mwkeefer
1 MW
Yes R25 should open the range on the top end... when you say you added the zener of 19v... you wired them properly (direction) and in series right? check your caps voltage rating too.
-Mike
-Mike
Switch mode supply mod
- Thread starter Jeremy Harris
- Start date
Vanquizor
10 mW
Hyena said:Awe, the fried guinea pig award, now I HAVE to investigate further
OK as requested here's the pics.
Firstly, here's where mine let the magic smoke out!
Maybe I missed it, but what was the final verdict for this failure? Bad bits or related to Hyena's mods?
I have several of the s-350-48's identical to the one here: http://img44.imageshack.us/img44/6478/mws35048.jpg with the exception that mine hs C36 populated with a 63x3300uF cap.
Per Mike's suggestion I measured r21 to be 9960 ohm (banded as 10k +-5%), R37 is 269ohm (banded as 260+-10%). Resistance from pin 16 to 15 on TL494CN reads 268 ohm. All were taken with my $7.00 multimeter so some tolerence should be allowed.
My plan is to use a few of the power supplies to fast charge my "touring" scooter... keeping them reliable is a must. Any other headaches I should be aware of with connecting 2 of these in series for charging?
Thanks in advance!
mwkeefer
1 MW
The mods work fine but we are / I am in the process of getting the current tapered off to a reliable power level (ie: 350w) which I have succeeded in doing, I will post logs and a plot of the charge curve I have obtained later tonight (when I empty my packs). I have also figured a method for adjusting the CC-CV cross over point so we can better tune these for our various applications. The only issue still remaining is a bit of harmonics, a slight buzz which is likely due to the bump up of a single units output voltage... Thing is my stock 48 would output between 57-58v so 62.25 isn't asking much more of it, I think adjusting the clock rate of the TL494 will solve the buzzing issue.
I believe the 48v mod to 60-70v is okay... I haven't tested at above 70v yet (but I will be soon enough) you just have to reduce the current maximums (had to anyway as the power the stock unit puts out will burn it up quickly).
That said... waiting a few days (or hours depending on my "real" workload) for me or Fechter (he should have one of these tommorow) until the various mods are tested and vetted by .. well us I suppose, would be the prudent move.
Connecting 2 of them in series is okay... what target output voltage / current do you want... you may be able to do it with your existing units.
Also, are your units identical to the ones you post the link image to? If not, you know the drill - open it up, remove the PCB and scan the trace side + photo the component side... plus, what version is it (should be stamped on the board or have a date at minimum).
Oh - one last note... the Diode and Transistors which use the case as a heatsink require that thermal coupling paste to work correctly (and the shims so they don't short out)... a cheap and available alternative which Radio shack should have is ceramic polysynthetic thermal compound, commonly known as "Arctic Silver". Be sure to re-apply some if you damage it. I use this same stuff on the Infineon controllers and have never had an issue with it.
Hope this helps!
-Mike
I believe the 48v mod to 60-70v is okay... I haven't tested at above 70v yet (but I will be soon enough) you just have to reduce the current maximums (had to anyway as the power the stock unit puts out will burn it up quickly).
That said... waiting a few days (or hours depending on my "real" workload) for me or Fechter (he should have one of these tommorow) until the various mods are tested and vetted by .. well us I suppose, would be the prudent move.
Connecting 2 of them in series is okay... what target output voltage / current do you want... you may be able to do it with your existing units.
Also, are your units identical to the ones you post the link image to? If not, you know the drill - open it up, remove the PCB and scan the trace side + photo the component side... plus, what version is it (should be stamped on the board or have a date at minimum).
Oh - one last note... the Diode and Transistors which use the case as a heatsink require that thermal coupling paste to work correctly (and the shims so they don't short out)... a cheap and available alternative which Radio shack should have is ceramic polysynthetic thermal compound, commonly known as "Arctic Silver". Be sure to re-apply some if you damage it. I use this same stuff on the Infineon controllers and have never had an issue with it.
Hope this helps!
-Mike
Indeed after changing the resistor R25 from 2k7 to 2k2 the range opened to 56V. GREAT!mwkeefer said:
I charged my 12s a123 battery and after cutting one of the three shunts the current is max. 8,35A at 44V so the watts up reads max. 367W. The fan did not start running, the case was hand warm.
mwkeefer
1 MW
manfred,
Thanks for the report back.. What was your current maximum before removing a shunt?
The shunt trick works, and fairly well however; I would recommend modifying error input amp resistor network instead of the shunts. For a quick hack, the shunts are easy but at higher voltages I am concerned with shunt mod (removing or increasing the resistance) will come increased heat dissipation on the shunts. The amount of issues this could cause are dependent upon the voltage and current output (obviously).
My other concern with shunt modification is because some of the units (even 48v) only have a single shunt.. modifying the shunt could lead to wrong operation or worse, a broken shunt and no power.
All the same, glad it worked!
-Mike
Thanks for the report back.. What was your current maximum before removing a shunt?
The shunt trick works, and fairly well however; I would recommend modifying error input amp resistor network instead of the shunts. For a quick hack, the shunts are easy but at higher voltages I am concerned with shunt mod (removing or increasing the resistance) will come increased heat dissipation on the shunts. The amount of issues this could cause are dependent upon the voltage and current output (obviously).
My other concern with shunt modification is because some of the units (even 48v) only have a single shunt.. modifying the shunt could lead to wrong operation or worse, a broken shunt and no power.
All the same, glad it worked!
-Mike
Vanquizor
10 mW
mwkeefer said:
I am looking for something in the 98-99 volt neighbourhood so on the voltage side I should be fine with 2 in series. I just want to get the current under control so they don't heat so bad while charging. My units are exactly the same board as I linked to (stamped: s-350&201 2005-09-19), with the noted exception of having one more cap installed in C36. Looking VERY closely the only other difference I found was my shunt is installed in J3 as opposed to J5 and my R37 is a diferent value. Perhaps R37 is also how the factory adjusts for different shunts.
Any reason a guy couldn't replace R37 with a 0-250ohm pot given that it currently measures at 269 ohms?
Hyena
10 GW
I'm not too sure, I didn't do anything particularly wrong...Vanquizor said:
I did have it powered up while nibbling away at the shunt trying to reduce the current. In the end I ended up cutting right through the shunt and had the 2 pieces just barely touching together and got 5a. I moved it around and the current dropped to 0 as the 2 sides of the shunt seperated, then poked with a screw driver making the connection again and the current came back up to 5-10a. It was still working at this point, and it was after this that I turned off the power and soldered in a new piece of wire to the shunt 'stubs'. When I turned it back on that's when it blew. Maybe those few seconds I had it powered up with no shunt connected upset it ? I don't know, it shouldn't really have blown IMO...
I have a replacement on the way for that same model as well as a genuine meanwell. It sounds like a few of us have the meanwell copies so I'll perservere with the mods when the new one arrives. If this one goes up in smoke too I'll be pissed!
The resistor mod is probably the best way to go but seeing as these ones have a slightly different circuit board to what Mike and others have sucessfully modded and mine made such a racket last time I'm reluctant to try again.
I think I'll test the current out of the box then power down, replace the shunt with some lighter gauge wire then power back up and see how it looks.
If it blows again:
A: I'll be pissed
B: We'll know not to mess with the shunts any more!
and
C: Hopefully Fechter will bestow another guinea pig award
Any updates with your testing so far Mike ?
mwkeefer
1 MW
Vanquizor,
You are correct with regards to the limited differences between units... I have documented them all on paper and am transposing them over to a simpler format for posting... additionally I have documented the PCB schematics for the current control circuit and voltage control / range portion.
That said... I've tried replacing the 280 ohm with a 250 ohm variable pot and had poor results with stability of output current and also voltage...
Once I have this finished up (tonight I hope, working on it now) I will be posting instructions for testing and then modification of voltage and current along with calculations to be done for proper voltage / current. There are a few other adjustments I will be suggesting also to enhance the safety and stability of these supplies.... if you can wait a short while, your chances of success should approach 98% with a good MTBF.
The other thing that has been holding me back is the lack of a decent scope, Fechter has one on hand (4 channel, lucky bugger) and now he has a 24v S-350-R17VAI to work with (he has been helping blindly for weeks).... there is only so much I can do with all the test points and monitoring of voltage but I am limited to a 10khz sample rate at best (not even close to a scope) so I can't really test the PWM on the scope to vary it properly.
BTW: on the 201 series, R37 is populated and decreasing it's value will lower the current... on our R18VAI and R17VAI units ( I have both, seems 24v is R18 and 48 is R17VAI) this resistor is not populated but it's location is there to form a fully variable (1-7A) adjustment... there is a transistor, diode and pot that aren't populated either which feedback into the TL494 error amp. Your units don't have this variable feedback with overshoot protection.
Although I am not finished testing (but working ferverently, even with my bank of headlights for rapid discharge @ 10-20A it still takes a while to cycle these test packs) I am including logging from the first charging tests I performed after my current mods... before this I was seeing peaks of 8.5-9A... this is much better...
A note, the peaks I mentioned earlier... I have figured them out, when the fan circuit engages it consumes too much current which causes the internal isloated logic voltage bus to drop and therefore causes the voltage presented to the ERROR AMP inputs, the DEAD TIME CONTROL or both to fluctuate (one is negative biased, one positive)... this in turn causes the supply to produce a higher PWM and as such more current, ie the spikes - I am working on this issue now...
[Configuration]
Meanwell S-350-48 (v S-350-R17) - Voltage output modified for 63.8v maximum, set to 62.25 (ish) and current modified by replacing my R33 with 2x 100ohm (Radio Shack) resistors in series.
3x5S Turnigy Lipo 5000mah 20/30C in series for 15S1P or 15S 5AH... pack starting voltage was approx 55.8v and initial (intentional) inbalance was 30mv +- 4mv (resolution accuracy but measured on a cell by cell basis).
Eagle tree has Meanwell as Batt and Batts as ESC to properly track positive AMPS and MAH accumulated. Connected to PC and running the EagleTree software in Live Mode. Also monitored the Power Supply Temps.
For those interested, here is the full EagleTree log file for your viewing pleasure: http://www.e-bikemike.com/reviews/M...Charge_1_15S1P_1C_Meanwell-S-350-48_62.25.zip
Without further delay... some pics(since Pics rule the world in ES).
Amps and Volts over Time - The peaks are from fan switching on, note the peaks in voltage correspond with the peaks in current... this means either both of the regulation circuits (Current and Voltage) or the DTC (which sets maximum duty cycle) MUST be seeing issue from this voltage drop.
Amps volts and watts over time (for those who don't want to do the math):
Since warm to the touch isn't accurate enough - here Amperage, Wattage and Temperature of the Meanwell over the charge process:
Configuration prior to connection of load (batteries):
These two screen shots are first few seconds (milliseconds?) after connecting the pack (load) and show the ramp up to current output... I have also been addressing this as I would prefer a slower current ramp up than this... (so I am playing with adding a ramp delay to startup):
First Sample grabbed on connect:
Stabilized after connect:
The final result:
You are correct with regards to the limited differences between units... I have documented them all on paper and am transposing them over to a simpler format for posting... additionally I have documented the PCB schematics for the current control circuit and voltage control / range portion.
That said... I've tried replacing the 280 ohm with a 250 ohm variable pot and had poor results with stability of output current and also voltage...
Once I have this finished up (tonight I hope, working on it now) I will be posting instructions for testing and then modification of voltage and current along with calculations to be done for proper voltage / current. There are a few other adjustments I will be suggesting also to enhance the safety and stability of these supplies.... if you can wait a short while, your chances of success should approach 98% with a good MTBF.
The other thing that has been holding me back is the lack of a decent scope, Fechter has one on hand (4 channel, lucky bugger) and now he has a 24v S-350-R17VAI to work with (he has been helping blindly for weeks).... there is only so much I can do with all the test points and monitoring of voltage but I am limited to a 10khz sample rate at best (not even close to a scope) so I can't really test the PWM on the scope to vary it properly.
BTW: on the 201 series, R37 is populated and decreasing it's value will lower the current... on our R18VAI and R17VAI units ( I have both, seems 24v is R18 and 48 is R17VAI) this resistor is not populated but it's location is there to form a fully variable (1-7A) adjustment... there is a transistor, diode and pot that aren't populated either which feedback into the TL494 error amp. Your units don't have this variable feedback with overshoot protection.
Although I am not finished testing (but working ferverently, even with my bank of headlights for rapid discharge @ 10-20A it still takes a while to cycle these test packs) I am including logging from the first charging tests I performed after my current mods... before this I was seeing peaks of 8.5-9A... this is much better...
A note, the peaks I mentioned earlier... I have figured them out, when the fan circuit engages it consumes too much current which causes the internal isloated logic voltage bus to drop and therefore causes the voltage presented to the ERROR AMP inputs, the DEAD TIME CONTROL or both to fluctuate (one is negative biased, one positive)... this in turn causes the supply to produce a higher PWM and as such more current, ie the spikes - I am working on this issue now...
[Configuration]
Meanwell S-350-48 (v S-350-R17) - Voltage output modified for 63.8v maximum, set to 62.25 (ish) and current modified by replacing my R33 with 2x 100ohm (Radio Shack) resistors in series.
3x5S Turnigy Lipo 5000mah 20/30C in series for 15S1P or 15S 5AH... pack starting voltage was approx 55.8v and initial (intentional) inbalance was 30mv +- 4mv (resolution accuracy but measured on a cell by cell basis).
Eagle tree has Meanwell as Batt and Batts as ESC to properly track positive AMPS and MAH accumulated. Connected to PC and running the EagleTree software in Live Mode. Also monitored the Power Supply Temps.
For those interested, here is the full EagleTree log file for your viewing pleasure: http://www.e-bikemike.com/reviews/M...Charge_1_15S1P_1C_Meanwell-S-350-48_62.25.zip
Without further delay... some pics(since Pics rule the world in ES).
Amps and Volts over Time - The peaks are from fan switching on, note the peaks in voltage correspond with the peaks in current... this means either both of the regulation circuits (Current and Voltage) or the DTC (which sets maximum duty cycle) MUST be seeing issue from this voltage drop.
Amps volts and watts over time (for those who don't want to do the math):
Since warm to the touch isn't accurate enough - here Amperage, Wattage and Temperature of the Meanwell over the charge process:
Configuration prior to connection of load (batteries):
These two screen shots are first few seconds (milliseconds?) after connecting the pack (load) and show the ramp up to current output... I have also been addressing this as I would prefer a slower current ramp up than this... (so I am playing with adding a ramp delay to startup):
First Sample grabbed on connect:
Stabilized after connect:
The final result:
tailwind
100 W
To get around the fan cutting in and out, I gave it its own independent power supply.
Greg
Greg
Hyena
10 GW
I did this on my passively cooled 145w meanwell. Why not just pinch 12v from somewhere in the power supply itself though ???
I had a poke around with the multimeter while powered up and found 12v off the back of a zener diode and much to my suprise found that when under load and supplying alot of current the voltage actually increased, so I ended up with a variable speed fan that increased with current draw. BONUS!
Cool graphs Mike. Why does it cut out at ~ 38, 47 and 54 minute ? Is it going into hiccup mode or did a lead come off or something ?
I had a poke around with the multimeter while powered up and found 12v off the back of a zener diode and much to my suprise found that when under load and supplying alot of current the voltage actually increased, so I ended up with a variable speed fan that increased with current draw. BONUS!
Cool graphs Mike. Why does it cut out at ~ 38, 47 and 54 minute ? Is it going into hiccup mode or did a lead come off or something ?
mwkeefer
1 MW
*redundant
mwkeefer
1 MW
Hyena said:
Hyena,
hmmm... 12v is increasing when under load... Think this through and look at the peaks in my graph (the out of band or odd peaks), those are when the fan kicks in and the unit produces more power becuase of the associated current draw on the low voltage rail. This is exactly the issue I'm attempting to correct for... I suspect if you analyze your meanwell 145 (I have one too) under full load with the fan off, then with the fan on and anayze the curve you will see the same spikes... since we want constant current at maximum power (safe maximum) these spikes could be deadly or at best reduce the average time between failures.
My thought is to either beef up the low voltage bus (maybe not a good idea, if additional load caused higher currents... what will additional available current do? Probably the inverse knowing these things) but for the stated reasons... I want to avoid that so we have:
1.) Isolated power (wall wart)
2.) Secondary step down regulator - tapped into the output of the primary stepdown transformer.
3.) Dedicated tiny SMPS .5A @ 5v-12v built into the meanwell just to power the fan and tapping the same input AC as the main supply.
Of these I think 1 is the best short term solution (some of the nokia and the motorola cell phone chargers are tiny enough to fit inside the meanwells) especially if we can embed them into the unit. Actually leaving the fan on all the time would be best so if we can embed a cell phone charger (or other properly small isolated supply) then it would be a fine long term solution (if not elegant).
Solution 3 is what interests me the most.. no parasitic loss from the power supply output power, low or no noise injection and yes... a more "elegant" solution.
Again... #1 is fine, I would let the fan always run... it will keep the temperature more stable and lower.
Greg - Nice... does the fan have NEON? I have tons of those puppies from cases, power supplies... small wind tunnels, etc = )_
-Mike
PS: I forgot the cutouts at 38, 47 and 54 minutes are me checking the pack voltage and adjusting the three switchable power levels I have (tuning for adjustable switch based 4.15v cutout, 4.17 or 4.19v)... What your seeing is me disconnecting the packs and reconnecting them a moment or so later with an ever so slightly higher voltage cutout... Basically I'm manually balancing by draining highest cells down in the gap, turning up the current and hitting them again... end results were 8mv across the 3 5S packs (if I remember).
tailwind
100 W
It did, but the red glow coming from the kitchen was stressing me out!
I just charge at home, so the setup works ok. Its not very portable though. Just too many bits to carry around.
I havent moded the power supply yet. I am interested in setting a 5A limit on the current.
At the moment I just use the wattsup to monitor the voltage and limit the watts to about 250w.
Every 10 minutes I go and tweak the voltage up until I reach my target voltage of 46.8v.
This gives me about a 1 hour charge time.
Greg
mwkeefer
1 MW
On another topic - the buzzing... here goes somthing really funny... on the last pack I topped off... it didn't buzz though it was over 60v and pushing 5A (it started at 4.11v per cell). I just connected up another 15S pack at 3.97v per cell and there's the freaking buzz again.
Funnier still... when I ran the top off the pack I just finished balancing (the one which didn't buzz) the power seemed noticeably better... I think it's time to break down the individual packs from that 15S and see if there is some issue - maybe I did fry somthing when I shorted it out a while back. That would suck!
-Mike
Funnier still... when I ran the top off the pack I just finished balancing (the one which didn't buzz) the power seemed noticeably better... I think it's time to break down the individual packs from that 15S and see if there is some issue - maybe I did fry somthing when I shorted it out a while back. That would suck!
-Mike
Hyena
10 GW
So you've been charging packs happily with it buzzing like that ? Or did you turn it off when you heard it ?
My packs were around 3.8-3.9v/cell when I tested and heard it too. Maybe it's to do with the voltage difference between the supply output and the battery that's causing it ?
Actually, now that I think of if I've got one of those universal laptop chargers that I was playing around with initially as lipo chargers and when I increased the voltage to more than the the pack rating (eg if I had a pack at 14v that should have charged to 16.8v but I set the voltage to 18v temporarily) it'd make that arcing/buzzing noise then shut down. Luckily power cycling fixed it, but maybe its the same deal in our meanwells ?
My packs were around 3.8-3.9v/cell when I tested and heard it too. Maybe it's to do with the voltage difference between the supply output and the battery that's causing it ?
Actually, now that I think of if I've got one of those universal laptop chargers that I was playing around with initially as lipo chargers and when I increased the voltage to more than the the pack rating (eg if I had a pack at 14v that should have charged to 16.8v but I set the voltage to 18v temporarily) it'd make that arcing/buzzing noise then shut down. Luckily power cycling fixed it, but maybe its the same deal in our meanwells ?
mwkeefer
1 MW
Hyena,
I charge right through the buzzing..
I must admit turning it down a bit (current) so it stays under 350 seems to cut the noise in half but again - I did a top off on a pack which I haven't tested using the meanwell yet and it was silent!
I suspect as we shift the voltage regulation circuitry upwards... we also change the duty cycle which is already way too high but more importantly I believe we are changing the operating frequency (drift and shift baby)... Fechter can scope this if he has time to see what's happening (or I can take it to my local electronics repair guy... who has recently developed a new respect for me, he loves this homebrew kit stuff!).
If it turns out to be as I suspect.... we are changing the Ct/Rt network timing through the voltage bump... we just need to recalculate the proper Ct/Rt for proper oscillator speed... the current limiting will handle PWM and DTC for us (Pulse width modulation and Dead time control) which will handle the total power output and curve point.
The next pack I charged was the BUZZER pack... When I was charging 5S3P using these at 20.7v they buzzed then too but hadn't been adjusted for proper current output of 350w.
Im going to split this buzzing pack apart after the next discharge cycle (5AM coffee run to wawa) but I should add that I blew a bit off the non-buzzer pack earlier (just a miles ride) and didn't get a bit of buzz out of that pack when I topped it off + I felt more performance out of it (I swear it was performance not the cold).
Who knows? I'll continue testing... I have even gone as far as running a 40' 20A extension cord from a dedicated breaker with nothing else attached to the mains just in case my 6+ pcs and other equipment in my home lab were causing feedback. The fan was causing some noise for sure when in the same breaker bar... I think the CFLs may have played a part also.
Do you have the cases grounded to earth? I never bothered (never do) but that could also be part of the issue I suppose... but yes I've done > 10 charges while monitoring current and temperatures (pack and charger) and so I just left it buzz... the source seems to be the largest yellow covered "transformer" or whatever it is, closest to the front of the supply.
I keep hearing my ion generator going buzzz... thinking I left the lipo in and charging = )_
-Mike
I charge right through the buzzing..
I must admit turning it down a bit (current) so it stays under 350 seems to cut the noise in half but again - I did a top off on a pack which I haven't tested using the meanwell yet and it was silent!
I suspect as we shift the voltage regulation circuitry upwards... we also change the duty cycle which is already way too high but more importantly I believe we are changing the operating frequency (drift and shift baby)... Fechter can scope this if he has time to see what's happening (or I can take it to my local electronics repair guy... who has recently developed a new respect for me, he loves this homebrew kit stuff!).
If it turns out to be as I suspect.... we are changing the Ct/Rt network timing through the voltage bump... we just need to recalculate the proper Ct/Rt for proper oscillator speed... the current limiting will handle PWM and DTC for us (Pulse width modulation and Dead time control) which will handle the total power output and curve point.
The next pack I charged was the BUZZER pack... When I was charging 5S3P using these at 20.7v they buzzed then too but hadn't been adjusted for proper current output of 350w.
Im going to split this buzzing pack apart after the next discharge cycle (5AM coffee run to wawa) but I should add that I blew a bit off the non-buzzer pack earlier (just a miles ride) and didn't get a bit of buzz out of that pack when I topped it off + I felt more performance out of it (I swear it was performance not the cold).
Who knows? I'll continue testing... I have even gone as far as running a 40' 20A extension cord from a dedicated breaker with nothing else attached to the mains just in case my 6+ pcs and other equipment in my home lab were causing feedback. The fan was causing some noise for sure when in the same breaker bar... I think the CFLs may have played a part also.
Do you have the cases grounded to earth? I never bothered (never do) but that could also be part of the issue I suppose... but yes I've done > 10 charges while monitoring current and temperatures (pack and charger) and so I just left it buzz... the source seems to be the largest yellow covered "transformer" or whatever it is, closest to the front of the supply.
I keep hearing my ion generator going buzzz... thinking I left the lipo in and charging = )_
-Mike
Hyena
10 GW
My buzzing ONLY came from the R37 resistor mod. The second I touched another resistor across in in parallel the current dropped and it started the arcing/buzzing sound. Remove it and the current shoots back up and the arcing stopped. The packs I was charging are also brand new, but as I said they happily charged silently at the full current output from the power supply so in my case atleast its not the packs causing the buzzing (and probably less likely in your case too, unless one cell is madly sucking current or something ? )
Oh and yeah I have the earth connected - well, the earth connector to the AC power lead earth. I haven't physically run a lead from that earth terminal to the side of the case. to the I have a CFL light over my work bench that buzzes quite loud sometimes, but a sharp whack fixes it. Hmmm, maybe I should try that with the PSU next time... :lol:
Oh and yeah I have the earth connected - well, the earth connector to the AC power lead earth. I haven't physically run a lead from that earth terminal to the side of the case. to the I have a CFL light over my work bench that buzzes quite loud sometimes, but a sharp whack fixes it. Hmmm, maybe I should try that with the PSU next time... :lol:
mwkeefer
1 MW
Hyena said:
Don't bother whacking, I tried it and used a hammer even ! (mostly kidding).
I had worse buzzing from the 48v unit before the current resistor mods, again I think it's the OSC rate being changed with the central voltage changing... im workin on it but I think fechter will beat me to a real diagnosis / fix.
-Mike
Hyena
10 GW
Well my new power supply arrived today and to my suprise when I opened the case I found that it's actually slightly different to the previous one I blew up. It sounds like it's exactly the same as the one Vanquizor has, with C36 populated this time and the shunt in J3. I didnt check the R37 value but I wouldn't mind betting its lower. Out of the box the current was only 9.6 amps this time, still more than the rated but not as excessive as before. I nibbled at the shunt a bit and got it down to 9.1a and I'm going to leave it at that. That's fine for charging a 10ah turnigy pack and although they can have bulk current belted into them I'd rather keep the charge current under 1C in the interest of longevity.
Annoyingly I found SVR1 only went to 900 ohms so I couldn't go below 43v. It was marked the same value as my previous supply so I guess it's poor QC / crappy component tolerance. I replaced it with the pot from the other power supply and managed to get the voltage down to 41.6v and it's now charges happily and quietly.
On the previous topic of the buzzing, I recorded a short video incase its of any interest. The quality is pretty crap off my phone but you can make out the noise. It starts and stops because I'm just bumping another resister across R37.
[youtube]PsTc2MVHy94[/youtube]
Annoyingly I found SVR1 only went to 900 ohms so I couldn't go below 43v. It was marked the same value as my previous supply so I guess it's poor QC / crappy component tolerance. I replaced it with the pot from the other power supply and managed to get the voltage down to 41.6v and it's now charges happily and quietly.
On the previous topic of the buzzing, I recorded a short video incase its of any interest. The quality is pretty crap off my phone but you can make out the noise. It starts and stops because I'm just bumping another resister across R37.
[youtube]PsTc2MVHy94[/youtube]
heathyoung
100 kW
Sounds like a cicada
Seriously though - that sounds like instability in the feedback loop - have you scoped the output? The frequency should be around 40Khz or so, you should never hear it, but when the frequency drops (or when you get weird harmonics) the ferrite cores 'sing' especially if they are gapped. This is a good tell-tale that the PSU is NOT happy with the load. When I was building my own, that was usually the point where my high-side switches perished, or got REALLY hot.
Have you post-mortum'ed the old PSU's? I wouldn't mind a look at some of your dead-'uns to see what happened... Adjustable SMPS'es are interesting beasts, they are adjustable upwards less so than downwards - reason being that you still have a step-down transformer with fixed winding ratios - if you have (say) a ratio of 340:15 (340 comes from the 340V DC of rectified 240V AC mains * 1.414 due to sinusoidal waveform). This gives 12V DC out easily. If you try to get 20V out - not a chance. You may see 20V at a low current capacity, due to rectified spikes/higher mains voltage locally, it will sag on load and will try to cook itself trying to get there on load.
There is some leeway (obviously) in the winding ratios to cope with output/input sag/surge etc. but there comes a point where a higher duty cycle just wont produce any more voltage.
Better to adjust down a higher voltage PSU thats close than adjust up a lower voltage thats close - you are running so much closer to the limits of design. Higher voltage down means lower duty cycle (+lower current) than it would normally run at, so more headroom when its called on 'to produce the goods'.
Best is to buy a PSU that is the correct voltage for the application in the first place
Seriously though - that sounds like instability in the feedback loop - have you scoped the output? The frequency should be around 40Khz or so, you should never hear it, but when the frequency drops (or when you get weird harmonics) the ferrite cores 'sing' especially if they are gapped. This is a good tell-tale that the PSU is NOT happy with the load. When I was building my own, that was usually the point where my high-side switches perished, or got REALLY hot.
Have you post-mortum'ed the old PSU's? I wouldn't mind a look at some of your dead-'uns to see what happened... Adjustable SMPS'es are interesting beasts, they are adjustable upwards less so than downwards - reason being that you still have a step-down transformer with fixed winding ratios - if you have (say) a ratio of 340:15 (340 comes from the 340V DC of rectified 240V AC mains * 1.414 due to sinusoidal waveform). This gives 12V DC out easily. If you try to get 20V out - not a chance. You may see 20V at a low current capacity, due to rectified spikes/higher mains voltage locally, it will sag on load and will try to cook itself trying to get there on load.
There is some leeway (obviously) in the winding ratios to cope with output/input sag/surge etc. but there comes a point where a higher duty cycle just wont produce any more voltage.
Better to adjust down a higher voltage PSU thats close than adjust up a lower voltage thats close - you are running so much closer to the limits of design. Higher voltage down means lower duty cycle (+lower current) than it would normally run at, so more headroom when its called on 'to produce the goods'.
Best is to buy a PSU that is the correct voltage for the application in the first place
Hyena
10 GW
The pics I posted earlier were of the one that blew - thats about as far as I got, I didn't bother trying to test individual components. For $40 delivered it's not really worth trying to troubleshoot. If there's anything in particular you want me to test or look at though let me know. I dont have a scope or any other cool toys, just a multimetre and a few other basic bits and pieces.heathyoung said:
I've played with the replacement some more and found that it does actually make the buzzing / cicada noise at higher current. I've got it set at 9.2a which it happily charges at silently, but when the fan kicks in the current spikes up to 10.5a and it starts chirping. When the fan goes off the current drops back it and it's silent again. As the current starts to drop off at the end of the charge it doesn't make any noise any more when the fan kicks in, so it seems to be when the current exceeds 10a it does it.
Here's a vid demonstrating the above
[youtube]E4iFaiTvVDA[/youtube]
(real cicadas can be heard at the end
)
Tiberius
10 kW
Hi,
I had buzzing from one of mine when running it in current limiting mode. It appears to be an instability in the control loop.
I have a variable electronic load and I used that to simulate a range of load resistances. With that it was always stable, but when charging batteries there was often a stage where it went unstable, usually near the beginning of the charge. So clearly, a discharged battery isn't a linear resistive load; well we knew that, but in some cases the effect is enough to upset the PSU control loop.
In my case the buzzing was with my external current control loop, so there was the potential problem of two interacting control loops and the solution was simple. Make sure the natural frequency of the two loops are well apart. It is of course possible for a single loop to go unstable with particular loads. The fix is often dead simple, but it really requires knowing the circuit to find out which are the dominant filter components.
http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=6749&p=209094#p209081
Nick
Edited to add: A battery has some of the characteristics of a capacitor, and designing circuits to drive capacitive loads is a well known problem. From the point of view of stability, a capacitive load is the worst thing to hang onto a circuit with a feedback loop.
I had buzzing from one of mine when running it in current limiting mode. It appears to be an instability in the control loop.
I have a variable electronic load and I used that to simulate a range of load resistances. With that it was always stable, but when charging batteries there was often a stage where it went unstable, usually near the beginning of the charge. So clearly, a discharged battery isn't a linear resistive load; well we knew that, but in some cases the effect is enough to upset the PSU control loop.
In my case the buzzing was with my external current control loop, so there was the potential problem of two interacting control loops and the solution was simple. Make sure the natural frequency of the two loops are well apart. It is of course possible for a single loop to go unstable with particular loads. The fix is often dead simple, but it really requires knowing the circuit to find out which are the dominant filter components.
http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=6749&p=209094#p209081
Nick
Edited to add: A battery has some of the characteristics of a capacitor, and designing circuits to drive capacitive loads is a well known problem. From the point of view of stability, a capacitive load is the worst thing to hang onto a circuit with a feedback loop.
Hyena
10 GW
Cool, I missed that part of the other thread, I'll have to try adding a cap and see if that works.
In the mean time, it just isn't my week, I just had another SWPS go up in smoke in a fashion that could only be described as truely epic!
At first I thought it was my lipos going up and was about to exit stage left at a rate of knots but then I realised the smoke was pouring out of my charger.
This one was a 24v 'JAE' brand and has been performing well for months as my every day charger. I have it set to 20.7v and charge up my 15S pack in 3 parallel banks of 5S. I have a 6 pin anderson plug on the battery and one lead for charging in parallel and another for discharging in series which is connected to the bike so it's almost impossible to get it wrong. However... to cap off a bad week, my bike frame snapped a few days ago so I stripped the bike and have it in pieces in my workshop. I'm used to the series discharge plug always being fixed to the bike so didn't even think when I plugged it in and sent 60v odd backwards in the charger. It wasn't turned on or plugged in but that didn't stop it. I've never seen so much smoke come out of an electronic device that wasn't actually on fire! When I opened it up I found the 2 caps at the front had blown, covering the most of the board in 'capacitor juice'. They are only rated at 36v so not suprising they blew with 60v.
Does anyone think they alone would have been taken out ? Nothing else looks phyically damaged. If it's likely just them that have popped I'll replace them otherwise I'll scrap it.
Also for the sake of documenting different board layouts - this is the 350w 24v JAE.
The board is quite different to the meanwell and the other clone I have. The obvious differences at a glance are only 2 caps up front with only an unpopulated 3rd (vs the others that have 4 or 3 with the 4th unpopulated), there's 2 big transistors heatsinked to the case up front, vs the others that only have 1 and other components are in different locations or absent.
In the mean time, it just isn't my week, I just had another SWPS go up in smoke in a fashion that could only be described as truely epic!
At first I thought it was my lipos going up and was about to exit stage left at a rate of knots but then I realised the smoke was pouring out of my charger.
This one was a 24v 'JAE' brand and has been performing well for months as my every day charger. I have it set to 20.7v and charge up my 15S pack in 3 parallel banks of 5S. I have a 6 pin anderson plug on the battery and one lead for charging in parallel and another for discharging in series which is connected to the bike so it's almost impossible to get it wrong. However... to cap off a bad week, my bike frame snapped a few days ago so I stripped the bike and have it in pieces in my workshop. I'm used to the series discharge plug always being fixed to the bike so didn't even think when I plugged it in and sent 60v odd backwards in the charger. It wasn't turned on or plugged in but that didn't stop it. I've never seen so much smoke come out of an electronic device that wasn't actually on fire! When I opened it up I found the 2 caps at the front had blown, covering the most of the board in 'capacitor juice'. They are only rated at 36v so not suprising they blew with 60v.
Does anyone think they alone would have been taken out ? Nothing else looks phyically damaged. If it's likely just them that have popped I'll replace them otherwise I'll scrap it.
Also for the sake of documenting different board layouts - this is the 350w 24v JAE.
The board is quite different to the meanwell and the other clone I have. The obvious differences at a glance are only 2 caps up front with only an unpopulated 3rd (vs the others that have 4 or 3 with the 4th unpopulated), there's 2 big transistors heatsinked to the case up front, vs the others that only have 1 and other components are in different locations or absent.
heathyoung
100 kW
$40 delivered? Who is your source - computronics want more than that...
