AlanB's Sled Meets Modamag's TK Monster - Review & Pix

This is another review of AlanB's Incan Regulator Sled. Previous review is in this post, but there are just too many pix I took tonight to cram into that Philips 5761 bulb testing thread.

This time we wanted to test Alan's regulator with something a bit more challenging, so I took the venerable Modamag TK Monster as a host, but did it even more beefed up. Instead of 8s x Li-Mn Ion cells as direct drive, I added a 9th for regulated model.

Otherwise it is using the same 64647 250W bulb, with settings in previous link. I had to do a little mod on Modamag's battery holder, and shorten the KIU bulb holder on Alan's sled. Didn't take that long to do, and I took lots of pix. It was one of those foggy New England nights when I went outside.

I pulled myself away from testing JimmyM's PhD regulator (sorry Jimmy), but wanted to get this done and posted, as Alan mentioned in a recent PM that there is a CPF get together tonight with Modamag and Alan attending. I'm just going to post these pix tonight....and can't think of much else to say anyway.

(All pix posted)

Re: AlanB's Sled Meets Modamag's TK Monster

reserved

Oh My God !

I just moved into this really nice quiet neighbourhood and some BAS***D keeps shining a stupidly bright light at my bedroom window.

Again and again and again . . .

:devil:

Great work Lux :thumbsup:

Alan's sled looks really nice.

Cheers
Pete

Nice build Lux!

Those pictures are quite impressive. HAHA! It must be bright if it can wake up petrev from the other side of the atlantic ocean.

Thanks for the great pics Lux.

We probably need to set the minimum battery voltage a little lower. The battery measurement is apparently indicating low battery and dropping one step to 3/4 voltage (about half power). The calibration may be a bit off as well, I had to calibrate this unit at 15 volts due to lack of higher voltage power supply. The 50V power supply arrives monday so this won't be a problem in the future.

But we may have to rethink our low voltage shutoff values a bit on these high current lights.

It would be good to collect the info on the socket standoff height for different situations. If you figure out the ideal height for that setup let's document it. Looks like a few washers would be about right.

Thanks again!

petrev said:

Oh My God !

I just moved into this really nice quiet neighbourhood and some BAS***D keeps shining a stupidly bright light at my bedroom window.

Again and again and again . . .

:devil:

Great work Lux :thumbsup:

Alan's sled looks really nice.

Cheers
Pete

Click to expand...

The funniest thing will be if someone starts renting or housesitting unbeknownst to me, and they come stumbling out in their pajamas screaming: "OK you found us.....we give up"

Yeah, Alan that reversion to about 75% seems about right. One other thing I tried before zonking out, was pushed ramp up and held it down at maximum, and there was a pulsing down to about that level and back to maximum since I was still holding ramp up, this would go on endlessly in cycles back and forth that each seem to last 1/3 to 1/2 sec rate.

Also a few times I had done the maximum ramp up and it cut it down as above, but then the click off did not respond...which got my attention. I had to ramp down manually then the click off responded. I never had to unscrew tailcap, but something was causing this locked on, and no response to any single, double, triple click. Only ramping it allowed me to regain control with clicks. I'll test it again with new bats to see if I can figure out what was keeping it on.

That's a new one. We'll have to keep an eye on that.

SN003 lives. Set up of Osram 64623. 6D Here We Come.

I found that the initially charged up cells battery saving reduction from MAX to 75% only happens with full charge cells.

If I drain down the batteries for about 30 sec on medium, and then go back to MAX, it will stay at MAX until the neck around switch area gets too hot to touch (from this 250W bulb & current), and freaks me out, or the 1 minute time limit without button pressing ramps it down and off. I did let it run at least three times for a full minute.

I touched the Fluke probe to the outside metal wall where head joins body, and it got up to about 140.2° F (60.1°C). On a more recent check, I did see the abrupt temp shutdown on second MAX run close to first.

I got at least 3 runs of 1 minute runs at MAX. After switching to MED and draining batteries to shutoff, I removed them after about 3 minutes, and put the Fluke thermistor inside between two batteries, and it gave 154.4°F (67.8°C). So yeah, these batteries got hot.

After full drainage shuttoff, and one repeat that again shutdown, I measured the ending pack voltage as 33.89V. When I took the pack apart and measured each battery, they were almost all exactly 3.77V which amazed me. This is great news on the low battery protection issue, as this represents a bounce back after the low voltage shutoff.

I have not been able to reproduce the button not responding while on on 2 more runs.

Man is this MAX bright!

I have a bright 500W underwater pool light that is dwarfed by this thing shining down into the deep end. Of course one is being shone down, and other through middle of water, but it is still impressive. No comments on the dirty pool, I just opened and shocked it, now have to vacuum up the settled debris.




.

Hi Lux

Thanks again for the data you are amassing on the PhD regulator project(s)

Good the get some empirical Vbounce data plus all the other results.

Cheers
Pete

I am very happy that Lux's testing has gone so well. This shows that the regulator electronics survives at 38 volts and approximately 13 amps, which verifies many features of the electronics design that I had not been able to verify in my lower power test setups.

Based on Lux's temperature measurements the present shutdown setting of 60C is probably too high. Keeping in mind that the temperature measurements are not precisely calibrated and they will vary some, I am reducing the default to 50C. This is a parameter that can be changed by the user, but I would like to see it shutdown a little sooner by default. Comments and suggestions on this are welcome, though perhaps they should go into the regulator's discussion thread.

Another thing that Lux's testing has shown is that the regulation is working. Not that the light output has been measured and shown to be constant (though that is an interesting test to try, albeit at a lower power level), but remember that if the regulation on this light allows full battery voltage to reach this bulb it would almost certainly instaflash. Even 8 cells (Lux is running 9) will flash the bulb if fully charged and not soft started. So the soft start and regulation is working at this power level. Excellent.

The other useful result is the control falling back to half power on the bulb when the batteries are fully charged. I don't understand completely what is happening there yet, but it does appear to be a software issue relating to the low battery voltage protection (or the ADC overvoltage protection) that is incorrectly tripping in these circumstances.

The fact that the effect only happens when the light is at full power, and only happens when the battery is fully charged (and not when it is slightly drained) is interesting and not quite what would be expected (though this is what the ADC overvoltage protection should do). In the coming days we'll get Lux set up to compile and load software into the regulator and we can start making some changes and develop code that works better. For lights drawing less than about 9 amps this is probably not a problem (I have not seen it on my 6 amp WA1164), but we want this to work properly up to the 13 amp level. My 12 amp test bed is almost ready as well, so that will allow me to test and reproduce these issues as well.

Edit - In thinking about this issue a bit further I suspect that it may be the ADC overvoltage protection code again. This is the same issue that we had with Lux's last test with the 5761, but in that case we were right against the max voltage so it was not surprising. In this case we should not be near the full voltage of the range. I have just changed this code to add some filtering to it so a single high reading will not trigger it. ADCs occasionally get bad readings and so filtering the data a bit before reacting is a good idea, and in almost every case that is already being done in the regulator, however this is the one case there was no filtering. So this filtering has been added and we'll have to test it to see if this helps out in Lux's testbed.

Thanks for the testing effort, Lux.

Nice work Lux, and Alan too :thumbsup:.

One of mine will be going into my Colossus with 8 cells. Will this be enough for reasonable regulated performance with the 64457 (perhaps at 26v?), or should I try something less ambitious like the 150w 64642 or even a 64458 at 20v?

I had some more thoughts about this and edited my posting above, adding the paragraph regarding the ADC overvoltage protection code possibly being the cause of the issue discussed there.

Holly Moly Lux, that thing really looks bright given 1 sec exposer. How does it hold up against the Polarion PH-50. Looks much brighter but I guess it get's hot very quickly...




.

Nice work and great pictures

Alan's design is the BOMB!!!

wquiles said:

Nice work and great pictures

Alan's design is the BOMB!!!

Click to expand...

Thanks, Will. Hope is doesn't blow up like the bomb...

Incidentally, your earlier work may explain what is happening here with the backoff. The battery test reading is taken near the end of the PWM pulse where voltage is lowest. Remember what is going on then? The timing is not that precise, but now and then it may catch part of the inductive kick and consider the ADC over-ranged. Making a second verifying reading should prevent that (a little filtering). We'll have to test that theory of course, but it is making more sense now...

Alan B said:

Incidentally, your earlier work may explain what is happening here with the backoff. The battery test reading is taken near the end of the PWM pulse where voltage is lowest. Remember what is going on then? The timing is not that precise, but now and then it may catch part of the inductive kick and consider the ADC over-ranged. Making a second verifying reading should prevent that (a little filtering). We'll have to test that theory of course, but it is making more sense now...

Click to expand...

Filtering should work, but also perhaps we can consider throwing away values/samples not within the "expected" range - kind of a hard low/high filter that would ignore unreasonable values.

That's essentially what the new code is doing - requiring the ADC to be over-ranged two readings in a row before reacting.

BSBG said:

Nice work Lux, and Alan too :thumbsup:.

One of mine will be going into my Colossus with 8 cells. Will this be enough for reasonable regulated performance with the 64457 (perhaps at 26v?), or should I try something less ambitious like the 150w 64642 or even a 64458 at 20v?

Click to expand...

To answer your question properly, you have to look at a couple of links. First is the 64657 250W bulb report here.

Then we need to look at how do the Li-Mn Ion cells which I used (AW sells his version of them) at this amp demand level. Best estimate is with this testing chart I did, although my leads had more resistance during testing than I wanted.

Now that I look at these charts again, the cell I used is the Sanyo (green plot line in my linked chart) which appears to have 3.0 to 3.4V output with this Amp rate, not including further resistance losses in the light components. (By the way, make sure you do the tailcap spring fix....which I forgot about, but re-discovered when opening the tailcap after a short run and felt the spring.)

So in reality, I was starting with 9 x 3.4V = 30.6V dropping to 27V as batteries drop to 3V rundown stage. Your 8 cell proposal would start 3.4V below those numbers, and not in the optimal overdrive range for this bulb. Same is true with the units Jonathan has sent out with AW's higher watt switch using 8 cells. (BTW, AW has reported that his soft starting, multi-level switch alone has 220 mΩ of resistance which at 12A is a drop of 2.64V)

Bottom line is 8 LiMn Ion cells are not enough to drive this bulb properly.

Lips said:

Holly Moly Lux, that thing really looks bright given 1 sec exposer. How does it hold up against the Polarion PH-50. Looks much brighter but I guess it get's hot very quickly...

.

Click to expand...

Got your PM. Mulling it over.

I wanted to take another shot at the vacant house at 1/3 sec on MAX, but ran into the auto-reducing to 75% issue that Alan is pretty sure he identified the source of. Now that I know I can get around it by first draining the power on MED for about 30 sec....then MAX works. So I could do another outside shoot. However that last set of images will be unique since it was a foggy night.

The swimming pool views were without any fog/humidity (unless you count the supersaturated humidity of the water in the pool :devil and at 1/5th sec. I can do a pool shootout at reduced shutter speeds between BB, PH-50, TK Monster. Yeah the Monster does get hot.

This is the hottest I have ever let any of my MagMods get, and you have to assume those readings I took from outside, and of batteries about 3 mins later were lower than actual temps.

In part I wanted to test Jonathan's belief in the unique quality of his reflector coating process, and the fact that he made it to screw into the head for better heat protection, which works very well. No evidence of any reflector peeling, discoloration, or damage....so this is also a :thumbsup::thumbsup: review of his creation, and hope he sees this thread.

Alan B said:

I had some more thoughts about this and edited my posting above, adding the paragraph regarding the ADC overvoltage protection code possibly being the cause of the issue discussed there.

Click to expand...

I exchanged PMs with Lux and he confirmed that the behavior was a reduction in brightness, not a blink and reduction. The reduction without blinking is the ADC over-range code that I have already improved. I suspect that will solve this issue.

We will get Lux set up to load code into his regulator and then he will be able to retest with the new code.