Extreme DIY.

[SteveTheShadow]

Been indulging in a bit of very stupid and silly behaviour over the past couple of weeks, which has resulted in this amplifier:

0A50399E-7A4B-4B03-8EE1-5B0D4A85291D.jpeg (109.08 KiB) Viewed 1563 times

Using the world-famous ‘45’ direct heated triode, it puts out 1.8 Wpc and is not a lot of use unless you have speakers of 100dB/W efficiency like my Fane single driver units. If you have the necessary speakers, this old triode is hard to beat for its sound quality. I’m smitten with it.

But the amp in the picture has a much more involved story behind its construction that its appearance may suggest. The output transformers are half professional and half built by me.

[SteveTheShadow]

The story is that the single ended output transformers I had were the wrong primary impedance, being 2.5K units. So I’d bodged the impedance to fit the 5K requirement of the 45 by configuring the transformer secondaries for 4 Ohms, then connecting an 8 Ohm speaker. The sound was very good, but my brain was under constant assault because of the fudged impedance.
Problem was I didn’t have a pair of 5K single ended output transformers, and couldn’t afford to fork out for new ones that would do justice to the transparency of the 45 valve….but I did have a pair of 5K push-pull output transformers. What if I were able to make use of those?

[Lindsayt]

Doesn't seem extreme to me at all. Seems like an entirely sensible use of the resources that you have at hand.

My 1.8 watt 45 baset SET amp goes plenty loud enough for me with c96 db efficient EV Sentry III's in a large room.

[SteveTheShadow]

Doesn't seem extreme to me at all. Seems like an entirely sensible use of the resources that you have at hand.

My 1.8 watt 45 based SET amp goes plenty loud enough for me with c96 db efficient EV Sentry III's in a large room.

You’ve not seen what’s coming

[SteveTheShadow]

Now then,
I had a pair of 5K push-pull output transformers, but that doesn’t mean to say they could be used in a single ended amplifier; in fact as they were….configured for push-pull KT88, they were completely useless.

Push-pull output stages differ from single ended, in that they make use of two opposite polarity signals to substantially increase their power output over a single ended stage. The two opposite signals are recombined in the output transformer to produce a single polarity signal that can be stepped down to feed the speaker.
This is achieved by feeding the high voltage DC operating current to a centre tap on the primary winding of the output transformer and connecting a valve to each end.

Because your two AC (music) signals are already coming in 180 deg out of phase with each other and the centre tapped primary winding puts them both 180 deg out of phase again which makes 360 deg in total, this means that they end up back in phase with each other and ready to be stepped down to drive the speakers. Clever huh.

Because the high voltage DC is fed to the middle of the primary winding, and the two output valves are connected to each end of said winding, the standing current flows in opposite directions. This means that there is a net DC current flow of zero in the primary winding due to cancellation. This is important because with a push pull output transformer, any appreciable static DC flow through its primary will magnetically saturate the core and create awful distortion when a music signal is applied. This is why matched valves and balancing up the output valve currents is important with a push-pull amplifier.

On the other hand, a single-ended output stage only has one valve. The high voltage DC operating current comes in at one end of the primary and exits at the other, where the valve anode connects. This means that the idle current will saturate the transformer core unless steps are taken to deal with it.

The way this is done, is to create a magnetic gap in the core, which prevents saturation on music signals. This translates to a physical gap being deliberately placed in the transformer laminations by the manufacturer.

But there is a price to be paid for dealing with this saturation problem. In simple terms, the gap in the core dramatically reduces the transformer’s inductance, which takes away the bass. To get the inductance back, the core needs to be made physically bigger, which means big and heavy transformers. When you make the core bigger you need a physically bigger set of coils, meaning you get more capacitance in the windings, which means rolled off treble, so by optimising one parameter, the bass, you degrade the other, the treble. So a single ended output transformer needs skilful winding to optimise the treble, whilst at the same time needing a big core to counteract the LF losses caused by the gap.

With a push pull transformer it is much easier for a manufacturer/designer to optimise the whole frequency range because he/she does not have to worry as much about core saturation, so collapsing inductance fecking up the bass is much less of an issue, as is high capacitance affecting the treble, because PP transformers are physically smaller than single ended. Winner, winner etc.

[SteveTheShadow]

So back to my situation. I had a pair of push-pull transformers that had the right impedance (5k) for a 45, but were no good for single ended because there was no gap in the core.

Most valve amp output transformers follow the E-I lamination method of construction below:

AE04F5F2-7940-4CA7-BA6B-910B19864BCE.jpeg (21.72 KiB) Viewed 1516 times

Push-pull output transformers, not needing a gap, are constructed with interleaved laminations, which means that they are stacked E-I, I-E, E-I, I-E and so on.
Single-ended output transformers on the other hand need a gap in the core, so that they have to have all the Es facing the same direction and the I laminations butted up to them, with a piece of insulating material between, to create the gap.

My 5K push-pull transformers had interleaved laminations, so if I wanted to use them in a single-ended application they would need to be completely dismantled, then put back together again with a butted stack, to create a gap. This is easier said than done, as transformer lamination stacks are vacuum impregnated with varnish and are nigh on impossible to dismantle. But being a crazed DIYer and an idiot to boot, I decided to set about the laminations with a Stanley knife and spatula, convinced I was onto a winner.

I knew, the manufacturer would have made the winding setup suitably wide bandwidth and the 45 valves only need 36mA of idle current. The transformers themselves were meant for 50W Class AB push-pull operation, so I figured the core would easily be big enough to support 2W max of single-ended power output, and with that low power and low idle current, the gap would provide the anti saturation characteristics without losing massive amounts of inductance, distorting the bass.

There were certainly a lot of assumptions there, even if I could have got them apart to test my hypothesis, but I was willing to take the gamble.

[Tony Moore]

[SteveTheShadow]

So, being the deluded fool, I set to work on one of the transformers.
Tools required:
Stanley knife with new, sharp blade.
Miniature spatula.
Endless patience care and perseverance

First one took four hours to dismantle.

524BC203-EF8F-46E8-A519-62F26D4EE35F.jpeg (154.03 KiB) Viewed 1476 times

Second one took three.

45E792B0-239E-4539-8EB4-C2D14CC43B19.jpeg (162.53 KiB) Viewed 1476 times

It was tedious, and tiring, but I got them apart, and separated them from the coils, despite the varnish impregnation.

[SteveTheShadow]

The reassembly underway on one of the transformers.

95657F83-9C15-424A-8872-FD008FEDF635.jpeg (84.18 KiB) Viewed 1464 times

Lams were stacked on a dead flat surface ie. a mirror, so that the mating surfaces for the gap would be as uniform as humanly possible.

D8C08E04-6D48-42F8-AAFE-E23A018C66AA.jpeg (166.71 KiB) Viewed 1464 times

E Lams were then varnished with a 50/50 mix of turps and yacht varnish.
The stuff was allowed to seep down through the structure until it came out all over the mirror surface.
The lams were then clamped up, squeezing the excess varnish out again.

I lams were dealt with the same, and three bolts used to clamp them up.

The other transformer was done by the same method and the pair were left for two days whilst the varnish set.

[SteveTheShadow]

Whilst the varnish was setting, a bit of research was done via the website of the late Patrick Turner. www.turneraudio.com.au
Here I found a ton of stuff about output transformer design, of both the single ended and push-pull variety.
To cut a long story short I managed with much head scratching and brain pain, to work out that I needed a 0.2mm gap in the laminations, to give the proper level of performance with a core of the size I was dealing with.

A paper gasket made from a pizza menu somebody shoved through the letter box, was just the right thickness to give the correct gap.

Much clamping and bolting, produced a pair of gapped, single ended output transformers.
Cost £5 for a small tin of yacht varnish and a lot of very careful labour.

AF84F69D-CF6B-45C1-B847-ECC37D9C2758.jpeg (99.76 KiB) Viewed 1459 times