Zeners not regulating -- I am repairing a crappy 13.8v/20a power supply (Chinese)

Thread Starter

Martyk

Joined Jul 23, 2009
35
Hello all

I have always tested zeners with a DC source and a simple series resistor-substitution box. No problem.
I am now checking some old 1N746A zeners (3.3v,½w) and they are not regulating at all; just acting like a resistor.
Are they all defective? Is there a better way? Should I be doing something different?

Thanks
Marty
 

BobTPH

Joined Jun 5, 2013
8,804
Low voltage Zeners have a very soft knee, so what you are seeing might be normal. Check the current at which the 3.3V is specified and choose a resistor to give that current, snd see if it looks better.

Bob
 
1N746 datasheet test current is 20mA for 3.3V but not sure how much current you are using.
What the part does at 5mA or 10mA I couldn't find any decent curves showing how sharp the knee is. For low V zeners it does get softer as BobTPH mentioned.
 

crutschow

Joined Mar 14, 2008
34,280
What voltage are you measuring at their 20mA rated current?

Below is the LTspice simulation of a1N746 Zener.
As you can see, it has a soft breakdown curve.

For a sharp breakdown point you can go to a shunt voltage reference IC.

1650474215322.png
 

Audioguru again

Joined Oct 21, 2019
6,671
A zener diode regulates well when its impedance is low because the impedance is in series with it.
A 1N746A max impedance is 28 ohms, then its voltage changes 0.5V when its current changes a little. A 6.8V 1N754A max impedance is only 5 ohms so its voltage regulation is almost 6 times better.

Also, the voltage produce by a 3.3V zener diode reduces when it heats. the voltage produced by a 5V to 7v zener diode does not change much by temperature change.

A zener diode is antique. Why not use a modern voltage regulator IC instead?
 

Thread Starter

Martyk

Joined Jul 23, 2009
35
Low voltage Zeners have a very soft knee, so what you are seeing might be normal. Check the current at which the 3.3V is specified and choose a resistor to give that current, snd see if it looks better.
Thanks for the quick reply Bob.
The specs say test current= 20mA. So I tried at 20mA, and yes, I get ~3.3v
But what if I didn't have the specs? I would not be possible to determine the rated voltage.
Marty
 

MrChips

Joined Oct 2, 2009
30,704
As already pointed out, 3.3V zener diodes make lousy voltage regulators.
If you need 3.3V power supply, use a three-terminal linear voltage regulator.
 

dl324

Joined Mar 30, 2015
16,839
But what if I didn't have the specs? I would not be possible to determine the rated voltage.
Vishay uses a test current of 20mA for all of it's zeners below 13V:
1650480045485.png

If you can't find a datasheet, which is impossible to believe, you can choose an operating current near the maximum power rating and reduce current from there.
 

Thread Starter

Martyk

Joined Jul 23, 2009
35
Thanks to everyone.

I've now tested all the semi's and found nothing wrong. Has anyone ever worked on one of these?

schematic - fix.jpg
 

dl324

Joined Mar 30, 2015
16,839
Has anyone ever worked on one of these?
It's a pretty basic design. Why didn't you just troubleshoot the traditional way instead of removing components to test them?

The 4 2N3055's in parallel looks like a bad design choice to me. They should have put ballast resistors on the emitters to reduce the likelihood of one of them hogging current and causing a cascading failure.
 
OP I have worked with these power supplies, they are very cheaply constructed.
The electrolytics all failed, I think because they (filter caps) are all rated 25V and I measured 22V on them, so not much margin for high mains voltage.

There's two traps when fixing these PSU's - the overcurrent protection/mains loss transistor V30 is actually a latch, once it trips because the output voltage is too low or no AC mains, you need to pull power and give it a cold boot to reset it. It makes troubleshooting extra tough and took me a while to figure what was going on, because any problems with pass transistors etc. giving low output voltage, the latch trips and you can get lost.
edit: To be clear, transistor V30 is normally on all the time but if there is loss of AC or DC output voltage sags low (overload) the transistor turns off and the circuit latches so V30 stays off until you pull power and the smaller filter cap C5 discharges.
Second trap is the main filter capacitors stay charged because the mains loss latch trip shuts off the pass transistors! So expect fully charged filter caps 4,700uFx4 which can be a big bang if you don't manually discharge with an i.e. 1k resistor before going in.

I ditched the zener/transistor regulation V31 and put in a TL431 for much better regulation especially over temperature. (also modded the compensation caps around that part for better response time).
 
Last edited:

Thread Starter

Martyk

Joined Jul 23, 2009
35
I ditched the zener/transistor regulation V31 and put in a TL431 for much better regulation especially over temperature. (also modded the compensation caps around that part for better response time).
Thanks for this.
I am getting a full ON condition.
Can you post a sketch of your regulator?
Tks
 
Last edited:
"full on" sounds like shorted pass transistor(s) or driver transistors usually, or V31 is not turning on.
Note that normally, protection latch transistor V30 must be on for the supply to be enabled.

Samlex PSU mods Rev1.PNG
For the mod I just pulled transistor V31, put in a TL431 like this.
 
Did you get yours fixed? The 3.6V zener V20 is part of the AC fail latch.

These 13.8V linear power supplies from china use the same circuit, sometimes small variations. It's widely copied and scaled up from 3A to 50A even just my adding pass transistors etc.
Sold under Samlex, Alan, PowerTech, Jaycar, Pyramid private labels as well. Popular for Ham shacks because they have no EMI compared to newer models using SMPS.

Common problems are failed transistors because the short-circuit protection is not so great and the power transistors are pushed hard. The electrolytic capacitors all leaked in one I have. Overall, the parts are pretty cheap.

As far as improving the regulation, my mod is to remove the one transistor V31 and replace with TL431, jumper out the 12V zener and change the two resistors on either side of the trimpot.
It's the same circuit as the ALAN/CTE International K105-K405 schematic you posted. On mine the 3.6V zener is not used, instead two 1N4148's to make a 2.4V part.
Another mod is to gut the board and install a 0-30V 2mA-3A power supply board (kit) and use the old power transistors/heatsink/rectifier/caps if you wanted to make this into a bench adjustable power supply. The kit board is about $8.

Again, for repair the protection/ac fail latch transistor V30 is normally on all the time, and latches off if AC mains is absent or DC output voltage dips too low (overload). You can temporarily disable it lifting one leg of diode V29, for repair troubleshooting.
Other than this, the circuit is straightforward. If you are still fixing it, maybe start a new repair thread.
 

MrAl

Joined Jun 17, 2014
11,389
Hello all

I have always tested zeners with a DC source and a simple series resistor-substitution box. No problem.
I am now checking some old 1N746A zeners (3.3v,½w) and they are not regulating at all; just acting like a resistor.
Are they all defective? Is there a better way? Should I be doing something different?

Thanks
Marty
Hello Marty,

As others have said, low voltage zeners do not regular very well so you might see some unusual behavior.
In fact, zeners of any voltage usually arent that good in modern design unless it's a cheapie.
Better is a voltage reference diode you can look up on the web. They are easy to use and you can aks someone here for help using one.

I personally hate zeners unless it is really really necessary to use one.

What you could do with your current zener(s) is run a current through one like you usually do and plot the voltage. then report the voltages and currents here and we can take a look at just how bad they are.
Maybe 100ua, 1ma, 2ma, 5ma, 10ma, 20ma would be good, and you'd get a different voltage for each of those levels.
Once we can see the voltages we can tell if it is bad or not or just plain lousy.
 
Top