Side Projects

[Mike H]

Finally put together a LTspice schematic for the 'Mystery Circuit'.

Woah I totally wasn't expecting a switching regulator, but that's what it is.

The bottom line is, it maintains a Voltage on the end of line smoothing cap, 2,500 µF (C5, will be replaced with 2,200 µF), that is only a few Volts more than the output Voltage, and which is present across the 2N457 (Q6).

Meanwhile, usptream, the other power transistor 2N4348 (Q5), is switching on and off at 2 kHz (in simulation, anyway) to keep the C5 topped up. All this makes for minimum power dissipation for both output transistors. How cool is that?

Schematic below, altho it's somewhat confusing because it's 'busy', there's a lot going on ... I had to include the conventional linear regulator part (ideal op-amp again) and its series pass transistor, or it wouldn't work. The choke is a critical component too.
-

Marconi DS 50_2 regulator-7.png (126.2 KiB) Viewed 17625 times

-

[Nick]

Not sure (could be wrong) that that doesn't look like a switching reg to me (at least not what I would expect today for that name, a buck or boost circuit), I would have expected more inductors. Maybe its the sim making the differential amp x1 (I think) unstable. Its running it seems open circuit. Maybe try a small 22pf or so cap around its output and -ve pin to see if it calms down. Unless Q1, 2, 3 and 4 are working as a multi-vibrator generating the 2kHz maybe. Maybe the first part is generating the 2kHz, and the the diff amp is working as a comparator varying the mark space ratio of the charging current. That might make sense, and would be cunning. As you say, reduce the dissipation of Q6.

Bit of thinking as I type there.

[Mike H]

OK, tried what you suggested, no difference.

Yes the left part is like a multivibrator, Q1 & Q2 form a Schmitt trigger. Q3 inverts, so its output is the inverse of Q1 input. It then oscillates up and down as C5 charges and discharges. But I agree the sim may be a wee bit ropey.

The diff amp & Q6 are the linear regulator. No oscillations around there.


Green, Q1 base, relative to Vs2 (-9V).
Red, Q1 collector current.
-

Switching-2.png (16.64 KiB) Viewed 17603 times

-


Red, output or Bk10.
Blue, C5 charge.
Green, Q1 base - all relative to Vs2.
-

Switching-3.png (10.23 KiB) Viewed 17603 times

-

The discharge time of C5 is much shorter hence the sawtooth. Not sure if that's correct or not, but if I get it working, we may be able to find out!

[Mike H]

Switch Reg Waveforms


Well the above is all cobblers - it's all change - remember 'mystery diode' 1S420..?
-

1S420 diode.jpg (19.06 KiB) Viewed 17540 times

-

We thought it might be for capacitor discharge at start-up and/or shutdown. Actually it's a critical component of the switching regulator. I plugged it into the schematic, and it radically changed its behaviour...

-

Switching-1.png (17.33 KiB) Viewed 17540 times

-
Violet - (flat line at the bottom) - regulated output to the black terminal (wire Bk10), relative to red terminal.
Green - point '(A)', which is between the two power transistors, and connects to C5 (2,200 µF) negative via a 1 Ω resistor.
Red - 2N4348 (Q5) collector current.

The transistor current is a squarewave, and it's 11 Hz (in the simulation anyway).

NOW THEN - all the times I've used it, it always made like a soft buzzing noise, which sounded just like a loose stack lamination vibrating. Increase the current output, it gets louder. But the frequency is lower than 50 Hz, it's not a 'hum'. I now believe this is the sound of the floating regulator oscillating, altho it's the choke making the actual mechanical noise.

I'm glad I've found that out because I was always curious about that.

[Mike H]

Switching-2.png (13.99 KiB) Viewed 17539 times

-
Red - Q5 (2N4348) collector current.
Green - current through 'mystery diode' (D2) 1S420.

-

Switching-3.png (14.81 KiB) Viewed 17539 times

-
And lastly, current through the choke.


I need a simpler version of the block diagram... don't go away...

[Mike H]

That took longer than I thought. (And I made and ate tea in the middle)

Hopefully this better shows the relationship between D1 (1S420), the choke (L4) and C4.
-

DS 50_2 Simplified Block Diagram.png (83.87 KiB) Viewed 17518 times

[Nick]

Aha, that's making far more sense. Remember my comment earlier about the absence of a inductor in a switching regulator

That's now a buck regulator https://en.wikipedia.org/wiki/Buck_converter shape followed I would guess by a more precise and lower noise linear regulator. The first part creating a high efficiency voltage a few volts over the target voltage, and its probably designed to track a few volts over the output voltage, then the second stage can be a linear reg knowing the dissipation in the output stage will be as low as possible because the voltage drop will be minimized.

[Mike H]

Right on daddy-o. Neat innit? I am super impressed.

The few volts over the target is 5, simply by how the Schmitt trigger Voltage divider resistors are configured. Not shown, but the linear reg power transistor has a pair of 9V Zeners in parallel, in case the overhead exceeds 9V. (?)

[Mike H]

Onto actual practical things...

Does anyone remember these?
-

Marconi-015.jpg (91.36 KiB) Viewed 17182 times

-
(Clue: DTB freebies one Owston meet.)

[andrew Ivimey]

I've got a fine selection of Tropical Fish caps by Mullard. Who knows where from!?

[Mike H]

If you were there, you might have got given some too...


It's coming along slowly - made awkward by not wanting to bend the wires on the board about too much (3 have broken off already!). I was able to attach one of the end brackets to the choke frame with a screw and nut, so that it was upright, that was super useful because I could then easily access both sides of the board without have to move it at all. Only issues after that was not enough light (as in, illuminating the right places, but a small LED torch helped), and my short-term memory.

Before and after photos - first, can caps and rectifier...
-

Marconi-006.jpg (157.54 KiB) Viewed 17169 times

-
-

Marconi-016.jpg (176.8 KiB) Viewed 17169 times

-
New rectifier is a SIL GBJ 2510.

[Mike H]

I was up last last night finishing the board. I decided to change all the small resistors too. A. they're carbon composit (Yay!) but B. they're 10% (boo!). (And C., they'rev old.) Sure nuff, some of the values were waaay off - to be fair a couple were spot on, but on average, most were adrift by greater or lesser degrees. I was late because I had just 6 left to change and thought why stop now. All the caps were replaced also...

Before and after...
-

Marconi DS 50_2 board-2b.jpg (164.87 KiB) Viewed 17157 times

-
Second blue 'lytic from the top is the one that caused all the damage. Top left - two metal rectifiers which I'm sure can't be original, the board holes spacing is wrong. A prior repair! ... the next pair of little ones were properly toasted by the shorting blue cap, and went zero resistance and that did for the transformer winding.

-

Marconi-017b.jpg (160.1 KiB) Viewed 17157 times

[andrew Ivimey]

Those boards remind me of Quad 303s ( a bit) A bugger to replace components as the pcb tracks would lift at the earliest opportunity. I gave up on my second 303 and bought already made up boards new output transistors and power supply board. I still have the second one and very fine it is too. I gave the first to a surgeon who passed it on to his son at 'Uni'. Apparently it refuses to conk out.

[andrew Ivimey]

Pasolini.... I've just remembered this word

[andrew Ivimey]

Oh ffs, I'm a fan of the films of Pasolini - well worth a yhoughtful watch bit I did NOT type that and just how cultured is my phone's predictive text. I typed...

PAXOLIN!!!!