# unregulated power supply - unstable output or no output at all

#### dsc

##### Member
This might be a super obvious question to most over here, but I know nothing about power supplies, so please bear with as I try to explain what is going on.

I have one of these SPS705 68VDC power supplies running inside a rather simple stepper motor control box:

This powers up a DM856 stepper driver (also made by Leadshine) which is then connected to a SY57STH56-3008B Nema 23 stepper. There's also a 24VDC indicator sitting on the 68VDC input line into the DM856, this is wired via a zener diode to drop the voltage to not fry the 24VDC lamp. The main AC supply is connected to the PSU via a rotary isolator and powers up both the SPS705 as well as a servomotor controller (there's an AC filter mounted just after the isolator to stop any noise getting back into the line).

The workload on the above is very minimal as it's only powered up 2-3 times a day max and only for a few minutes. The Nema 23 stepper never really works under any heavy load either (indeed the 68VDC PSU is way overkill for what this stepper does), it only rotates a few steps left or right and that's it, the servo runs for a minute or so and the whole ring gets turned off. This used to work fine for the last 3 years but I've started noticing that the 24VDC "DC ON" indicator is dimming a bit, then dimming a bit more, then not turning on at all. I've measured the voltage output from the PSU and it's fluctuating between 20-40VDC, often dropping out almost all the way and barely reaching 20VDC. What is interesting is that if I leave the rig powered on for a longer period of time (>10-15min) the PSU eventually reaches high enough voltage levels that the indicator lights up and all works as before.

At first I thought the diodes in the rectifier are dead, but now seeing the way it springs back to life after 10-15min I'm not so sure anymore. I'm planning to get the PSU out of the rig chassis this weekend, but I'm not 100% sure what to check for. One possibility is that something inside the PSU simply warms up and then starts working properly? can it be a capacitor somewhere that takes ages to charge? looking at a diagram of an unregulated PSU there's not much inside, a transformer, a rectifier bridge and a cap on the output line to act as a filter. Another thing which I'm not 100% sure about is why it failed in the first place as there might be something in the entire set up that is causing premature failure. Would the AC being turned on / off often and the fact that the servomotor controller and the PSU are on the same line have something to do with it? could it be the zener diode sitting on the output of the PSU?

Any help is much appreciated as usual : )

Cheers,
T.

#### Diver300

##### Well-Known Member
It's often electrolytic capacitors that loose capacity or get larger ESR (Equivalent series resistance) over time. It's a bit like how rechargeable batteries go as they get older.

As a start you could change all the electrolytic capacitors. Would that be practical?

#### dsc

##### Member
Thanks for your reply Diver300! yes this would be doable, I just need to get the PSU out of the current enclosure which is a bit tricky due to the limited space and existing wiring, but it needs to come out anyway, so it just has to be done.

Cheers,
T.

#### KeepItSimpleStupid

##### Well-Known Member
#1. Switching and unregulated don't go together. The lack of complexity proves that.
#2. a 24 V Zener diode across the supply really stresses it. You need to use this circuit: https://www.electronics-tutorials.ws/diode/diode_7.html If the zener is wired in series to subtract some voltage that's OK. Series?
#3. A oscilloscope could easily tell what's wrong.
#4. Idec makes some nice 24 V LED replacement lamps and indicator holders that will work on AC and DC. All you would need is a series resistor and the lamp won;t burn out. Just get the same color lamp as the lens. What's the hole size? This is my recommendation.
#5. Put a properly sized ZNR across the capacitor to increase reliability.
#6. Thanks for the uk location. Could reccomend a parts source.
#7. Is it in a hot location? You generally have a choice of 85 or 100C.
#8. The cap likely has screw terminals. Make sure they are tight. Secure with Loctite 222. Thus getting warm could improve the connection. This, actually makes the most sense based on your symptoms.

Without a scope, try measuring the DC voltage across each diode in the bridge in both directions.

Parallel a capacitor and see if the DC voltage improves. Capacitors in series behave as 1/Ct=1/C1+1/C2+.....1/Cn, but the voltage would add.
So two 100 uF 50 V caps in series would be a 50 uF @ 100 V for a test.

Aside: KISS works instead of the entire handle. An @symbol is used to tag someone. e.g. dsc.

The "dsc" underlined in blue is typed as @dsc Like for Diver300 in a previous post.

Last edited:
dsc

#### dsc

##### Member
Thanks for your reply @KeepItSimpleStupid (in the last post I didn't use an "@" on purpose) ! Some further info below:

#1. Forgot to mention that this is the other servo controller connected to the AC line:

There's piggybacked AC cables running from the plug on this unit straight into the SPS705 DC PSU.

#2. Yup series, 43VDC zener (DIODE, ZENER, 43V, 5W; Diode Case Style: 1N5367BG ) on a small PCB with a lot of copper left over to dissipate heat.

#3. Don't have one, but might be able to borrow one from work.

#4. I'm using these indicators: https://uk.rs-online.com/web/p/indicators/7001928/, one AC for the AC supply and one 24VDC for the DC side, but as that's running on 68VDC I had to come up with a way to drop the voltage

#5. Are we talking on the PSU itself?

#7. Nope, room temp. It's in a tight spot inside a metal chassis, but it's only on for a minute or so 2-3 times a day or sometimes only once a day, so I doubt it really matters.

#8. Will pop the cover off and have a look / measure tomorrow.

The SPS705 is way overkill for this application, I'm planning on switching to a 48VDC as I don't have to push the stepper too hard (or at all really), but those PSUs aren't exactly cheap, so I'd like to have a go at fixing this if I can.

Cheers,
T.

#### KeepItSimpleStupid

##### Well-Known Member
Take a photo of the power supply inside and get the info off the rectifier and capacitor.
A ZNR across the caps will go a long way. I suspect it's a screw terminal capacitor and I do suspect the screws are loose. Locktite 222 is what i use. It's a low srength threadlocker liquid.

You might be looking at a capacitor similar to: https://uk.rs-online.com/web/p/aluminium-capacitors/8526566/

I used a "rule of thumb" of 1000 uF/Amp and upped it significantly, Many times these are called "computer grade" and you can find them in the surplus market.
I bought four 9600 uF at 85 V computer grade caps in the 1980's for like $3.00 USD each surplus for an amp I built. Today we're looking at$65.00 USD each. https://www.mouser.com/ProductDetail/Cornell-Dubilier-CDE/CGR962T075W4C?qs=nEDK3UM2Sgi3n9/7Qo5L4w== See here http://www.surplussales.com/Capacitors/Electrolytics/10000uF-300000uF.html for some idea on surplus prices.

#### KMoffett

##### Well-Known Member
#1. Switching and unregulated don't go together. The lack of complexity proves that.
....
Just as a side note. I gut wall warts for use as embedded power supplies in projects. I've run into quite a few SMPS wall warts that only have a pair of diodes or a bridge and a capacitor on the output side with no feedback.

Ken

Last edited:

#### dsc

##### Member
Here's some photos:

There doesn't seem to be anything clearly wrong with the board, none of the caps are bulging and nothing seems scorched / fried, the underside of the PCB looks clean as well. The only thing I've noticed is that the two left transistors (IRFB11N50A) sitting by the enclosure wall (they are pushed against that wall using that thin aluminium bar and three bolts) have slightly discoloured heatsinks, but just noticed that they are a different type to the two thyristors on the right (MCR310-10G), so it's probably nothing. I've not got anything to hook up to this bare PSU to act as a load, so for now I could only measure the output without any load and it's jumping between 62 and 64VDC.

EDIT: forgot to add, as you can see none of the caps are screwed down, the big ones are soldered on and glued in place (none of the glue is falling apart, it all seems intact).

T.

#### Western

##### Member
I'm inclined to suggest it's likely to be one of the smaller electros. This is a classic case for using for an esr meter ... but an alternative is to warm the supply up till the output is ok ... then switch off ... then use freezer spray on the caps, one at a time until the fault reoccurs at switch on.

#### rjenkinsgb

##### Well-Known Member
From experience I'd try swapping the high voltage input smoothing cap first, the 330uF 400V one in the first photo.

From quite a few varied switch mode PSUs I've repaired that were in a semi-working state still, the input caps have been the single commonest problem.

#### Western

##### Member
From experience I'd try swapping the high voltage input smoothing cap first, the 330uF 400V one in the first photo.
Haha ... my experience is the exact opposite (no disrespect intended). May have something to do with me being on the opposite side of the globe perhaps.

Anyway, I'd still put my money on one of the smaller electros ... only one I can see in the photos is a 220uF 25v.

From quite a few varied switch mode PSUs I've repaired that were in a semi-working state still, the input caps have been the single commonest problem.
I've changed plenty of those as well, but rarely for the symptons mentioned ... and only in maybe 5% of smps repairs

... still I've been wrong before ... so will be interesting to hear the outcome.

#### KeepItSimpleStupid

##### Well-Known Member
Well, definately a switching power supply.

Last Picture.
I see 5 possible solder joints. Starting at the "s" in Leadshine (top 1/3 of the last pic) moving to the right and down.

Right top corner, capacitor pads. To the left of the square pad.

You might want to examine closer and touch up the joints.

They are generous with the amount of ZNR's in the supply.

#### dsc

##### Member
Sorry for reviving an old thread, but I've only just got around to fixing this (only 1.5 years later).

In the end I just bought a new PSU, this time a lower voltage 42VDC as nothing in the circuit actually needed 68V.

The main question I have now is whether there's anything I can improve in terms of wiring to increase the life of the components inside the box. At the moment this is how things are wired in:

Main AC supply -> rotary isolator -> AC filter -> 68VDC PSU and piggy backed into a 400W servo drive EL5-D0400 [http://www.leadshine.com/UploadFile/Down/LSC_High_Voltage_Servos.pdf]

(same) Main AC supply -> switch -> 5VDC PSU for uC and OLED display

The drive drives a 400W servo (see the above PDF for details). On the AC line there's also a small DIY 5VDC supply for electronics, this has a small PCB mounted transformer and powers up a uC and an OLED display (*more about that later). As mentioned before there's also a 24VDC indicator sitting on the 68VDC input line into the DM856 (again piggy back from the DM856 terminals), this is wired via a zener diode to drop the voltage to not fry the 24VDC lamp.

So the 68VDC will now become 42VDC and I'm putting an additional switch for the AC going into the 42VDC PSU, so instead of the above this will now look like this:

Main AC supply -> rotary isolator -> AC filter -> servo drive EL5-D0400
(same) Main AC supply -> switch -> 42VDC PSU + 24VDC indicator via a zener
(same) Main AC supply -> switch -> 5VDC PSU for uC and OLED display

The idea here is to separate the servo drive and 42VDC as I was thinking that perhaps when I switch both on / off at the same time the servo drive might be doing some wreckage on the line inside the box. Also the OLED display constantly drops off I have a feeling this is due to the 5VDC dropping down to low levels which aren't low enough to kill the uC, but enough to "scramble" the OLED (a quick power cycle of the 5VDC via a small switch resets everything back to normal).

I can fairly easily add other components, I just don't know what else might be useful or actually if any of the above is needed at all

Thanks in advance for any tips!