Switch mode supply mod

[Hyena]

Yeah computronics charge like wounded buffalo's. I'm all for supporting a local business/importer but when they're triple the cost it's a bit hard. Although if you really want a giggle try pricing them at jaycar!

I've had a few now, but my latest ones I've been getting and posted about earlier are from here:

http://cgi.ebay.com.au/ws/eBayISAPI.dll?ViewItem&item=150387972110&ssPageName=STRK:MEWNX:IT

The seller is very reasonable too, I had one arrive with a dead fan and although it's an easy fix I thought I'd email them to see what they said. I expected them to say they'd replace it if I sent it back (at which point I'd go down the street and buy a fan for $5) but instead they replied 10 minutes later with "we'll send you another one" and a week later it was on my doorstep. I didn't feel so bad when I blew it up a day later :lol: (although maybe it blew because the fan wasn't the only problem with it ?)

If you want a genuine meanwell they're $58 delivered from sure electronics (which is still cheap)
http://cgi.ebay.com.au/ws/eBayISAPI.dll?ViewItem&item=250538261153&ssPageName=STRK:MEWNX:IT

 

[mwkeefer]

Hey all,

Sure-electronics is a good spot to get the meanwells and if you order enough (5+) the price comes down a bit and S&H is reasonable.

Just in case anyone cares.. I toasted an alligator clip tonight... my parallel adapter on the positive side I had to use black shrink wrap on and I knew the moment I built the pack that it was a potential disaster waiting to happen... the good news, the supply and packs are fine and though I had blackened my fingers I suffered no burns. My son was on the other side of the room and it even served as a good teaching example of why he isn't allowed to touch daddy's stuff when I'm asleep. All in all it was a single Female 4mm gold connector and the tip of the aligator clip that vapoorized in my hands - I love indoor fireworks.

About the buzzing... I don't have a scope worthwhile enough to measure the OSC of the various portions of the PCB but I know a few guys out here who do.. I have managed to nearly eliminate buzzing in my 48v to 63v covert model... the trick was not "EXPANDING" the range but moving the range up. Once I re-populated the voltage regulation resistor network with values to give me roughly 55-65v of output (zener limited and tested to 63v) so I keep the same basic "range" just move it up the voltage spectrum. I think nick is right though, that it is because of the two limiting functions of the supply working at too close a frequency... I just don't have any viable way to test.

The current jumping (or peaks) when the fan comes on is exactly what I explained previously - when the fan kicks on, it draws power from the same isolated and scaled voltage power source that the TL494 uses... this additional current draw causes the voltage being supplied to the DTC (Dead time control) to drop. This pin works in INVERSE mode so 0v input = MAXIMUM CURRENT, 3.3v on the pin = 0 CURRENT. Since the additional load of the fan causes a voltage drop, that is the reason the current spikes up when the fan switches on.

I've solved this problem in my own units by building a small internal supply (which uses the output voltage as a source) which is wired up much the same as the Infineon ANY VOLTAGE modification to an LM317 and outputting 12v (very little current). Since this is pulled from the main outputs instead of the logic VCC bus... there is no drop on the DTC pin with the fan running... this means a flat curve without the peaks.

I will share the schematic once I have any issues worked out.

Hope this helps!

-Mike

 

[Tiberius]

The current jumping (or peaks) when the fan comes on is exactly what I explained previously - when the fan kicks on, it draws power from the same isolated and scaled voltage power source that the TL494 uses... this additional current draw causes the voltage being supplied to the DTC (Dead time control) to drop. This pin works in INVERSE mode so 0v input = MAXIMUM CURRENT, 3.3v on the pin = 0 CURRENT. Since the additional load of the fan causes a voltage drop, that is the reason the current spikes up when the fan switches on.

I've seen a similar effect on my ones. The parameters change slightly when the fan goes on or off.

But....I decided to live with it. We're trying to make battery chargers here. If I were building a supply to power some sensitive electronics then I'd be doing things very differently. What led me to these PSUs was not their performance in terms of accuracy or noise, but simply the W/$, W/Kg and W/cuft.

Nick

 

[mwkeefer]

Nick,

For my own personal use... I am in 100% agreement with you, just use them with proper nominal current output and the voltage needed. I thought it a good idea for others if I test and push this on all fronts, trying to make it perfect.

Personally... I've been linking a pair of voltage modified 24v units for a 650w nominal charge into a 10A pack of 15s, this works out to a bit over 1C. My only concern here (and why I got into the fans in the first place) is that running close to maximum ratings is bad enough but add an amp overage (it's roughly what type of jump I was seeing) and things could get really bad, really quick - imho.

These things are great for the w/$ - even with the ones I've sent out to various people (gratis) and blown up, I've still paid less for these all together than for 2 iChargers (without DC supplies or S&H). What I have ended up with is one portable (semi?) 5 - 7.5A charger (10-15S) and a set of "base station" chargers one for 10A @ 15S charging (1C) and the other for 2C or 20A charge of 15S (yep, 4 supplies in series / parallel and trimmed to 300w each nominal).

I can't tell you how freaking cool it is to be able to load up my 15S2P to 90% within 20-30 minutes!

Again thanks to you, Fechter, Jermey and Methods (I'd not have gone this way without you guys!)!

-Mike

 

[Vanquizor]

With the concerns about instability with the fan cycling wouldn't the quick and dirty be to just force the fan to run 100% of the time? Would still give predictable output and it can't hurt life cycle to keep things a little cooler.

Suggestions on the best way to accomplish this?

 

[Hyena]

This was mentioned earlier, either find 12v on the board on use a small phone charger or similar, powered off the same AC input but running the fan seperately.

I agree with Nick too, as battery chargers they don't need to be uber precise to the 10th decimal place and they're pretty good as is. I doubt a few spikes as the current tapers off is going to be the end of the world and drastically reduce pack life.

Also, I managed to repair my other supply I blew up the other day. The blown caps were 2 35v 2200uf units and I replaced them with 2 63v 1000uf units from my dead 48v supply (nearest match I had around) I was tempted to fit an additional cap to the spot along side that was previously unpopulated (given the original units were 2x 2200uf and I only replaced them with 2x 1000uf) but after a 10 minute test it seems to be ok so I'll leave it alone.
So I guess that's handy to know - if you happen to be a dumb ass like me and have a lapse in concentration that sees you connecting too high a pack voltage, it's only those caps at the front that will blow.

 

[mwkeefer]

Hyena,

12v on the board is the logic supply rail... that's what is being tapped now and would cause the same issue.

The 5v-12v supply (wall pack style) is fine... but I want to continue researching and testing, refining which includes a very simple 12v regulated supply drawn from the output stage... it's a few resistors a diode and a LM317 and can source up to 1A if needed. Also very efficient and tiny enough to be installed inside each unit.

While I agree with you both, peaks to the packs won't matter but... peaks for the supply when already configured for 350w continuous output power push up over 480 and without initial current reduction were approaching 10A (that's 580w output)... that was without a peak caused by increased load on the 12v internal rail...

Im sorry but until someone here corrects me with regards to my estimation that these supplies should last quite a while with proper power limiting and fan not driven by the internal logic 12v bus, but if just run silly nilly (it takes an hour of continuous operation to charge in many cases) they will surely give up the white puff in short time.

Please do correct me if I am wrong or someone has experience with this... I am basing this on my own experiences and also the specs of the most probably points of failure (transformer, FETs, Diode, Caps).

*With that aside, I'm curious and would like to ask the experts a question...

As everyone knows some models (more recent) of the 350 have an unpopulated section of the PCB just aft of U1 (TL494CN)... In reviewing this circuit, I understand how 1/2 of it works but I am at a total loss for the purpose of the second half. A bit more specific:

If R33 is removed from the PCB and R37 is populated in addition to SVR2 it forms a variable current or adjustable current feedback loop.

The part I just for the life of me don't understand (read that as I'm not sure what value components to install here) is involving D18 and Q7. In my conversations (limited as they were, the tech guys at meanwell are awesome - sales at Meanwell USA isn't even worth bothing with and I was inquiring with regards to 1000 unit purchase, the sales biat** referred me to mouser who had 2 in stock?) with Meanwell tech, he was very hesitant (I think he could get in trouble) but at least informed me the transistor is of type NPN.

I have the main parts were interested in (opamp, shunts, voltage reference and current reference) in schematic form and I'm just in the process of double/triple checking I didn't miss anything important (Like as if I know what important is) - once I've verified them I'm happy to post but I just thought maybe one of the EE or just all round geniuses could figure out the purpose (if they have a model with missing parts on the PCB).

Thanks in advance!

-Mike

 

[fechter]

The part I just for the life of me don't understand (read that as I'm not sure what value components to install here) is involving D18 and Q7. In my conversations (limited as they were, the tech guys at meanwell are awesome - sales at Meanwell USA isn't even worth bothing with and I was inquiring with regards to 1000 unit purchase, the sales biat** referred me to mouser who had 2 in stock?) with Meanwell tech, he was very hesitant (I think he could get in trouble) but at least informed me the transistor is of type NPN.

-Mike

As far as I could tell, that was some kind of backup overvoltage protection. If D18 is a zener and it turns on Q7 above a certain voltage, it can pull on the same op amp that the current limiter does and turn down the duty cycle.

 

[mwkeefer]

With the concerns about instability with the fan cycling wouldn't the quick and dirty be to just force the fan to run 100% of the time? Would still give predictable output and it can't hurt life cycle to keep things a little cooler.

Suggestions on the best way to accomplish this?

Not sure how I missed this post... but your right, if we enable the fan by default... the load to the circuit will be constant then we just adjust for the new offset in the R33/R37.

I'll try this in the morning...

Right now I have replaced R33 with a piece of perf board holding 2 100 ohm resistors (about 196 ohms) in series followed by a 100 ohm 10 turn pot. The factory value was 360 ohm on this 48v unit. This gives me a wide range of currents but I detect noise in transformer next to the case mnted diode... Funny part is, in most cases... adjusting the current up a turn removes the noise. The only issue here is that I didn't even run the fan (had the case open and a huge house fan going full clip on it) and was maintaining 450+ watts but the noise would dissappear for a while then creep up again (not nearly as bad) as the pack voltage climbed (and I suppose the resistance of the load, since the Ri should increase as they become full right?).

Additionally I replaced R25 with a 15k and SVR1 with a 1K 10 turn, in addition to the 1K... i tagged another POT on top of SVR1, a 100 ohm 10 turn for fine adjustments (now I can keep steady voltage between 62.25/62.27).

Maybe I should just replace my R33 variable circuit with a single 260 ohm precision (1%) and a higher precision 10 turn 100 ohm pot in series. If I leave that calibrated for 360 ohms (where R33 was) I can adjust from 260 ohms up to 360 ohms to give myself a taper (as I have now)... then remove the thermistor style fan control and replace with just simple voltage divider calculated to provide 2x the current required to run the fan. My bumps in current amount to approx 1A when the fan goes active - my existing adjustment circuit (200 ohm + 100 ohm pot) would manage just fine tapering to account for the draw from the fan load. I think the existing fan is 22ma or somthing (been a while), can't exactly drive a CPU or Case fan from that (22ma @ 12v) but... I suppose if that little fan they come with were on all the time and the components are properly sinked to the case (ie: artic silver on both sides of silicone spacer pads) then the temperatures would stay pretty darn low.

I have yet to try to move any of the caps around... I'm still trying to wrap my mind around it.

-Mike

PS: anyone tried any SICK voltage mods yet, ie: 72 or 83 or 88 or higher with these?

 

[fechter]

Part of the problem might be if you're increasing the output voltage too much over stock. It really needs a new transformer wind to push it really far. A sidestep approach is to simply use two or more supplies in series to get the desired voltage. With two in series, I suspect you could get away with just keeping one stock and mod the current limit on the other one. The one with the lowest limit will do all the limiting anyway. You may actually need to turn down the voltage on one in this case to get what you want.

 

[Hyena]

Mike, I picked up some 1k, 2k and 200 ohm 10 turn pots + a handful of 100 and 200 ohm resistors yesterday afternoon to do pretty much the exact thing you just described! I was just going to put the 2k in svr1 to allow the voltage to go a bit lower. What was R25 before you changed it ? Mine is 22k I think. If yours was the same and you've dropped 7k that sounds like a fair bit (assuming your R25 is in series with SVR1) Or is this big drop what's enabling you to go to 63v ? (plus upping the zener under the transformer of course).

As for higher voltage mods, if the transformer is even up to it you'd have to swap out the front caps atleast. Mine are only 63v units so you're running right at the upper limit of what the 48v models can do.

Part of the problem might be if you're increasing the output voltage too much over stock. It really needs a new transformer wind to push it really far

I don't think so Fechter, at least not completely. Mine has all the same symptoms Mike is describing and unmodified it's running at the lower end of its range (42v on a 48v unit)

 

[mwkeefer]

Part of the problem might be if you're increasing the output voltage too much over stock. It really needs a new transformer wind to push it really far. A sidestep approach is to simply use two or more supplies in series to get the desired voltage. With two in series, I suspect you could get away with just keeping one stock and mod the current limit on the other one. The one with the lowest limit will do all the limiting anyway. You may actually need to turn down the voltage on one in this case to get what you want.

Wouldn't the "correct" way to do this be, connect 2 in series (each of higher than 1/2 required voltage nominal so for 62.25 - 96v use 48v units) and then configure so each is identical in voltage first? Then adjust one of the 2 units to limit the PWM or current output?

I've tried that too... honestly the noise I'm getting now is different than the cicada noise everyone (including me) have been pointing to, it is much more subtle humming or buzzing and really i need to quiet all else in the room to hear it. The same humming occours when I put 2 48v units adjusted for 31.125 each... regardless of power levels I run them at (R33 mod limits).

So I think by moving the voltage range instead of just increasing the range helps and gets me/us closer to a well working charger...

Here is an graph of the 6.7mi WOT (40mph max) discharge trip:

Here is the charge plot:

Balance was nice at C/20 the pack was within 5mv (4.155 - 4.150).

The maximum temperature of the charger FETS durring operation was 89 F (not bad given an ambient of 72) - Probe attached to outside of heatsink portion of case... arcticsilver compound for better readings.

The buzzing or humming was adjustable out (a little tweak to the resistance of the current limit - silent after first adjustment to 7.02A) until the CV phase cross over point then I could hear the hum (nothing like before) until the current dropped below 6.8A (when I configured for flat 6.8 A, the hum is still present but lower down the voltage range and actually dissappears before CC/CV cross over).

Back to the refinements for tommorow... I will have a nice 6 or 7A portable charger if it kills me (it may).

-Mike

 

[mwkeefer]

Mike, I picked up some 1k, 2k and 200 ohm 10 turn pots + a handful of 100 and 200 ohm resistors yesterday afternoon to do pretty much the exact thing you just described! I was just going to put the 2k in svr1 to allow the voltage to go a bit lower. What was R25 before you changed it ? Mine is 22k I think. If yours was the same and you've dropped 7k that sounds like a fair bit (assuming your R25 is in series with SVR1) Or is this big drop what's enabling you to go to 63v ? (plus upping the zener under the transformer of course).

As for higher voltage mods, if the transformer is even up to it you'd have to swap out the front caps atleast. Mine are only 63v units so you're running right at the upper limit of what the 48v models can do.

Part of the problem might be if you're increasing the output voltage too much over stock. It really needs a new transformer wind to push it really far

I don't think so Fechter, at least not completely. Mine has all the same symptoms Mike is describing and unmodified it's running at the lower end of its range (42v on a 48v unit)

Actually... my 48v units come with 100v caps on the output already (sure-electronics) so they can handle it... the PCB is stamped with a 94-0V which in chinese means 0-94V output. This just means the design can handle it... I haven't measured the voltage output at the transformer taps on these units yet (I need to though) and I am going to stick a temp probe on there tommorow too (SNOW DAY, LOTS OF CURRENT DUMP).

One thing I did note... if you look at Meanwells specs (Like they are useful for anything) the higher the voltage, the more efficient the unit... I wonder if they don't source with transformer at high end range and tone it down? Tommorow.

My unit was different, R25 was 2k I think... I don't remember anymore = )_

But you have the relationship correct, lower impedance on my R25 (in series with SVR1). With the stock 1K pot... the extremes of range were:

Testing minimum output: 39.1v @ 2.82k
Testing maximum output: 58.1v @ 1.92k

And the voltage wasn't very steady..

(I did add a 5v zener to the stock 62v unit that came with my chargers...)
By using 1.5K + 1K pot + 100ohm pot... new maximum impedance is = 2.6k (so only 400 ohms or so lower than the default)... by default I could tune up to 58v so I needed just a little lower impedance in the range...

Rough low range is now 1.6k but the 1K and the 100 ohm were started dead center of 10 turns... so the outcome was tuned (before power up) to the midrange value of 2.050 K

I lost a bit of low end (more than I had planned)... I expected to have dial in between 41v and 63v but overshot the low end a tiny bit (not enough resistance to get the range low enough) but with the main SVR1 + the 100ohm trimmer... I'm getting much more stable voltage output (+- 3mv)

One last bit.. tommorow I'm replacing all my resistors in the voltage and current reference and adjustment circuits for 1 and 2% types... I think this may solve the remaining fluctuation with the output... if not, I will add more 100v caps.. lower ESR... that should flatten the output.

When I have all this "figured out" I am thinking that I will just setup a switch to select what voltage range I want (ie: 40-42, 49v-50.4, 61-63v)... once I see how stable I can get the voltage using better resistors in the network... I will look into configuring these as "chargers".

I'm thinking a 3 power level switch might be good also (400w, 350w and 100w)... this would be useful for bringing stubborn packs into balance over a longer charge curve.

Oh well... I'm rambling, I'll look at your PCB pics in the AM and figure out what points to check for impedance... then we can bump the unit either way or just expand the range.

-Mike

 

[mwkeefer]

One last bit...

Hyena - if your stock R25 is 22K then assume approx 23K max range with SVR1 existing... Just lift SVR1 pin #3 and wire up one of your 200 ohm pots in series (off pin #3, the independent unjoined pin).. that should increase the low end by about a volt or so.

Hot glue is your friend. These tiny 10 turn pots aren't stable just soldered into position with the legs bent, I find that mine are much more stable and user friendly once you add gobs of hot glue around them (not where the wires are or under, but around). You can always seperate the glue and it does a good job of holding things steady.

If you want to add a "fine tuning" adjustment... adding a smaller value (100, 200, etc) in series with SVR1 will give you minute adjustment (100 ohms / 10 turns = 10 ohms per turn) and you can just stack it ontop of another 10 turn but with the adjuster screw opposite. I will post pics tommorow when I can get to my camera.

-Mike

 

[Hyena]

Actually... my 48v units come with 100v caps on the output already (sure-electronics)

Ah ok, that's another difference in the cheaper knock offs then.

My unit was different, R25 was 2k I think... I don't remember anymore = )_

That's exactly why I take pictures of everything before I pull it apart
Mind you, taking lots of photos didn't stop me having 2 left over bolts and a broken tensioner the first time I swapped the camshaft in my car... :lol:

I'll look at your PCB pics in the AM and figure out what points to check for impedance... then we can bump the unit either way or just expand the range.

Don't spend too much time on it, I have a meanwell on the way from sure electronics that should be here in a few days - that's the one I'll do all the modding to so it should be exactly the same components and values as what you're using.

 

[mwkeefer]

Hyena,

i just checked (my own darn posts on the S-350-48) and it was 2k to start (R25) so a rough range of 2 - 3k with SVR1 pot (I did take photos just way too tired last night to get up and look).

Wondering how well the bike is going to run in this snow (I was expecting 3-4", we already have 10" and they are saying 15" - I don't think I have ground clearance to the bottom bracket sufficient for this?)

I should have a few chances to charge today and I'll try to work out the buzzing/humming issue (it may just be AC ripple).

With regards to the "knock offs"... The circuits aren't really different (trace level) by enough to make it complicated... A few tests with the DVM and you can locate and calculate the required resistors for a "safe mod", as of now though... tuning the output voltage and current is a manual "by ear" process.

-Mike

PS: I'll take some photo's of the snow for you, it really is SIC - I don't have sidewalks or grass anymore (12") when I look out the window.

 

[monster]

wouldn't it be cheaper to just recycle computer PSU's and connect them in series for the desired voltage rather than try and over-volt a commcial 48V power supply? 5x 12V PSU's would give you 60V, perfect for 48V lifepo4. PSU's can pull some really high amps too.

 

[momo]

That is fine but how big and bulky would that be. The whole idea of this is portability.

 

[Hyena]

Only they're not particularly stable or adjustable. Most PC PSU's I've played with are very twitchy too and quickly shut down when you try to pull high power from them. I had 3 that I tried to use as a highish current 12v source for powering an IMAX charger and 2 of the 3 shut down quickly, one dropping to 10.8v for a while before doing so.

 

[mwkeefer]

Most PC power supplies - ATX, ATX2, etc... and also most 12v style (automotive) supplies will only drop down to 10.8 (+- .2) before they self protect under load...

If you were going to use them as chargers, the issue would be with the initial pack empty charging... they would consume/produce too much current and voltage would drop below 10.8 cutout - this too can be modified but also remember they are not isolated by default like the meanwells are so they can't be used in series or parallel without some modification to isolate the power sections... it's a really big PITA!

I still have some testing to do on the meanwells - I've come up with a few more mods and believe I've solved most if not all of the issues with the meanwells but I don't want to put the info out here until I have a chance to test it more (wouldn't want anyone blowing a supply or pack on my account!)

-Mike

 

[mwkeefer]

Hello again,

Just an observation - as I have worked to make these Meanwell supplies (and hopefully the clones) stable for use as a CC/CV charger (ie: beyond the buzzing and overpower and heat and so on.) with the hopes of acheiving years of service from the units (yep, I'm nutz)... I've monitored and logged each charging session (and many of the iCharger, Hyperion, Triton, etc chargers too)...

As I have managed to further reduce the ripple voltage (when measuring 10 samples per second, before my most recent modification I was seeing 6-8mv of ripple, you may not see this with normal DVM) now to 3-4 mv at the worst case (I've actually got this working better, details to come after safety testing) and the strangest thing seems apparent from the logged data:

It seems that the lower the ripple voltage is, the better the end balance of the packs (without balancer) wind up... I say this because at each step of the process, each charge I realized the maximum ripple was also the final charge deviation or imbalance between cells. This may be coincidence or it may be cause and effect (could also just be the pack breaking in finally as I have hit about 18 cycles on this new pack 15S10P which was precharged once with an iCharger and since has been cycled on the meanwell S-350-48.

-Mike

 

[Hyena]

Hey guys,
I got my latest power supply this arvo - a genuine meanwell s350 48.
I plugged it in and it got super hot after 5 minutes of pumping out 9.5a and the fan never kicked in. It sat there silently the whole time until at last the fan finally kicked in, only to have the fuse blow a second later. I replaced the fuse and now it keeps blowing. This was straight out of the box too. I just have no luck with these SMPSs... Hopefully sure will replace it for me under their "100% DOA warranty"
One thing worth mentioning is it had 3 very tall coppery looking shunts - about 1/2 inch high. They look easy to cut/bypass . Well, if the thing wasn't stone dead

Here's a picture for reference, to add to the growing collection in this thread.

 

[mwkeefer]

same as mine, mod one resistor to limit to 350w and no prob, also hte grrn thermistors, bend the one closer to the flyback coil.... it will pick up the heat better!

And yea we can trick it to be on always but I want a better fan and better stability when setting voltage and current, working on it now.

Dont give up on that supply yet, I have one from my first days testing that does the same thing.... I'll figure it out, meanwhile email meanwell.


-Mike

 

[Hyena]

Dont give up on that supply yet, I have one from my first days testing that does the same thing.... I'll figure it out, meanwhile email meanwell.

If you can work it out that'd be great. At the moment its another pile of spare parts. I pulled the board out and unlike last time the under side looks brand new. No signs top or bottom of anything having let the magic smoke out. I emailed sure, they asked for a picture. I'm not sure what they hope to see other than a blackened fuse (which I told them) but I also video'd it blowing fuses with a flash of light so hopefully that'll be enough for them to replace it

 

[wookey]

This is all good stuff. I need to charge a 72V, 20Ah (24S2P) pack, which actually means ~87V max to fill the cells. So far as I can tell that's too much for a single supply so it looks like 2 in series would do the trick, but I'm struggling to work out from this thread weht I should be getting 2x36V (72V nominal), or 2 48V (comfortably up to 96V) or one of each (and then try to get them both putting out the same current?). I'm happy to go swapping components about but my experience of messing with SMPS in the past has generally been that they blow up if you don't mess carefully...

Presumably getting the 48V units guarantees higher-voltage caps, but lower-voltage ones have 'spare' shunts?

Any advice greatly received - I have my finger hovering over the buy-it-now button. (36V ones not available on ebay uk currently - I have to go via .au or .com - not that it really matters).

Cheers