Simple PSU for a few valves

[Mark Hennessy]

Yes indeed. Grundig employed thyristors in line output stages for a while - thankfully I never needed to fix them! And who could forget Sony's GCS???

Interestingly, the drive to raise switching frequency might well reduce the size of the energy-storing components, but "switching losses" become a problem. Given that a switch, be it bipolar, MOS-FET or whatever, doesn't switch instantaneously, the current flowing during switch time leads to heat dissipation in the switch. This switching time is fixed by the device and surrounding conditions, so as you drive the frequency up, the switching time becomes a significant part of the overall period. Luckily, developments in silicon over the last 3 decades have reduced switching time, enabling higher switching frequencies. On a computer motherboard, the DC-DC converters for the CPU run at upwards of 2MHz, which is pretty impressive!

All of the above also explains why a CPU runs hotter at higher clock speeds

The above is "switching loss"; the other source of inefficiency in a switching supply is called "conduction loss". Sources of this include diodes (which obviously drop 0.7 volts, or, more likely, 0.3 if it's a Schottky), MOS-FETs (which have a defined value of resistance when conducting called RDS-ON), and the resistance of the wire in the inductors. These losses are constant with switching frequency.

Any practical design has to juggle these parameters. For example, a designer of a mains switched-mode supply might be tempted to go for a MOS-FET switch because they tend to be faster than bi-polar transistors. But, high voltage MOS-FETs have relatively high values of RDS-ON, so the conduction losses might outweigh the savings made in switching losses. On the other hand, at very low voltages, you'd be mad to use anything else because bi-polar transistors drop ~0.3V when saturated, and it you're turning 12V into 1.5V @20A for a modern CPU, that's about 0.75W lost in the switch. And for that reason, designers replace the diode with another MOS-FET (that's called "synchronous rectification).

Switching power supplies are probably the most underrated thing around. Reckon you'd have iPhones without the magic of switched-mode power supplies? Think about all the parts in there that need different supply rails, and then the nightmare of producing them, then the energy losses involved with linear regulation. The battery would be massive and only last 2 hours :-/

Sorry to go off-topic

Mark

[tiv48]

i cant forgive sony ther gate controled swich witch needed to be turned both on and off making a cetain 18" colour set one of the most unreliable sony ever made
rob t

[Pamphonica]

My two-pennyworth.....
I used to try to use switched-mode supplies in my amateur radio days. And very nice chunky things they were too. Made by people like Advance. Chunky but very compact and cool running. Ideal for chucking out 20A at 12V without breaking sweat.

But (and there is always a but) switch-mode supplies always radiate a lot of "hash" no matter how well the designers filtered every input and output. A homemade thyristor supply will produce variable voltages, but the switching speed is quite fast, so you get a sharp "edge" to the waveform. As you probably know, anything looking "sharp" on a scope means lots and lots of harmonics. In this case well up to 30MHz and beyond.

So if you have any radios on the bench, or even nearby, your thyristor PSU (which will not have industrial-grade filtering) may well be a rather effective broadband transmitter as well!

I'm afraid I'm a luddite - I like nice linear PSUs that glow red hot but are very electrically quiet.

Two pennyworth now over...
all the best
Jeremy

[Pamphonica]

By the way, if you run a shortwave receiver anywhere near PCs, fax machines and lots of other devices with switch-mode psus you can hear their hash all over all the bands.
Jeremy

[glowinganode]

My two-pennyworth.....

A homemade thyristor supply will produce variable voltages, but the switching speed is quite fast, so you get a sharp "edge" to the waveform.

I guess some of the benefits of thyristor control include high slew rate and lower drive power requirements.
Ok till it commits suicide and goes short circuit.
Actually not really ok as the chopped, half-wave rectification makes a right old mess of the mains supply waveform.
Rob.

[Shaman]

Nice usefull power supply, great around the workshop, well done Al.........regards....Alan.

[beam me up schottky]

Thyristor pre-reg variable DC lab PSU's were quite popular in the 60's, 70's and early 80's. I have a few examples, the Farnell H series was quite notable for this technique. I have a Farnell H30-100, 0-30 volts at upto 100 amps !! This is a real monster, even with the pre-reg which limits dissipation in the pass reg by some margin, it still weighs 90KG !! The TX is massive, around 280mm square. Also have the H60-25, 0-60 volt at upto 25 amps, about 40KG from memory. They are superbly built units. Also have a Solartron from the 60's which is SCR pre-reg, thats quite early for such a technique.
I bought the Farnells when I was doing some experimental work on high performance car fuel pumps. May seem strange but when a turbo engine is pushing numbers like 750 BHP it needs alot of fuel !! Pumps capable of maintaining upto 60 PSI and 600-800 ltrs per hour they need alot of current. Initially I thought the H60-25 would be fine but when I saw pumps routinely drawing 20 amps I knew I needed something much bigger. Which is where the H30-100 came in, that coped ! The facility to vary and maintain high supply currents within fine limits allowed me to plot flow -V- pressure graphs for prospective pumps and also test for thermal stability.
I've seen some pumps draw 70 amps, which at a nominal 14 volts is about 1.3 HP !!
Noise has been cited as a problem with these SCR pre-reg supplies but I've scoped the output of the H series units on full load and there is less than most true linears !!

[Mark Hennessy]

Wow! The units I saw were small bench units - 10 amps max. Can only imagine what 90kg of PSU looks like

Similarly, I can only imagine what those fuel pumps looked like! I've got a mental image of a pump with a starter motor hanging out of one end... Thanks for posting an impressive set of numbers and firing my imagination

Mark

[beam me up schottky]

They are 19" rack size, about 60 cm deep and about 6U high. Internally they are very nice indeed with avionic style forced air cooling, using 4 x 120mm fans in the case of the H30-100. Packed very tightly though, I wouldn't want to replace a pass transistor very often ! They are massively paralleled up for reliability though, 20 x 2N3055 in the case of the H30-100 !

You are quite close in your visualisation of the fuel pumps, but the highest performance pumps are all billet alloy and cnc machined beauties, engineering porn ! Standard road car type pumps are immersed in the fuel tank for cooling and noise suppression, but these high performance pumps are external mount so run quite hot. Shockingly expensive too !

[Ian E G7OLT]

The reason that thyristor power supplies were dropped from domestic TV, was down to x million TV power supplies all turning on at roughly the same point in the +ve half of the mains cycle. This caused the power companies problems and damaged generating equipment by introducing asymmetry into the supply waveform. Effectively, DC was injected back into the grid, causing saturation of transformer cores and overheating. By the end of the 1970s, TV manufacturers had dropped thyristor power supplies in favour of switch-mode.

Switch-mode power supplies that are fed from a full-wave bridge and use higher than mains switching frequencies that are not locked to 50Hz, don't cause the same problems. They're also more efficient and can provide mains isolation, important with so much ancillary equipment connected to the TV nowadays.