813 PP Part 1
I’ve wanted to build a power amplifier for ages but until now, the EAR 509 has /not been considered a weak link. However I’ve always liked the idea of a pure class A triode power amplifier but the problem has been most do not have the power to drive a full range ESL. The EAR sets a high mark to beat too!
To begin with, it was more or less a copy of Gary Dahl’s Aurora. Lynn Olson’s discourse on push pull amplifier design is enlightening. Then he proposed an all PP/differential type amp which was more in line with what I wanted.
300Bs and full range ESLs do not make a good mix and the 300B is way too “popular” and very expensive. I had heard good things about the 813 and unlike 211/845, can run well from lower HT, requires lower anode-anode resistance due to it’s low Rp and is easier to drive than either. I was able to pick up 4 813s circa 1943 quite easily and cheaply. And of course these big transmitter tubes are fantastic looking when lit up. The GM70 is probably another good choice and is a pure triode.
A good read is Western Electric – A Rosetta Stone for Triodes by Lynn Olson. In this and then his Aurora article, the why’s and wherefores of this design are explained. Finding Common Ground is worth it too! I hope in time to make the design my own though as usual, there is little new in the world of valves. Another huge inspiration is Sakuma san’s Direct Heating site – his work and design approach.
Grand Aurora Mk1
The Mk1s were going to be monoblocks; using one supply for output stage and resistors to drop HT for second stage. In practice, this didn’t work very well requiring heroic efforts to keep the resistors cool. The amp couldn’t run for long but it provided enough encouragement to further the design and refine the construction.
the guys beforehand, overall view and side views of the audio bit – first amp
Grand Aurora Mk2
Looking at the iron and glass round these parts made me realise a better design could be had if a mono supply was used with seperate supplies for the output and driver stage. I’ve noticed a certain something, a musical seemlessness using a stereo amp compared to monoblocks in a number of systems over the years and this amp has that too. The one problem was going to be its size and weight. Never mind, build it in situ. The chassis is 80cm/2ft 8in square and I can’t lift it. Some thoughts occured to me while building this boatanchor.
- you can sit IN the chassis
- you walk to the other side – don’t forget to take your tools
- sometimes the soldering iron lead isn’t long enough to get to the other side
Thanks to the guys on the Boatanchors Amp Club for the inspiration to get going on this thing.
I was told and kinda knew the SN7 first stage would not have the required gain nor would it drive the IT properly but hey, it would go – it did. However gain was low so popped in triode strapped D3a for mu=75-80, rp=2k or so. Although the amp was good, it didn’t quite have the authority I was after; compared to the EAR509s which set a high bench mark in this area and the bass, so tried the LL1671 as driver IT – much better.
At this point, tried the ultrapath cap connection on the driver. That improved things all round opening up the sound and dynamics however trying the same on the input did virtually nothing.
Next up tried direct coupling the input to the driver. There were some benefits but undecided overall if this works best with the current tube compliment. The shunt regulated driver was also dispensed with so unsure as to whether its loss contributes to the disadvantages or whether its the use of the IT in this manner. It’s possible that the character of the input stage is being heard more. Time and more fiddling around will hopefully make this clearer.
The gain structure of the system would mean being limited to a small group of valves to try especially in the front end. Therefore a line stage addition to the preamp will be needed rather than having a super sensitive power amp.
side view, day and nite glow and Mk1 driver transplanted into Mk2
Having made the new line stage, was now able to try the ECC99 as input however it didn’t work very well direct coupled using the LL1635 as a choke. After a lot of mucking about, arrived at the following schematic which let’s face it, is only a ECC99 transplant of the original 6SN7 minus the shunt reg for the driver.
Now I had an amp that was a big step forward and something I could develop further. The shunt regs on the driver were removed so that I could try to see what the effect was when adding them back in.
It sounds very powerful, dynamic and dramatic while staying rock solid and smooth. Despite having so many transformers in the signal chain, many people would think it to be lacking however it sounds very extended at both ends and certainly the equal of the EAR in this regard. Probably something to do with modern transformer technology not being as limiting as it has been in the past. The pleasure so far is listening to how it presents the music and I look forward in getting to know it over time.
Now the basics are done, progress will be in refining it further. Given that it’s early days, the potential seems huge. In the pipeline is direct coupling input to driver (again) but with the proper choke load, LL1668 and trying triode strapped DHT 46 as driver and re-introducing the shunt regs for the driver.
Audio Circuit Design Choices
The preamp journey got me thinking a lot more about the “little guys” that do the work, electrons, rather than volts and amps; their representatives. That translates into looking at the AC loops and reference points, keeping them clean, short and as unambiguous as possible. Although the amp was developed from the output backwards, as any amplifier should be, the description will start at the front.
So starting at the beginning, the input tx splits phase accurately as possible. For this I chose the LL1544a as I’d heard good things about amorphous core transformers and had seen others use it with good success. As the preamp now has a balanced output, I could have split phase with a resistive divider across the grids with the center at ground. However that means, well in my head anyway, that the grid circuit effectively extends back to the preamp including the cabling which could pick up noise and RF. The common mode rejection capabilites of the input tx helps reduce this noise and keeps the grid circuit short.
The input tube, ECC99 again has a good rep. Could have used the 5687 but already using one of them in the pre. It drives the LL1635/5mA IT; selected as it can take a small imbalance in current. To begin with, I’ll stick with transformer coupling coz it sounds good and it allows more variation in the driver with the PSU options available at present. Again, it keeps the input circuit and earthing for the driver neat, short and simple and completely seperate from the input circuit. I kept the VR shunt tube and as yet didn’t audition it’s effects compared to RC decoupling plus it glows nicely.
The driver to start with is a 6BX7 as I knew it would work well enough. Only using a single valve so limited to about 17mA or so at 250V as the whole valve can only dissapate 10W max so 5W max per section. That doesn’t seem enough current to drive the 813s optimally although it seems to do a decent job. It is the limiting factor so far therefore something that can run at double that will be tried next. As a much higher current driver was planned and as Kevin Carter had used the LL1671 successfully in his amplifier, it was an easy choice.
The output stage proved to be the easiest part of the design and offered very few choices. Paul, a fellow LLDIYer had been championing the 813 on AA and as I’d heard his lovely sounding amplifier and system, the op point of about 880V or so at about 100mA seemed an easy choice. The Sowter U076 was designed for the 813 and had seen it used successfully. From the curves, that looked like about 75V. 75v @ 200mA set the bias resistors of 375R (120 + 150 = 370) and the HT at about 955V. In practice it turned out to be 80V @ 220mA with the 370R bias resistor.
Power Supply Design Choices
The power supply has to provide a wide range of voltages so it was inevitably going to be heavy especially as I was keen to use big MV rectifiers having found a pile of GU20s going cheap. As I was going to build monoblocks, had at hand 2 MV rectifier heater transformers so the front end supply could use them at no extra cost for starters. The design came about due to the iron I had lying about and the desire to keep the channels seperate despite using a single HT supply hence the chokes feeding each channel. This enabled further ripple reducation along the way.
One gamble was the choke fed output stage without an additional decoupling capacitor. Time will tell whether that works properly but it seems to sound OK.
In PSUDII, I aimed for about 1V of ripple hence the 220u filter capacitor as simulation of the EAR supply seemed to produce a similar figure and it was very quiet so proving the hum cancellation properties of a PP output stage. Most would have chosen to use something other than electros in this position but given the individual choke fed O/P stages and the cathode decoupling, I figured these caps were not in the AC loop.
The front end transformer has many taps (260, 300, 340, 380, 420V) so I can fiddle with a variety of supply voltages. That plus the use of VR tubes enables many options for experimentation. Wish I’d done the same for the ouput stage HT.
So what Now?
Last summer we experienced some long periods of hot weather and as can be imagined this amplifier puts out considerable heat. During one listening session with a friend, we had to listen shirtless with the windows open. OK it sounded great but it was uncomfortable. This and my change of speaker that requires some 20dB less power meant this amp was too over the top and totally impractical. To reclaim some of the floor real estate in the listening room, I pulled the amplifier to bits. I thoroughly enjoyed listening to the amplifier and will revisit a similar design some time in the future. Next time it will be more practical – physically and electrically.
In hindsight, this project was something that I had to do – to prove something to myself. In many respects I couldn’t recommend that anyone copy this design as is.
Ian’s 813 PP amp
A few years ago (2004), I starting corresponding with Ian as he was interested in building a design similar to the above. Then in 2005 he sent me photos of the nearly completed design. I was to say the least, stunned. I asked him recently (late 2007) how it was going and whether I could post his design so here it is. It makes me want to build mine again in a more sensible fashion. Inspiration indeed!!
Inside view, first one going, front and side view
Audio Schematic
Driver Stages PSU Schematic
O/P Stage PSU Schematic
From Ian …
Yes, I am still running the 813 amps which are performing beautifully. I use them several times a week and have performed faultlessly for the past two years.
I used a linear DC regulated power supply’s for the 813 valve filaments. In the future I will abandon them in favour of the much simpler centre tapped transformer arrangement providing 5-0-5v AC as you used in your prototype.
I originally used 813’s with pressed steel anodes but have subsequently obtained graphite anode NOS items. I cannot really say that I can tell any difference in the sound quality.
The aesthetic design I chose was a mixture of old and new. As you can see I have provided L.E.D. peak programme meters on the front panels. This was merely a question of personal taste.
In total honesty I have never regretted building these monster amps as they out-perform anything which I have previously built and they look great! As with yourself it was something I just had to do. I listen to both CD’s and Vinyl through the amps and just love closing my eyes and listening to the nearest thing I know to a live performance.
Many thanks for the inspiration!