C3g-Aikido Phono Stage
My WAD PHONO II has given sterling service and, apart from a dodgy Ei ECC83 which went west soon after I built it, is still going strong, and is currently in use as I write. But hearing some of the DIY designs by Andrew Lahane and particularly Nick Gorham's LCR phono I was interested to try a new phono.
A beer-inspired conversation at a Witham-fest in February 2007 lead to a plan - a C3g pentode first stage and a 6N1P/6N6P second stage after the RIAA. Andrew very kindly offered to do the clever bit designing the audio circuit, leaving me with the easier job of prototyping the build and sorting out the PSU.
Philosophy of the Design
The first stage was to be a C3g, which is a German telecoms pentode reknown for its linearity. It also happens to sound pretty good too. A triode first stage has the disadvantage that it changes over time as it ages changing the impedance and affecting the RIAA. The pentode should be more stable and so better long term. Andrew's design cleverly allows for the insertion of LCR units in place of the RIAA without further modification.
The choice of the Aikido was mine - I was interested to see what effect an Aikido would have, and we hadn't tried it, so why not give it a go? Hopefully the PSRR noise rejection would have a positive benefit too. The choice of 6N1P and 6N6P was straight from John Broskie's Tubecad website (substitute 6N6P for 5687).
Fortunately (for me) Andrew is fascinated by phono stages and quickly knocked up the first draft.
I guessed that trying to get the audio circuit and PSU into the same chassis was going to be a little tricky, so I decided to go for separate chassis linked by an umbilical. Chassis sides were made from some offcuts of 18mm birch ply and some 2mm thick ally sheet for the top plate. I figured this would give a reasonably domestically acceptable solution. With hindsight though I really wish I'd built it on a large breadboard as building the audio circuit the first time was fine, but modifying it subsequently was painful!
I wasn't sure about using an umbilical with HT, heaters and earth all bundled together, but I thought I'd give it a go and see what happened. I simply threaded the wires through a length of black expandable nylon mesh sleeving which was attached to the audio circuit end. Definitely don't want to connect a flying lead to the HT end as a few hundred volts lying around by mistake is not a good idea.
Although I have a second chassis for the PSU I never got beyond a small breadboard, my attention concentrated on the audio circuit...
So, first job is to mount the hardware - valve sockets, phono sockets, earth post etc. I also thought I'd add a couple of 12 way rotary switches for selectable load resistors. No doubt the switches degrade the sound a little but I thought it was worth trying for the convenience. The values of the resistors were selected so as to give a good spread of load resistors assuming my Lundahl MC stepups are set at 1:10.
Then it was a question of wiring up the rest of the circuit. Andrew had gone for an unbypassed cathode resistor on the C3g figuring that there should just about be enough gain without.
The RIAA stage is critical as this provides the filtering to restore the RIAA signal on a vinyl record to a "line" signal. I had a few Russian teflon 47nF caps and I found a pair that were very close to the required value. For the 470nF I got a couple of matched Soniqs caps from WD. Resistors are cheapo carbons which are reasonably accurate.
The Aikido is straightforward, built from John Broskie's website. This could be replaced by a simple common cathode stage, but I was interested to see what happened with an Aikido. The coupling caps are Russian K40Y PIOs I had in stock.
The PSU was a simple and functionary affair; something to get me going. As a phono has an awful lot of gain compared to a power amp a lot of smoothing is required, generally requiring three stages and lots of C and L. So modelling in PSUD II I came up with something that looked like it should work with the bits I had in stock.
So then it was time to plug it in and see how it sounds.
For the first switch-on of a prototype it wasn't bad really. There was more hum than was really tollerable, probably from the AC heaters, and the sound was pretty good but no better than my WAD PHONO II. Gain was a little on the low side too.
Checking a few voltages the C3g screen was a little down. I tweaked this with a lower value of resistor but I found that this influenced the voltage on the anode. I went round in circles and couldn't achieve both values and settled on the best compromise.
The sound was a touch bass heavy so I tweaked the RIAA, amending the series resistor till I found the best balance. Nick Gorham was kind enough to check the RIAA using a square wave generator and inverse RIAA network. The square wave was indeed square with just a little loss of shape in one of the corners; a pretty good result really.
The next tweak was to try to increase the gain by concentrating on the C3g cathode resistor. Clipleads are wonderful things for quickly and easily changing components. It's very easy to end up with rather too many though... Discussing with Andrew, he made a suggestion of using an LED instead of a resistor. A red LED gives a volts drop of 1.2V, just the job, and thought to sound better than a resistor. I can't really make a conclusive judgement on this though.
Regulated DC Heater Supply
Perhaps the most significant tweak was to change the heaters to a current regulated DC supply using a humble LM338. This is very similar to the example in Morgan Jones which uses the smaller current capacity LM317.
The regulator works by "throwing away" voltage so, aiming for a voltage of 6.3V the starting point is a 9V transformer. The total heater current of all the valves comes to approximately 3.3A, so 9V x 3.3A = 30VA. So a 9V transformer with a rating greater than 30VA is required; I used a 9V 50VA tranny from Rapid.
For the bridge rectifier I used Schottky diodes. Schottkys are supposedly better then cooking 1N... series diodes as they have much lower switching noise. They drop less voltage than the 1N... series which drop 0.7V each. Instead of using fixed resistors to fix the output voltage I used a variable resistor so I could play around. This isn't especially useful in this application, but in a wider context I can set the output voltage at 6V or 4V if I want to play with some old British triodes, or the like.
More to come...
|Bluebell Audio "Loftin White" 2A3 Direct Coupled SET Amplifier|