I have a Harrison 6475 10kW 100 Volt 100 Amp power supply that uses basically the same circuit. They called it a SCR regulator. I bought mine as a parts not working unit and found mine had a bad SCR. The HP Journal really helped with troubleshooting it. I'm sure you will have this filmed and repaired by now but for anyone else interested in SCR regulator circuits the HP Journal Vol 13 #11 dated July 1962 which is available on the HP archive website is hands down the best technical document on the circuit.

There is a big mistake you made while talking about SCR preregulator. This is called phase-angle cut preregulation, and works inverted from the thing you draw. There are two reasons for it. First, and the obvious one, is the usage of scrs. Scrs are the components that once triggered stay on until the current that passes through SCR drops below something called minimum latching current (in this case we can call it 'zero' since it is typically some small value in order of mAs or tens of mAs). So the scrs are triggered some time during the half wave via the pulse from preregulator controll circuit and the zero crossing of the mains turns them off. The second and more subtle reason is that the transformer has inductance and if you start your cycle at the zero-cross moment and try to cut the current off while it is at its peak, you'll get quite a high voltage spike since you inadvertently made a boost converter. 😂 Also in this circuit there is a hidden gem. Imagine trying to recharge the high-value capacitor through scr, and from a tightly coupled transformer that acts almost as an ideal voltage source... Since the capacitor will do its best to resist the change of voltage, you'll get a short circuit with the current defined as peak voltage (at the moment of the scr turn on) devided by thermal resistance of the transformer secondary plus capacitor ESR (this is simplified, ther is also the mirrored impedance from the transformer primary, and the scr on impedance, but there is no need to go that deep). The resulting current will be huge, resulting in heavy transformer heating, capacitor heating and detterioration... So the engineers, in order to reduce the mentioned current, put the air-gapped choke between the SCR bridge output and the capacitor. And that was done like this from the time of thyratron phase angle cut rectifiers... But engineers from Harison lab got the more elegant idea! They purposefully made the transformer with loose coupling so that the leakage inductance of the secondary winding would act as the current-limiting choke. Ideas like this made the companies like Harrison and later HP standard in high end electronics. Also got me into EE. 😄

I remember lusting after these PS's at work. HP made some wonderful art.

Wow, scr's in a bridge rectifier. One of those things that seems obvious once I know about it, but never occurred to me. I learned something! Cool.

I ran across a similar circuit in a lot of 4 bad Kikusui PAL-35 power supplies I bought on Fleabay years ago. I wasn't familiar with the circuit or technique at the time. No schematics or service information was available which always makes for a fun time trying to repair something. These things were packed inside, I didn't understand how they complicated such a simple thing. At some point I ran across the SCR and assumed they were using it to control the voltage going into the shunt regulator to keep waste heat down. I assumed it was a clever circuit that Kikusui came up with. They were bad in 2 of the 4 supplies which took out more stuff downstream. I recently came across another use of the SCR bridge circuit in a Bridgeport Series II milling machine. It uses it in the DC motor speed controller which drives a 90v DC motor that powers the x/y/knee power feeds.

I have a newer version of one of these (E3633A) do the same thing to me after I cleaned all the dust bunnies out. I'm pretty sure it spent it's previous life around continuously running brushed DC motors and a bench grinder. It was an intermittent problem that showed up after a couple minutes, going into Unreg with no purring. I fixed it by launching the parts cannon at every component in that preregulator circuit except the control IC. The design had changed to where the SCRs just selected between Low and High range and the phase control was being done with a single MOSFET on the rectified DC. (E3633A/3634A parts list/schematic has errors, BTW.)

I can't remember to say 100%, but I believe this is how Bob Carver uses the SCR in the " magnetic field " amplifiers

Projects spawning sub-projects is also known as Yak Shaving. “Why are you shaving the yak when you were supposed to be repairing the tractor?!”

That's a neat hack with the SCRs in the bridge rectifier. Wonder if you could use that to help manage the inrush current when the power supply is turned on and is charging that big-ass capacitor? I am curious how doing that switching on the secondary side of the transformer affects the AC line side power factor, noise, etc?

Actually, we use the part of the sine wave AFTER we fire the SCRs. If the filter capacitor voltage minus the output voltage is too low, then the control circuit will fire the SCRs earlier and deliver more charge to the filter capacitor. If the difference is too high, then the SCRs are fired later to deliver less charge.

This will be interesting. I have one of these supplies so I'm eager to learn.

I believe the word you were looking for is duty cycle. Almost identical to pwm, with the main difference being that pwm is specifically concerning signals. But technically pwm is still a specific type of duty cycle, and commonly used for duty cycle purposes due to the ease of using digital means to produce it. That's why the words are often treated pretty much as interchangeable nowadays, despite there being a slight distinction between the 2

I think the SCR gets turned on part-way through the cycle and turned off at the AC zero crossing point (if I remember my solid state electronics from years ago). More or less correct but different cycle.

very cool! I would have assumed it was an issue with the series pass transistors and associated components until you showed us that SCR circuit. Never seen that before. Looking forward to the next one!

I love a Purring Power Supply....

The 12 volt, 3,000 amp electroplating rectifiers we have at work function in a very similar manner to the pre-regulator on this power supply. I have the 10 volt, 10 amp version of this power supply. I had problems with the voltage and current adjusting wire wound potentiometers themselves. Sometimes the voltage would just randomly drop to nothing, sometimes it would fluctuate back and forth. The current output would do the same at times. I thought it had a more serious internal problem, but the wiper was distorted and was not making contact with the wire wound element correctly. After disassembling and bending things around a little bit on both of those pots, I haven't had any problems since then.

An interesting control circuit. I wonder how it affects the noise.

If I remember correctly, Harrison Labs started using that SCR pre-regulator circuit since the mid 1960's. I have never measured the actual operation but, I assume it limits the series pass transistor headroom to about 2 volts, substantially reducing dissipation. Since this power supply is a series of control loops wrapped around more control loops, studying the theory of operation first would be quite helpful. Although as a possible time saver, you might want to check for dried out electrolytics first. Your symptoms suggest a dying SCR but, that is just a guess. One final comment: The standard rule of thumb for linear power supply sizing for the rectifier bulk filter cap is 3,000 uF per amp so, the 33,000 uF computer grade cap sounds about right.

That big blue cap I believe the brand name is The Compensator, and it is rated at 1.21 jiggawatts.

A slightly different implementation of the "SCR bridge rectifier" with discrete voltage levels can be found in the more modern HP/Agilent E3632A (0-15V/7A, 0-30V/4A; service manual available).

Interesting.. i need to learn this scr rectification system... This would reduce the heat dissipation in my circuit...

Power supplies: the offensive linemen of electronics.

I need couple 22000mfd caps for my Lambda PS. The second output is taking twice as long to supply 40v and if I short the leads at 1 amp it won't hold voltage above 10v. I think one is bad.

I have one of these that blew that giant 33k cap. I have not figured out why it did that. Can a physically smaller cap with the same specs be substituted? Those caps are difficult to find.

interesting... is this the same PS as your video #766?

inject power at the cap from another supply to see if your assumption is correct. (?)

I agree. Wow. When I worked for a living, many years ago, I worked in hospitals and worked on medical electronic equipment. Our goal was to make sure that HP was always purchased because they don't break - mostly. When opening up a HP schematic it was almost a religious experience. Equipment from other manufacturers, well it was like it was designed and built by engineers that had barely passed engineering school. So when you said that your HP power supply was broke, I went into denial. I got better and have been delighting in your going over the schematic. I was going to get on my knees the wife was in the same room.

That is some audiophile's wet dream cap 😅 I'd love to see a series on high current linear PSUs like that working principles.

I think you haven’t considered ohms law, your power supply is working ok except possibly for the humming sound. If you put a short across the power supply terminals the load resistance is zero ohms, so you should always expect zero volts and any amount of amps. I believe this is what you got. The only way to get some voltage is to supply 10’s of amps- that short + clip leads is probably about .02 ohms, put 100 amps into it and you will get some voltage before you melt the insulation off the connection leads. I f you now put a (high wattage) 10ohm resistor across the power supply terminals and set the voltage to 10V, you will only get exactly 1A (10V/10ohms) if you set the current limit above 1A. If you double the voltage, the current will double. I believe this is what you got. The power supply hum might be normal at high current, or it might indicate some loose transformer windings.

All projects are fun...and side projects are the most fun, or not!

You short the output and 'complain' it measures 0 volt. 0V is what I would expect for a short. With the resistor connected the Amps are dictated by Ohm's law, that is exactly what happens. I guess the resistor is 10R given the 1A measured at 10V and 2A at 20V