RIAA Preamp Build (part 4): 2-stage standard

Lets look at a basic 2-stage RIAA preamp from the early 50s.

As expected, the RIAA EQ will be inserted after the first stage, where headroom can be found and behavior controlled.

There are actually two parts to the circuit in regards to phono cartridges.

(1)  The Input Impedance of the Unit. (R in).  This will be set to 16K for ceramic cartridges, and otherwise, for magnetic cartridges, it should match that specified by the maker of the cartridge.  Thus you could have a couple of resistors here to switch in and out depending upon the cartridge you were using, or a Ceramic/Magnetic setting.

(2)  The RIAA Network Proper.  This is a 3-terminal (in/out/ground) interchangeable or adjustable network, activated by selection switches.   Note the pre-specified .0068 uF input cap, and the 1 Meg output load to the next stage.  These are chosen as a base for the variations to follow.


Circuit Operation:

The 6SL7 is chosen for its high Mu (gain).  With all high Mu triodes, care should be taken to prevent HF rolloff from Miller capacitance.  Here with a gain of 30 and a MC in the order of 100 uuFd, the output Z of the source shouldn't be higher than 75K to keep the 3db point above 20kHz.  Luckily both ceramics and magnetics generally have Out Z lower than this.

We now design the Out Z of the RIAA circuit etc. to keep the signal to stage 2 low in size (and distortion).   The EQ circuit will have an Amplification factor of .022 at 1kHz, even at 20 Hz the Amplificatoin (after losses) will be only about .33.  This means you can have a bias as low as -2 volts on the 2nd stage, and a typical input signal of .1 volt to the first stage.

1K R Cathode resistors will typically be bypassed with a 47 uF cap to keep up the gain and most importantly to reduce heater hum.   To keep the signal clean, this (electrolytic) can itself be bypassed with a small polypropolene for signal clarity, slew-rate and reduced noise.

In these low-signal circuits you need a really good, well filtered Power Supply.  In my view this makes the Mu-Follower the ideal method for supplying the HV.


Design of the RIAA Circuit (EQ):

(1)  First, the input Z which the first stage sees must be high enough not to load the tube too much, (which would cause distortion).

(2)  Second, the Out Z presented to the 2nd stage should be low enough not to lose high end through Miller Capacitance. 

(3)  Finally, if you have more than one RIAA circuit, they should be matched in overall attenuation losses so you can switch easily between settings without volume problems.

Flat Turnover EQ:

A basic L-circuit does the trick for flat recordings with a fixed turn-over point (i.e., where the gain is about 0 db).

(A) 250 Hz Flat Xover:  For British H.M.V. and other European pressings (Mittell, B.E.B., etc)

(B)  500 Hz Flat Xover:  Older American Vintage pressings.

(C)  1 kHz  Flat Xover:  Some rare European recordings.

Tube Biasing 101 (Pt 2): 12AX7 bias chart!

What you need really is my ideal Bias point chart:

This is the Bermuda Triangle of Tube Biasing:



There is really no single ideal bias setup,
however, you can intelligently select operating points and ranges based on intended use.

Here are some examples:

(1) Hi-Fi setup: High Voltage B+ (500v) and Large Load (420k), to minimize distortion and maximize voltage transfer to load (this is not a power circuit, so voltage is more important than power transfer efficiency). Although higher voltages raise risks, they give more horizontal loadlines, which means current remains stable (think CCS). Headroom is not paramount here, because typical input signals will be played at low to mid-level volumes, to further limit system distortion, and protection against high voltage swings can be built in via -db pads for input of stage.
A lower bias point is selected, to better center swing in zone of maximum linearity.

(2) Universal Soldier Setup: Slightly Higher Voltage B+ (400 v range) Here is the mid-zone. Reasonable voltage and current excursion is expected, and more current can give both stability and a current source for subsequent stages which might need draw. a bias-point of -2v gives a nice centering for a balance of headroom and linearity.

(3) Guitar Maniac Setup: Here overloading is expected and harmonic distortion (non-linearity) is actually desired. Lower plate load gives a nice steep 'dive-bomb' loadline, while lower plate voltage (330 B+) protects tube from HV shorting, and allows pushing tube into cut-off safely. The higher bias point (-2.5 to even 3.5) is chosen to give maximum headroom for wild guitar antics and easy sliding into non-linear 'sweet-spots', without driving tube into grid-conduction, which is a non-musical type of distortion.

One thing that many people will notice, is that a large number of circuits load and bias 12AX7s nowhere near the design-center Triangle depicted above.

It will be a good exercise in fact to take both your favorite circuits (and your notorious pet peeves) and put them on my chart, to see how and where they are screwing up.

As I maintained in another thread, few people know how to properly set up a tube for the intended purpose.
 

Some things to observe:

(1) proper bias and setup is first dictated by intended use of the circuit. This must be nailed down first.

(2) Next, appropriate B+ voltages and loads are chosen, to fix the slope and position of the load line.

(3) Now, the bias-point is selected based on the balance desired between headroom/input range and non-linearity/harmonic distortion.

(4) To force the bias-point, the correct self-biasing cathode resistor is chosen, or better, several tubes are set up in a rig, and the resistor is selected by experiment to put the bias-point in the best compromise position between the acceptable range of tube samples.

(5) The performance of the circuit is tested under realistic conditions, including input signals, and output loads from following stages. Attenuation or amplification is adjusted at the input, and impedance matching is done at the output, to conform to expected conditions.

Finally, notice that the ideal bias-point slides along the load-line to the left, as the load-line tilts toward horizontal and slides to the right. The Bias Point traces the beautiful mathematical curve known as the sea-shell spiral:

Tube Biasing 101: (Part 1) Biasing a 6922

oh oh, I've seen schematics like this before,
and he's driving the 6922 too hard: taking it over its dissipation-wattage rating.

But others have also pointed out that there is nothing really to gain from that, except shortened tube life.


OK this schematic came from here:

DIY 6922 / E88CC Tube preamp


On that page you'll see his loadline for the 6922:

If you plot the Watt Dissipation line over this, you'll see the problem.

The solution is to alter the loadline (by putting a db-pad or volume pot on the output) and also the quiescent-point (idle-current) by adjusting the bias via the cathode resistor, and possibly the voltage.


The 6922 is a great tube, but it is a low-noise fragile signal tube, not meant for this kind of manhandling.

The Dot which represents the idle-spot should be well below the max-dissipation curve.
The voltage swing, which will meander over the line anyway, should never spend more than half its time outside the boundaries.

Although Philips allows 2 watts dissipation / triode,
The Telefunken sets 1.5 watts as the real limit. Obviously we go with the lower rating.

Also, it is generally recommended that your design stay at 70% of max wattage rating for safety and to allow for variation between tubes and brands.

Here's a better loadline and bias0-point for the 6922.

It sits at 70% of the conservative max dissipation (1 watt).




With this new bias-point you stay in 70% zone, and tube stays out of trouble most of its life.

The load resistor becomes about 12k5, the cathode resistor is 830 ohms, the bias voltage is -5.
The cathode floats about 150 volts above ground.

The tube coasts in its comfy spot, with very little loss of headroom and great gains in tube longevity. 

Tube Charts without Heat-dissipation and max rating boundaries are almost useless for setting tube loads and bias points.