Sold/Expired Discussion - Programmable Hotwire Regulated Driver drop-in for D-M@g

[LuxLuthor]

My problem with 623 bulbs may also be the voltage I selected. Perhaps LuxLuthor might comment on this. I selected 16.5 volts. That may be a bit much. This bulb seems to have a somewhat soft failure zone according to Lux's tests, but I note that he chose lower voltages for his lifetime testing. So I have a combination of a high voltage, short 100 mS quick soft start, and very low impedance lithium-iron nano-phosphate batteries that put a lot of stress on the bulb, apparently.

During development some folks were not happy with the long natural soft start that my algorithm has - as Jim mentions it is up to slightly over a second, depending on the battery vs bulb voltages. So I added code to move in larger steps when the error was large. It reverts to small steps when the error is lower, that is when it approaches the correct output voltage. This is an easily changed compile-time parameter.

I am considering doing two things to reduce the problem I had with the 623. One is to change to 16.0 volts default max on this bulb. The other is to lengthen the default fast soft start to 200 milliseconds.

Comments??

Yes on both strategies. Personally, I would never push the extreme edge of my flash points for any regulated setting. For the 623, I would set regulation at 15.3V and see how long it ran, and how the bulb output quality held up. If it seemed to be more than 6-8 hours, then I would go up to 15.5 to 15.8V and watch it again.

I'm sure that 16.5V is the primary source of your failure. I also think 100ms is too quick, especially for higher Watt bulbs. Although AWR's Hotdriver seems instantaneous (so is likely 100-200ms). I would have to go back over his threads and notes to find out what he had, but his mechanism of soft starting and regulation is totally different.

Take a look at this new topic where I posted download links for the three Excel files that can also give you some idea of predictive work that AWR and I worked on after he left.

KiwiMark, there is no harmful downside to making the softstart longer. I think people raised an issue early on not wanting it to be too slow so it was significantly noticeable. Longer time of softstart represents safer cushion protection from surge of current rushing into the cold filament...but at some point, the delay becomes cosmetically irritating...like...."Turn On ALREADY!!!" So Alan picked 100ms, but I think that is a bit too fast for more demanding bulbs that need more current like the 100W 64623. As a comparison, I think AW's soft starter multi-level driver/switches are around 1,200-1500ms just eyeballing some I have.

 

[JimmyM]

Yes on both strategies. Personally, I would never push the extreme edge of my flash points for any regulated setting. For the 623, I would set regulation at 15.3V and see how long it ran, and how the bulb output quality held up. If it seemed to be more than 6-8 hours, then I would go up to 15.5 to 15.8V and watch it again.

I'm sure that 16.5V is the primary source of your failure. I also think 100ms is too quick, especially for higher Watt bulbs. Although AWR's Hotdriver seems instantaneous (so is likely 100-200ms). I would have to go back over his threads and notes to find out what he had, but his mechanism of soft starting and regulation is totally different.

Take a look at this new topic where I posted download links for the three Excel files that can also give you some idea of predictive work that AWR and I worked on after he left.

KiwiMark, there is no harmful downside to making the softstart longer. I think people raised an issue early on not wanting it to be too slow so it was significantly noticeable. Longer time of softstart represents safer cushion protection from surge of current rushing into the cold filament...but at some point, the delay becomes cosmetically irritating...like...."Turn On ALREADY!!!" So Alan picked 100ms, but I think that is a bit too fast for more demanding bulbs that need more current like the 100W 64623. As a comparison, I think AW's soft starter multi-level driver/switches are around 1,200-1500ms just eyeballing some I have.

AWRs softstart time was deliberately low, 100 mS MAX, to keep from incinerating the FET. We don't have that problem. We can softstart the bulb over days if we want. My softstart is in the range of 300-600 mS. Depending on Vbat and Vbulb settings. This may even be too quick for the 64625 on 13.8V. I've added compile time options for standard or aggressive starts. The difference between 100mS and 300mS is barely noticeable.

 

[Alan B]

Yes on both strategies. Personally, I would never push the extreme edge of my flash points for any regulated setting. For the 623, I would set regulation at 15.3V and see how long it ran, and how the bulb output quality held up. If it seemed to be more than 6-8 hours, then I would go up to 15.5 to 15.8V and watch it again.

I'm sure that 16.5V is the primary source of your failure. I also think 100ms is too quick, especially for higher Watt bulbs. Although AWR's Hotdriver seems instantaneous (so is likely 100-200ms). I would have to go back over his threads and notes to find out what he had, but his mechanism of soft starting and regulation is totally different.

Thanks for your comments. Perhaps 100mS is not the problem and I'm just pushing it too hard. Generally I've been looking for a voltage near but below the 10 hour life point. 16.5 was about 22 hours on the chart but this bulb didn't seem to have as sharp a falloff and flashed before getting to 10, and it flashed at 16.9. So 16.5 is a bit close to the flashpoint.

One other thing that I can do is to taper the end of the softstart to the low rate sooner, and leave a bit more time for tha landing phase of the ramp. My softstart actually has two rates, the fast rate and the slow rate. The transition point can be controlled separately from the fast rate. The slow rate is currently fixed at one PWM step per PWM pulse at 250 hz.

I might tend to make the test top down rather than bottom up. If we set the voltage low the experiment will take too long. We can recognize too high a lot faster than too low when the bulb fails sooner than we'd like.

 

[JimmyM]

The 623 is stellar at 15.6V (Lux's 13 cell Mag623 pack).
The difference between 14.4 and 15.6 is stunning.

 

[Alan B]

The 623 is stellar at 15.6V (Lux's 13 cell Mag623 pack).
The difference between 14.4 and 15.6 is stunning.

It sure is impressive at 16.5!! I lit up my whole backyard. It is a bit floody with the small head, MOP reflector and the large filament.

It is really impressive except when the filament melts...

The fact that the filament calmly melted on one side (the support/conductor on one side of the filament melted) indicates that it is more likely an overdriven situation than a soft start. Both bulbs look the same, too. So it wasn't a massive failure such as hard starting often causes, but a very calm melting.

If I'm going to push it this hard I better order some more bulbs.

Maybe I'll just try 16.0 for awhile and see how it does. The regulated voltage is at the sled, the bulb voltage is a bit lower due to resistance downstream, in the socket etc.

One other thing I've thought about is compensating for this downstream resistance to get more accurate regulation at the bulb. The downside of doing this is the requirement to put in an estimate for the socket resistance, and for the bulb resistance or current. From this we can calculate the drop and dynamically adjust the voltage up a little at the sled to compensate.

Also we can do the same on the battery cutoff (which may be even more important than the bulb compensation). If we put in the flashlight resistance and possibly even cell resistance we can adjust cutoff to compensate for the drop and get a more accurate cutoff at these high currents. I have noticed with the A123 M1 cells that it cuts off sooner than I'd like at high currents. So this is more interesting to improve.

The downside to doing this is more data to put in, but the upside is more precise operations. If I do this it will be an option so folks that don't want the feature can leave it out.

 

[Starlight]

If this will be a programmable function and we have an option, I would still like mine to be slower. .5 seconds is plenty fast enough for me.

 

[Alan B]

If this will be a programmable function and we have an option, I would still like mine to be slower. .5 seconds is plenty fast enough for me.

Fast Soft Start is programmable. If we turn it off we get approximately 1 second ramp-up, so that's as slow as it goes. From there the FASTSTART option makes it go faster. Specify what you want in your ordering parameters and we can set it near there.

 

[KiwiMark]

Maybe I'll just try 16.0 for awhile and see how it does.

This is the brilliant thing about a programmable regulated driver, I love the idea of being able to change the parameters if you blow bulbs too often or want to drive it harder or whatever. :twothumbs

So when are the first units likely to be shipping? There better be some reviews & reports - I definitely expect them from Lux!

 

[Starlight]

Ordering parameters changed.

 

[Chodes]

It sure is impressive at 16.5!! I lit up my whole backyard. It is a bit floody with the small head, MOP reflector and the large filament.

It is really impressive except when the filament melts...

The fact that the filament calmly melted on one side (the support/conductor on one side of the filament melted) indicates that it is more likely an overdriven situation than a soft start. Both bulbs look the same, too. So it wasn't a massive failure such as hard starting often causes, but a very calm melting.
If I'm going to push it this hard I better order some more bulbs.
QUOTE]

I believe 15.0V for the 623 would be fine. Even lower.
I've been taking vbulb measurements of some of my hotwires , I've done Osrams 64440,64458,62138 but waiting to cycle my 64623 batt pack to test it.

It's a 15.6v 1500maH pack and I'm not expecting more than 14.5v at the bulb. I thought those cells (Elite 1500s) only manage about 1.15v at 10A so that would be 14.3V pack - so around 14.0v bulb?

 

[Alan B]

Calibration is compensating for changes in:

1) chip characteristics - selected voltage reference (one of which is the voltage regulator onboard, the other two are in the cpu)
2) resistor characteristics - their actual values
3) jumper position
4) changes to the regulation algorithm such as averaging, etc

Changes such as bulb voltage or interface options do not affect calibration. Changing battery voltage may change the range and may require re-calibration.

Calibration can be done at any voltage but it is necessary for regulation to be working - ie not direct drive - so battery voltage must be higher than calibration voltage. It is generally best to calibrate at or near full output voltage. That is where the calibration accuracy really matters, and where it is most sensitive and accurate. Calibration affects both bulb output voltage accuracy and battery shutdown accuracy.

The sled has an internal load resistor, so no load is required for calibration purposes. Just connect the meter and go. I make sure the input voltage is above the output so it is regulating, and use double click to go to max output. With the AC+DC RMS meter you read the actual output and put the desired output and actual output into the code and recompile and reload to calibrate. Double click and read it again to verify it.

 

[Alan B]

I believe 15.0V for the 623 would be fine. Even lower.
I've been taking vbulb measurements of some of my hotwires , I've done Osrams 64440,64458,62138 but waiting to cycle my 64623 batt pack to test it.

It's a 15.6v 1500maH pack and I'm not expecting more than 14.5v at the bulb. I thought those cells (Elite 1500s) only manage about 1.15v at 10A so that would be 14.3V pack - so around 14.0v bulb?

Your quote above broke because the left bracked slash is missing from the ending quote...

The 64623 is impressive even at the factory rated 12 volts, so anything above that is just better. Since it melts at 16.9 according to Lux's tests, and it occasionally melts at 16.5 clearly those are too high. I'm still running at 16.5 but plan to change to 16.0 when I next reload the software in that flashlight. 14.0V is going to give you a much longer bulb life and a few less lumens but it will work great.

 

[LuxLuthor]

I believe 15.0V for the 623 would be fine. Even lower.
I've been taking vbulb measurements of some of my hotwires , I've done Osrams 64440,64458,62138 but waiting to cycle my 64623 batt pack to test it.

It's a 15.6v 1500maH pack and I'm not expecting more than 14.5v at the bulb. I thought those cells (Elite 1500s) only manage about 1.15v at 10A so that would be 14.3V pack - so around 14.0v bulb?

CBP's performance 20A graph is overly optimistic for practical uses, but it should be above 1.2V @ 10A for most of the run. They test by significantly heating up the cells to enhance current output, and use battery leads that are soldered to the cells, and use the Competition Electronics CE-35 to measure. Here is their Elite 1700 AA graph.

 

[Roland]

Regulation creates lots of battery possibilities.
I am thinking about a 623 driven at up to 16 volts.
I want to run it on Low Self Discharge batteries so it is ready when I want to grab it, even after several months of neglect (not charging batteries).

But what are realistic safe chemistry LSD options. How well can for example 18 x AA eneloop's hold up?

I believe the current is drawn in pulses from the batteries which gives strange capacity, voltage and current draw. Nothing that is tested before. I therefore suspect battery performance charts to not be accurate and the current draw will be high for LSD nimh's.

Does anyone have an opinion on this?

 

[KiwiMark]

I want to run it on Low Self Discharge batteries so it is ready when I want to grab it, even after several months of neglect (not charging batteries).

But what are realistic safe chemistry LSD options.

Does anyone have an opinion on this?

That's why I prefer Li-ion. They are LSD and safe. I am going to use AW's IMR cells with one of my programmable drivers - safe chemistry and with the drivers ability to shut off when the battery voltage drops below a certain level it is easy to look after the cells. The IMR cells could otherwise be harmed by discharging too low - but that would only harm their capacity not cause danger, they are a safe chemisty (no 'venting with flame').

The protected Li-ion cells are pretty safe, but the protection circuit greatly limits the maximum amperage you can draw from the cells - this make them only useful for lower powered lights. My protected D-cells are OK with my ROP Low, but not my ROP High bulbs.

Unprotected D cells are OK with the programmable driver and a smart charger - but you should avoid using them with a driver that does not cut off below a threshold voltage as this could damage the cell and lead to a 'vent with flame' situation during charging. I am planing on using these with my 2nd programmable driver and will take care not to use them with other drivers. These are safe cells if you are careful, make sure they are safe from being shorted though.

 

[Alan B]

Regulation creates lots of battery possibilities.
I am thinking about a 623 driven at up to 16 volts.
I want to run it on Low Self Discharge batteries so it is ready when I want to grab it, even after several months of neglect (not charging batteries).

But what are realistic safe chemistry LSD options. How well can for example 18 x AA eneloop's hold up?

I believe the current is drawn in pulses from the batteries which gives strange capacity, voltage and current draw. Nothing that is tested before. I therefore suspect battery performance charts to not be accurate and the current draw will be high for LSD nimh's.

Does anyone have an opinion on this?

First of all, I suspect that the battery chemistry pretty much sees RMS current. But in any case you can predict the peak current. If the battery voltage is double the bulb, the peak current is four times. If the voltage is 1.414 times the current is doubled. If the voltage is 10% over the bulb the peak current will be about 20% over.

Best cells for this type of standby use are Lithium Iron and Lithium Manganese. Both safe, both high current capable.

Low self discharge NiMH may have a problem. I have heard from one source that the small trickle standby current required to run the micro may negate the low self discharge capability of LSD NiMH cells. I have not verified this.

Aside from that Eneloops would be fine if their terminal voltage was not too much higher than the bulbs, if the peak current is a problem at all, this would limit it.

 

[Alan B]

The production build is starting. The first unit has passed testing. A new configuration has been added to the price list. The semi kit. Check the price list in the sales thread for more information. The Assembly Notes for the Semi-Kit are in post #4 of this thread.

 

[donn_]

Hi Alan,

A few questions for you:

1. I'd like to use one of these in a battle lantern Mac is building out for me. The idea is to stuff as many 26650 A123s inside as it will hold, and regulate down to appropriate voltage for a good hotwire bulb. Is the 40V maximum limit for this switch Vbatt or Vbulb?

2. I've read about your trials and tribulations with the 623, and wonder if you've reached a happy solution. Have you lowered the voltage and/or removed the fast soft-start?

3. You also mentioned a 2D 5761 build. Have you come up with a perfect configuration for this lamp? I'd use either two AW IMR C cells, two 26670 Emoli cells, or 3 26650 A123s.

 

[Alan B]

Hi Alan,

A few questions for you:

1. I'd like to use one of these in a battle lantern Mac is building out for me. The idea is to stuff as many 26650 A123s inside as it will hold, and regulate down to appropriate voltage for a good hotwire bulb. Is the 40V maximum limit for this switch Vbatt or Vbulb?

2. I've read about your trials and tribulations with the 623, and wonder if you've reached a happy solution. Have you lowered the voltage and/or removed the fast soft-start?

3. You also mentioned a 2D 5761 build. Have you come up with a perfect configuration for this lamp? I'd use either two AW IMR C cells, two 26670 Emoli cells, or 3 26650 A123s.

40V is is max for the FET and 5V regulator, so Vbatt. I have not tested this but I believe the 5V regulator will go to 60V at this low load. We can get a different FET. The input cap is 50V so might want to change that as well. There is some risk in trying it but we can set up a board that way if someone wants to try it. Probably be August before I can do that.

I have had no problem with the OS64623 since I avoided full-on with cold filaments (using the original 16.5V and 100mS fast start), and now I've also lowered the voltage to 16.0, and turned fast-start off. I think we have good settings for the 623 now.

For the 5761 I use two KD D Li-Ion at 4.8AH and set for 7.2V. I've built one and one of my local testers copied it. LuxLuthor had something similar for his first test. Lux used 7.3V.

 

[donn_]

5761 vs 1164 vs 1185?

That's the choice I'm having trouble with right now.

One of the regulators will be a '623, for sure, with whatever parameters you've found best.

One will be for the Battle Lantern, and I'll wait for that one.

But one will be a low-power hotwire, using one of the 3 listed above. I currently have all of them in a variety of configurations, ranging from AW's soft-starters to AWR's regulated switch.

I like them all, but tend to grab the 1164s more often than the others.

How about other user's opinions. Which of these would benefit most from the regulation offered by Alan's Sled? I'll probably use 26650 A123s most often.