• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

For RIAA preamp: Large value caps vs. Regulated/Stabilized PSU

You can use LED's if you apply global feedback which you clearly don't in a pasive riaa configuration.

Why is that?

That picture is not mine...

Why does that make a difference?

Difference is that you can clearly hear the distortions in a phono preamp when using led biaing with D3a at a gain of 77.ECC83 has an even higher gain of 100.never tried biasing ecc83 with LED but there should be the same effect.

In what way does LED bias cause distortion when used as cathode bias for a high mu triode? That is not documented (at least not that I could find) and you have not supported that assertion with data of any kind. Actually, all the data seems to point in the opposite direction. Like in the measurements tabulated in the graph below (from Ale Moglia). It shows that an HLMP-6000 (red LED) has fairly constant internal resistance even between 0.5mA and 1mA:

Diode-Bias-Examples.png


Actually, as far as distortion is concerned, the choice of operating points has more to do with it than LED vs bypassed cathode resistor, at least as far as I can tell. I could put an IR LED in the cathode, make the bias less negative thus biasing the tube hotter (higher Ip), and get less distortion (and more gain) that way.

Other more subjective concerns might trump that for you, as in 'I like the sound of the bypassed resistor bias better than LED bias', which would be perfectly fine if that's how you hear it.

LED's might need just 100nf for bypass to lower the noise , they have a very low dynamic resistance at all audio frequencies anyway so they don't need any electrolitic capacitors around.

It was Salas who brought up the desirability of bypassing the LED with 1000uF. I have 220uF in there now only because I ran out of small enough 100uF caps. They're only in there to bypass whatever little bit of noise the LED might generate. The idea is that a bypass cap is only desirable in input stages where signal levels are at their lowest, so any kind of noise is a concern.
--
 
Last edited:
Account Closed
Joined 2010
Regarding the use of grid stopper resistors of a few kilo-ohm as a way to filter off RF, taking the thermal noise for granted:

You have to keep in mind that the Miller capacitance that you see at the input has an ESR of approximately 1/gm, where gm is the transconductance of the first valve. That will be something like 700 ohm for a 12AX7 triode biased at 0.75 mA...0.9 mA (see post #156), which will much reduce the efficacy as an RF filter. If the frequencies you are worried about are much greater than 10 MHz, adding a 22 pF capacitor from grid to ground should help a lot.

If you don't want extra noise, you could consider using a couple of high-loss ferrite beads in series instead of a resistor, BLM18RK102SN1D for example. They are inductive at low and resistive at high frequencies.
 
Account Closed
Joined 2010
My first d3a phono preamp had only one ferrite bead on the input, but proved to be very stable and free of any noise due to shielding , the second one based on d3a used 220 ohms welwin wirewound resistor...Being a triode strapped triode i had ferrinte beads on G2 as well.I saw that the similar to Luxman Marantz 7 C preamp has no resitor at all on the input...https://www.ampslab.com/SCHEMATICS/MarantzStereoConsole7.jpg As I already built the Luxman cl34 clone i can't really hear any white noise...nor i hear anything weird with it.
 
@dreamth - Well, you're in a much higher league than I am with this stuff. I'll defer to your judgment here on the LED vs RC bias question. I do intend to replace the red LED with a 2k ohm resistor at some point and give that a listen. You might be interested to know that I only stuck with LED bias in a 5687 line stage, years ago. In every other project, I've eventually settled on plain old RC cathode bias. I generally prefer the sound of RC biasing.

@MarcelvdG - How do you physically install those SMD beads? Tack-soldered onto the tube socket pin? Tack-soldered in PCB pads meant for the grid stopper resistor?
 
Last edited:
Account Closed
Joined 2010
Usually i make no real measurements on noise or distortions , i never have an appropriate setup for that, simulations are good to give me a direction, but the LED bias gave audible distortions in 1v2 preamp that dissapeared when replaced with RC on cathode.On the other hand you might be true cause i saw an old statement saying that those ditortions weren't present in a a different version using LED as bias and the difference between those version were that the one not exhibiting audible distortions had much greater headroom so it might be that i heard some hard clipping garbage in the 1v2 version which had indeed less headroom in the first stage.
 
@MarcelvdG - How do you physically install those SMD beads? Tack-soldered onto the tube socket pin? Tack-soldered in PCB pads meant for the grid stopper resistor?

I never actually tried this (*), but you could use the BLM21RK102SN1D, which are 0.08 in rather than 0.06 in long, and solder them on the copper side of a small piece of perfboard with 0.1 in pitch and one hole per copper island. One end can then be connected to the grid connection of the valve holder using a short wire and the other side to the shielded wire to the input connector.

(*): I have mounted SMDs on perfboard like that, but I never used that in a valve amplifier.
 
Last edited:
Taking a break from kibbitzing over parts dimensions and non-linearities, how about some listening impressions?

Today is the first day I've been able to listen to the preamp since the subsonic resonance and hum issues were sorted out.

I'm not bothered by noise levels, and I don't hear any obvious problems, so I'm not going to obsess about RF. It would be nice to implement whatever is generally agreed is the best way to inhibit interference and oscillation. The standard recommendation is grid stopper resistors, and I so have those in place. The question is whether or not the 120 ohm stoppers I used are adequate. It's a common choice of value in this type of circuit.

I'm a little traumatized by the subsonic resonance issues from last week, leaving me sensitive to every little rumble. However, I have the farthest thing from a quiet listening room, with the central air conditioning whooshing (it's old and noisy). However, I figure it's better to put up with some environmental noise than to roast in the summer heat!

Cutting all power to the TV, which is practically on top of the turntable and preamp, got rid of the hum problem. I'm satisfied that's been resolved. Hum levels are acceptably low now.

Now I'm left with questions like:
- Did I do the RIAA EQ correctly?
- Do I like the tone of this thing?

I find the preamp to sound a bit bright, which could be a consequence of LED biasing the first stage (in which case dreamth is correct that LED biasing might not sound the best in this type of circuit/application), or it could be that I don't have the RIAA EQ implemented 100% correctly. I have a space to put in a capacitor from 2nd stage grid to ground, which would attenuate the 10kHz region. I have a collection of small value polystyrene caps (100pF, 240pF, 270pF, etc.). Over the next few weeks I can tack example values in and listen for what sounds right to me. That will take some time.

I have a 'scope and signal generator, but no inverse-RIAA network. I can wire up the Hagerman one, or the one from Morgan Jones' book. I should do that.

I ran into a big surprise while trying out the different cartridges in my collection. I have a bunch of reasonably priced carts of different styles and vintages, including a Denon DL110 (HOMC), an old (white plastic bodied) Grado Signature (MI), a Pickering XV15 (MI), a Stanton 680 (MI, the same thing as the Pickering XV15), a Shure M35X (MM) and an Audio Technica AT-VM95E (MM).

So far, I've tried the following:

Audio Technica AT-V95E - This one sounds wretched in this preamp. The high frequencies are 'hot' sounding. It's like an old Altec theater system. Too much output in the upper mids. SY measured the freq resp of an AT150MLX into a 12AX7 (assuming 500pF total Cin + cable capacitance) and found it had a +3.5dB bump around 10kHz (with 47k loading). That sounds like what I'm hearing from my AT cart. AT recommends only 100pF to 200pF input capacitance. A 12AX7 has that all by itself, right at its grid. Add 150pF of cable capacitance and the minimum input C is around 350pF, and probably a bit more. Oh well. The AT-VM95 is just ill-suited for this preamp. I might try it again with 22k in parallel with R1 (47k grid leak/input load) for about a 15k load. See how that sounds. But I don't have high hopes for that.

Shure M35X - This sounds pretty good, and sounds better with 100k in parallel with R1 (load = 32k). Smooth, but not very detailed. It's a DJ cart with a spherical needle, so that's to be expected. The recommended load is 47k and 240pF, but this is an older-style MM with higher coil DCR than the AT-VM95. I could live with this, with the 32k ohm load in place.

Denon DL110 - Now we're talking! This sounds far better than the other two. Detailed, lots of 'texture' to the sound, good bass, very nice high frequencies, vocals sound clear and clean, smoother, not 'hot' sounding. I think it's because the high input C of the 12AX7 doesn't affect a moving coil cart, allowing the DL110 to comes out ahead of the others in this particular situation. As other users have done through the years, I load the DL110 with about 2.1k ohms (2.2k in parallel with 47k).

I haven't had a chance to try the Pickering XV15 yet. I'll get to it. Its recommended load is 47k and 275pF. It's a moving-iron type, but I don't know how those react to higher input C.

The Grado is waiting in the wings, which is also an MI type.

I've read in various places that MI cartridges (like HOMC carts) are not sensitive to differences in input capacitance. Perhaps this will become a Denon/Grado-only preamp.

It's clear that this preamp is super-picky about which type of cartridge goes in. Different cartridges going into the Hagerman Bugle sound much more alike than different, while those same cartridges sound night-and-day different from each other going into this 12AX7 preamp. I'm blaming the 12AX7's high input C for that.
--
 
The discussion of miller effect on input capacitance brings up a question with regard to feedback. That is does the reduction in gain from negative feedback reduce the Cmiller? if so is it possible that a feedback style of RIAA might work better for cartridges that are sensitive to excess capacitance? Many classic phono preamps use a feedback network to the cathode of the input tube which reduces gain at high frequencies. If this reduces C at high frequencies could it not reduce the high frequency bump?
 
Negative feedback to the input stage cathode (that is, in series with the input signal) increases the *amplifier's* input impedance in the same proportion as it decreases gain. Note that this doesn't apply to the (usually 47K Ohm) load resistor, because it's outside of the amplifier.


Another approach to the high inductance cartridge loading issue is to load with a much smaller resistance, even approaching the resistance that would provide the 75uS pole by its LR pole. First popularized in National Semiconductor app notes, then refined by SY and available on his webpage, in _Linear Audio_, and here on diyAudio.


All good fortune,
Chris
 
All good points!

[Input Capacitance]
- The tonearm cable I'm using measures 75pF. It's a special low-C one. It's the only one I have like that. My other tonearm cables all measure 100pF or 150pF, depending on length. I'd like to make a shortie tonearm cable out of 75 ohm coax to reduce cable C as much as I can. (MarceldvG, do you know a part number for a suitable coax cable for something like this?)

[NFB Reduces Input C]
- If NFB applied to the input stage reduces the effect of Cmiller, then could this be why active RIAA was chosen for use in the classic 12AX7 phono preamps like Marantz 7C and Dynaco PAS2/3/3X, as well as the EAR 834P?

- A simple local NFB loop would do the same, yes? As in an anode follower (feedback from plate to a virtual earth point at the grid). A phono preamp could conceivably be designed to be 1st stage anode follower -> passive RIAA EQ network -> output stage. The feedback would have to come from *before* the RIAA EQ of course. Possibly... (Of course a pentode or cascode input stage would be more effective, but... also more expensive than a single 12AX7 per channel.)

[Input Load Resistance]
- I've been experimenting with reducing the load R (R1) by putting various resistors in parallel with R1. I rigged up a pair of Y-cables and a collection RCA plugs with different value resistors soldered in. What I've tried so far that's sounded good to me:

Denon DL110 - 2.2k in parallel with R1 (R1 = 2.1k)
Shure M35X - 47k in parallel with R1 (R1 = 23.5k)

I also have plugs with 100k (yields 32k), 150k (36k), 22k (15k), 1k (980R).

I have yet to try out my Grado, Stanton and Pickering carts to see how those sound. I need to return to the AT cart and try the lower Rload values to see if that tames it.

In simulation (using a series L and R to model the MM cartridge) using lower values of load resistance does reduce the Q of the 5kHz to 15kHz resonant peak. However, going too low also reduces the high freq F3 to well short of 20kHz. It looks like it's a balancing act with a difficult cart that wants a low-C load, with no ideal solution possible from simply lowering the value of Rload.
 
Last edited:
According to Hagerman Technology LLC: Cartridge Loading ,

Being that the AT-VM95 has Lcoil = 550mH and Rcoil = 485R,
its resonant peak happens at 10.7kHz.

No wonder it's so clearly audible!

With 400pF input C (I'm estimating 250pF from the 12AX7 itself plus 150pF for the tonearm cable) that 10kHz peak could be over +8dB in amplitude centered right around the 'sizzle' frequency.

That also means loading with lower value Rload will bring its -3dB down well below 20kHz. I'll try this, but I don't have high hopes for the AT cart into a 12AX7 with passive EQ.


Stanton 681 with Lcoil = 930mH and Rcoil = 1.3k
Resonant peak = 830Hz (way low)

I'm not sure how that will go. Will be interesting to find out. It could be that upper frequency bandwidth will be quite limited.


Shure M35X Lcoil = 425mH and Rcoil = 975R
Resonant peak = 12.2kHz

Well, no wonder that one works! The peak is up where my old ears start rolling off steeply. Add a little parallel R to the loading and get just a little boost from the resonance, but not too much.


Grado Prestige Lcoil = 45mH and Rcoil = 475R
Resonant peak = 37.5kHz

Theoretically, this should work fine with the standard 47k Rload. I suppose some tweaking of Rload could reduce any ultrasonic peak. This may be the easiest cart to use with this particular circuit. Best? I dunno...


The only electrical spec I could find for the popular Ortofon 2M Red is that its Lcoil is 700mH. That would put the resonant peak at about 9.5kHz, so again really low.

It looks like a cart with low Lcoil is what you want for a preamp with 12AX7 or D3a-triode at its input. Grado MI and various HOMC fill that bill.
 
Last edited:
Since I basically like the way this preamp sounds, except that I find it be a little bright and brash sounding with the cartridges I happen to have, I think I'll tweak the EQ a little to suit my particular tastes/requirements.

I'm going to add 270pF (actually a measured 265pF, matched pair of silver mica caps) for C5 (2nd stage grid to ground, in parallel with C4). According to simulations, that will roll off another -0.6dB at 20kHz. That may not be enough to make an audible difference, so I'm only going to tack solder these in place and audition it that way.

If I can't hear the difference, I may tack in 470pF for C5, which simulation says should reduce the response at 20kHz by -1.1dB. If I can't hear the difference with that in place, then I'll just leave it that way for good, figuring that a high frequency deviation from 'ideal' response can't hurt me if I can't hear it!

If I find that tacking in 470pF still doesn't mellow out the high frequencies enough for me, I'll replace the 1st stage cathode bias LED with a 1.5k ohm resistor for R3, paralleled by 220uF for C2 (because I don't have any 100uF available right now). Or maybe I'll use those fancy 1000uF 16V organic polymer caps. They're supposed to sound better than aluminum electrolytics. That's the only value I have in those caps.
--
 
Two things to keep in mind:

1. You normally want the combined mechanical and electrical response to be more or less flat - when there is a response peak in the mechanical transfer, the electrical response must be overdamped.

2. The ESR of a cartridge is not necessarily constant over frequency. It can increase a lot at higher audio frequencies, leading to a more damped electrical response.