DIY Mains Regulation - Oops, I forgot to factor in heat dissipation

[Deke609]

My mains VAC dropped as low as 115VAC over the summer, causing the filament voltage regulator to dropout and hum in my Beepre with EML 300Bs (1.3A filament). And the low mains voltage must certainly have been mucking with my calibrated filament voltages in the Kaiju as well - in which I also use EML 300Bs.  My temporary solution has been to use a variac to bring up the voltage. But I've wanted a better solution - one that would be fun to build.

So I've been planning to build a DIY mains voltage quasi-regulator built around a big 2000VA Hammond auto-transformer (AT). The AT is intended to take an input of 115 VAC,  and has output taps for 85, 95, 105, 110, 115 and 125 VAC. I planned to use the 115 tap as the fixed output and have switchable inputs  at the 105, 110, and115 taps for nominal +10, +5 and 0V choices. I figured this would be fine since I'd have a 9-10A fuse on the AT that would limit the VA to max 1350 (in the off-chance, worst-case-scenario that my mains went to 125V and I had the AT input selected for +10V): so well below its 2000 VA rating.

PB warned me that the output voltages might be a bit unpredictable if only low current were drawn from the AT - I probably wouldn't ever draw more than 5 amps from the thing - and quite likely less. But I like the idea of having the headroom to scale up if I ever want to.  I did some reading and saw that the "regulation" of ATs is pretty good - meaning that output voltage of an AT is less dependent on current draw than is the case with a traditional primary/secondary transformer. So I figured I'd give it a shot, and if the voltage increase was a bit off I could trim it with my planned "diode voltage trimmers". And here is where my problem was born. My thinking was that since Schottky diodes, unlike resistors, provide a relatively fixed voltage drop regardless of current, I could use them to trim the output voltage from the AT.  The idea was to put a pair of reverse paralleled diodes on each of the hot/live wire and the common wire, with the eventual load (amps) sitting between them: when voltage was positive, one pair would drop a fixed nominal 1.7VAC before the load, and vice versa. I planned to have two such "diode trimmer switches" so that I could choose 0, -1.7 or -3.4 V, on top of the zero, +5 and +10 V from the AT.

Problem: I forgot to take into consideration heat dissipation of the diodes. If I wanted the DIY regulator to handle up to 10A, that would mean each diode would be generating 17W of heat. 8 diodes @ 17W each is 136 Watts! That's a lot of heat to get rid of from inside an enclosure.

So my tentative plan B is to build the the AT box without the diode trimmers and just install the output switches (0, +5 and +10V) and a small 17A rated common mode choke pcb for taming whatever nasties might be on my mains.

BUT, I'D STILL LIKE TO DO THE DIODE TRIMMER THING, IF POSSIBLE, ON A SMALLER SCALE: 3A instead of 10A. I have three 3A rated combo common/differential mode choke pcbs that I want to pair with the diode trimmer switches. I don't want to run the chokes at their max rating - from what I've read, inductance will be a bit higher at lower currents and that should be good for filtering out noise. My plan is to limit each combo filter to 1A of current. So 3 filters * 1A * 1.7V drop * 8 diodes (4 per switch) = 40.8W.

So my question is: Do people think it's feasible to dissipate 41W of heat (max) from inside an enclosure? - I'm willing to implement whatever combo of heatsinking, air holes/screens and even battery powered fans that might be necessary to make this work. [Edit: and just to be clear - I'm not asking for people to tell me exactly how to solve my heat dissipation problem (although that would be awesomely generous); I just want to know whether this is even worth me trying to figure out - or is it just a lost cause?]

many thanks in advance, Derek

[Paul Birkeland]

That is a lot of heat, I would be shopping for an enclosure for a DIY solid state amp with sides that are heatsinks.

You could also talk to a custom winder about having an autoformer made with a few more taps.

[Jamier]

Derek,

     diyaudio store, Dissipante 5U.

Jamie

[Deke609]

Thanks PB. Those are both good ideas. I will look into them. Hammond is "just up the road" - I think about a 2 hour drive from Toronto. And they do custom auto transformers - I should have thought of that to begin with!  But the diode trimmer thing is a neat idea. If I can find a suitable enclosure for not too much money, I'm kinda inclined to build the thing.

@Jamie: thanks much. I was just about to post and got the "Warning ... new reply" thing. I was thinking to look @ the very same enclosure.

cheers and thanks, Derek

[2wo]

I think you could  do what you're looking to do with  existing taps and a smaller transformer wired to buck or boost  A couple of volts.  A few switches and you'd have fine and course control with less heat...John

[Deke609]

Another fine idea. Many thanks John. I will contact Hammond tomorrow and ask for quotes for both a 115VAC nominal in and 112, 115, 118, 121 VAC nominal out, and a simpler and smaller buck transformer with -2 and -4VAC rated for 500VA.

cheers and thanks, Derek

[Paul Birkeland]

a simpler and smaller buck transformer with -2 and -4VAC rated for 500VA.

You were talking about 10A on the secondary, so a 4V AC transformer would only need to be rated for 40VA.

[Paul Joppa]

The diodes will create a wobble in the sine wave - it will go flat for a bit at every zero crossing. This will create energy at all the odd multiples of 60Hz (1.e. 180, 300, 420, 540 ...Hz). It will likely make noise, both electrical and mechanical.

Long ago, GR made a regulator that was a motor-driven variac with a feedback loop. It's an antique now, but interesting.

The modern version is a regenerator.

[Deke609]

You were talking about 10A on the secondary, so a 4V AC transformer would only need to be rated for 40VA.

Well that stopped me in my tracks. I figured that the VA rating for an AT works the same way as for a transformers with two separate windings - so if I wanted, say, a max of 125 VAC @ 10A, I'd need 1250VA plus some additional headroom .... hence the 2000VA AT.  But it's only the difference between input and output voltages that is used to calculate the required VA rating of an AT???!!!! If that's the case, then a nominal +10V difference @10A only requires a 100VA AT, plus some headroom, so maybe 150VA to be totally safe?

Which means my 2000VA AT is sufficient for an entire 200A panel?   Good grief!   

Can someone please confirm? If so, that was a big  mistake (literally - it's huge and weighs 23lbs!).  I guess I should have known something was off when I was ordering it from a local electronics parts supplier: the owner remarked something along the lines of "Oh, we've never had someone ask for that one before." Which I thought was odd in a city of about 3 million with iffy voltage regulation - but not odd enough to question things. 

The diodes will create a wobble in the sine wave - it will go flat for a bit at every zero crossing. This will create energy at all the odd multiples of 60Hz (1.e. 180, 300, 420, 540 ...Hz). It will likely make noise, both electrical and mechanical.

Many thanks PJ. That's the final nail in the coffin for this project. I'd read about the forward/reverse switching noise of diodes and that's why I was going with Schottkys, in the hope that they wold be quiet enough, at least after being run through some AC filter chokes. Oh well!

So, subject to someone confirming the point about VA rating of ATs above, it looks like I need to find a much smaller AT, ideally with the taps I actually want (perhaps custom from Hammond) or also find an even smaller buck/boost transformer for trimming 0-2-4 V.

Man ... and I was just about to start drilling holes in the enclosure! Sheesh.

Well ... I am disappointed but undeterred. I will figure out and build my own homegrown half-ass solution to my mains voltage problems ... even if that means I end up spending almost as much as I would have paid for one of those expensive regenerators that would  offer a complete solution. I'm not sure whether it's mostly enthusiasm or obstinacy that's driving me at this point ... sometimes they're hard to tell apart 

cheers and thanks, Derek

[Paul Birkeland]

Well that stopped me in my tracks. I figured that the VA rating for an AT works the same way as for a transformers with two separate windings

You want to go with current.  10A across the autoformer means 10A across your buck/boost transformer (you would wire it and use it as a transformer, not an autoformer), so a 4V/10A transformer is called for, which is 40VA.

Which means my 2000VA AT is sufficient for an entire 200A panel?   Good grief!   

If your mains voltage is 10V, then yes.

[Deke609]

Many thanks PB. I was just about to post a follow-up when you posted your response. But I think the following already typed post ties in with what you wrote.

I did some quick reading on "bucking transformers" this morning, and if I'm understanding things (big "if" here), it looks like I could use a high current 2.5VAC filament heater transformer as a bucking transformer - is that right? The Hammond 166S2 is 25 VA rated, primary is 115VAC, secondary is 2.5VAC C.T. From what I read it sounds like I can wire one end of the primary to the secondary so that the secondary is out-of-phase and hence "bucking" 2.5VAC. And since it's only bucking 2.5VAC, the 25VA rating means it good for a max of 10A. I think I'd feel more comfortable going with 9A or even 8A just to be on the safe side.

Is this right?

cheers and many thanks, Derek

[Paul Birkeland]

From what I read it sounds like I can wire one end of the primary secondary to the secondary autoformer so that the secondary is out-of-phase and hence "bucking" 2.5VAC. And since it's only bucking 2.5VAC, the 25VA rating means it good for a max of 10A. I think I'd feel more comfortable going with 9A or even 8A just to be on the safe side.

You should use the 10A rated model since you have the 10A rated autoformer and you've mentioned that you may hook a lot of gear up to this.  The primary of your bucking transformer is wired to the incoming mains as well, don't forget about that. 

[Deke609]

Many thanks PB. I'm not quite sure what configuration you are suggesting. Here is what I was thinking. Is this the same or equivalent to your suggestion? [Edit: you can scroll across the image with arrow keys to see the right hand side]

[Paul Birkeland]

No, the primary of the bucking transformer needs to be connected to the incoming AC, not the output of the autoformer.

[Deke609]

So: Mains -> Bucking -> Autoformer?

many thanks, Derek