Summary and context:

Several years ago I bought a linear power supply from Pyramid, model PS-3KX.

  1. I would like to understand in detail how the overload protection really works, as said by Pyramid manufacturer (red box in the Specification part).
  2. It seems the PS-3KX is capable of 2 A in continuous service (blue underline) but triggers the protection above 3 A (maximum current - purple line).
  3. An AC component of the secondary transformer seems to be "detected" by the components marked in red (D5, R3&R4, D6&D7, C3, and Q1), which triggers Q2 through C4 and D8 - marked in purple.
  4. Severe current overload reducing the stabilized Output voltage, or a short-circuit in the output, would lower the polarization voltage in Q1 (green makings) that would raise the voltage at C4, conducting Q2 and shutting down output through Q3 and Q4.
  5. Voltage feedback regulation is obtained by the circuitry marked in blue, but I have concerns if the ripple detection (red and purple circuitries) would not worsen voltage stabilization or even induce a higher ripple.

After studying the replies and simulation from recently accepted answer and the results of my "Update#1" (Waveforms in a PS transformer), I will comment with UP#2, certain paragraphs in my original post, to avoid misguiding future readers in EESE here.

TLDR: The simulated original circuit does not work as expected and the manufacturer's performance figures for protection could not be confirmed.

I made a collage of the PS-3KX I have, to share the general view of the internals:

enter image description here

== Resuming Original Post ==

The user manual, which can be obtained here, shows:

Specification Summary, with some highlights for my PS model.

enter image description here

Circuit Diagram, with circuitry features color-marked:

enter image description here

Doubts and questions:

Based on the presented information, and assuming the actual PS-3KX unit is not available here for further testing, please help me to solve the following doubts and questions.
If I understood something inadequate or simply wrong or missed any hidden feature, please share your views:

{1} Short-circuit & Overload Protection #1 - Seems to be effective only if the output voltage is lower than 4.1 V. What happens if a higher current is required at higher voltages?
{1A} For example: Let's say, a high-powered LED array starts to conduct at 8 V and requires 4 A @ 12 V? The voltage could be high enough to not trigger green and Q1 circuitry.
{1B} Did I miss something for this circuitry feature?

{2} Overload of the Transformer - Protection #2 - An apparent ingenious circuitry, marked in red and purple, seems to detect higher frequency (harmonics) if the transformer core is overloaded.
{2A} These overload, magnetization, and higher frequency are the most intriguing features for me. I would like to understand properly/completely how this happens.
UP#2: Scoped waveforms in Update#1 and some simulation results (answer), made my mind that the red and purple-marked circuit, as it is, is NOT reliable overload protection. The overload protection circuit could be improved using Antonio51's answer and associated discussions as a guide and starting point.

{3} Effectiveness of the Transformer Protection #2 - This feature could be a feature we could use on an unknown power-supply transformer. Curiously, I have not seen this kind of circuit being used elsewhere.
UP#2: As mentioned above, the existing protection is not reliable/effective to protect either the electronics or any overload on the transformer.
{3A} Is it too good to be true? (UP#2: It is "Not True"...). What would be the hidden trade-offs?
{3B} Would it be problematic, eventually lowering the output voltage even when the current is within the continuous 2 A limit?
Please observe the ripple at this current is specified as 150 mV RMS (in the blue box).
{3C} It is specified the overload protection is +10% ~ 15% with auto-reset (see red box in spec).
I assume this would be above 3 A = maximum stated current.
How precise could this be as Pyramid's stated, if there is no current being directly measured?

{4} Semiconductor Protection - How safe is this schematic to protect the series-pass transistor and rectifier diodes from overcurrent?
{4A} Transformer Protection - How safe is it to avoid severe overload of transformer?
UP#2: Original circuit does not provide the specified protection.
A R_shunt and 1 or 2 BJTs working as a classic current limiting circuit, or working as hiccup protection (see answer) are necessary additions for PS protection.

I'm sure my description was a bit repetitive (or recursive); sorry about that, but see it as a shared brainstorming to understand the circuit and its potential problems.

UPDATE#1: Waveform in Linear Power Supply Transformer

After seeing simulation work made by Antonio51 (Thanks!) I decided to check how a Transformer inside a smaller power supply I made 40 years ago (CV = 1-25 V; CC = 0.13-1 A) would behave from:

  • No/Minimum Loading = 76.5 Vpp @ 0 A & 71.0 Vpp @ 0.13 A.
  • Average Loading = 58.5 Vpp @ 0.5 A
  • Light Overloading = 46.5 Vpp @ 0.94 A.
  • Severe Overloading = 11.5 Vpp @ 2~3 A.

The photo collage shows the tests and waveforms.

enter image description here

This test was made with a different linear PS, made with a repurposed car slot transformer (12+12 VAC, 2 A) ~ about the same VA magnitude of the PS-3KX. So I believe the transformer in the PS-3KX would behave similarly.
I did not see any unexpected spikes, just the sinusoid AC wave being flattened when able to charge the capacitor and be drained under constant current.
Even when the transformer is severely overloaded, the waveform is qualitatively similar to regular loading.

Conclusions (by UP#1 & UP#2):
Unless it is missed something, as my hobbyist Scope does not do FFT (it would be great to see), I did not find any transformer behavior in real life (not as model/simulation) that would justify the manufacturer's protection method to be effective or safe.

If any of you has a different view or had another experience, please let us know!

  • \$\begingroup\$ @jonk This may be a site that scraps stack exchange though… \$\endgroup\$ Apr 29, 2022 at 4:35
  • \$\begingroup\$ Humm ... Interesting thing ... to be simulated ... because no "apparent" measure of current load. \$\endgroup\$
    – Antonio51
    Apr 29, 2022 at 7:04
  • \$\begingroup\$ @jonk My "Norton" does not like this site link (in RED)! \$\endgroup\$
    – Antonio51
    Apr 29, 2022 at 9:33
  • \$\begingroup\$ @Antonio51 I'll remove it then. I think kuba for it right. Thanks \$\endgroup\$
    – jonk
    Apr 29, 2022 at 10:10
  • \$\begingroup\$ Little problem ... Schematic shows Q3=2N1384 is a BJT-NPN ... I found a datasheet that shows 2N1384 is a Germanium BJT-PNP ??? web-bcs.com/transistor/tc/2n/2N1384.php \$\endgroup\$
    – Antonio51
    Apr 30, 2022 at 7:54

1 Answer 1


I would like to understand in detail how the "OVER-LOAD PROTECTION" really works ...

Simulated that "thing", EE&O in the schematic ...
Some names BJT transistors changed, don't have searched for "equivalent".
Transformer parameters are "guessed", with no "non-linearities" ...

Made with free software microcap v12. Dropbox link1. Dropbox link2.

enter image description here

Done in three cases

  • normal use max 3 A, @ output = 13.8 V (it seems a power supply for "transceivers")

enter image description here

  • with "light" overload

enter image description here

  • with "big" overload (power supply "pumps").

enter image description here

Here, changed 2 BJT to BC109C, added windings resistors ... Quasi same behavior.

enter image description here

Conlusion: "Overload protection" is very a bit "strange" and "not efficient".

UPDATE: I have added some overload "protection" ... EE&O.
Sorry, the post is becoming "long" :-)

enter image description here

enter image description here

enter image description here

enter image description here

As asked by OP, I add the picture for severe overload.
The peak current is high, but the time of the first pulse is short, (about 500us), probably the "discharge time" of C6 (confirmed \$timePulse = fullLoad * C6 * 3 \$).
I guess the "modified protection" circuit will be ok. (in any case, need to check all voltages, currents, powers ...).

enter image description here

UPDATE 2: Changed R22 to 10 kOhm. Power supply functional after overloading disappears.
More interesting ...

enter image description here

  • 2
    \$\begingroup\$ the method is not immediately obvious. To me, it looks like it is measuring the ripple on the output and summing that with a signal from the transformer. If there is enough ripple then Q4 starts turning on. C2 slows down the action. Interesting for an old-time circuit but we have much better methods these days. \$\endgroup\$
    – Kartman
    Apr 29, 2022 at 10:36
  • \$\begingroup\$ @Antonio51, Thanks for your quick work. Interesting simulation results. Pyramid does/did linear PS for transceivers, but mine @2~3A might be more for powering Car Radios (at low volumes). I saw, but not not follow, why the output is being 100% cut at 1R overload and oscillating On-Off during this “short-circuit”. You also said transformer is ideal, so no magnetic-saturation effects are considered in the simulation and I didn’t find if/where Winding Resistance was simulated; in case it was not, how a 500 milliOhm (total) in the secondary would affect the results? \$\endgroup\$
    – EJE
    Apr 29, 2022 at 11:56
  • \$\begingroup\$ Yes. Not obvious. happily, 2N3055 (my Q1) has a high current capacity (15 A, if I am not wrong). I will search for a means to implement another type of protection. \$\endgroup\$
    – Antonio51
    Apr 29, 2022 at 11:58
  • 1
    \$\begingroup\$ @EJE Until 3 A, the power supply seems to be ok. Why it oscillates when big overload, don't know yet (searching...). No magnetic effect, not taken into account (perhaps later, model to be found and included). Winding resistance, ok, I will insert easily, but I think it will not change the behavior (to be simulated, ok). \$\endgroup\$
    – Antonio51
    Apr 29, 2022 at 12:04
  • \$\begingroup\$ @Kartman I had a similar feeling, but the DC output values are considered to polarize Q4 and just the AC component from transformer drives Q4; it made me remember of an RF detection of AM radio. I’m a Mechanical Engineer by profession, but study&deal&love Electronics for 40+ years. As this design resulted in a real line of products with a good name, it should Not a hobbyist quirk in their design and may consider hidden features to “somehow” protect the PS, but simulation is not catching that (yet). I agree with you that other methods could be better, but that is not the challenge we face here. \$\endgroup\$
    – EJE
    Apr 29, 2022 at 12:08

This site is temporarily in read-only mode and not accepting new answers.

Not the answer you're looking for? Browse other questions tagged .