APRIL 2009 - PART 1


Hi Peoples. 

I really must tell you this story from years gone by. From the age of around nine, I was first introduced to valves (tubes), and at about age eleven I first clapped-eyes on an 807 tube. Now, please don't condemn me here (some people collect stamps, do train spotting etc!), but…..I was fascinated by the shape of the 807. Why, I don't really know the answer to that! I remember distinctly back then in 1961 (I am 59 now…..born February 1950) that I wanted to build an amp with 20 of these in the output, for some strange reason, thinking how magnificent this would look! However, I never really did it. I have built many amps over the years with just a pair, a few with a quad, and just one with an octet (please see this article on the CBA-225 ).

Now, some 48 years later, I have actually gotten off my arse and made a real start on it! One important factor here is the original tube data sheets. A lot of people out there might not actually be aware of this, but all tube data available….for any tube is always expected to be at the given voltages of the said data sheets. This basically means in English…….stabilized!!

Apart from very few acceptations, most Hi-Fi amps and guitar/musical instrument amps don't actually do this! The basic amp generally runs the output stage voltages considerably higher than is actually required then, when the amp is under full load conditions, the output power is now achieved as the HT supply drops to around the desired data sheet figures. In almost all cases this is fine to do without any major or serious damaging consequences. However, in reality and again in most cases, the output tubes are running unnecessarily hard under no signal conditions. Most tubes handle this quite well though without any complaints.

Now, welcome to the world of regulation! One simple way to achieve voltage regulation in high voltage tube equipment is to use a choke input filter….this is very reliable and successful but…..the choke is usually about the size of the power transformer and adds a massive amount of weight and extra size to the said equipment, which is not always practical or cost effective. This is why most amps don't have this. As we now live in the "solid-state" world this has become so much easier to do, and we can use something like a 1,600 volt / 25 amp IGBT and get the regulation we require, quite easily with a small chip to control the IGBT. This though deviates away from the tube world and, although this kind of regulation is in no way in the "audio signal path" it still nevertheless is "solid-state" at the end of the day!

Now, as it happens, around this time, my good friend John Wood was in the process of designing a regulated power supply for a Hi-Fi amp that he had also designed and was already in prototype form……sounding really brilliant! He was explaining to me about the amount of current this regulated power supply was capable of with a pair of PL509 tubes. His prototype Hi-Fi monoblock was 200 watts with four KT88’s. This being the case I asked him if he would mind me using his idea for my 1000 watt monster. John duly replied "no problem" and kindly made me the two PCB boards that I needed for the job.

When I got down to the nitty-gritty of designing this humongous monster and having decided to use John’s regulated power supplies for it. John’s plan in-fact was to use tubes (valves) for the main regulation and a chip plus a small high voltage Mosfet for the control of the said valve(s). On data sheets for the 807, and on not pushing them into AB2 (which would have required many more tubes for the drive and more complicated circuitry too), I decided to simply do the amp in AB1. Under these conditions with 750 volts plate and 300 volts screen (grid 2), 12,000 ohms plate-to-plate, fixed bias, we can expect from these data sheets 72 watts for a pair. However, under real-life test conditions with just a pair of tubes, John and I found that this was actually closer to 82 watts and seemed to be quite consistent too from various manufacturers of different tubes. There is also the added advantage that we could squeeze the plate voltage up-to 800 volts and the screens (grid 2's) up-to say 325 volts. This then means the pair of tubes could possibly make the 100 watts, even in AB1!

After all this numerous testing on just a pair of 807's at the suggested previously mentioned voltages, I accepted the fact that in AB1 the amp should really be 720 watts for 20 tubes, 820 watts is likely without any tweaking but……a 1000 watts is certainly possible with the said tweaking!! I will let you know the final results at the end of the project.

Moving back to the regulated power supply(s) issue, as previously stated these were designed by my good friend John Wood. There are quite a good few hefty tubes out there that can be used for voltage regulation but for many of them it is not always too practical. What John had ended up with was the PL509. This also includes many of it's counterparts/equivalents, PL519, PL505, 40KG6, EL509, EL519, 6KG6 etc. All these tubes have around a 500+ m/a cathode current and are readily available out there in many different variances. Having settled on this family of tubes for the job we set-out doing loads of experimenting as, if my 20 x 807's amp was expected to reach the 1000 watts, we would have needed a 750 volts supply capable of around 2.4 amps for sine-wave drive. (About two-thirds of this in reality on music power.) In actual "real-life" testing we achieved the full 2.4 amps from 4 of the PL509 tubes at 750 volts, and the supply dropped only .1 of a volt throughout all the testing……...bloody wicked!

So now having achieved all the various "real-world" testing, the plan was put into action. The final amp will have: 20 x 807's for the main output of hopefully 1000 watts (or thereabouts); 4 x PL509's for the main regulated 750 volt (plates) HT supply; 2 x EL36 (or any of their family equivalents) for the regulated screen grid (grid 2) supply (about 300 volts); and 4 x 6SN7 for the pre-amp, tone circuit, splitter and push-pull driver. Thirty tubes all-told……….absolutely cool!!

As in all other projects and repairs on the site, I have always added the relevant page to the website when all is finished and working fine. For once I thought it might be a good idea to present this latest crazy project of mine in stages……as an on-going thing until it is finally finished. I though by doing it this way it would attract a lot of continued interest in the project.

I have done all photo captions/information and comments as always.

Big thanks for dropping by and as they say, "watch this space" for Part 2!

Cheers, John.

As I have said before with a new project……here is the initial idea! I have already used these tag boards before with 807's, and the 24 pairs of tags on each board work out perfectly for all the components to do this. Each board drives four tubes (two above the board – and two below, though the tubes below are not shown here). Knowing I would need five boards for the 20 tubes, here I am getting the basic idea of chassis size. With all the boards laid side-by-side and the tubes on a sensible 3" spacing in all directions, everything worked out perfect for a 31" wide chassis.

The five drive boards fully built-up ready. In fact, the component configuration works out that well on these boards that they can be turned around and still be correct. In other words each half is a mirror-image of itself.

Close-up of just one of the five boards. I thought all of you out there might like to see this layout close-up, in case any of you would like to build/copy this.

The underside wiring of the same board, again in-case anyone out there fancies building them. This one board will drive four of the 807's. The un-sleeved wires are the ground (buss) rail. The red links are for the four screen grids' 300 volt supply, through the screen grid resistors which are also used as a link from this board to the tube bases. The four blue links are taking the output of the bias pots' wipers to the grid 1 return resistors. From these said tags the 4 x grid 1 stopper resistors are also used as links from the board to the tube bases. The yellow links are the negative bias-in. The two white links are the in-coming push-pull drive. Lastly, the four tubes' cathodes are connected directly to the four on-board 1 ohm, 1% resistors.

The simple reverse-bias diode board, using 8 x UF (ultra-fast) 5408's in two series lots between the output tubes' plates and ground.

With using regulated power supplies for all the main HT supplies, it would be pointless / stupid to do this and leave the bias supply un-regulated! As the AC inlet power fluctuates, the bias would then drift whilst the rest of the supplies would remain stable, hence the regulated bias board shown here. I need about -35 volts for the tubes' bias. I also, as always, intend to run the four pre-amp/drive tubes heaters in series and on DC (50.4 volts). There are five 12 volt, 100m/a, 3" cooling fans, also in series (60 volts, but these run fine on 50 volts… a nice gentle air-flow). And finally there is the soft-start and bias protection relay. What I have simply done here is to use 2 x positive 24 volt, 2 amp regulators in series and with a IN4007 in each ground leg. This just creeps the voltage up a notch to around the 50 volts, for running all the aforementioned. The total current pulled is around 300 m/a and the heat-sinks shown in the photo are fine for this. However, a little later on in the project I decided to change the four pre-amp/drive tubes from ECC82's (originally planned) to 6SN7's…..looking better & beefier! The problem with the heaters on these though is that they are 6.3 volt only, and at a 600 m/a as opposed to the 150 m/a of the ECC82. I have ended up putting the four tubes' heaters in series (25.2 volts, 600 m/a) and these will now be connected to the first regulator only. On bench testing this up-front, the heat-sink on this first regulator proved to be insufficient with the now 600 m/a+ load. I have simply removed the regulators from this board completely, and they will now be mounted on a separate substantial heat-sink……ribbon-cable connected back to the board. This will be evident on the photos coming in Part 2…..soon. Oh yes, and just in case you spotted the fact that I used "positive" regulators in this supply….you just simply swing the outputs around for the negative!

The very simple underside wiring of the bias board. Realistically, this is just two independent 24 volt regulator circuits stacked on top of one-another.

Preparing and checking the cooling fans plate.

Laminating the massive power transformer. I have always found that when taking photos of transformers, it is always difficult to scale the size. I think my tobacco tin helps a bit and the slogan on it is very apt here! There are sixteen secondary windings on this beast (many extra windings are required for the two regulated PCB's) and I calculated the windings using the formula for low-grade laminations. Then I loaded it with M6 grade. This really makes the transformer very efficient and "over-wound". It will also run cooler too.

Just tying to scale the transformer again. The PL519 & 807 tubes in the photo are standing on their pins; which makes them look taller and deviates away again from the true size of the power transformer. The PCB in the lower/right of the photo is the regulator board for controlling four tubes……the PL509's for the adjustable 750 volt plates HT.

Right-hand front view - carefully checking and marking everything out. The output transformer is not shown here as at this point (April 2009) as I haven't wound it yet.

Left-hand front view. Same again……checking and planning everything out.


And now…..the laborious bit……drilling and hole-cutting of the 2mm mild-steel chassis. Please stay tuned for Part 2…..coming soon. I will also be providing the schematic at some point in time. 

Many thanks indeed for looking, John.




Click HERE to view Part 2 of the Champ CBA-20807 Project