ANK 300B Interstage Monoblocks

I've become a big DIY enthusiast over the past couple of years, rekindling an interest I had in my college years and early 20s (~40 years ago). At the time, career and family got in the way, and I never got back to it until recently.

In the past two years, I've built a pair of speakers, a pair of subs, three pair of solid state monoblocks (class A, class D and a chip amp based design), a class D stereo amp, and a preamp. And I close to finishing another preamp.

For my next project, I decided to try something with tubes, and the 300B SET amps have always intrigued me. Since I've never done a build with tubes before, I decided to start with a kit. But I can't leave well enough alone, so my build will have a lot of embellishments.

Below is a photo of what the standard build looks like. Mine will look a bit different since I'm designing my own chassis/enclosure. This will give me room to add the extra stuff I'm planning and to move the connectors to the rear (making it easier to connect my stiff interconnects and speaker cables).

If there is interest, I can turn this into a build thread showing progress along the way.

 

Jay, I'm interested (and I imagine others will be). I'd love to see your progress & maybe even hear the results when you're finished. Carry on!

 

Nice! There was another build thread on the ANK 300B here a few years ago (wow, just checked and it was actually in 2014), but didn't really get into build details and progress pictures, it was a full system build with speaker and preamp too. Be fun to see your progress along the way, whatever you may want to share.

 

Thanks guys. Yes, I read the previous thread when I was thinking of doing this.

Mike - I hope you're doing well. I'd be happy to have you over for a listen when they're done.

I ordered the kit about three weeks ago. Brian gave me a nice discount because he was putting together another of the same kit for another customer. Then he took a bit more off when I told him I didn't need the chassis. Brian told me everything will ship today so it will hopefully arrive by the end of next week.

In the meantime, I've been ordering parts and building the extra PCBs I'm going to incorporate in the build. I'm planning the following changes to the basic kit.

• Move all connections to the back panel. As I mentioned - my interconnects are super stiff (Iconoclast balanced cables) and my speaker cables are stiff enough that I'd have a hard time routing them to the side of the amps.
• Add a balanced connection (using a Jensen JT-11P line input transformer).
• Add a second balanced input with Miflex KPCU caps to roll off the bottom two octaves - my current speakers cross over to subs at around 70Hz, and the new speakers I'm building cross over even higher.
• Use higher quality connectors for power (Furutech), RCA (WBT) and speaker (EIZZ binding posts as well as GR-Research Tube Connectors).
• Add a soft-start using the Neurochrome ISS. This not only reduces stress on the rectifier tube, but also allows me to add an illuminated anti-vandal switch on the front panel, as well as a 12V trigger input (not sure if I'll ever use this).
• Add a speaker protection circuit with 30-second start-up delay. I'm using the PCBs from the DIYAudio Store.
• Replace the power supply circuit for the driver stage (which uses two resistor/cap stages to drop the voltage) with a regulated PS using the Neurochrome Maida regulator.
• Use a regulated DC supply for the driver stage filament (the kit uses AC from a transformer tap). I'm using PCBs from Pete Millet.
• Replace the 300B DC-regulated filament PS supplied with the kit with two TentLabs current-source filament supplies (one for each 300B).
• Use Audio Note 2W non-magnetic silver tantalum resistors for the three gate resistors (replacing the entry-level AN tantalums).
• Add a cap between B+ and 300B cathodes. This injects a small amount of B+ PS noise into the output stage to null the noise on the output (see this tubecad blog post (BCF-2 & Aikido Single-Ended Output Stage ) about half way down.

To fit all this stuff, my chassis will be a couple inches wider and deeper. I'm also using much heavier panels (6mm top plate, 10mm side and front panels, 4mm back and bottom panels) to create a stiffer chassis. I'm also planning to make a cover for the power supply since I don't consider these transformers particularly attractive. The C-core interstage and output transformers will remain exposed because I think these look pretty cool. I'm also keep the copper plate shown in the photo above.

I'll post more detail on the build as I progress. Thanks for your interest.

 

I'll be really interested to watch along and others have said with pictures of build progress will be a great thread ... count me in to follow

 

Do it!

 

Always like me some diy. The original amp looks great, can't wait to see your version.............

 

All these tweaks and additions will make for a really interesting and educational thread. Looking forward to what you come up with!

 

Jay I would love to follow your build and lots of pics please!

 

+1

 

Here's a bit to whet your appetite. Here's the schematic of the standard power supply that comes with the kit.


The part highlighted in yellow is used to reduce the main B+ voltage (~425V) down to about 180V for the driver stage. It does this by relying on the current through the two 15K resistors to drop the voltage the required amount. This works OK, but the voltage will be dependent on the driver stage current draw, which will be modulated by the signal. Of course, the large 100uF caps will filter this down to nearly DC, so the voltage will change slowly based on the average current draw.

In addition, the input of this part of the power supply is shared with the output stage, so the input voltage will be modulated to some extent by the fluctuating current demands of the two 300Bs.

And finally, even though this type of supply does a surprisingly good job of reducing power supply ripple, it's possible to make this even lower. In this kind of zero-feedback design, the power supply is essentially in the signal path, so anything that can be done to clean up the supply will help.

So I've decided to replace the highlighted part of the circuit with a Maida regulator circuit. This type of circuit was originally shown in an application note by Michael Maida, then of National Semiconductor, in 1980. It uses a combination of an integrated circuit voltage regulator to provide fine voltage control, and a separate power transistor to handle the high voltage drop. Tom Christianson from Neurochrome has developed an updated version of this concept which provides an extremely low output impedance (< 0.06V @ 1Khz) and very low ripple and noise (20 uV). I purchased boards from Neurochrome that had the surface mount components pre-installed, and built up the rest of the board with the appropriate components to provide the desired 180V output (shown on the right in the photo below). One board is used in each monoblock.

The other change I made to the driver circuit was to use a DC regulated supply for the filament. Normally, the filaments would be connected directly to the transformer filament taps shown above. But using AC for the heater filaments provides a path to couple power supply hum into the audio signal. Since I will eventually be using these amps with my new speakers (in progress) that are efficient (98db/w) line sources, I want to make sure the background noise and hum is as low as possible.

So I purchased a couple of blank PCBs from Pete Millett through ebay and built them up (shown on the left below). These are designed with soft-recovery schottky rectifiers and a low-voltage-drop low-noise linear regulator, followed by a CLC circuit, to provide a very low noise DC filament supply. The circuit is designed to work with minimal voltage drop, so their is enough voltage from the normal AC filament supply to generate the necessary DC voltage. One board is used in each monoblock.

 

These amps take a while for the voltages to stabilize and can create quite a bit of hum and noise for the first 10-20 seconds. So I decided to add a speaker protection circuit with a delayed start. This basically uses a relay to engage the speaker outputs after a period of time, and will also disconnect the speakers if any significant DC is detected.

Since the speakers are connected through a transformer, there isn't going to be any sustained DC, but it's possible this circuit might help in a catastrophic situation. The main purpose, though, is to keep the speakers disconnected for 20-30 seconds when the amp is first powered up.

I purchased a couple of blank PCBs from the DIYAudioStore which provide what I need. These are designed for a stereo amp, so I've only populated one of the two relays. Since tube amps with transformer coupled outputs can create damaging voltages on the output tubes if no load is connected, I have added a connection to the normally closed (relay not engaged) connection that I will connect to a 25W 8ohm resistor load. So when the speakers are disengaged, the amp will still see a normal load.

This circuit is primarily designed for solid state amps which reach equilibrium faster than a tube amp, so I've adjusted the turn on delay by increasing the size of the cap in the RC circuit. It now takes about 25 seconds to engage, which I think should be enough. If not, it's not hard to modify.

The completed board is shown on the left in the photo below (one used for each monoblock). The two resistors and green shrink wrap you see on the right side of the board is used to replace the missing relay coil that would normally be there in a stereo setup and provide the correct voltage for the 12V relay that is populated.

This board is designed to run off it's own AC power source. It doesn't use a lot of current, but the voltages from the transformer included in the kit were either WAY too high, or too low, so I am going to add a small 12VAC 5VA transformer on it's own PCB (shown on the right in the photo below) to power this board.

 

Most of you probably know that the 300B is a Directly Heated Triode. This means that the cathode is the filament.

The ANK 300B SET uses what is referred to as cathode bias. This means that the cathode is connected to ground through a resistor paralleled with a capacitor. The current through the resistor sets the bias. Since the two sides of the cathode must also be connected to a power supply of some sort to heat it up, this supply must be floating relative to ground.

The simplest approach is to use a separate transformer winding and use AC to heat the cathode. The problem with this is that the AC couples with the audio signal and will generate some hum on the output (generally around 5mv for a 300B SET) which is easily heard on any speakers efficient enough to use with this amp.

The AC hum also creates intermodulation distortion which is approximately 20db higher than using a DC filament supply. This will reduce detail and imaging, and make the amp less pleasant to listen to.

The DC regulated supply that comes with the ANK kit is a big improvement over using AC, but is still not optimal. This supply uses a bridge rectifier that generates a fair amount of switching noise. Further, the low output impedance of this supply means that a portion of the audio current will run through the supply, so any noise or nonlinearity of the supply at audio frequencies will affect the output.

The preferred approach is to use a high-impedance current source to drive the filament. TentLabs has developed such a supply which has received excellent reviews. From the comments I've read, this supply results in a bigger sound stage with better focus and detail.

A separate supply is required for each of the two 300B tubes. These supplies come fully assembled and are shown below. There are numerous surface mount components mounted to the bottom side of the board.

 

@Jaytor, these last three posts of yours have been really great. Not only have I learned quite a bit from the text already, but it's also exciting to see you utilise many of the PCBs I've been curious about.

That Tentlabs supply is quite a surprise. I have always been confused as to what makes it so noteworthy, never having seen the underside until now.

How does one know the impedance of a power supply output?

 

Looking forward to seeing the finished version!

 

Thank you.

Most simple regulated DC supplies are designed to have as low an output impedance as possible, since they are designed to maintain the voltage with varying current demand. In the case of the DHT filament supply, the current demand is pretty constant (once the filament comes up to temperature), so the only thing that is important is to have a low impedance at DC, but ideally a higher impedance at audio frequencies so that it has minimal affect on the cathode voltage as the audio currents fluctuate.

So this board is designed to ramp up the current until the voltage hits the reference point (5V in this case), and then maintain enough current to keep it there. A secondary advantage of this is that the filament is slowly heated to temperature. With a simple voltage regulated supply, the current would spike on turn-on since the filament has lower resistance when cold. So this also helps to put less stress on the tube.

The Tentlabs supply is spec'ed at 10Kohms at > 5 Hz. Many power supplies will include an output impedance spec, but a typical regulated DC supply will have an output impedance in the low milliohm or even microohm range.

 

The last "extra" board I am adding to the design is the Neurochrome Intelligent Soft Start board. This is sold fully assembled. It connects between the mains power (after the rear-mounted power switch and fuse) and the mains transformer. I'm using this board on my current preamp project and I like how easy it is to use and how well it works.

When an amplifier is first turned on without a soft start, there is a high current surge to energize the transformer and charge the power supply capacitors. This puts a lot of stress on the transformer and rectifier, ultimately decreasing the life of these components.

A simple soft start can be implemented by just inserting an NTC (negative temperature coefficient) thermistor in series with the mains connection to the transformer. This is basically a resistor that starts out as a high resistance, but becomes a resistance as it heats up. So it gradually allows more current to flow into the mains transformer. The problem with using such a simple approach is that the thermistor will always have some resistance, and also takes some time to cool down again so if you turn off your amp and then turn it back on before the thermistor has cooled down, you will no longer have a "soft start".

A more sophisticated soft start, such as the Neurochrome board, will connect the ouput directly to the mains using a mechanical or solid-state relay after a predetermined amount of time. This eliminates the voltage drop and allows the thermistor to cool back down so that it can do its job the next time the amp is powered up.

The Neurochrome board adds some additional nice features such as a low voltage (3.3V) connection to a power switch (either toggle or momentary), a dimmable standby and power indicator connection, and a 12v trigger input. I am using a momentary contact anti-vandal switch as shown in the photo below. I will probably set it so that the ring illumination on the switch is fully dimmed when in standby, so the switch will only be illuminated when the amp is on. This switch will be mounted on the front panel of each monoblock.

The board also provides an optically isolated 12V trigger input. I was assuming that'd I'd probably not use this input, but now I'm thinking that I might wire the trigger inputs up to my preamp so that I can turn the system on using the preamp remote.

 

I have to admit I usually lazily bung in an NTC thermistor and call it a day!

I'm looking forward to seeing how you incorporate all these boards into your chassis layout. Keep 'em coming!

 

A bit confused that you are putting in a speaker protection board. I thought as valve amps don’t pass DC they are safe for speakers. What is the reason you are fitting them?

 

Yes, you are correct. The output transformer will block DC. The reason I'm adding this board is that the ANK design generates some hum and buzzing on the output during the first 10-15 seconds on power up. I'd rather not have this coming through my speakers. The speaker protection circuit will also quickly disconnect the speakers when the amp is powered down in case there are any transients generated by the amp as the various power supply voltages go to zero.

 

The illustration below indicates the general layout that I am thinking about. This is oriented as if looking down from the top. Since the transformers mount on the top of the chassis, circuit boards and other components can be mounted underneath the top plate as long as they don't interfere with transformer mounting holes and through-holes for wires. I may need to tweak the positions a bit after I've had a chance to review and measure the transformers once they arrive.

I am going to rearrange the position of some components compared to the standard kit. I need one less large capacitor since I am replacing it with the Maida regulator. Since my chassis is a bit larger than the kit, there is enough room between the mains transformer and choke to mount the caps there. But the main reason for the rearrangement is that I want to use a cover over the mains transformer, choke, and power supply caps, and I'd like to have the rectifier tube outside this cover.

This arrangement is similar to the current Audio Note Conquest monoblocks. The rectifier will now be in front of the mains transfomer and choke, centered just behind the c-core transformers. This area slso gives me a place for some venting through the top plate (left and right of the rectifier socket).

The grey area around the back and sides indicates the back and side panels. The interior dimensions of the enclosure are 280mm wide and 550mm deep.



All printed circuit boards are mounted to the underside of the top plate using threaded studs. I also need to leave room in a couple areas for tag boards to mount a few components. This will be required near the input (for the transformer dampening circuit and input loading resistors) and near the power supply (for the voltage dividers to set the input tube heater voltage reference). There will also need to be places to mount some additional capacitors I plan to add between the B+ and 300B cathodes (will be explained in a future post), so these will probably be placed to the left and right of the copper plate.

If you look at the photo I posted in the first post, you'll see the copper plate which has most of the actual audio circuitry mounted to it.

 

I've got my first enclosure panel back. I started with the backpanel, since this is the only one that's likely not going to be affected by my final measurements of the transformers and other mounting hardware which I have to wait to receive the kit to take. Brian hasn't sent me the tracking number yet, but I'm hoping it will actually get shipped this week.



Even though I am building a pair of amps, I like to make one panel at a time in case I have to make some modifications to the design. For example, on this panel I forgot to adjust the RCA connector hole size for the connector I chose. The WBT connector requires a hole a bit bigger, so this connector is installed as a placeholder until I can drill out the hole.

There are two XLR inputs. The second connector, labeled High-Pass, is connected to the first one through a pair of Miflex KPCU-2 capacitors. Since I will be using these amps with speakers that use subwoofers for the bottom octaves, this reduces the load on the amp (and the speakers), but reducing the low frequency information handled by the amp. My (very rough) hand-drawn schematic is shown below.



The transformer is a Jensen JT-11P-1. The resistor/cap network on the output of the transformer is a dampening network to prevent resonances since the input impedance of the amp is a lot higher than the 10Kohms the transformer is optimized for. A toggle switch is used to select between the transformer-coupled balanced inputs and the RCA input.

 

I really like your back panel.

Have you ever had an opportunity to directly compare standard binding posts with GR-Research's tube connectors? I know many here will scoff at them making any difference. Hell, I would probably have been one of them, but I saw/listened to a YouTube video by New Record Day where he recorded a pair of speakers while swapping between the two types of connector. Hearing that made me far more open-minded.

Another thing I was wondering about was if you had any opinion on delaying the turn-on of the B+ supply until the heaters have warmed up. Some seem to swear by it, while others say it doesn't matter. Or is that not even applicable for directly heated triodes?

 

Thank you.

I haven't done a direct comparison using the same type of cable with high quality spades and binding posts vs male/female tube connectors. I have built other amps with both tube connectors and binding posts, and found the speaker cable with male tube connectors sounded better (a bit more detail and focus in the soundstage) than plugging the tube connectors into binding posts that were banana-plug compatible, but I don't think this is the optimal interface with binding posts.

For the price, the tube connectors are clearly better than the cheap binding posts you could buy for the same money, but I suspect a high quality binding post (plated solid copper) and high quality spade would probably delivery similar results. But I've been very happy with the sound of the tube connectors using speaker cables terminated with the male tube connectors.

The downside of the tube connectors is that you lose some flexibility (can't use with spades or bare wire), and they don't grip as well as spades or locking bananas, so if your speaker cables are in any danger of being jostled, there is much greater risk the cables will pull out.

First, I need to stress that I am a tube neophyte. This is literally the first tube project I have ever done.

While I have an electrical engineering background, I haven't done any circuit design as part of my job since the 80s and then it was almost exclusively digital circuits and chip design. So I am learning along the way.

I've been reading up tube circuits in the past six months, and when I decided to do this project about 6 weeks ago, I started digging into what could be done to improve what I believe is already an excellent sounding amp.

That said, I have spent a bit of time researching the B+ supply vs heater supply timing. From what I've read, it appears that this is a bigger problem with tube circuits that run with a much higher B+ (>1000V). The current spikes on heater start up are more likely to cause long term damage, and I have addressed this with the soft-start circuit and TentLabs supplies. I think the combination of the softstart, relatively slow rectifier, and 40ohm series resistor on the output of the rectifier, will slow down the ramp up of B+ enough to minimize any problems. Hopefully I'm not wrong.

 

I finally received the tracking number for the amp kit from ANK. It was shipped today and is expected to "scheduled" to arrive on Wednesday. But with all the logistics challenges these days, I would not be surprised it was delayed a day or two (or more). Now, on to the main point of this post.

Besides the changes to the filament supplies and input stage power supply, the only other change I am making to the actual audio circuitry is the addition of capacitors between the B+ supply and the 300B cathodes. I believe this "trick" trick was used in the early days of SET amps when power supply caps were much smaller than today, but it is described in the first issue of the Tube Cad Journal from 1999.

Since this amp (and most SET amps) does not utilize any negative feedback to improve power supply rejection, any noise on the B+ rail is passed directly through the output transformer to the speakers, although it is reduced by the transformer winding ration which I believe is around 150:1. That is, before adding these additional capacitors.

The idea is to inject a small amount of the power supply noise into the output stage such that the noise is nulled from the output. The optimal amount will result in the same noise on the plate of the 300Bs as on the B+ connection to the transformer, resulting in the same noise signal on both connections to the transformer primary which will result in no noise on the output.

Since each 300B tube amplifies the cathode signal, the amount of noise injected at the cathode must be adjusted for the gain of the tube. If you want to learn more about how this works, take a look at the tubecad article I linked to in the first post. But the bottom line is that the added capacitor value will be equal to the cathode capacitor value divided by the mu of the tube. In this case 220uF / 3.9 = 56uF. From what I have read, this approach reduces power supply noise on the output by approximately 20db.

Since this capacitor has to handle fairly high voltage (I'm using 450V), I didn't have a lot of choices for high quality caps that were a reasonable size and cost. I picked some Nichicons that look fairly good, and I am bypassing them with Dueland copper foil 0.01 bypass caps.