Zener diode clipper

[Al (astral highway)]

Since the Olympic Park was built our line voltage often goes up from 230V to as high as yesterday's 246. This is causing me a headache with an unregulated power supply powered by a 70 v step down xformer, since the full wave rectified voltage exceeds the electrolytic capacitor's 100v rating.

My remedy is to put a Zener diode clipper with two 68v rated devices back to back, to clip the over Over voltage (68v plus the 0.7v forward v drop of the series device is close enough to 70v). Only problem is that this clipper can't have any series resistance as the current drawn is always high and also variable. So I need hefty zeners! Any comments or advice before I proceed, please?

Thanks!

[Miguel Lopez]

I don't think the zener ideas would be OK. You need to replace the capacitor with another one rated at a higher voltage or build a series voltage regulator at 70V.

[G8HQP Dave]

Zener clippers are not the right solution. They will necessarily need to be hefty and will waste power.

If the problem is 'mains too high' then the solution is a buck transformer.

If the problem is 'mains too variable' then the solution is stabilisation of the DC.

Note that a 70V AC secondary will routinely output almost 100V DC after rectification. Nothing dire happens to a 100V capacitor at 101V, which wasn't already happening to a smaller extent at 99V. The issue is capacitor life. The cap is already under-specified if you want reliability.

[Herald1360]

If you've got some leeway at the lower end of the input range, you could drop the the AC input voltage enough to keep the unreg output below 100V at max supply level.

A reasonably elegant way to do this is with a bucking transformer- find a TX with a secondary voltage equal to the volt drop you need and secondary current rating adequate for the full load mains current into your unreg psu. Connect this TX's secondary in series opposition with the psu ac input and its primary across the mains. The secondary volts are then subtracted from the mains volts to the psu.

Anything else gets into same complexity league as making your psu regulated.

[TrevorG3VLF]

You could consider a couple of zener diodes in series with the mains or the 70V.

The UK mains is 230V+10%-6% or 216.2V to 253V. You aint seen nuttin yet !

[Al (astral highway)]

Originally Posted by G8HQP Dave

Nothing dire happens to a 100V capacitor at 101V, which wasn't already happening to a smaller extent at 99V. The issue is capacitor life. The cap is already under-specified if you want reliability.

Hi, the circuit was designed for 100VDC off the 70V Xformer. Yesterday I had 109VDC as a result of the line overvoltage. As you know, the next jump up in size of electrolytic is physically much bigger and I can't easily mod the circuit board now.

Originally Posted by Herald1360

If you've got some leeway at the lower end of the input range, you could drop the the AC input voltage enough to keep the unreg output below 100V at max supply level.

Hmmm... I get you, and thanks, that would be an ideal solution. However, and it's a big however, I already have this set up, so don't want any more transformers involved!

1) 1KVA step-down xformer rated 230 down to 115 VAC This obviously doesn't sag under the average load (!) and yesterday its output voltage was 123V

This is a hefty thing indeed with a short circuit rating of 100,000 Amps. Not a typo. It is obviously not part of the circuit board!

2) A 115 to 70 V xformer on the board.

Because my 1KVA transformer has has so much more to give before it is anything close to being under duress, the clipper still seems the way to go unless there is an electronic engineering objection? I mean, the zeners would be fitted inside this transformer's housing - there's plenty of room for them not to affect the windings.

They can dissipate as much heat as they like - the thing is massive and enclosed in a metal box that would at a pinch make a good heatsink.

I similarly can't retrofit a regulated power supply on the board, except something like a simple zener. Same trouble applies - can't put a series resistor to protect it as the current drawn by the circuit is too variable.

Originally Posted by TrevorG3VLF

The UK mains is 230V+10%-6% or 216.2V to 253V

Indeed... but ours was rock-solid 230 for years, really well regulated. It hardly ever sagged and it never went over 235. Siemens built a huge new distribution station just 1 mile away and even on a cold afternoon at the weekend (yesterday) it was way up. Never seen it at 246 before!

[Al (astral highway)]

I should add to my post #7 that I've missed out some of my train of thought...

I originally posted about putting the zener clipper at the output of the 70V transformer, but you'll see I've now come to the conclusion that it would be better of inside the 230:115 kit.

However, I'm still not sure whether, say, 120V zeners exist in a high enough wattage to deal with this application! Any leads, please?

Also, what's the effect on the transformer of clipping the sinuisoidal waveform? Does it cause losses?

[glowinganode]

Could you remove a few turns off the secondary of the 230:115 transformer?
I think the idea of hard clipping with zeners is potentially a bad idea and could generate substantial mains borne interference.
The design should allow for variations in mains voltage, 246V is well within limits.
All sounds a bit marginal to me.
Rob.
p.s the zener voltage would need to be around 155 V not 120 V, as it is the peak voltage you are clipping.

[McMurdo]

If we take into account that commercial DC power supplies will use (say) a 35V rated capacitor for 12Vdc, a 63V capacitor for 24Vdc and a 400V capacitor for 340Vdc, the unpalatable solution is to upgrade the caps or downgrade the transformer secondary, whichever way you look at it! I know that's not the answer you want but sailing so close to the wind is going to end up with vented tops....

[Al (astral highway)]

glowinganode:

Yes. The design was wrong and didn't allow for these variations. Lesson learned! I'll definitely blow those electrolytics with +10v over on a long-term basis.

McMurdo: in similar vein, yep, I'll replace the caps before they explode.

Thanks folks.

[Herald1360]

I suspect your problem with any sort of zener limiter will be power dissipation- not from the viewpoint of your transformers etc, but the zeners themselves. The only realistic place to use them would be in series between the rectifier and the capacitor. You don't say what your dc load is or whether it's constant, variable or intermittent so it's impossible to figure out how much dissipation would be involved. As a starting point, if you need to drop around 10V, the zener will have to cope with at least 10*Idc watts continuous and be capable of coping with the repetitive peak currents as the capacitor is topped up each cycle. You won't find a zener good for much more than 75W at Tcase=25C and that will require decent heatsinking to cope with even half of that in reality.

Have you any access to remove secondary turns or add primary turns to either transformer?

Bear in mind that a bucking transformer would only need be about 100VA for a 15V or so drop in the input ac at 1kVA loading- it could be set up as a simple "bulge in the mains lead" type of installation.

[Radio Wrangler]

The zener diode will be taking current in order to create voltage drop in the effective source impedance of your supply, which is going to be your mains supply impedance, plus the copper resistance of your transformer primary divided by your turns ratio squared plus the copper resistance of your transformer secondary…

Say mains impedance is 0.1 ohms, transformer primary 2 ohms, secondary 0.5 ohms (just guesses but you do seem to be looking for 10,000uF capacitors so I suspect things are a bit beefy) so doing the sums gives 0.68 Ohms. Call it 1 Ohm to allow for the resistance of the bridge rectifier.

Now, you want to drop 10V. So the zener is going to be taking 10 Amps to pull the supply and transformer down by 10v. The Zener sees 70V at 10A which is 700 watts.

Essentially, you're trying to handle a capacitor working voltage issue by trying to pull down the mains voltage, taking as much power as is needed to do it. It could be expensive as well as making a bad mains waveform for everything else in your place.

David

[Billy T]

Originally Posted by astral highway

I originally posted about putting the zener clipper at the output of the 70V transformer, but you'll see I've now come to the conclusion that it would be better of inside the 230:115 kit.

Sounds like being too close to the point of supply is going to be problematic for quite some time to come, but there is no point in throwing power-consuming loss technology at something a simple as a high mains voltage. Should future loads or supply management reduce the over-voltages, you then have to undo your mods.

The simplest answer is to upgrade the voltage rating of the cap in your power supply, or you could build in a voltage regulator (still a bit lossy) but I prefer a switchable buck/boost supply that can handle both the highs and the lows.

Any power transformer with 5 volt and 6.3 volt RMS windings will give you a range of correction voltages, in your case -5 volts rms,-6.3 volts rms, -11.3 volts rms (-7.1/-9v/-16vpk) and being filament windings they will have ample load capacity and a simple reversing switch will give you buck or boost.It is a long time ago now, but one of my electronics lecturers had a simple design mantra: beware of creeping elegance!

Cheers

Billy

[Al (astral highway)]

Herald1360 - thank you. Removing turns is too messy and the xformer is beautifully made and hard to source here, so not my favoured way forwards. Your buck xformer idea is great...thanks

RadioWrangler: thanks for doing the detailed sums. That's too absurd a power dissipation to contemplate! Cheers David for the lucid analysis...

BillyT... Again, thanks. Looks like buck xformer is the way forwards!

[maninashed]

Might be easier to get back to basics. What exactly is this unregulated psu supplying, and what is the current requirement? Once the guys on here have some idea of what you are trying to do, its far easier to provide a solution.
Bill

[G8HQP Dave]

When something is wrong the best way to fix it is to correct the thing which is wrong. Anything else is likely to lead to more problems.

The thing which is wrong is that you have underspecified the cap voltage. You have to change that. If that is really impossible then a buck transformer is the next best option. Adding zeners is simply daft, as a back-of-envelope calculation (see post 14) shows.

[mhennessy]

Although I agree with the advice so far, another possible solution might be an SCR pre-regulator - the type that is sometimes used with high-voltage bench power supplies to reduce power dissipation in the main regulating transistors.

If doing this, you could take the opportunity to drop the DC a little bit more - if the rest of the system allows - to improve the safety margin on that existing capacitor...

Mark

[Al (astral highway)]

Originally Posted by G8HQP Dave

Adding zeners is simply daft

Steady on! That came across as a bit sharp. I know I've got into an interesting position, but please don't knock me for thinking creatively. I'm not claiming to be the big expert here, or I wouldn't have asked for advice.

Anyway, I have a huge, 1KVA rated transformer that is external to the power supply. Yes, of course I knew very that the Zeners would be dissipating a lot of power, but I thought it no big deal, if inelegant, to mod the AC source off-circuit.

My mistake - and it was plain ignorance - was not to know that when electrolytic capacitors are rated 100VDC working, it means exactly that, with no tolerance. I somehow assumed it was plus 20%. I only found out when it was too late and I investigated further. (ie, when I found the DC rectified voltage on board was 109 instead of 100, I had an 'ahah' moment and thought, steady on, better check that.).

Also, and critically, (see thread #1), for years - like 7 years - my supply has been stable at 230V nearly any time I've tested it. It sometimes went up to 235, sometimes sagged to 228, but that's about it.

Anyway, I think we've moved on somewhat as the consensus seems to be that a buck transformer is the way to go, or simply higher rated caps. Alternatively, I can put a linear regulator on board in this power stage at little expense and a tiny component count - but wasting some energy that is really needed in the circuit. Either the caps or buck xformer will cost a bit, but lesson learned.

[Tim]

This is why I always design assuming mains is 250 volts...............

[G8HQP Dave]

A clipper is a potential divider. You always have to think about both parts, the source and the clipping device, and the current which the latter will cause in the former. Forgetting this is not 'creative'.

In my post 16 I refer to post 14. That should now be post 12 - I assume two posts have been sent away to consider their position.