Switch mode supply mod

[steveo]

That's nice!.. you are doing the same exact opposite that i'm doing.. reducing voltages!..

Your PSu will probably be able to handle that mod if you dont exceet the total output power of 400W and is you respect the component rating safety referenced to that new voltage... it's normal and you must keep that safety range about voltage rating of componint like caps and diode

ex: if the caps was 63V rated for a 36V psu (max 40V) then that mean the ratio is 63/40=1.57.. so this ratio must be equal or above.

this is easier for a psu to have a voltage upgrade instead of current upgrade. but you must keep under the max 400W range cause the primary of the transformer is made to compensate in dutycycle and that mean that it,s the current that will vary to keep the CC at output..

let suppose that the transfo is 100% efficient.. so you would need 400W input power on the primary.. so the input voltage is nearly constant and will always be V x I and should keep the same current for a 400W output... but if you expect to increase the voltage output and to keep the current as before.. the total output power will change and... the input too.. but with the input voltage constant it's the input current that will increase to compensate and it is not built to handle that

So that mean you must decrease the output current if you increase the output voltage to stay within the 400W max range.

The transformer is made to handle 400W no more.



Just to avoid you to make things going wrong into a Method's mushroom of smoke ( i'm kidding you Methods )

Doc

Hey Doc,

I was actually thinking of this when i was modding the supply;

The psu, before modification could output 15amps at 35v (500w) instead of the rated 400w. This is psu is underrated i think .. i could reduce the current to be safe which was my intention .. but i need to test on a watts up meter with load to find out the exact max out-put .. i'll probably try tonight ...

Do you think by removing one of those (shunt wires) in front of the capasitors . it would reduce the current ... I think it will ... they say jumper wires .. but i believe there current wires IMO..

-steveo

[Doctorbass]

There is a chance that they are shunt resistor to sense the output current but i would be surprized

If they are paralleled try removing one and if the current limit drop then you'll have your answer.

I would say China made PSU are OVERrated if taking account of their MTBF life...

Doc

[monster]

if i wanted to use a switched mod psu as a DC converter (48v to 12) how could i mod it so that it accepted 48v and didn't refuse to switch on because it was below 110v?

i cant remember if this has been discussed? i don't think so.

[Doctorbass]

if i wanted to use a switched mod psu as a DC converter (48v to 12) how could i mod it so that it accepted 48v and didn't refuse to switch on because it was below 110v?

i cant remember if this has been discussed? i don't think so.

Some 100-240Vac PSU dot need to be modified.. and ... some need. I never tried to mod one to accept the lower voltage cause i never had to do that.

But I guess it's with the shutdown pin on the PWMcontroller chip (the IC that generate the pulse current to the mosfet) and then to the transfo.

Doc

[Hyena]

Do you think by removing one of those (shunt wires) in front of the capasitors . it would reduce the current ... I think it will ... they say jumper wires .. but i believe there current wires IMO..

Have you tried this yet steveo ?
I have ordered a similar power supply (THIS smaller 24v 6A version) but if they put out more current than rated I'll definitely want to keep an eye on mine, considering I'm going to use it to directly charge lipos (ala the methods method )
It hasnt arrived yet, so I don't know if it'll have the same shunt wires. Anyone got any other suggestions for limiting the current if it doesn't ?

[steveo]

Do you think by removing one of those (shunt wires) in front of the capasitors . it would reduce the current ... I think it will ... they say jumper wires .. but i believe there current wires IMO..

Have you tried this yet steveo ?
I have ordered a similar power supply (THIS smaller 24v 6A version) but if they put out more current than rated I'll definitely want to keep an eye on mine, considering I'm going to use it to directly charge lipos (ala the methods method )
It hasnt arrived yet, so I don't know if it'll have the same shunt wires. Anyone got any other suggestions for limiting the current if it doesn't ?

Hey

I will do my best to test this mod out if i can soon enough .. I'm in the midest of finishing assebly of 2 bms board

I'm sure it will work .. worst case i think you could use a shunt .. !

I need to first see how many amps the psu is pulling at 44v currently is where i have it set! .. i'm using 2 in series for 88.8v .. i think i don't have to mod anything .. but doc is right.. for longer life i should limit my current further to around 10amps! ..

-steveo

[rkosiorek]

if i wanted to use a switched mod psu as a DC converter (48v to 12) how could i mod it so that it accepted 48v and didn't refuse to switch on because it was below 110v?

i cant remember if this has been discussed? i don't think so.

Some 100-240Vac PSU dot need to be modified.. and ... some need. I never tried to mod one to accept the lower voltage cause i never had to do that.

But I guess it's with the shutdown pin on the PWMcontroller chip (the IC that generate the pulse current to the mosfet) and then to the transfo.

Doc

i don't think that it has, either. the problem in a nutshell is how do you provide enough intital voltage to power the pwm controller chip to get it going without a separate supply? most PWM chips operate at 8 to 30V (depends on the specific chip) some supplies have a resistor that feeds the voltage from the main supply rail rectified AC to a capacitor in parallel with the chip. the resistor limits the voltage ad current to a capacitor. the voltage across this capacitor provides the initial charge to power the chip through a couple of pulses. after that a tertiary output winding on the main transformer supplies the power needed for the chip to operate. if that is the case one would need to identify this resistor and replace it with a new value that will allow the capacitor to charge to a high enough voltage for this sequence to start.

maybe if someone had reverse engnneered a schematic for one of these that they could share? or even if we knew what the switching chip was? TL494 or SG3524 are most common for 16 pin PWM chips and UC3842 for 8 pin. knowing which chip would let you figure out where to start loking for this resistor.

rick

[Hyena]

My power supply arrived today, I set it to 25.1v, hooked up my 4 6S lipo packs in parrallel and away it went. It started off at 8 amps (rated at 6a) at around 23v then progressively got lower as the voltage went up. At 24v it was 5a, 2a at 24.5 and I turned it off when it was reading 700ma at 25.0v. Turning the voltage adjust pot up to around 26v while charging momentarily increased the current but obviously I can sit there tweaking that to keep the where I want it.

I was under the impression these power supplies were CCCV, but that's not terribly constant current

Any suggestions ? Is this no good for charging my lipos as intended ?

[GGoodrum]

My power supply arrived today, I set it to 25.1v, hooked up my 4 6S lipo packs in parrallel and away it went. It started off at 8 amps (rated at 6a) at around 23v then progressively got lower as the voltage went up. At 24v it was 5a, 2a at 24.5 and I turned it off when it was reading 700ma at 25.0v. Turning the voltage adjust pot up to around 26v while charging momentarily increased the current but obviously I can sit there tweaking that to keep the where I want it.

I was under the impression these power supplies were CCCV, but that's not terribly constant current

Any suggestions ? Is this no good for charging my lipos as intended ?

Actually, this is might be normal. I say might, because although it is normal for cells to slow down the amount of current it lets in, as it gets fuller, it depends on the rate. Normally, the cell voltage rises at a steady rate, as it becomes fuller. There is a point, however, where the rate of this voltage rise suddenly changes to a much faster rate. At that point, the cell is about 85% full. For LiFePO4 cells, this "knee" in the curve is about 3.65-3.70V. For LiPos, this number is 4.2V. What needs to happen to get the last 15%, or so, into the cell, is hold the voltage at this point (CV mode...), and let the current taper off. Once it is below about 25-50 mA, the cell is about as full as it is going to get.

With multi-cell packs, however, it gets more complicated. If all the cells are perfectly balanced, they will all reach this crossover point at the same time, so you can treat the pack like it is one big cell. Once the cells get out of balance a bit, you end up having a case where the cells start reaching the crossover point before the others, and the cell voltage starts to quickly rise, while it also starts limiting the current it lets in. The thing about current is that all the current has to go through all the cells, so if one cell is limiting the current, it is also limiting it for the rest of the cells, which still can accept a higher amount of current, but now can't get as much. The net result is that these cells don't end up with a full charge, which further increasesthe amount of imbalance.

Compounding this further is if you are simply connecting the main pack leads in parallel, and not the individual cells, because now you have cells that are not only not in balance with the other cells in its own pack, but between packs as well. The first thing I would do is check the voltage for each cell in each pack, and see how close they are. If there is more than about .05V difference between them, I'd use a balancer on the pack. Next, you can reduce the amount of potential imbalances during charging by connecting all the balance plugs on each pack together in parallel. This ends up giving you 6 blocks of 4 paralleled cells. When cells are put in parallel, they automatically equalize to the same voltage. Then when you charge, it will be like one big pack.

If you are not regularly using balancers, or a BMS board, you need to check the individual cell voltages for all your packs on a regular basis. If they get too far out-of-balance you can end up with one cell hitting the cutoff way before the rest, which can cause catestrophic results. If Lithium-Cobalt cells are allowed to get above about 4.3V, they will explode in a fireball, and will burn white-hot until all the cells in the pack have cooked off. That's the big difference between LiPos and LiFePO4-based cells. Tha latter can be overcharged to extremes (10V+...) and they will not explode, or catch fire. Most RC chargers these days either individually charge each cell, or use integrated balancers and check the voltage level for each cell during charging, shutting down if it detects an overvoltage condition. By simply using a 25V supply, you don't have any such protection, so you need to take extra precautions.

-- Gary

[fechter]

Yes, that behavior is normal. When the pack voltage equals the supply voltage, the current will go to zero eventually. It sounds like it is doing a fairly good job. It will only be in the constant current mode when the load (batteries) are trying to draw more than the current limit.

[Hyena]

Gary, thanks for the detailed post.
I fully intend to use a balancer, I'm getting a 10a charge through type that will hopefully limit any overcharging of individual cells. I'm just waiting on the JST connectors I've ordered to arrive, but was inpatient and wanted to test the PSU straight away
I turned it off at 25v so there was a lesser risk of it overcharging cells. I did measure them afterwards and they were all fairly well balanced at around 4.14 - 4.16v.

I was aware of the risks of charging in parallel with without a balancer and how one cell can potentially charge up quicker, but I didnt occur to me that that too would cause the current to drop. I doesnt sound like that's the case here anyway.

It will only be in the constant current mode when the load (batteries) are trying to draw more than the current limit.

Ahh, I didnt realise this! My little 50w RC charger I was using previously held close to it's max 2.2a charge rate most of the way through so I assumed that's what CC type chargers do, but I guess it was more a case that the batteries being fairly hefty were trying to draw more than that all the way.

Thanks for the info guys, I shall go forth and make this into a nice charger First job is to insulate / cover those bare exposed AC in terminals so I dont electrocute myself!


EDIT: I was just thinking about people who talk about charging lipos at 2C, how do they get on when the batteries limit the current like that ? Are they only getting the 2C rate for a few minutes before it drops down to a much lower rate ? My lipos wired as they are are the equilivent of a 16ah pack, yet were quickly drawing less than 1/2C

[Hyena]

After playing with the PSU some more and making it into a proper charger, I bought a 10a panel meter today and found that it did actually hold the current fairly constant, around 8.5a and only started to slowly fall as the battery neared 25v.
I wonder if it was my multimetre doing funny things when measuring it with that previously ?

[Tiberius]

I took a gamble and purchased one of these: http://cgi.ebay.com/350W-48V-7-2A-Switching-Power-Supply-CCTV-Radio_W0QQitemZ310040654028QQihZ021QQcategoryZ48632QQssPageNameZWDVWQQrdZ1QQcmdZViewItem to see if I could mod it to run at about 60V. The idea is to use it as an on-board charger for my motorcycle build, which is using a 48V, 40Ah Headway LiFePO4 pack.

Today I took the PSU apart and spent a couple of hours reverse engineering it. I believe that these supplies are available from several different suppliers, all with essentially the same design, usually with a number like "S-350-48". The supply is based around the TL494 PWM chip (see here http://focus.ti.com/lit/ds/symlink/tl494.pdf for details). It's easy enough to up the voltage range, by simply adding a 15k resistor across the 3k9 resistor adjacent to the voltage trim pot. The snag is that the over-voltage protection circuit kicks in at about 53 volts, so seriously limiting the ability to trim the output up.

The fix for this is easy. Underneath the main ferrite transformer there is a 53V zener diode, labelled ZD1. Simply unsolder one end of this and add a 10V zener in series. This ups the over-voltage protection to 63V, which is the same as the rating of the main output filter caps.

The supply is good for over 6 amps (as high as I've tested it) at 60V, with no problems. I reckon it will make a neat on-board charger. All I need to get on and build is a shunt regulator type charge management system, like the excellent one from Gary/Bob. I've already got one of Gary's LVC cut-off boards, to switch off the controller in the event of a cell going low.

I love it when a plan comes together...................

Jeremy

I can confirm these mods work. For anyone picking up this thread and working on these PSUs - there is some more info in this thread http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=6749&p=209094#p209094

Nick

[mwkeefer]

Hey all,

I too can confirm these mods (and some others work)... I have suggested removing shunts in the past to limit current but for the S-350 there is a much simpler way via the resistors in the back of the supply.

I tested this "mod" which Fechter was so kind as to help me work out, I measued R33 (next to the TL494) and it came up at 360 ohms (if memory serves me)... I ran a test charge on a 15S pack with supply voltage set to 62.5 and it screamed through the charge cycle (it was a 5AH pack) at just below 10A right up and over 60v... Once the cells began to hit 3.9-4.0v across the board the current begins to taper... the proper charge curve would be C/20 so if charging at 10A when you reach 500ma your done... (sort of.... it should float here for the CV portion but too complicated to explain here).

With my mod in place I set out again with the second half of my pack 15S1P, guess what... 7.7A or so to begin which I ran to it's natural conclusion of 350ma (actually I let it drop down to 0 but don't know if I should have).

So by cutting my R33 by about 30% I reduced the charge current by about 2 AMPS.

I still have some testing to do on this mod across the various S-350 models I have here to ensure consistency and I would like to implement adjustable current limiting since these can do it but... that will be tommorow as it's far too late tonight already.

-Mike

[Hyena]

(actually I let it drop down to 0 but don't know if I should have).

So by cutting my R33 by about 30% I reduced the charge current by about 2 AMPS.

I always let mine go to 0, largely because I put it on the charge then walk off and forget about it! My packs are still going strong after 150+ cycles but I guess in the long run it'll shorten their life in theory ?

The R33 mod is good to know! I might solder in 2 resistors with a switch on the second one so I can have fast and slow(er) charging modes. I was going to do the same thing with shunts but not having alot of current running through the switch is a good thing.

I've actually just made a charger for a mate who's only running a 10ah 10S pack and it's currently charging at 12A so it'll be good to pull it back to 1C or below (he's doesn't need it to charge any faster than 2 hrs anyway)

On that topic, does this look like the 350w meanwells you guys are getting ?
I suspect it's a knock off as it wasn't advertised specifically as a meanwell, and the only brand markings are MW and some chinese characters.
This is the board, pics are a bit crappy from my phone but you can make out the general layout. It seems different to my other 350w PSU and although it's actually a 24v unit I would have thought the boards would be near identical. It only has one shunt too and suprisingly puts out 12A at 42v making it more like 500w!
edit: searched another thread and seems its identical except for the v-adj pot and only 1 shunt seems the be the norm for the 48v models.

[mwkeefer]

Hyena,

Though that unit is a bit different than mine, they are fundamentally identical (parts are slightly moved about) and running at that power level you will kill the thing in short time.

Since they use a single shunt, you will need to measure R33 and R21 manually with DVM and adjust - post back with measurement and I'll help you calculate a proper replacement.

I've noted the voltage is difficult to adjust with voltage mod for 67v (though Im set for 62.2X) and the current does find a peak. My unit is limited and I pulled a 1.5ish (8A peak) charge to 62.25v and let the current drop to C/20 or .25A before cutting... many hours later the final result is a pack of 15S1P charged in 38/39m and within 9mv difference between cells across the entire pack. And yep my supply is running to high also I am working on recalibrating and gaining better voltage adjustments.... it's looking more and more like each batch of these supplies is built in a different manor (Ive seen quite a few of the same supply now) which means upgrades are going to require measurements to be certain of what is happening.

One final warning... the S-350-X series becomes obsolete end of this year, mean well has gone to the S-320 to obtain UL/AU listing / certification - I don't have a 320 yet but will soon.

-Mike

PS: You hook your packs to a meanwell and walk away... damn I can't decide if that's the ballsiest thing I've heard or the dumbest = )_

[Tiberius]

PS: You hook your packs to a meanwell and walk away... damn I can't decide if that's the ballsiest thing I've heard or the dumbest = )_

Yeah, why not? I've only ever seen these PSUs fail to 0 V, not to overvolt.

Hyena, what we keep finding is that these PSUs have a decent current limiting mode but its set higher than the continuous rating. So it has to be turned down or eventually they blow up. That's a fail to zero mode so if its few feet from the pack, it will be ok.

Nick

[liveforphysics]

It only becomes a problem in those events where something abnormal happens, like a cell puffs and has it's Ri triple, or one of the little aluminum pouch connection strips gets bumped and torn, so one cell group is 5Ah were the others are 10Ah or something goofy like that. The charger does it's job perfectly, but the pack still fails into a fireball.

I think at this point we can begin to see a nice trend in the use of E-bike LiPo. There are many folks using it now, and though we've had our share of shorts and bad cells and problems inherent to any battery type, I don't think we've seen a single non-intentional LiPo fireball yet on our forum. Come to think of it, even on the RC forums the horror stories they always bring up all seem to be from about 5 years ago or so, and pretty stale by now. Abnormal things still happen once in a while, but loss/damage has been very minimal in the last few years. Perhaps better education about LiPo? Perhaps better chargers? Perhaps better tools for easy cell monitering? Perhaps it's the newer cells with better manufactureing techniques and 1/8th the Ri of the older cell types?

[Hyena]

Since they use a single shunt, you will need to measure R33 and R21 manually with DVM and adjust - post back with measurement and I'll help you calculate a proper replacement.

Well, there's my first problem, R33 isnt populated on my board! Is yours up the top left corner ?
I may have to nibble away at the shunt if it's not straight forward...
I guess I'll pay the extra for a genuine meanwell next time for the ease of mod compatibility. I'd like to be able to get it up to 62.5 for my 15S pack (I currently charge in banks of 5S on a 24v meanwell)

PS: You hook your packs to a meanwell and walk away... damn I can't decide if that's the ballsiest thing I've heard or the dumbest = )_

Probably a bit of both I wouldn't do it if I still charged in my kitchen. I have a $5 timer on the power supply now, I just set it to 1.5hrs and walk off.

[mwkeefer]

The timer works - I thought you were just walkin away from it...

Can you take bettter pics of the PCB in your supply... I need to see the rear section where the 2 FETs are heatsinked - right near the 16pin dip chip 494CN. Does your camera have Macro?

This may be a later model than the ones I have, MeanWell engineers informed me that the later S-350's didn't have the same PCB (no SVR2 provision) is there an unpopulated SVR2 on your supply?

-Mike

[fechter]

Well, there's my first problem, R33 isnt populated on my board! Is yours up the top left corner ?
I may have to nibble away at the shunt if it's not straight forward...
I guess I'll pay the extra for a genuine meanwell next time for the ease of mod compatibility. I'd like to be able to get it up to 62.5 for my 15S pack (I currently charge in banks of 5S on a 24v meanwell)

If you can post a good picture of that area and the corresponding area on the bottom of the board, I might be able to figure it out.

[mwkeefer]

fechter's help is the only reason I stopped overthinking the whole thing and figured it out, without his assistance I would still probably be trying to figure this out... That said, there are variances between different models of supply... all mine are S-350-R18-VAI, yours look to be S-350-R20 or somthing ... I can make that out but I couldn't see the TL494CN chip or the current control feedback loop (that fetcher spotted and made quite simple for me to understand!).

Fechter - yours is leaving Monday, it should go Priority and you should have it by end of business Tuesday. I got a bit slammed end of week with other projects and the usual family chaos and just couldn't get it done.

That said... on the units I have and have sent you.... Ive figured out how to get variable current from it, in addressable modes (ref v control signal, 40-50ma) to switch between various current levels. I have tested (prelim) using 2.5A, 5A and 9A rates from my 48 units reconfigured for 62.5v - I figure we can go a bit lower but there is some minimum duty cycle and I haven't measured temps durring testing yet, just made sure it worked on the prototype.

I kept the switching inputs limited to shorting a single rate at a time, this test unit defaults at 5A and has 2 control leads and a single ref VCC (from TL494CN) which the control leads are tied to for selecting 2.5 or 9A modes.

Knowing you have a scope, perhaps you will have time to refine this and see how low she will go. I figure for proper balancing using most (including yours) balancers / BMS systems, we need trickle mode right?

It's on a protoboard right now and done from the hip so give me a day or two to refine it just a bit and I will post the schematic (probably by the time you have your unit). The other concern is using units in series and/or parallel... the current limit switching gets a little more complicated in these cases (I think) but... it's workable.

-Mike

[Hyena]

OK I dragged out the SLR, here's a pic of my unit


http://img44.imageshack.us/img44/6478/mws35048.jpg

Thanks guys

The timer works - I thought you were just walkin away from it...

Well the timer is set long enough that the current falls to 0, but atleast the PSU isnt powered up all night if I forget about it.
I've noticed it stays quite warm when just sitting there so I'm sure it'd reduce its life span.

This may be a later model than the ones I have, MeanWell engineers informed me that the later S-350's didn't have the same PCB (no SVR2 provision) is there an unpopulated SVR2 on your supply?

Maybe it's actually an earlier model? On the board it says S-350&201. 2005-09-19. I don't know if that second number is supposed to be the date, if so it could be 4 years old. Although it looks like there is no SVR2 which suggests it may be the newer model ? Who knows...

[GGoodrum]

Here's a shot of one of mine, which appear to be just like what Mike has:

S-350-24-01.jpg

(200.25 KiB) Downloaded 41 times

In the upper right, I've circled R33 in red. On my 24V/14A S-350, R33 is labeled as a 750 ohm resistor. I measured it at 735 ohms. With it, the max current it tries to pump out is a bit over 20A. I added an 820 ohm resistor (810 measured...) to this one, in parallel, which brought the total resistance down to about 385 ohms. This dropped the max current to about 10.4A

Looking at yours, if I had to guess, I'd say R37 is the equivalent to R33 on our boards. The reason I say this is because on mine at least, R33 is the only one that was less than 1k ohms. I can't tell if your R37 is 29 ohms, or 290 ohms.

-- Gary

[Tiberius]

My 36 V ones look like the picture Gary just posted, with the blank SVR2 position.

R33 is 1K2 on one of them and 680 on 2 of them.

Nick