My favourite valves together
Recently I revisited a beloved amp, the SE 45. This time I will share a more orthodox design without sand in play. Surprised? Well, I love lots of iron as well and here is a design I’ve been playing around for some time as I have all the components at hand.
Driving the 45
Driving the 45 isn’t an easy task. You can get outstanding performance with the gyrator (hybrid mu-follower stage) and nice triode stages (or triode-strapped pentodes) like 4P1L, 6J52p, D3a, C3g, 6J49P-DR, 6C45p-E, etc. If you want to go full DHT, then you have a challenge. The gain of DHTs is relatively low and unless you want to go crazy and spend money on an Emission Labs’ EML 20 or 30 valves, you will need to resort to a step up transformer to obtain the gain needed if you want to avoid more than 2 stages.
I looked back into my favourite list of valves and obviously the 46 came up at first thought. I have used it before with great success in my 814 SE Amp. However, implementing filament bias with that valve was crazy. I will not do that again, therefore a more sensible approach would be either fixed bias or cathode bias.
Let’s look at the 45 driving requirements first:
In my experience, you can get the best out of the 45 by loading it with +5KΩ and biasing it at 300V/30-34mA. The actual load with a 5K6Ω transformer will be closer to 6KΩ if you factor in the primary and secondary winding resistance. Either way, you’re looking to a 120Vpp driving requirements to get the 2W out of this stage. So you driver has to easily deliver 150Vpp with very low distortion. This rule out many options of course. I’m not going to dwell around this as it has been covered extensively by many people out there.
So if we look at the 46 in triode mode, you will only get a gain of 5. How do we achieve the voltage gain needed? My recommended way is to include a step up input transformer which also provides a galvanic isolation and load the 46 with a 1:1 (or 1:2) interstage transformer. I have used before the LL7903 which can do 1:8 with lower distortion at high signal levels. It sounds really nice and is a fantastic transformer. Recently I reviewed a great input transformer from Dorin Bodea here. This can do 1:6 but I assume that if you ask Dorin, he can get you a 1:8 version. You can implement this design with 1:6 though.
With a 1:8 input transformer you will get about a gain closer to 40 overall. This means that with just over 1Vrms you can drive the 45 to full tilt.
A lot of iron
So here is the initial design, you may critique this. That’s great. There is plenty of iron and oil caps for the fanatics:
The design is minimalistic and classic. The input transformer drives the 46 directly and has a Zobel network to compensate the HF response. The 46 is cathode bias with a 910Ω resistor and a 45μF oil cap. The choice of the interstage is the Lundahl LL2746. This nice transformer has proper primary inductance for this valve as well as can be wired on 1:2 or 1:1. In my case I prefer the 1:1 stage. It performs really well, I have tried it with the 4P1L to achieve a high gain stage before, see here. The operating point is reflected below:
The 46 is biased to 30mA and has enough headroom to deliver the +120Vpp without getting closer to grid current and with very low distortion.
The output stage is fixed bias. I will use Rod Coleman’s fixed bias regulator. He has recently released this and I’m in the process of completing my tests. The secondary of the LL2746 is loaded with a 220K but i suspect that a proper Zobel network may be needed to flat the response. Need to test on the bench and optimise this a bit.
The LL1623/60mA is my favourite 5K6:8 transformer I have used with both 45 and 2A3/6C4C stages.
The HT supply is being reused from my 4P1L PSE amp, but can be adapted as you need.
So if you have high efficient speakers and want the best sound of the first Watt, then you should look into this amp.
I’m interested in hearing different experiences and comments about this amp. I’m sure everyone has a view.
Happy New Year!