DEER B350ATX PSU - tips on making it safer and "better"

[kotel studios]

Hi,

Since it looks like my excellent power PSU chances at living are now close to zero, I decided to give a shot at upgrading this DEER B350ATX PSU.
At first glance it looks like, well.... a shitty cheaped out DEER garbage (cause it is). Although it does have an LF 1250CD00 filter transformer (at least I think that's what it is) 😆
Tiny heatsinks, small transformators, single-pole ON/OFF switch. You name it.
Now since this one works, I'd like to know how can I make it safer and better (besides burning it with napal 😃). My plan is to make the switch double-pole for neutral and live, add bigger heatsinks and maybe transformators from my LC psu (same model).

[DrewPhillips]

First thing I'd consider would be to trade in the DEER for BEER. 😊. Barring that, a few items bear mentioning. Regarding your misgivings about the switch - ATX PSU don't actually need a "hard" switch in the AC line input. A great many ATX PSU have no hard switch. I did not see any AC line switch at all in your photograph, let alone the SPST you mentioned.

In ATX PSU, the +5VSB is always ON and the main regulator for the other voltages is switched on-off by grounding the green lead. If having some part of your PSU always ON bothers you, yes, you can install a DPST-connected hard switch in the PSU's case. Me? I'd just plug it into a switched power strip. Note that having line power removed (as by a hard switch) will also turn off the +5VSB and thus will disable your computer from being able to "wake up" on some external events, such as a packet coming in over a LAN, or getting the FAX that just came in over that old 14.4K modem connected to your landline☺☺

Cheap PSU? Well, at least the heatsinks are aligned with the airflow which will improve cooling efficiency to a marked degree in this application. Note that some metalworking would be required were you to intall bigger heatsinks.

Of greater concern would be the quality of the electrolytics (recap time?), and the effectiveness (or even presence) of supervisory circuits to automatically shut down the PSU in the event that any of its output voltages tries to go too high. Another item is whether this PSU has a real PG (Power Good) circuit driving that output, or did Deer just put in a pullup resistor and nothing else. Or beer intead of Deer.

[dmill89]

First thing I'd consider would be to trade in the DEER for BEER. 😊. Barring that, a few items bear mentioning. Regarding your misgivings about the switch - ATX PSU don't actually need a "hard" switch in the AC line input.

^This, beyond that as far as "upgrading" it goes the most viable option would be to rip out the "guts" and replace them with something better. If you just want a "fun" project and aren't concerned with it being financially viable (you could likely just buy a better PSU for what upgrading this one would cost), caps would be a good place to start as these had notoriously sketchy caps, and generally a self-oscillating flyback 5VSB rail that will go high and fry the motherboard if the "killer cap" on it fails (optionally you could potentially do a 5VSB mod like many do on the notorious Bestec ATX250E to fully correct this, but it is a bit "involved" to build out an almost entirely new circuit). After that replacing the 4-diodes for a bridge-rectifier with a "real" bridge rectifier would probably be next on the list followed by upgrading the switching transistors/MOSFETs and secondary Schottky (or possibly just fast-recovery on one of these) diodes. Even after all of that you'll probably still only have a ~250W (actual) PSU unless you also upgrade the transformers (definitely cost-prohibitive unless it is a true "hobby" project, and you don't care at all about cost).

[stj]

if it actually works then a full recap with something good like FR's and a new fan will do wonders.
i like to replace the switches with ones that have a neon in them - so i know if the cable is live if the system doesnt start.
you dont need double-pole because the iec connector is polarised - the switch will cut the live side.

[kotel studios]

...a self-oscillating flyback 5VSB rail that will go high and fry the motherboard if the "killer cap" on it fails..

What exact value this "killer cap" is. Those sketchy 1000uF 10v caps marked with an "F" logo show value within +/- 50% of case value (with 0ohm esr due to lcr t7 limitations). The smaller 47uF 50v cap had around 3.4ohm esr and the 1uF 50v cap near the 5vsb cable had around 2ohms esr.

[dmill89]

What exact value this "killer cap" is. Those sketchy 1000uF 10v caps marked with an "F" logo show value within +/- 50% of case value (with 0ohm esr due to lcr t7 limitations). The smaller 47uF 50v cap had around 3.4ohm esr and the 1uF 50v cap near the 5vsb cable had around 2ohms esr.

On the Bestec ATX250-12E it is C1 which is a 10uf 50v cap located near the base of the 5VSB transformer. While I'm not completely familiar with the design of your unit 22uF/50v and 47uF/50v are often used as the "killer" cap so given what you listed the 47uF/50v one would likely be the main suspect (though it would be wise to re-cap the whole thing possibly aside from the input caps).

[PeteS in CA]

If those blue ceramic caps near the top tight of your picture are DC-rated instead of safety agency-approved and rated for an AC voltage, remove them, and if you can replace them with proper Y-caps.

While the heatsinks look possibly OK for 250W-300W, the 330uF input 'lytics and output inductor probably limit the realistic output power to 150W-200W. 470uF or 560uF input 'lytics would be needed for input filtering adequate for 250W or 300W, but that output inductor would have to have a larger core and probably heavier gauge wire, and the output caps might not be adequate for those power levels.

[momaka]

Just fully recap the PSU - that is, replace all caps except the big 200V caps... and that's about as good as you can do in terms of making it safer and better.
More specifically about the recap, make sure to change all of the small caps. And if a rail has 2 cap spots, make sure those two cap spots are filled. Ideally, the 3.3V and 5V rails should get no less than 2x 2200 uF caps each. The 12V rail will benefit from having 2x 1200 uF caps... or if there is space for only one cap, use a 2200 or 3300 uF cap. And lastly, the 5VSB - use 2x 1000 uF caps minimum. 2x 1200 or 2x 1500 uF will be fine too.

Caps aside, check if the "PI" coils (rod inductors) are actually connected. FWIW, I had one late Deer / L&C PSU like this where the PI coils were installed but were bypassed underneath with the PCB, essentially making them completely useless. I'll see if I can dig out that post later for you. It's one of my older posts back from around 2010-2012 or so.

If those blue ceramic caps near the top tight of your picture are DC-rated instead of safety agency-approved and rated for an AC voltage, remove them, and if you can replace them with proper Y-caps.

While the heatsinks look possibly OK for 250W-300W, the 330uF input 'lytics and output inductor probably limit the realistic output power to 150W-200W. 470uF or 560uF input 'lytics would be needed for input filtering adequate for 250W or 300W, but that output inductor would have to have a larger core and probably heavier gauge wire, and the output caps might not be adequate for those power levels.

^ This 100%.

Simply said, there isn't much you can do to make this PSU "better". All of the parts were selected for a target range of around 150-200 Watts tops, with the primary side and main transformer probably good for 250W max. But that's not necessarily a bad thing. Most basic older PCs (low-power CPU and low-power GPU or no GPU at all except onboard) will happily run on 120-160 Watts. What matters more is really what type of system you want to use this PSU in and also what the PSU is actually capable of handling - i.e. old 5V-heavy design or newer 12V-based one. Some of these older PSUs are better-suited towards 5V-heavier designs, while others towards 12V-heavier designs... and some a little bit of both or little bit of neither. To get an answer to that, I usually look at the secondary side rectifiers. If the 5V rail has something relatively big (i.e. 30 Amp Schottky), then the PSU is likely to do well with a 5V-heavy old PC. And if the 12V rail has a 16 Amp or better rectifier, then it might be able to handle 12V-based PCs too. Of course, the output rectifiers only tell half the story. The other half is to just take the PSU and connect it to a junker motherboard, then run it and see how it handles itself. In particular, you want to look at the 5V and 12V rails to see if one sags while the other one is high under load, and vice versa. I have some PSUs that look like they would make nice units for 5V-heavy PCs, but actually do terrible in such PCs, with the 5V rail either sagging hard (under 4.8V) or the 12V rail going very high (over 12.6V).

Once you find in what type of system this PSU works better, you can then go and upgrade the rectifiers on that rail to increase the efficiency a little. For example, if you find the 5V and 12V rails very stable when using this on an old Athlon/XP system, then consider upgrading the 5V rail rectifier to a 30 Amp part (if it isn't using one already.) Conversely, if you find the PSU works well with a Pentium 4 PC and the 5V and 12V rails are more stable in such setup, then upgrade the 12V rail rectifier. Most of the time, these cheaper Deer / L&C PSUs come with a 10 Amp rectifier on the 3.3V rail, 16 or 20 Amp rectifier on the 5V rail, and 10 or 12 Amp rectifier on the 12V rail.
For the 3.3V rail, upgrade to 20 Amp rect. always. For the 5V rail, use either 20 or 30 Amp rectifier, depending on what system this will go in. And for the 12V rail, aim for minimum of 16 Amp rectifier, 20 preferred (and make sure it is rated for 60V minimum, but preferably 100V.)

if it actually works then a full recap with something good like FR's and a new fan will do wonders.

Agreed about the recap, although I wouldn't suggest to use Panasonic FR... well, not in post places. For the 5V and 12V rail, probably better off with something a little more conservative like Panasonic FC, Nichicon PW, PS, PM, PJ, and HE, United Chemicon LXY, LXV, LXZ, KY, and KYB, and Rubycon YXJ, YXS, and YXG.
As for Panasonic FR/FS/FM - those would be OK for the 3.3V rail and the 5VSB.

With good caps, these PSUs aren't that bad at all.
FWIW, I'm still using my recapped 250W Deer from (now close to) 15 years ago to power my Pentium 3 PC. It's an older model Deer, though, and built slightly better (with dual-pole ON-OFF switch, proper Y2 and X2 -safety rated caps, and etc.)

The fan you can leave. The stock fans in these PSUs are OK once oiled properly and also if the PSU has a fan controller to not keep the fan running at full 12V all the time - that's bad for the fan bearings and also makes too much noise. On my 250W Deer, I think I put something like 30 Ohms resistance between the fan's red (+) lead and 12V rail, since there wasn't a fan controller in it. IIRC, this keeps the fan running at around 8 or 8.5V, which is both quiet enough and at the same time fast enough to keep the PSU cool under full load.

i like to replace the switches with ones that have a neon in them - so i know if the cable is live if the system doesnt start.

Heh, good idea, but a new switch like that probably costs more than what this whole PSU costs.

you dont need double-pole because the iec connector is polarised - the switch will cut the live side.

In the UK and France, you're right - the power cord remains polarized, so the Line and Neutral are always on the same side. But for many other places in Europe that use the "standard" Schuko plug, this polarization of the cord is not maintained, so Line and Neutral can be switched depending on how one plugs the AC cord in the wall. This can make it slightly more dangerous when working on live electronics.

What exact value this "killer cap" is. Those sketchy 1000uF 10v caps marked with an "F" logo show value within +/- 50% of case value (with 0ohm esr due to lcr t7 limitations).

OK, anytime you have caps showing capacitance OVER 20% of their printed case value, GET RID OF THEM., regardless of what ESR they show.
Abnormally high capacitance = high internal leakage current inside the cap... which means the cap electrolyte is breaking down and attacking the foil layers... which really means, the cap will go bad soon.

For crappy cap brands, I don't even wait for them to go that high on the capacitance. If they go even close to 15%, I already know they are well on their way to breaking down.

That said, those "F" logo caps (some here call them "planet logo" caps, others "YC" caps) are utter garbage . You can leave the 200V primary ones, but everything else has to go. They really are nothing but trouble.

[kotel studios]

How does one identify fake Y2 caps? Mine look to be JNC, exact specs are unknown (checked them in a rush). Looking online there aren't any posts about fake JNC Y2 safety caps.

[stj]

if they say X or Y on them then o.k.
by "fake" they just mean ones that arent rated/marked as such

[kotel studios]

On mine there are weird symbols and under them there's
400~/X1 250~
250~/Y2 50C
I think these ones are proper Y/X caps right?

[stj]

yes

[kotel studios]

After some careful considerations I didn't replace the rail filter caps since it wasn't worth it at all. Instead I replaced all the smaller caps and the one bulging 5v rail cap. Sure, it's nor safe and better, but it's waaaay cheaper to get like an proper 300W FSP unit or even better to just buy an whole PC and steal the PSU. Although I might switch around the caps from my excellent power PSU if it turns out to be something irrepairable on it.

Oh and I crossed out "MAX 350W" and wrote 200W 🤣

[kotel studios]

Just to make me sleep better at night I added the 5v zener diodes (55C5V1) as protection on both the 5vsb and PS_ON. Saldy the voltage drop is too big. PS_ON is okay-ish (4.35v to 3.45v) but 5vsb isn't (5.05v to 4.55v). I think motherboards will work fine with the low 5vsb, but before I try it I'd like to ask the specialist here if the voltage is okay.

[stj]

use higher zeners - 5v6 atleast, maybe 6v2
read the datasheet, they start to conduct before the rated voltage.

[momaka]

Just to make me sleep better at night I added the 5v zener diodes (55C5V1) as protection on both the 5vsb and PS_ON. Saldy the voltage drop is too big. PS_ON is okay-ish (4.35v to 3.45v) but 5vsb isn't (5.05v to 4.55v). I think motherboards will work fine with the low 5vsb, but before I try it I'd like to ask the specialist here if the voltage is okay.

It appears you are using the Zener diode incorrectly. By the looks of the voltage drop, you have the 5VSB going through the Zener diode as a regular diode and then out to 5VSB - that, unfortunately, won't do anything for voltage protection. Instead, what you're doing is you're only making the circuit less efficient (by loosing an extra 0.5-0.7V through the diode... hence why your 5VSB dropped from 5.05V to 4.55.) Also, at high output current, your Zener diode will likely overheat and short out... so it really CANNOT work the way you have it wired. and 4.55V is out of spec. Lowest allowable for 5VSB is 4.75V (i.e. -5%).
One more thing: PS_ON is derived from the 5VSB, so there's no need to put a Zener diode on it. If the 5VSB is protected, then so will be the PS_ON line. Moreover, the pull-up resistor between 5VSB and PS_ON will limit current (and thus voltage) going to the PS_ON line even if the 5VSB is NOT protected. So absolutely no need to waste a Zener diode on the PS_ON line.
-----
Now, to use a Zener diode as a voltage protection device, you wire the Zener directly between the output and ground. That is, wire the CATHODE to 5VSB output and ANODE to GROUND. No resistors or anything.
STJ mentioned a very valid point, though - Zener diodes tend to conduct a little before their rated Zener voltage, so you run the risk of your Zener diode shorting out as soon as the 5VSB is powered up. So before placing the Zener diode as instructed above, first either LOWER the voltage of the 5VSB output (4.85-4.9V should be OK) -OR- use a 5.6V Zener diode, as STJ suggested.

If you go with option 1 to lower the 5VSB voltage, look for a resistor divider circuit connected between 5VSB and ground (essentially two resistors in series - one connected to 5VSB, the other to ground, and their "middle point" tied together and going to either a TL431 shunt regulator or the PWM controller IC, depending on what that is.) Once you find this resistor divider circuit, you can place a high-value resistance resistor across (in parallel) to the "upper" resistor (the one that connects between 5VSB and the "middle point" / TL431 Feedback pin) and that should lower the voltage accordingly. Here's an example:
Take a look at the 5VSB circuit schematic shown below, at the two 1-KOhm resistors all the way on the right of that schematic.
https://www.badcaps.net/filedata/fetch?id=2068236
If the 5VSB is exactly 5V, then the voltage at the middle point (the one going to the TL431 shunt regulator) will be half of that (2.5V), since the two 1-KOhm resistor are equal.
If the upper resistor resistance is increased over 1 KOhm (but not too much), then the 5VSB voltage will become higher.
And if the upper resistor resistance is decreased below 1 KOhm (but not too much), then the 5VSB voltage will become lower.
On the other hand, if the lower resistor resistance is increased over 1 KOhm (but not too much), then the 5VSB voltage will become lower.
And if the lower resistor resistance is decreased below 1 KOhm (but not too much), then the 5VSB voltage will become higher.
Now, the Feedback (FB) voltage going to the TL431 shunt regulator is always assumed to be 2.5V. Assuming we keep the lower resistor the same (1 KOhms), the current going through it will be 2.5 V / 1 KOhms = 2.5 mA. To adjust the 5VSB to 4.9V, we do the following calculation to find what the upper resistor (let's call it R_u) needs to be:
R_u = (4.9 - 2.5) / 2.5 = 2.4 / 2.5 = 0.96 KOhms = 960 Ohms.
Now, I'm not going to go explaining how to calculate the resistance of parallel resistors (you can a good explanation of that in many places online), but if you put a 27 KOhm resistor across (in parallel) to the upper 1-KOhm resistor, the total equivalent resistance will be about 964 Ohms (or 0.964 KOhms, if you like), which should make the 5VSB pretty close to 4.9V.

On mine there are weird symbols and under them there's
400~/X1 250~
250~/Y2 50C
I think these ones are proper Y/X caps right?

Yes, they are supposedly legit if they have those... but even more so if on the other side, they have the logos of various safety agencies, like UL, CSA, Nemko (N), Demko (D), Fimko (Fi), TUV, and etc.
As for what the above text means...

400~/X1 --> this cap can be used as X1-type (directly across Line and Neutral on the AC line) up to 400V AC lines.
250~/Y2 --> this cap can be used as Y2-type (either between Line and Ground or Neutral and Ground) up to 250V AC lines.

Oh and I crossed out "MAX 350W" and wrote 200W

Always a good idea.

Been there, done that too VV
https://www.badcaps.net/filedata/fetch?id=2086917

Probably also a good idea to label the maximum allowable currents on each rail, as limited by what the output rectifiers are capable of (or less).

[kotel studios]

Now, to use a Zener diode as a voltage protection device, you wire the Zener directly between the output and ground. That is, wire the CATHODE to 5VSB output and ANODE to GROUND. No resistors or anything.

This doesn't sit right with me.... If I will connect it from 5vsb output (right where the cable goes) to gnd then how can :
1. Cut the power from 5vsb? Do they just go short and put all the load on ground?
2. Where do I connect the 5vsb cable after I place the diode?

Sorry for sounding dumb but with my knowledge it doesn't make any sense.

[PeteS in CA]

This picture is of ceramic capacitors with quite a few safety agency logos:

[momaka]

This doesn't sit right with me.... If I will connect it from 5vsb output (right where the cable goes) to gnd then how can :
1. Cut the power from 5vsb? Do they just go short and put all the load on ground?

Yup, exactly.
5VSB goes over the Zener diode's voltage and the Zener diode shorts out, protecting everything else connected to the 5VSB output (motherboard and peripherals) at the expense of killing the 5VSB circuit (usually.)
Obviously, this is not the best way to have a protection... but at the end, if you had to choose between destroyed motherboard or destroyed PSU, I think the choice is usually the latter.

Now, you could make a better protection circuit by having the Zener diode trigger an SCR on the primary side and make the 5VSB just turn off... but that's a relatively complicated circuit. If going out of your way to make 5VSB safer, the choice is to usually replace the primary MOSFET and driving transistor with a PWM-FET IC, like DM311, TNYxxx, VIPer 22A, and etc.

2. Where do I connect the 5vsb cable after I place the diode?

5VSB stays connected where it is (i.e. you don't move it.)

[kotel studios]

Okay so I connected it like you've said and there's no smoke or anything inside. It is an 5v1 zener tho... Couldn't find any 5v6, but according to the datasheet for the 55c5v1 absolute max voltage it can do is 5.4v (which should be 5.2v average) and I get 5.07v with load on 5vsb. And yes I know about the voltage spike at the 5vsb.
I have also checked out the diodes and mosfets on 3.3v (16A)/5v (30A)/12v (10A) and they actually add up to max 330W! Weird how they went with such small primary caps and transformators tho.... I bet it's planned obsolescence by cheaping out on parts
Athough who would wanna run more than 200W on this sad piece of shit as it is now.