300B SE Amplifier

Introduction

It’s been long since I had a 300B around. Yet, I always found it very interesting to hear all sorts of comments about the 300B single-ended amplifiers. The vast majority of the negative comments about them are misleading in my view as very few designs out there are able to get the best out of a 300B in SE. Mainly because the output transformer isn’t optimal for the amplifier due to cost compromise decisions and because the driver has to be able to push the 300B to its full swing and many of the drivers implemented are poorly designed to achieve very low distortion at maximum swing (e.g. 150 or 200Vpp).

For this reason, I always discouraged people from implementing a 300B unless they are prepared to invest on the amplifier properly. I mean, good iron, power supplies and a sterling driver which can do the job without being the one dominating the sound and level of distortion of the SE amp. This is exactly the same design principle I apply for any transmitting valve in SE mode. No exception. Some transmitting valves are even more demanding. However, the 300B is yet very much so.

After many years of failing often to hear good 300B amps and that I wanted to increase a bit the power of my current 4P1L PSE, I got my hands busy again in designing a new amplifier. As it is my style, it was somehow no compromise with the only exception that I designed and built it around the parts I had at hand (which have to admit are plenty) due to the COVID-19 pandemic situation. Many of the parts I managed to procure them over the years and some were intended for a 300B amplifier anyway.

My intention of this write-up is to share my experience of this 300B SE Amplifier design and build journey, which has taught me a lot of new things, like always. It may encourage other builders to do so, I hope. There are a few compromise decisions along the way which is always the case. You may chose to improve this in any way so would be keen to hear your comments and contribution to the DIY Audio community. Please don’t take this in black and white.

Design

You may well know by now that I’m a fan of fixed bias designs. On the other hand I don’t fancy much autobias and prefer not to have capacitors around where I can use them in better places which are less sonically present. Fixed bias is synonymous of power supply complexity. I don’t go simple in audio, so space and complexity wasn’t a constraint from the start.

After considering most of the various possible topologies, I settled down with the one I’ve had the best results sonically over the years. There are few circuit designs which can’t beat the mu-follower when driving many volts into the output stage. You should know by now that I really like the hybrid mu-follower stage for its sound as well as its performance. I experimented over the years with several triode-strapped frame grid pentodes, triodes and DHTs for the purpose.

In my design, the 300B with a 3.3KΩ OPT will need about 150Vpp to deliver the maximum output power rated at about 8W. The driver therefore, needs to be able to provide 150 to 200Vpp comfortably at very low distortion. This isn’t an easy task and many (many) valves fail to do so. Several design will add far too many odd harmonics when pushed to the limits.

From the IHT valve short list, I have a few which are my favourite: D3a, 6C45P, 6E5P, 6E6P-DR, E810F, E282F, C3g, 12HL7, etc. You should be able to find most of my driver experiments in the blog if you search carefully around. Here are some examples.

If we are going the fixed bias route, we can either use a good quality IT or use a source follower to drive the 300B grid. I opted for the later so I could have flexibility on the driver stage to experiment. That is something I set myself to do with this amp. I want to be able to exchange the driver stage easily. As you will see later, there is a modular aspect of this amp which is for this reason.

The grid of the 300B isn’t one of the worst loads to drive. At around 100pF total capacitance including parasitic ones, not that hard to be driven. I designed a Source Follower which could drive the grid of the 300B easily whilst providing very good signal swing: 200Vpp. Enough headroom for excellent transient response and no slew rate distortion. The main driver circuit is shown below:

The fixed bias circuit operates out of a raw 200V supply for negative swing as well as a (+25 or 50V) for positive grid excursion and or proper bias of the SF’s MOSFET. Let’s look at the design above. I implemented a fixed bias voltage reference which is shown here using OSRAM HV LEDs. They are very bright but I run them at 2mA which makes the light dim and very enjoyable. The voltage reference (Reg1) is an independent PCB and is decoupled by C5 and C6 given the high impedance at the input of the pass MOSFET part of the cap multiplier included in Reg1. The reference voltage is adjusted by a trimpot and provides the same reference to both L and R channels. There is a resistor divider to generate the bias voltage which is R9-P1-R10 and R11-P2-R12. P1 and P2 are 10 turn trimpots which allow for independent bias regulation of both channels. The fixed bias circuit has a source follower to provide a constant output impedance to the coupling circuit (C1 and R4) regardless of the bias resistor network position. This Source Follower is made up of T1, R1 and with a tail CCS (T2, R3 and R2). The tail CCS bias the PMBTA42 at 5mA to keep it within its SOA. I used SMD components here, but there is no reason why you shouldn’t implement it with standard through hole equivalent parts. There are a couple of additional capacitors, C3 which needs to be right close to T1 and T3 as well as the output cap C1.

The driver input is AC coupled via C1 (or C2) into the gate of the Source Follower. The output has a buildout resistor of 330Ω into the grid of the 300B.

The source follower is biased at 15mA. This should give plenty of current to drive the 3ooB effectively. Here is the design of the SF, which fits into a tiny PCB:

In some of my initial debugging tests due to the issue I described here, I changed the SF board and removed the tail CCS (T3 and T4) with a simple 10KΩ 5W wirewound resistor when I suspected that the tail CCS was causing the issue I experienced, I was wrong. If you want, you can implement a simpler SF this way and works well. The SF design uses T1 as the pass device. The 3LN80K5 is a great device as it has a plastic package, built in protection diodes and more importantly very low capacitances which makes it perfect for wide bandwidth operation. Please note that the protection zeners are shown in the diagram above in case anyone decides to use a different MOSFET. T2 and R2 provides surge current protection (or current limit in A2 operation). The decoupling of the SF is key with C2 and C4 on previous diagram. C2 is shown as C1 in the SF circuit and C4 is omitted. These caps should be soldered right at the +B/-B points of the PCB to ensure they do the proper decoupling job and avoid the SF to turn into a Colpitts oscillator. These SF boards have nearly 8MHz bandwidth as far as I could measure it!

The SF also provides a way of isolation of the driver with the output stage. This is great when we are looking to experiment with different drivers without worrying into creating slew rate distortion. A very nice driver is the 6SF5 which operates at sub 1mA anode current. Clearly, it can’t drive effectively the 300B grid directly, so best to couple it to the SF and let the SF to do the current drive job for him.

The main amplifier circuit is below:

The driver is V2. In this case a D3a. It has 3 SiC diodes for cathode bias. At 200V quiescent anode voltage provides enough swing to ensure 200Vpp swing at low distortion. How low? Well, my tests before shown 0.22% when driven hot and with cathode degeneration with a 270Ω cathode resistor instead of the SiC diodes:

I’m running it cooler at 200V / 10-12mA. The hybrid mu-follower is formed with a Rev1.0s (small gyrator PCB) which has a top IXTP08N100D and lower BSH111BK in the cascode stage. I also prefer heating the filaments with DC. I use a low voltage regulator for the purpose and heat both D3a valves with same reg. As am planning to build a modular amp, the driver power supply is prepared for DHT. I have one per channel and in this case, I use only one.

I’m running the 300B a tad hot. It’s on 375V/80mA whereas other will go with a cooler bias point at around 75mA. The D3a needs to swing at least 150Vpp in class A1 to deliver full power to deliver 8W of class A sound.

The output stage is very simple and revolves around the great OPT which is custom made by my friend Alexander. Here are the excellent specifications:

Power Supplies

The power supply becomes more than one when you switch to fixed bias, there is no free lunch am afraid. I implemented the HT supply with a surplus power transformer I had from Thomas Mayer. Excellent build quality. The use of damper diodes (6BY5GA) provides a smooth rise of the output voltage. I also use  WIMA DC Link film caps as the last caps. One on the power supply PCB (C11) and one in the amplifier board (C1) which is a massive 80μF/900V one.

The other power supplies include the fixed bias ones:

Some time ago I’ve got a set of custom-made power transformers by JMS Transformers in the UK. They have multi-taps in the secondaries and split bobbins. Also it comes with a copper screening layer. They were always intended for fixed bias purposes. One secondary is for the negative bias and the other one is for the positive one to ensure the SF operates at good VDS when the 300B grid is approaching 0V. Or even if the winding tap of 50V is used, it can deliver some positive grid current to push the 300B a little more beyond the 8W. Not needed in my case, but the design is adaptable for future use with transmitting valves (e.g. 845). The 2 power supplies are self explanatory. I use a PCB made for them which speed things up and provides a neat and compact finish. There are many ways to skin a cat so the same applies to the power supply design. I played with the filtering network based on the components I had at hand. You can do your way, of course.

The LT supply is shown below. This is for the 300B filaments, however I have another pair implemented for DHT drivers to be used in the future:

 

Build

I have posted several blog entries as shown below describing the building experience. I hope they provide enough insight to the modular (LEGO-like) approach used:

1. Build part I- project starts
2. Build part II – power supplies and SF boards
3. Build part III – Layout
4. Build part IV- board 1 finished (power supplies)
5. Build part V-fixed bias
6. Build part VI- fixed bias board
7. Build part VII – D3a driver
8. Building part VIII – cap boards
9. Building part IX – final layout  
10. 300B Amp finished

Measurements

It took me a bit of time to do this as measurement gear isn’t next to the amp as you can imagine. One needs to move to the other. I moved the measurement gear, simpler than moving the full cabinet of course!

I was pleased to see that the measurements correlated with what I was hearing. The amp is superb, in terms of response for a SE amplifier:

The frequency response is really good from 6.3Hz up to 71kHz @ -3dB points. This is great and most of it due to the output transformer characteristics. The D3a Driver has 37dB gain with 3-4MHz bandwidth and the Source Followers even 8Mhz. The excellent HF response is due to the transformer, an amazing piece of gear!

From a distortion point of view, it behaves as expected:

The harmonic decaying profile is as expected from the triodes involved. Only when you push it hard enough you get to see odd harmonics increasing a bit. There is no harmonic cancellation which I don’t care  as much as it’s hard to manage or replicate easily. Yet its only 0.3% higher than the simulation which is great. For example at 3W output the THD is 2%, mainly driven by H2.

My test setup was limited when I measured it today so couldn’t play to push the amp to clipping so didn’t measure it beyond 5W. The room was extremely hot as it was 32C in London today. This made me quit abruptly the work and decided to cool myself down somehow outside the loft room.

Sound

Oh yes, everyone is asking for this. Again I will make the caveat that this is very subjective and it depends on so many aspects (you all know this, why I’m trying to teach you to suck eggs!). Considering same system involved and room which hasn’t changed for some time, I have to admit that the sound of this amplifier is overwhelmingly good. It works better in my system compared to the 4P1L. Likely as it has a powerful bass response, it’s very fast and has an amazing bandwidth. Dynamics are great. The detail is near perfection. I have to admit that the D3a as driver is not failing to surprise me given I’m a DHT lover and my previous system was 100% DHT.

I’m currently playing EML300B, they do sound superbly. Previously did a few tests with Psvane 300B, which are good as well. Haven’t done enough listening yet to judge any material difference am afraid.

A digression in this story: the D3a fate

There is a funny story (if I can say truly that) to tell around the D3a. I had a nice collection of 10-12 pairs of NOS Valvo and Siemens D3a valves. A couple of years ago my daughter got her way into the workshop and started to poke around as kids do. She manage to throw from the workbench the entire set of D3a which were on the bench after some tracing and testing experiments at the time. I lost them all as they landed and damaged heavily on the floor, I think a pair survived if I don’t remember incorrectly. It wasn’t fun at the time I have to admit.

Last year at ETF.19 auction a beautiful set of Siemens NOS quartet of D3a was ready for bidding. I did not resist the temptation and bid for them. Luckily I won them without ending up in financial meltdown. Unfortunately some “clever” fellow at ETF stole them before I could pick them up. Still to date it surprise me that there can be educated people (as you would have thought they are if the are able to attend ETF) that can steal things from each other. They  were never found and the individual went quiet, like a snake. Shamefully enough the guilty fellow couldn’t stand up when asked to return them by the ETF organisers. I hope he or she will read this at some point. Let me point out that there is no intention on this comment to discredit ETF. I’m proud to be able to attend them and is probably one of the best audio gatherings out there.

Having being defeated and with no D3a at hand, I ended up buying a couple of pairs over the last 12 months. They are expensive these days. Anyhow, feels like D3a is resisting to be working on my system!

What’s next?

The modular approach of this amplifier and the fact that it has been built over IKEA chopping boards inside a dedicated cabinet was planned from the outset to play like LEGO with it. I’ve been developing a series of modular PCBs to speed up the building process and provide flexibility. Here are a few examples if you want to know more about them.

What comes next is the return of the DHT drivers. Currently I’m building a 46 (or 47) triode strapped driver with a SUT. It will be fun!

Perhaps trying the 6SF5 driver would deliver a 2 stage amplifier with enough gain for my DAC and system requirements. Currently I’m ok with the D3a unless I want to play really loud. In that case a x2 gain would be all I need. Adding a preamp is an option, but then the system becomes a 3-stage one.

Acknowledgements

There are a few people I have to mention here which made this project a reality. COVID-19 stopped my business weekly travel across Europe which it turned out to give me some evening time at the workshop instead after putting my little girls to bed which I didn’t have before. Trying to get something positive out of this pandemic year, I think at least it made this project possible in a short period of time:

• Rod Coleman – for his ongoing support. In particular when I ran out of ideas, he’s a man of great experience who unconditionally is there to help when needed!
• Pete Millett – for his amazing panel meters. More people should buy these.
• Alexander Naydenov – what can I say. Thank you for these great OPTs. Outstanding job.
• Simon Mears – made the modular building cabinet dream true.
• DHT Rob – for his encouragement and exchange of ideas. Love to learn from this man.

Now is time to spin some records and enjoy this amp in full. I hope you find this article useful.