Assembling the 6S45P amplifier, part three

The Christmas holidays are coming and the darkest part of the winter is here, only 4 hours of light each day. I have time, again, to work on my 6S45P amplifier continued from part two.
I´ve finished it in this part so here are all the links to this story:
The simplest amplifier obsession Part 0
Breadboard amplifier Part 1
Chassis building Part two
Assembling the 6S45P amplifier Part three – this one.
Finalizing the 6S45P Part four

Her goes the story:
Today I began soldering the output transformer to the banana sockets in the back of the amp. The wire was bent and zip-tied in place to give a neat appearance. I am hoping to finish this amp really neatly inside since it is so simple. I also mounted the signal input wires from the RCA plugs. I used CAT 5 twisted pairs encapsulated in thick aluminum foil to avoid hum. The wire package was tacked to the inside frame to be as far from the power supply as possible.

Now I had to decide on an operating point. The datasheet recommends 30mA at 150V, but several builders have tested the tube to be robust far beyond that point. After experimenting on the breadboard earlier this year I decided on operating the 6S45P at 175V and 40mA which will put the operating point just below the recommended power dissipation limit shown as the purple curve. In the diagram below the red lines point at the operating point, and the green line shows the voltage swing with an input of 2 Volts, typical max from a modern CD player. The slope of the green line is calculated by finding how much current is lost over 100Volts when loaded with the 5000ohm transformer. I=U/R so 100/50000=0.02A. So the slope is 20mA per 100Volts.

 Next I built the power supply, which is the most components in the design. This amp is almost only a power supply and 5 other components. I simulated the power supply, roughly, in Duncan Amps PSUD2. Since I was using the Russian 5Ц4С  I found nearest equivalent 5Z4P (unfortunately not G) data on Duncan amps pages and added to PSUD2´s rectifier list. The transformer is 240 to 190-0-190v and I used rectifier load of 5K to simulate my 5K output transformer. Usually it is recommended to use current draw for the operating point, but it did not seem to match my breadboard results. A confounding factor was not having the precise manufacturers data on the transformer or 5U4S in PSUD2.

Closest approximation in PSUD2 with Resistive load 5k.

I built the power supply´s smoothing section on a piece of experimenters board bolted at 90 degrees to the chassis. I could have isolated the board creating a better star ground point, but was impatient to get going. The power transformer has several optional taps that I cut short, isolated with heat shrink, and zip tied. At this point the dream of clean wiring started to crumble – as it usually does!

First capacitor 3uF to the left, then 100uF, 100R, 100uF, 100R, red wire i B+.

Angle brackets keep the board in place. Green/Yellow wire is heater for the rectifier.
Important: the bleeder resistor is connected between the two 100uF caps to drain them at power of.
Overview of the power supply. The switch is not mounted to the brown wire yet. I prefer switching through a ground fault protection switch.

Next I wired up the amplifier tubes. The 6S45P has several connecting points for cathode and grid. you only need to use one, so I chose the ones that reduced crossed wires to a minimum.
for novices pins are counted clockwise as seen from under the socket. starting from the gap on nine pin mini or from the key on octal sockets. 

Socket wiring:  Yellow pin 4 and 5 is heater, Orange pin 6 is cathode, Blue pin 7 is plate, Green pin 8 is grid and black in the middle is a local start ground for this channel.
The mA meter is connected to ground and then the cathode bias is created by series connecting a 22ohm resistor to a 50ohm trim potentiometer. A 470uF 35V capacitor is connected in parallel to the resistors. the light in the mA meter is fed 6.3v ac in parallel to the heaters.
Here is the first attempt at wiring. Not so neat after all. As usual. The yellow ac heater wires should not be routed out into the circuit, but kept along the side.

First power up. Oh why do I never get round to finishing that variac and light bulb load? I connect three multimeters to B+, filament 6.3v and 5 volt, the onboard mA meters show me the current. Connect the amp through a ground fault switch,  and hope for the best …. NO Smoke! The rectifier takes about 10 seconds to conduct properly. All voltages pretty good. I need to reduce minimum cathode resistance to get to 40mA. I can´t adjust it with the 33ohm limiting resistor I have now. Test with sound. One channel has a bad contact. After powering down and waiting for capacitors to drain I locate it by poking at contacts with a screwdriver. Its a resistor in the cathode circuit.  Now all I have to do is mount the switch and build the bottom.

I use some pieces of wood to support the transformer

Three voltmeters and the dummy load connected for first power up. Also I have a automatic ground fault switch connected to the mains cord, and switch mains of remotely.

The final schematic with measured voltages in red. A perfect beginners project!
UPDATE: I am not completely happy with Musical Power supplies for this application due to to the magnetic field inducing hum in trafo cover. I did have some hum, and when reworking the layout hum in the transformer also was reduced. you might consider equivalent Edcor or  even Sowther power transformers.

There a often discussions about what is the best beginners tube amplifier on forums like DIY Audio etc. I think this amplifier is the best place you can start because voltages are relatively low. (still deadly tho). You can follow up the amplifier build with a nice 95dB/w full range speaker and spend some quality time wondering why people spend tens of thousands on high end equipment.

If you want to build this amplifier really simply, take a look at how I did it on the breadboard version. Drop the ampere meter and the trimmer pot on the cathode.  Drop the tube rectifier and go for a cheaper lower power output transformer. With a cheap 220v to 120v transformer in a 240v country you will get close to 175v B+ using a normal bridge rectifier, 1 amp 600v should do, and capacitor resistor capacitor (CRC) smoothing, but make sure the transformer is a little over size to handle the extra voltage. Maybe 500mA or so. For heater use a 3 Amp 240 to 12 volt transformer and wire the tubes in series. And never leave out the grid stopper resistor soldered right up to the grid tag on the socket. The 6S45P can become a high frequency ADHD squirel if you do 🙂

Here are all the links to this story:
The simplest amplifier obsession Part 0
Breadboard amplifier Part 1
Chassis building Part two
Finishing the 6S45P amplifier Part three – top of this one.

Music Angel XD-800MKIII

Since my Audioromy 383 FU13 died almost a year ago I have been researching numerous amplifiers capable of driving the MagnepanSMGa ´s 90db/W. I don´t want to repair the Audioromy and continue using it because its B+ voltage of 870V is to high for comfort. After some research and posting on DIY audio I decided I needed at least 20W. Ideally an amplifier with B+ of approximately 300v would be nice. My first though was the 6C33C by  Dmitry Nizhegorodov, but in single ended it needs some huge and expensive transformers and does not really reach 20 Watts without parallel single ended or push pull, I considered an OTL, output transformer-less version like Tim Mellows which is popular on DIY audio now. But the Magnepans low 4ohm resistance is not really suited for OTL tube amps. So my attention turned to Push Pull amplifiers, classic Mullard 5-20 circuits, with output tubes like 6L6, 807, EL34, KT88. The problem was that B+ needed to be higher than 300V to get enough power. I was pretty interested in the EL84 for a while as it is cheap and has a nice operating point. But I would probably need Paralell pushpull for power reserves which means cost and complexity rises.
My studies resulted in slowly loosing interest in building from scratch. I looked at the Engineers Amp/DCPP by Pete Millet which was very well suited, and I looked into building a Dynaco S70 based on PCBs from Triode electronics, another great amplifier. Both would drive the Magnepans well and have a lot of support for sorting trouble or upgrading. But at this point I had moved into territory I wasn’t really feeling engaged about. I need a sense of exploration and wonder in my projects, not just walking beaten paths. So if I was going to work on a typical push-pull I might as well buy a finished amp. I decided to buy a Music Angel XD800 MkIII. This is a Chinese amplifier based on the classic Williamson circuit of 1947. it has a B+ of 450V, a 100V higher than I wanted but still half of the Audioromy. This particular Music Angel has a strong following in Norway, where I live, and is a cheap basis for experimenting. DO NOT confuse it with the YC808 – which is all about trouble. This amplifier I consider sort of a kit. There are several versions and a lot of experience to be found on “improving” or changing the amp for personal preferences. I looked long and hard at this amplifer when I bought the Audioromy a few years ago, and now I feel I should have gone this rute from the beginning.

Anyway, a Music Angel XD800 mkIII has been bought, used, and is in the mail to me now. It is the version built until July2008, the last version based on 12AT7 as preamp tube.
This may be the amplifier for transplanting into the Empire Chassis
Here is an older circuit diagram:

Variations to the build. (this is just a sketch  for now – more details will follow):

Preamp tubes. Both versions use 12AU7 (ECC82) as the driver tube. For several years the MA KT88 came with 12AT7 (ECC81) preamp tube – this is the original version. In autumn 2008 there was a change the amplifier was modified to use the ECC85 in the input – typically it is now the Chinese 6N1 sitting there (because it is cheap)
Grid stoppers. Early versions had grid stoppers, then Lampizator interfered and made them produce without gridstoppers introducing instability. Later they reemerged.

Chokes. Although all MA XD800´s have the pots from chokes, not all models actually have chokes.

Straight/slanted cabinet. From 2010 to 2011 the cabinets were built with slanting fronts, although some were made using left overs of the opposite type.

I got my Music Angel in the mail yesterday. It looks and works well, but today I decided to open it up to take a look around and adjust bias, if necessary. I found that the cathode resistors to the KT88s were wired unusually. both were connected to the same cathode resistor. I reconnected them, properly, and readjusted bias, discovering that one tube was not conducting current properly. it cut out after about two minutes. So now a new set of power tubes need to be bought. See picture:

Both cathode wires go to V7 instead of one to V7 and one to V8

Same with V5 and V6

I also found there was no safety earth, So I soldered one to the mains socket and connected it to the choke-pot bolt with a crimped ring and lock washer. There are no chokes in this amplifier, just their holes and cover.

Follow this thread for updates

Fooling arround with acoustics inspired by Geneva Labs

A colleague at work bought a Geneva Labs model xl this insipred me to take a closer look at the grandest of all iPad docks. I had always assumed that it had a huge bass speaker in the middle and a couple of angled tweeters firing from the outer corners to create a degree of stereo effect. At least this is what the semantics of the design said. But in our time semantics are for telling stories, not for Describing functionality. Inside the Geneva xl has a stereo set of three way speakers placed as close together as possible. What was going on here?

A quick search found a thead on DIYAudio which researched the Geneva. There is a professional speaker using the same technology the EMES Owl. The technology behind both speakers acoustics comes from a company called Embracing Sound Experience AB located in Sweden. Embracing Sound have published a paper describing the principles involved, called (PDF download) “Single Bipolar Loudspeaker System for Stereo Reproduction”

So… I had to find out what was going on here. I had a couple of Fostex FE103´s lying around, and glued together two closed speaker boxes side by side. In between the speaker elements I cut the MDF so I could slot in a piece of aluminum or plastic. Listening tests showed that this divider contributed somewhat to dividing the output of the two speakers. When the listener was positioned in front of the speakers.

The 30mm dividing strip is not in place in this picture. But you can see the slot it goes into.

By reversing polarity on one of them I got the typical effect of floating the sound out on each side of the listeners ears. I used Rouge Amoeba´s Audio Highjack Pro to hijack system output from my MacBook Pro. I set up a high pass and a low pass filter where the lowpass was bass boosted a bit and sent straight to output, creating a in-phase bass line, while the high pass was divided in left and right channel and phase shifted using the Flux SST stereo tool. This combination gave a nice mixture of center image and wide open sound. My Good Wife was impressed. Though not by the looks. “That´s not coming into the living room, right?” I don’t think this is the trick Embracing Sound Experience is using, but it sounded quite OK. Only problem was the SST tools were pulling 100% of the computers system resources. Oh, and the FE103´s don’t deserve the name full range – they are mid-high tweeters.

Screen dump of Audio Highjack Pro and the Flux SST tools.

PP Amplifier for my Empire Scientific turntable

In place of the Audioromy FU13 amplifier I have used as my main amp for two years, I am planning to build a push pull amp with EL84 or 6L6 or 807 class tubes. I expect to aim for up to 20W output power to drive MagnepanSMGa´s. In this project I am starting out by designing a cabinet to match the Empire Scientific turntable. Here is my design proposal alongside a picture of the empire turntable. Colors will of course be matched closer in the final design. By the way, if you want to make renderings like this, use Goggle Sketchup, and Kerkythea

Update january 4. 2012:
Recently I have also been looking at 829 based amps,since they are really cool looking. They have pretty high B+ though, I like it as low as possible, and I cant see tha 829 bringing enough power to the Maggies without voltage in the 400´s.

My recent favorite is a simple EL85 PP. Though it will only supply 10-15Watts it can have a low(ish) 320V B+. Here are a few reference links:
This schematic is enticingly simple and well documented, with variations and “improvements” and the cool thing is it runs in class A, although most of the harmonics are evened out. But without negative feedback some harmonics may get through the push pull stage. This is also a well documented design a development on the diyparadise build. For pushing towards 20 Watts I´ll explore parallel push-pull.

30.jan 2012 I decided to abandon diy on this amplifier. I bought a used Music Angel XD800 mkIII We´ll see if its suited for a new cabinet in Empire style

Building the 6S45P amplifier part two

Here are all the links to this story:
The simplest amplifier obsession Part 0
Breadboard amplifier Part 1
Chassis building Part two – this one
Assembling the 6S45P amplifier Part three
Finalizing the 6S45P Part four

The summer vacation is over and the weather is turning grey and rainy again here. Time to start making a finished version of the 6S45P breadboard prototype I built in April. My follow through ability has been too low, there are several projects floating in limbo right now. We´ll see what the winter brings.
The plan with this amp is to do a couple of new things, at least for me. I want to build the chassis in the traditional style of using an aluminum sheet for mounting all the components, and then fixing the sheet in a wooden frame. The second thing I want to try is to use the 5Ц4С as a hybrid rectifier bridge. this uses an octal socket so if it does not work I have several other options.

For output transformers I am still using the Edcor GXSE 10-6-5K transformers from the breadboard, but for the power supply I decided to buy a transformer with all the necessary supply’s rather than to patch together something with several simple power transformers I already have. I decided to take a chance on a no-brand transformer from China, sold through on Ebay.
Type nr. PT190.2i: primary voltage 220 or 240V, secondary 190-0-55-190 centertapped, 6.3V x 3A, 5V/6.3Vx2A filament winding.

Edcor GXSE 10-6-5K, 6 ohm output. Painted all white.

Musical Power Supplies PT190.2i:

190-0-55-190 Centertapped Output 120mADC, 220 or 240Vac in,  (1) 6.3V x 3A, (1) Combo 5A/6.3Vac x 2A filament winding, I painted the transformer white and polished the Zinc Plated finish

For materials I found some scrap aluminum at work, which was painted white with a clear finish and of-cuts of beech, which were cut and finished like this:

There are several considerations when planning an amplifier layout.
First you should make sure the power transformer is far from the sensitive input signal. Then the magnetic field of the power transformer should be far from the output transformers, and preferably oriented 90 degrees to avoid inducing noise in the output. All the components should be arranged around a star ground, preferably with a sub star for the noisy power supply. And then there is the issue of routing the filament wires away from the signal wires and positioning things to make it easy to do point to point wiring. Finally the design should be visually pleasing and the controls should be placed in a position were they are easy to reach and see.

Finally the design is starting to take shape! I will have the power transformer in the front left with a  rectifier tube behind it. The output transformers are in the middle and right rear and create a symmetrical space in front for the tubes and the mA meters. In the middle front is the volume control. Power switch and fuse is on the rear panel left.

More to come…
Update11 october 2011: Another two months passed by before I managed to continue the project. In the meantime I have painted parts. The only part left now is the back panel so its time to mount all the hardware and see what I have created. Before I show all the pictures, Yes, I know I didn’t turn the power transformer 90 degrees like I said. It didn’t look good and took room from the rectifier. Since its pretty far away, I´m taking my chances. Here is the eye-candy!

One of the screws for the amp meter is missing this is the hole for adjusting current.
Volume knob is temporary, but black works pretty well, I think.
power transformer is slightly out of line – on the to-do list

Dismantling a Pioneer SE-L40 Headphone

I was so lucky to buy a great Pioneer SE-L40 on the Norwegian trading site
These phones have a strong following among headphone addicts and command high prices on Ebay. A nice boxed example might sell for between $125 and $200 US. The main reason for buying a pair is their looks, they may be the most beautiful headset ever made!

There was only one catch with my phones, There was no sound in them, because of this I got mine for $70 including freight. I took the chance of buying a non-player because it´s usually the cable that breaks either at the headphone or at the jack plug, Otherwise there would be clearly visible trauma to the rest of the headset or cable.

I did some searches  to see if there were any dissasembly instructions anywere. The only one I found was at The post showed that the pads  were either glued, or twist locked on.  UPDATE 2015 January: This post proves that there are two versions. One is glued on, the other, like mine, is bayonet locked by twisting. The photo below shows the glued version. END OF UPDATE

Photo by sweatyk at
Photo by sweatyk at

I was a little apprehensive of using force on the pads, so I decided to study the phones closely with my background as an industrial design engineer. I found the pads are made to twist of using a kind of bayonet lock. In the outer shells inside there are two small semicircular tabs that correspond to flat areas on the earpad base edge. Just twist the earpads until the base lines up and they lift out without any tearing of the fabric.

Screw driver is pointing to flat area on ear pad ring
Screw driver is pointing to tiny semicircular tabs
on inside of outer shell

After dismounting the ear pads I removed a metal grill covering a speaker.  They are a loose fit over the speaker. Just lift them of.

The speaker twists out. The two spring loaded brass lugs sit under a shelf molded in the inside of the outer case. There is a hole in the top of the selves, which help you loosen the lugs by pressing them down while you twist and work them out. The plastic holding the lugs on my headphone was brittle, and two of the broke during disasembly. One was regluable with cyanoacrylate, the other crumbled into small parts.

A bit of the cloth covering on the ear pads had come of. I glued it on with a small drop of contact cement.

To keep the headphones apart I used their box and later on, a book. I measured continuity between the speaker elements. No Problems there. Then I measured between the jack plug and the headphones speakers. There was trouble. Now I had close to a 50% chance that the wire break was at the jack plug or the headphones. It is easier to solder in a new jack plug and keep the original cloth covered cable. I took a deep breath and cut of the lower 7 cm of the cable over the jack plug. Peeled back the cloth and measured continuity in the wire. No Luck. The break was either at the head phone side or some were along the wire.  I cut the wires inside the head shell leaving as long wires as possible

There is a small metal clamp that keeps the cable from sliding out. pry it open and keep it for reuse.
Now the cable is free so it can be worked out of the head shell hole.

You will need precision and a very sharp scalpel to cut the cable reasonably well. The cloth is very resilient, and you may end up with a frayed end that is difficult to rethred into the headphone casing.

Now its time to cut of the first 5 to 10 cm of the cable, strip it and measure continuity. Success!
I had found the break. It´s late and this is precision work, so I´ll stop for today. Tomorrow I will resolder the wires in the headphone, and mount a new jack plug.

Time for soldering and assembly
I cut 10cm of 4,6mm heat-shink and threaded the cable through. leaving 3cm of wire protruding. As things turned out I would probably leave 4 cm of wire.

Pull the heatshrinked cable through the headphone shell. Clip on the small metal cable-pull preventer, and bend the cable into the groove. This is probably the break point for the cable. The wires should protrude from the cable at the point where they can make the 90 degree bend into the speaker cavity.

solder together new and old wires, or solder directly to the speaker terminals.

Cover with heat shrink.

I use aluminum foil to protect against heat. This also speeds up heat shrinking as the heat is deflected better back at the wire.

Finally the clip is set in its place with epoxy glue since the plastic holding the clip had crumbled from age.

Audio jack plug wiring: Red is right and fit the ring of the plug. (RED-IS-RIGHT-IS-RING) The tip is the left channel and the sleeve is earth.

Picture from
Picture from
Listening impressions: My first impression is playing Flac and mp3 files out of the headphone socket on the MacBook Pro, I tried movies, rock, jazz, heavy rock. I used Audio Hijack pro and Blue Cat Audio´s Tripple Eq plug in to explore different equalization of the music. The first thing I experienced was surprise that these phones are written of as bad sounding but great lookers. The sound was full, with plenty of bass, and a rich lower mid tone. I found that they improve by amplifying the high frequencies about 3-5 dB with a 2,5 dB/octave filter centered at 2kHz. Unfiltered they have a prominent lower mid that favors jazz, male voices and bass music but imparts a veiled blurry quality to the music. With extended listening they are comfortable and light. I like the fakt that you can open up the dual headbands so they sit better over your head, but they have rounded pads that feel like the headphones are pushing a bit to the inside of the ear. My daughter felt they tickled her in the ear.

Follow TAO as he messes around with tube amps, speakers, etc.