside.
But hey, they have bigger capacity than EBL, and they are cheap

On a more technical note, in reaction to your measurements.: I tested one iPowerUtra 10 times.  The spread between these 10 discharges was max 13mAh. This includes discharges straight from the charger, but also discharges the next day. I also found not much difference between the first discharge of the battery and the 10th discharge.

So probably iPower is doing some selection of the cells. But clearly you have to pay for the added value. And I am not sure you need this for use in the PDVS2mini.

regards, Gertjan.