PT and PS design question

[krair]

So I am designing an amplifier, and am having a company build a custom power transformer to my specs. The current one I am using is working well, but has been getting extremely hot. After looking at the specs again, it looks like I am over-drawing one of the windings. The transformer is very large and heavy, so my goal is to create a new transformer with the current rating to handle my circuit and hopefully no larger or heavier. I am using 4 KT88's and 3 12AX7's.

Currently: 3 windings. One at 400-60-0-400 (KT88 plates and bias voltage), the second at 275-0-275 for KT88 screens and 12AX7 plates, third is the heaters. Right now the 2 HT windings are rated for 460mA EACH. The OT is 1.5k, and the KT88 plates are at 560V. Graphing the specs, I see a maximum potential swing of approx 310mA per tube, or 620mA for 2 in parallel.

I am a little unsure about the best route to achieve the above goal, but here are my two ideas. A little bit of input/criticism/etc. would be extremely helpful!

Option 1: Single winding, ~800mA, taps for 400-275-60-0-275-400. Probably about the same size and weight, using a special steel with better heat dissipation.

Option 2: 2 windings, ??? mA, One at 275-60-0-275, one at 125-0-125. Connect the center tap of the 125-0-125 to the rectified side of the 275-0-275. This would float the ground and hopefully give me a 560V HT for the plates and ~380V for the screens and pre-amp plates. The only thing I am not sure about in this situation is how I would need to setup the transformer as far current rating, and also how these 2 options might differ in performance.

See here for what I have in mind:
http://www.drtube.com/schematics/hiwatt/hwpsu3.gif

Please let me know your thoughts on this. If you have any questions I will do my best to answer them.

[defaced]

1.5k is a little on the low side for 4 x KT-88 at that voltage. I run a similar setup (400v AC winding, typical plate and screen setups (screens about 5v below the plates)) setup with a 1.9k primary but would prefer to even up it a bit more for safety. Maybe 2.2 to 2.5k.

I'd do option one. The screens will be alot stiffer, which you may or may not like for your design goals. See some of the other threads here regarding "dual rail" supplies and such. The topic has been hashed through pretty thoroughly.

[krair]

cool thanks, I searched around but didn't look specifically for "dual rail". I will check that out.

[defaced]

BTW, that's not the proper name for it, but anyway, this is the thread you want to start at:
http://music-electronics-forum.com/t19103/

[R.G.]

So I am designing an amplifier, and am having a company build a custom power transformer to my specs. The current one I am using is working well, but has been getting extremely hot. After looking at the specs again, it looks like I am over-drawing one of the windings. The transformer is very large and heavy, so my goal is to create a new transformer with the current rating to handle my circuit and hopefully no larger or heavier.

You have a built-in conflict here. I'm pretty sure you also want it to not be any more expensive, too. If it's getting too hot under existing operation, you either have to increase the operating insulation class and let it get that hot safely, or make it bigger to get the temperature down. Both of those will be more expensive. The MBA-disease solution would be to measure/test it and figure out a way to be OK with it being the same size and heat it is now, but decide "I'll probably get promoted and not have to deal with the field failures". If that last option is open to you.

The power a transformer puts out without dying depends on three things:
- core loss at the operating primary voltage and frequency;
- copper losses at the maximum load;
- what temperature the winding insulation system will withstand.
Core loss and copper loss are the heat inputs, heat transfer to the air and/or chassis is the heat output, and the temperature will rise until heat out equals heat in. It can do that on small surfaces with high temps or larger surfaces at low temps, or with fancier cooling methods than natural convection, like fan cooling, heat pipes, heat sink fins, etc. In general you get off into the weeds pretty quickly if you get away from natural convection.

You can run transformers hot enough to cause instantaneous second and third degree burns on skin without damaging them if you specify a high enough temperature rating on the insulation. They'll do it happily. Whether this has consequences on your amplifier meeting the UL or other testing body safety specs or not may have other consequences; maximum "may-touch" temp is about 60C I think. It's been a while since I did this, so I'd have to look it up. If the surface gets over that, you may have to prove that the surface cannot be touched by the standard "may touch" probe.

You ... are ... getting it safety inspected if you're going to be selling it to non-technical musicians, aren't you?

Neither the iron nor the copper care much as long as the insulation does not break down and let shorts and arcs happen that *can* melt the copper. The first temp you'll hit that will make the iron or copper quit might be the Curie temperature of the iron and the insulation will be toast long before that. So temperature rise is not the problem - it's a combination of transformer losses and temperature rating on the insulation. The simplest solution to your problem is to
(1) measure your maximum operating loads, including at high AC power line voltages and maximum heating on the transformer which may or may not be at max power out depending on the operating class of the output stage
(2) measure the surface temperature of the core after it's heated long enough not to keep rising; this is taken to be five time constants, and the time constant for a big transformer may be hours; also measure the internal copper temperature by the change of resistance method on the copper windings after the external temp quits going up.
(3) tell the guys who are the transformer pros what the maximum **measured** operating currents are, and the **measured** temperature rises, especially the internal copper temp rise
(4) let *them* tell you what your options are based on the actual facts and their experience
(5) tell the transformer pros which of their advised paths you want. It will be one of (a)the temp rise is really OK, it can take it even if it feels awfully hot; they should be willing to back up that opinion with warranty coverage; or (b) we can keep it the same size and weight, but you'll have to pay more for higher temp rated insulation; or (c ) we can make it bigger and heavier to get the temp down and at the same time keep the cheaper insulation system; that will cost a little more too.

I am a little unsure about the best route to achieve the above goal, but here are my two ideas. A little bit of input/criticism/etc. would be extremely helpful!

Transformer design is a specialty which depends on a lot of history with the materials and operating procedures of the place that will produce the goods to give good performance at the lowest reasonable cost. It is not just a volts and amps kind of thing.

I advise you to not get fancy with the volts and amps, but to tell the pros what you have to have, let them tell you the options they can provide, then decide which of those ways to go. Don't make them go build something you think is better without you personally having the detailed history with transformer design.

There is a hidden tradeoff here: warranty cost. If you get just a ... little... too cheap on the transformers, it will show up at some later time as an increased failure rate in the field. Saving a few bucks per transformer versus having to buy more whole transformers for warranty repairs or even worse getting a bad rep for stuff going up in flames in public is a very tricky tradeoff. Maximizing your profits over that kind of tradeoff is not simple, and again it's not solely a matter of volts and amps and clever rectification circuits. I think there's probably a KISS principle hidden in there too.

You have the attention of the transformer pros who are doing this for you. Use their expertise to solve this. You go make the amp work and sound good.

[krair]

R.G. Thanks for the insightful post. I was hoping to get a straight answer, but I guess I should give a little background since I am new to this forum.

I may end up selling a few, but it is too early to tell. Cost is not an issue right now, making it work right and for a long time is of concern however. Not trying to sound like an a**hole here, but I wanted to let you know I have gone through the thought process of trade-offs and cost analysis. I am actually a graduate with a business degree, not an EE degree. (Dad is an EE from duke though, so i've been engineering oriented my entire life lol). So without regards to the business and legal aspects of my amp, I need some technical help .

Talking to the company winding my trannies, the dude was of ZERO help. Even though he is supposedly their technical expert, when I came to him with these questions he gave me the "i just build them" treatment. Saying he had no knowledge of circuit design or how the transformer is used by me. Am hoping for some expertise here. When I mentioned the outside operating temperature, he did seem a bit concerned however. And as you mentioned, the last thing I want to deal with is worrying about the longevity of the amplifier as a whole or injuries caused by my amplifier.

Part of the reason I have the aforementioned goal, is because currently the winding set for screens and pre-amp plates is VERY over-rated for what the amp will actually draw. I haven't done current measurements on those parts of the amp yet, and I will shortly, but either way the theoretical draw from the plates already smashes the rated current at the moment. Either way, my 28lb tranny should have more than enough iron in it, but one of the windings should hardly be drawing more than 200mA, even though it is rated for 460. So with regards to weight and size, I feel that there must be a way to simplify the design so that the plate supply has more current handling, and less for the screen/pre-amp.

I understand that transformer design is quite an art, and at the moment I do not have the time to dive into the specifics, so I am indeed looking for help from experts here.

[R.G.]

Good. Thanks for clarifying. I'm going to knock off and go to bed now, and give this more and proper attention in the morning when I'm fresh.

I'm sorry to hear about that response from your transformer maker. It makes it much harder on you, and I find it disappointing in general. I did design transformers at one time, but the final iterations were done by vendors that did nothing else. They were extremely helpful, and I hope that is not going by the wayside.

You'll need to do the actual current measurements in any case, so get started thinking about doing that in a solid way. More tomorrow.

[Enzo]

Just for the experience, find a few amps, oh say a fender Twin or similar, a PV 5150, whatever, and once they have been running an hour, grab ahold of the power transformer just to see what an average amp transformer feels like. They can run darn hot, but none of them are known for eating power transformers. Heat is a legitimate concern of course, but we also don't want to be overly worried about what turns out to be quite common and not a likely failure point.

I claim no particular knowledge of transformer design. What all heats them up? I assume magnetic losses or eddy losses in the iron, and also IR heating from the winding wire resistance? SOmething else I haven't considered?

[trobbins]

You didn't indicate what winding was being over-loaded on your existing PT ? Was it for the heaters?

You don't need to still use a V-CT-V type winding if you are going to use silicon diode rectifiers - that is historically linked to valve rectifier diodes. I suggest it would be simple to just use a single section winding for each HT supply, and use a diode bridge, then link the negatives at a single star point. Tappings are messy for a transformer builder. I agree that it would be easier to use a separate HT winding for PP OT supply, and a separate supply for screens and preamps - which can be much lower current rating, and use more capacitance for hum reduction.

You can even use two or three heater windings - which don't need CTs, as easier and better performance to use humdingers.

Ciao, Tim

[R.G.]

Just for the experience, find a few amps, oh say a fender Twin or similar, a PV 5150, whatever, and once they have been running an hour, grab ahold of the power transformer just to see what an average amp transformer feels like. They can run darn hot, but none of them are known for eating power transformers. Heat is a legitimate concern of course, but we also don't want to be overly worried about what turns out to be quite common and not a likely failure point.

That's a very useful and practical suggestion. I can tell you the most likely answer. By pure accident, the human anatomy is set up so that most people will not hold their fingertip on a metal surface at over 130F/55C. This happens to be the most likely surface temp for typical transformers made to "Class A"/105C insulation practices. If you don't know what's inside a transformer and it's not marked, it is extremely likely to be Class A/105C because that's cheapest and doesn't present a burn hazard.

That's one reason I wanted him to measure the surface temp on the transformer. It's possible that it's hot, but fine.

I claim no particular knowledge of transformer design. What all heats them up? I assume magnetic losses or eddy losses in the iron, and also IR heating from the winding wire resistance? SOmething else I haven't considered?

All you missed are hysteresis losses. The iron heats from eddy currents and hysteresis from working the magnetic circuit back and forth. Then it's IR loss in the copper. Good transformer design starts with those being about equal per unit volume; or with iron loss and copper loss equal, to avoid hot spots.

But you can safely drive the iron and copper at temperatures that will sizzle water on the iron if your insulation can handle it. That's a *bad* idea in most cases, but it's possible.

You didn't indicate what winding was being over-loaded on your existing PT ? Was it for the heaters?

You don't need to still use a V-CT-V type winding if you are going to use silicon diode rectifiers - that is historically linked to valve rectifier diodes. I suggest it would be simple to just use a single section winding for each HT supply, and use a diode bridge, then link the negatives at a single star point. Tappings are messy for a transformer builder. I agree that it would be easier to use a separate HT winding for PP OT supply, and a separate supply for screens and preamps - which can be much lower current rating, and use more capacitance for hum reduction.

You can even use two or three heater windings - which don't need CTs, as easier and better performance to use humdingers.

Good points. That's why I wanted him to measure the actual operating currents. That lets you know what the currents are in the windings, and is the starting point for designing out hot spots in the transformer.

That business about equal iron and copper losses and equal power density is what drives the answers. Once you set the magnetic operating point on the B-H curve for the iron, its power loss density is fixed. Then you can apportion the copper losses. To a first approximation, these are proportional to the RMS current density in the wires. So you set the current density in each winding to be the same; since each winding has different currents and numbers of turns, setting equal current densities automatically sets the wire sizes to be close to right. The real variables after that are all manufacturing ones. You can't fill a winding window perfectly full, there has to be insulation. And since the insulation takes up space, and the wires don't perfectly fill layers, you can't use the entire window for copper. So you iterate back and forth trying to get a balance of copper area and current density along with winding techniques and insulation materials that will actually work on your assembly line.

The rectification and filtering on DC-making outputs matters. Notice that I said "RMS currents". For AC outputs like heaters, the current is already RMS since it's used as AC. For outputs which are rectified to DC, the current flows in the winding only for little blips of time near the peak of the AC waveform, charging the filter caps. The time average of these pulses must equal the time average of the DC current out, so the blips are much, much higher currents than the DC out. How much higher is very difficult to analyze with any accuracy and usually must be measured. There are rough guides though. And because RMS involves squaring the instantaneous current, the RMS value of a highly peaked current in the winding is larger than the average DC current out.

For full wave centertap windings, the RMS-to-DC-average current ratio is about 1.2 to 1.4. But the number of turns is twice as big as for a full wave bridge output because only half of the winding is active on any half-cycle. For Full Wave Bridge rectified windings, the RMS to DC average ratio can be 1.6 to 1.8 and in both cases this depends on the *size of the filter caps* (!?). And since the whole winding is working on each half cycle for FWB rectification, you get to use half the turns and presumably twice the wire diameter based on equal-power density considerations.

It's this interaction of load, rectifier, capacitors, and power densities that all have to be simultaneously correct that make transformer design tricky. Then it all has to be economically buildable.

It gets back to the aphorism: Good, Fast, Cheap; pick any two.

[krair]

Enzo - Understood and agreed. I have other amps and have built other amps where the PT does get hot, however I know that the specs of the tranny can handle the current and heat. According to the load line, I'm not comfortable leaving it "as is". I have used the amp for 10-12 hours straight at a time with no problems or changes in performance, but from a longevity standpoint I would feel better knowing that there is a safety margin in the specs.

Tim - As far as I can tell, it is not the heaters. Spec wise it is rated for 8A and I believe I should be just below that. Again, I will measure to be sure, but I left the amp on standby for about 2 hours and the tranny was cold. It gets HOT in about 5-10 min when on. Considering each of the other two windings are rated at 460mA EACH, I would be surprised of 3 AX7 plates and the screens are drawing anything over 200mA (primarily the screens). Thanks for the insight on the windings and taps.

R.G. - I will measure temps and current in the next few days and get back to you. Right now I am using a full wave center-tapped and 2 220uF filter caps in series with bleeder resistors for the plates. The screen supply goes through a 220uF and then a 50 + 50 split cap. The pre-amp is further filtered by some smaller caps.

As you are talking about power densities and how the winding process differs between the bridge vs standard FW rectification, how does this affect the transformer's overall size, weight and current carrying capabilities? It seems as though you could use a smaller piece of iron with the full wave CT, but more turns, where the bridge would use a larger piece of iron but half the turns for roughly the same current carrying?

I know a lot of the questions with regards to this stuff seems pretty rudimentary, but again I have only scratched the surface of EE. Do you guys have any good books you recommend with regards to transformer/PS design? Couldn't find a ton of info online.

Again, either way I will be redesigning the transformer/PS, I am just looking for the correct method and ratings.

[R.G.]

I will measure temps and current in the next few days and get back to you.

OK.

As you are talking about power densities and how the winding process differs between the bridge vs standard FW rectification, how does this affect the transformer's overall size, weight and current carrying capabilities?

Not much.

It seems as though you could use a smaller piece of iron with the full wave CT, but more turns, where the bridge would use a larger piece of iron but half the turns for roughly the same current carrying?

The iron turns out to be mostly the same. You do use twice the turns of smaller wire for FWCT, but mostly the losses and inefficiency in winding the copper, space lost to insulation and edges of layers double up too. That may mean you can't keep twice the number of half-sized wire in the same space. And the RMS current of half-wave rectified currents in the FWCT is not exactly half of the RMS of FWB, so the turns are not exactly half-diameter. So you may have to use more iron to get fewer turns to get that into the winding area. More iron translates to fewer turns per volt, so the turns for the same voltage can go down, and that can get you into more iron by a back door. They all interact.

I know a lot of the questions with regards to this stuff seems pretty rudimentary, but again I have only scratched the surface of EE. Do you guys have any good books you recommend with regards to transformer/PS design? Couldn't find a ton of info online.

You won't. Transformer design is not easy to calculate an exact solution to; everything affects everything else and so a large amount of it's from history charts of what worked last time we did this. Frankly, an EE degree is only the start of a beginners admission to transformer design, and most EEs are ignorant of the issues.

Again, either way I will be redesigning the transformer/PS, I am just looking for the correct method and ratings.

If it were me, I'd measure the loads carefully, voltage and current, then design backwards from that. Find out what you need, then go work on the cheapest and most reliable compromise to get it.

[Diablo]

What type of steel have you specified for your transformer? I would specify a steel grade, thickness of the lamination and the type of insulation coating to insure quality and reliability. There are lots of choices in the lamination steel that can effect the core loss and permeability. Lower core loss results in lower heat in the PT, with no other changes. Grain oriented steel, like an M3, would give you lower temperature rise than say an M43 non oriented steel.

[R.G.]

What type of steel have you specified for your transformer? I would specify a steel grade, thickness of the lamination and the type of insulation coating to insure quality and reliability. There are lots of choices in the lamination steel that can effect the core loss and permeability. Lower core loss results in lower heat in the PT, with no other changes. Grain oriented steel, like an M3, would give you lower temperature rise than say an M43 non oriented steel.

That is correct. Grain oriented iron increases permeability, letting you use fewer turns per volt, and helps with size. It also sharpens the knee of the B-H curve and costs more per pound. Thinner laminations can dramatically reduce eddy current loss, reducing core loss. They also cost more to buy, stack less compactly because of the insulating layer on each one, and up the labor cost for stacking a lot, not least because the really thin ones have to be hand stacked.

One thing to consider is that human design factor again. The most effective, efficient core materials will be the ones which the transformer designer has lots of experience with. You really, really want him (they are preponderantly male) to be familiar with the material and procedures and not have to experiment with yours to get it right. I made this very mistake myself once - demanding that materials I though were better be used. It got done, but was much more expensive, including the three prototype iterations to get it right.

[J M Fahey]

Fully agree with measuring actual current loads and designing backwards.
As-is it's grossly unbalanced, I'd suggest you lower the screen+preamp current consumption a lot and apply the copper/space saved to the main winding.
Go for bridge rectifiers on each of the HV ones for space efficiency, again "invest" saved space in thicker copper where it does most good.
For maximum efficiency and stifness, build two (single winding each, bridge rectified) power supplies, of the same current handling each, in series.
The first one should go from ground to screen B+; the second one from that point to actual plate B+.
Your screens will live in heaven.
Hiwatts are known for using some variation of this.
Since you are at it, increase heater current capacity by 30/50% , not because you really need it but to lower resistive losses.
Anyway heater *power* is but a fraction of total needs.

Now let me play Devil's advocate here: can you trust these guys when they say that a winding is rated for such and such current?
I know a lot of guys who have an established factory, sell a lot, and yet know little or nothing, design-wise.
The bulk of their output is copies of established designs, which may even be very well made, but when you ask them original or custom work .... tsk tsk.
The "I just build them" answer let me thinking.

My personal experience favors minimizing *copper* losses, buy using less turns of thicker wire, and slamming iron with high current densities.
It's worked much better for me than the other way out.
Obviously I use very good iron, the end result is smaller and cheaper transformers, with very good regulation (low internal impedance), with the only problem of higher turn-on current and needing somewhat larger fuses.
They behave in a similar way to toroids.