Iron & Glass.
Audiophile quality, hand built, stereo and mono block valve amplifiers.
Stereo Valve Amplifier Project. ( March 2016 ).
A long time ago, in a galaxy far, far away, I spent a few years working for a company called Q.C. Electronics as a wireman building HIWATT guitar valve amplifiers. Pat Gorman was the owner of Q.C. Electronics, and he took me on as a spotty 17 year old with attitude. He was the best boss I’ve ever had, giving me chance after chance to become reliable. In the end he was successful and I became a reasonably useful part of society (Thanks Pat). At that time, the name HIWATT was synonymous with quality, due to the owner, Dave Reeves, insisting on the highest quality components and an incredibly high standard of assembly. Sadly the quality dropped with Dave Reeves’ untimely death and licensing problems but the company has been rejuvenated and is now back at the forefront, building these amps as they should be built.
I didn’t know how these amplifiers worked. I just routed the wires and soldered them in the neatest way possible. Cable routing and neatness are essential in valve amplifiers as there are high and low voltages, A.C and D.C, plus highly sensitive signal cables running everywhere. A carelessly built valve amp will hum and buzz as signal leads pick up AC noise from nearby cables and transformers.
Recently my interest in valve amplifiers has been rekindled, I wanted to build them again but this time I wanted to understand how they worked.
Well, eight months have passed since I started my foray into the world of valve amplifiers on the internet and I’ve learnt a lot. I could bore you with the details, but I won’t. Suffice to say, it was mainly frustrating, interspersed with lightbulb moments when something sunk in.
Having no inclination and certainly no ability to learn the guitar at this late stage in my life I decided against building a guitar amp. I can’t play music, but I do listen to it all day long, so I decided to build a stereo valve amp. I have a pair of Mordaunt Short speakers which are reasonably good quality and a large collection of music on the computer. I decided to retire my NAD solid state amp and replace it with a valve amp.
Now, thanks to many hours on the internet, I have enough savvy to build and test a valve amplifier, but nowhere near enough to design one. There are hundreds of valve amplifier schematics on the internet and groups of people willing to help you. So after asking advice from many people I was steered toward a Low Power, (2.7Watts per channel), Class A, SET, (single ended triode), Stereo Amplifier designed by Matt Renaud. What I wanted was an amplifier that could reproduce recorded music 100% accurately with absolutely no hum or distortion.
I found the schematic and read up on it, then I got serious. A large amount of money later I had everything I needed. Transformers and Chokes from Edcor in the U.S.A., Valves from Watford Valves in England. All other components from different companies within Europe, always looking for the highest specifications.
I wanted the components inside to be hardwired, no printed circuit boards, so trying to find paxolin board and turret lugs to make my own turret boards took some time.
As well as the sound quality, I wanted a case design that would really show off the valves and transformers but hide the rest of the electronics. I also wanted a more old time look. Something that looked more like early valve equipment. So, after looking on the internet yet again, I found designs that could be modified using alcoves for the valves and transformers and panel ammeters for monitoring the amplifier. (These meters read cathode current, they don’t move with the music like VU meters). I decided to build the case from MDF and cover it in brown leather, trying to get a look similar to the old brown bakelite cases. As I work with leather for a living it seemed the logical way to go.
I bought a sheet of polished aluminium to make the bases and tops of the alcoves. I could mount the valves and transformers on the alcove bases. The aluminium could also be used to make heat vents in the top of the alcoves, and also in the base and back of the case.
HIWATT at its best in the seventies.
Photo 1: Input phono sockets and fused 230V mains socket. You can use a smart phone, an iPod or similar music player, a computer or a CD player directly into the inputs. A microphone or magnetic cartridge in a turntable would require a preamp.
Photo 2: Starting at the top. Left and right channel output transformers and two 6EM7 double triode valves. In the middle. Left and right channel ammeters and volume controls. Bottom left: Power transformer. Bottom right: On/Off switch and valve rectifier.
Photo 3: Overall shot of inside of amplifier. Note the red and black twisted pair of wires on the right. These are the signal wires at their most sensitive, so they are kept well away from all other cables. Everything in the bottom of the case is the power supply. The power transformer and valve rectifier are in their respective alcoves. Three large chokes and six electrolytic capacitors smooth out any noise that could be picked up and amplified. The long alcove at the top houses the valves and output transformers. The turret board holds the electronics that run the valves.
Photo 4: Underneath the power transformer and rectifier alcoves, all high and low A.C. cables are pushed toward the corners of the case to ensure maximum distance away from signal cables.
Photo 5: Handmade turret boards allow for neat layout of wildly differing component sizes. Also handy when testing and fault finding as all connections are easy to see, unlike printed circuit boards.
Photo 6: This tag strip is where power supply meets amplifier. I’ve used a junction point like this for my own safety. On this tag strip I can check most voltages, it requires a steady hand and no distractions. Note the voltage on the large blue capacitor.
Photo 7: To make sure I didn’t blow my speakers by inadvertantly pushing two or three hundred volts D.C. through them, I built a pair of 8 ohm load resistors with heat sinks as speaker substitutes. I connected up and switched on. No bangs, sparks or smoke. I checked voltages across the tag strip whilst shaking a little, again all seemed fine. Both ammeters had the correct readings, it looked like I was home and dry. I substituted the load resistors with my Mordaunt Shorts, plugged in my computer to the inputs, put on some Stevie Ray Vaughn and wound up the volume……NOTHING…not a whisper of sound, not even a hum or buzz. Two days later after a few transatlantic emails to the circuit designer he had helped me find the fault. A small adjustment to the turret board wiring and everything burst into life. The sound was beautiful. Clear, precise, crisp…. I could talk about it for hours without being able to describe it accurately.
Photo 8: The amp with the Mordaunt Shorts, sounded beautiful but didn’t look too good. I knew this would be a problem before I started the project but had convinced myself I could live with it. Well, it turned out I couldn’t. I had read a lot about Fostex full range drivers whilst reading up on this particular valve amp. Because the amp is low powered (somewhere between 2-3 watts per channel), Fostex drivers are very well suited to it. They’re not ridiculously expensive and they supply all the data required about cabinet size, reflex port length/diameter and acoustic damping material. I found a 6½ inch driver that had a specified cabinet size that was close to the height of the amp and ordered a pair. Then I had to find and buy reflex ports, acoustic damping and speaker connection posts. Finally, a trip to the local carpenter with a cutting list for the MDF for two cabinets. Once everything had arrived I was able to construct them over two weekends. I tested them out over the following week. The midrange and top end on these speakers were far superior to the Mordaunt Shorts but although the bass was crisp, it lacked something. An 8 inch subwoofer run by a medium powered plate amp solved the problem. I listened to this system for two weeks and couldn’t find any flaws with it so I went ahead and covered all three speaker cabinets in leather to match the amp.
I first posted a version of this design in the forum in September 2011 and then in a slightly improved version in December 2013. But I wanted to make sure that the design was really what I wanted. So I decided to go back and revisit the whole design process. The first step in this process was to review and update the power stage design.
The power stage in this amp is formed by the "Unit No. 2" triode. This triode has a peak plate dissipation of 10W meaning that in a SET design we should be able to get at least a couple of watts very cleanly. Below is the revised load line design for the power stage based on the plate characteristics from the RCA data sheet dated August 1960.
6EM7 Plate Load Line Design - Triode 2.
This 6EM7 design is very similar to the one presented back in December 2013. The only differences is that the cathode bypass has be returned to the original 100µF value. This was done solely to provide a more solid bass response. Although the bias excursion recovery time constant is a little longer than I usually prefer, I will correct for this by using a 10kΩ grid stopper resistor on the power stage to control excursions.
I am very comfortable with this amplifier design. The power stage plate dissipation is down at 9 Watts allowing for some design to build variability. The low end frequency response is good with provisions for controlling bias excursions. And finally, the distortion is well controlled at 2.7%/W (almost entirely 2nd harmonic). I could have gone with a lower load impedance and gained a little bit of power, but the distortion would have suffered. Overall the 5kΩ load impedance is a very good sweet spot for this triode.
So what about the 6EM7 driver stage? Well, the power stage is biased at about 35V. This means that the driver stage will need at least this much output voltage swing and it will need to do it from a line level input. This means a gain of at least 25dBV. This really isn't too much of a problem for the section 1 triode of the 6EM7. The power stage calls for a B+ voltage of approximately 260VDC. So I decided to use a 250VDC B+ for the driver stage. This will give me about 10 Volts of drop for an additional stage of B+ filtering. Below is the final load line design for the 6EM7 driver stage using the same data sheet referenced above.
6EM7 Plate Load Line Design - Triode 1.
The other thing that I really wanted out of this design was excellent channel separation for a deep and wide sound stage. This requires that the power supplies for the two channels be well isolated at their output to the circuits. To achieve this I decided to use three chokes (one main filter and one additional one for each channel) and then provide the gain stage B+ from an additional filter stage after the B+ supply filter.
Please note that the capacitor immediately following the 6CA4/EZ81 vacuum tube rectifier must never exceed 50µF.
This power supply design puts ≈75dBV of B+ filtering between channels at 20Hz. At the primary ripple frequency (120Hz) the B+ channel to channel isolation is ≈165dBV and by the time you get to 1kHz the isolation is a (theoretical) ≈276dBv. I say theoretical because at these levels just putting the wires in the same chassis causes more coupling than this. Regardless, this is a VERY well "channel-to-channel" isolated power filter. The use of the 390Ω dropping resistor and the 100Ω rheostat was to trim the power stage B+ voltages to exactly +260VDC.
Warning: This vacuum tube amplifier project uses high voltages. Contact with voltage potentials of this magnitude can cause serious injury or possibly be fatal. If you do not know how to build high voltage vacuum tube projects or you are not comfortable with projects that use these voltage levels, it is strongly recommended that you do not build this vacuum tube amplifier. Follow your governing electrical codes for all wiring and connections.
For those of you who wish to know more about the amplifier, I have included the text and diagrams from the amplifiers’ designer Matt Renaud as shown on his forum. It reads as follows:
Photo 9: Fostex 6.5 inch Full Range Driver with 2.5 inch bass reflex port. Once the pair had been built and tested I covered them with leather to match the amp.
Photo 10: Dayton Audio 8 inch Subwoofer. I built this driver into a 22mm thick MDF cabinet with internal bracing with dimensions supplied by Dayton Audio. Again, once tested this cabinet was covered in leather to match the amp.
Photo 11: Dayton Audio Subwoofer Plate Amplifier. This plate amplifier powers the subwoofer, allowing me to choose what frequencies can be amplified and by how much. It is fitted into the rear of the subwoofer cabinet.
Photo 12: 19mm thick, 100% Herdwick Sheeps Wool Felt from QTA Systems was used for all acoustic damping inside the three speaker cabinets.
Photo 13: Everything tested and working perfectly. All leatherwork complete.
Photo 14: For the final cosmetic touches I dyed all sewn edges black and then blended them into the brown dyed leather.
Photos 15, 16 and 17: I also liked the idea of detachable grills so as to have the choice of seeing the loudspeaker drivers or not. I found a pleasing to the eye simple acoustically transparent cloth from Fabric u.k. and made them over a weekend using black dyed cowhide as grill trim. Once these were finished I decided to put a matching panel on the amplifier.
Photo 18: Spikes fitted on the speaker cabinet bases and leather cups for the spikes to sit in gives the choice of coupling or uncoupling the cabinets to the surfaces they sit on.
This is the little Fiio DAC sitting on top of the right hand speaker in the photo below.
A few by the ways:
I use a Fiio DAC (digital to analogue converter) as an external sound card for my computer as the internal sound card is low to medium quality. I store my music as FLAC files. MP3 files tend to lose something being compressed.
Over half a kilometer of waxed thread was used to sew the leather onto the four cabinets. That took me over fifty hours of hand sewing accompanied by numerous blisters.
Thirty four square feet (3.4 square meters) of 3.5mm thick, Vegetable Tanned, Full Grain Cowhide was used to cover all four cabinets.
Photo 19: Everything connected up and working. The speakers really need to be spaced further apart and the subwoofer should be on the floor but with the limited space available in the workshop it’s the best I can do. It still sounds perfect.
The story continues click here to see the final setup.