• 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

I did a search through the threads and didn't see this specific topic addressed, so if you know of a thread that already went through this, I'll accept the shame of being inept at internet searches. Anyway, here goes...

I'm putting together a simple 12AX7-based RIAA phono stage with a source follower on the output. This is the first phono stage I've attempted with lots of SPICE and PSUD2 modeling to guide me. I'm set with the audio circuit, and had a PSU all rigged up, but the vintage Dynaco PAS2 power transformer I was going to use mysteriously died.

I have a Broskie passive power supply stuffed and tested, using UF4007 diodes and a CRC of 47uF_470R_47uF. After that I was going to use an LR8 per channel for rectified plate supplies to the audio PCBs.

The Dyna transformer would have given me about +360VDC, and the audio PCB was designed to get +320VDC. That left me with lots of volts to drop across the LR8 regulators. However, the most DC voltage I'm going to get from a 500VCT 40mA power transformer is maybe +330VDC, or more likely +325VDC.

To change the PSU design so that it mates well with the already stuffed audio PCB, I need to do something that drops very little volts. It looks like the LR8-based PSU I had in mind is going to need to drop at least 15V, or 20V to be safe. That reduces the available B+ to the audio board to down near 300V. That would be OK, but I did design it for 320V.

I did a little more modeling of the audio PCB and PSU combined, and found that if I can get all the ripple reduction I need by adding an RCRC network of 220R_800uF_220R_800uF after the Broskie power supply PCB. (I have a bunch of 800uF 330V electrolytic capacitors I was given some time ago.) PSUD2 predicts residual ripple will be in the microvolts.

So finally, the question:

What would be the disadvantages of using a 'passive' PSU employing these monster 800uF capacitors as compared to something like a single LR8 per channel, or a Maida reg set at 310V?

I figure that the impedances of a 12AX7-based circuit are high (rp of about 75k ohms), so an ultra-low impedance PSU is not going to be all that beneficial, right?
--
 
LR8N3 itself is rather high impedance, and will perspire hugely -- while rated at 20mA, the TO92 variety is only good to 0.74W.

I don't think anyone has characterized it for "noise" in this application.

You can use it as the control element of a HV Mosfet. In this case, you won't need an heroic amount of uF's on the output to maintain stability.
 
Completely opinionated opinion... regulators are THE best way to deliver darn quiet, outstanding DC power to especially noise-and-operating-point sensitive stages, such as those implied in your RIAA front-end situation. As I said, totally opinionated.

Thing is, regulation isn't just about regulation... but with a temperature-compensated voltage reference, regulation is also about throwing off the chains and fetters, the ideosyncrasies and wiggly silliness of the wall power itself. If you tune up a well regulated PS to deliver 325 VDC, guess what ... it'll do that regardless of whether The Wall is sourcing 210 VAC or 255 VAC. (Europe) ... or 108 VAC to 135 VAC (N. America)

That kind of independence means that to whatever degree your RIAA preamp is 'tuned and working' when you finally close its Faraday cage, 5 weeks, months, or years later, it ought to be performing almost-exactly the same -- or at least "the same, less component ageing"

So yah. Regulate! And follow the regulator(s) with a low value resistor (less than 100 ohms), and a nice sturdy capacitor (100 uF, I generally use.) IT goes a long way to decrease the AC impedance to milliohms over much of the audio band.

GoatGuy
 
  • Like
Reactions: 1 user
I've used LCRC supplies, stabilized supplies with mosfet over zeners, and supplies with shunt or Maida regulators. I go with the Maida regulators consistently now, I like the perfectly stable B+ voltage (no low DC variations leaking through to the output). Easy to build if you buy a PC board (I like the Pete Millett boards available on the bay).
 
Completely opinionated So yah. Regulate! And follow the regulator(s) with a low value resistor (less than 100 ohms), and a nice sturdy capacitor (100 uF, I generally use.) IT goes a long way to decrease the AC impedance to milliohms over much of the audio band.

100 ohm resistor in series, like an output 'stopper'?

Or are are you describing an RC decoupling stage after the regulator, RC of 100R_100uF?

Edit to add:
Incidentally, the audio PCB has two stages of RC decoupling between audio stages, 470R_47uF then 4.7k_33uF.
 
Last edited:
100 Ω resistor in series, like an output 'stopper'?

Or are are you describing an RC decoupling stage after the regulator, RC of 100R_100uF?

Yah…
VUNREG → REG → 100 Ω → 100 µF → VOUT

⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅

PS: in my world, the regulator acts as

• An absolute voltage stabilization point
• A rather remarkable CLC replacer (CLCRC becomes CXRC)
• A temperature-and-load independence-i-fy-ing element
• A cheaper-and-lighter (as well as the above) circuit improvement
 
Last edited:
In my opinion, the main disadvantage of a series regulator is that it produces noise, often quite a lot of it.

The main disadvantage of an CRCRC...-like supply is that it doesn't reject deep subsonic voltage variations due to mains voltage fluctuations.

A special low-noise regulator or a plain old regulator followed by an RC filter to suppress the noise, like GoatGuy recommended, should give you the best of both worlds.

About a decade ago, I built a moving-magnet RIAA amplifier with an unregulated supply with a CRCRC filter: 47 uF, 1 kohm, 47 uF, 2.2 kohm, 50 uF, last section built twice, once for the left and once for the right channel. Each channel had another 2.2 kohm, 47 uF between its second and its first stage.

Whenever I wanted to do low-level measurements at the output with an oscilloscope, the signal randomly kept moving out of the picture. Mind you, I also had unregulated, AC supplied heaters, so that could also be a path from mains fluctuations into the circuit.

Still, when I use it to play music, it just works fine. I never saw the loudspeakers move back and forth at random or anything like that.
 
Last edited:
I've used LCRC supplies, stabilized supplies with mosfet over zeners, and supplies with shunt or Maida regulators. I go with the Maida regulators consistently now, I like the perfectly stable B+ voltage (no low DC variations leaking through to the output). Easy to build if you buy a PC board (I like the Pete Millett boards available on the bay).

Excellent choice ... I also believe that "inventing it yourself" has very little upside for all the potential downside of doing it wrong. Just get the PCB!
 
About a decade ago, I built a moving-magnet RIAA amplifier with an unregulated supply with a CRCRC filter: 47 uF, 1 kohm, 47 uF, 2.2 kohm, 50 uF, last section built twice, once for the left and once for the right channel. Each channel had another 2.2 kohm, 47 uF between its second and its first stage.

Whenever I wanted to do low-level measurements at the output with an oscilloscope, the signal randomly kept moving out of the picture. Mind you, I also had unregulated, AC supplied heaters, so that could also be a path from mains fluctuations into the circuit.

Still, when I use it to play music, it just works fine. I never saw the loudspeakers move back and forth at random or anything like that.

I built an RCA-style phono stage for my brother, over 25 years ago, on the now-unavailable Handmade Electronics PCB, with its output DC-coupled to a 12AT7 cathode follower. He's still using it. The B+ is a simple passive supply, with a basic full-wave rectified supply using UF4007 diodes with something like 100uF_1k_100uF_1k_100uF for the raw B+ with additional RC decoupling per channel. The heaters are simple DC-rectified with a couple of 10,000uF caps for filtering. The thing sounds really good with a Denon DL110. You can hear a little hum if you crank the volume way up, but it's negligible at normal listening levels. It's been perfectly well behaved for nearly three decades.

One of the nice things about an over-spec'd passive PSU is that it's very reliable. No feedback so it doesn't oscillate. Yes, there are low frequency resonances to contend with, but there are workarounds for the worst ones.

I just remembered that I have a Maida board rigged up in some drawer somewhere. I should be able to reconfigure it for +310V out and give it a go. It does mean drilling new holes, though (grrrrrr....).
--
 
What surprised me was that all hum issues I had were due to stray field from the supply transformer. With the transformer placed at some distance, I didn't hear any hum at all (with the volume turned up enough to clearly hear the amplifier's noise), despite the 6.3 V AC on the heaters and the unregulated supply.

I used triode-connected EF86 valves with AC grounded cathodes for the first stage and used shielded twisted pair cables for the heater wiring. The EF86 is very suitable for low-level stages with AC supplied heaters because of its unusually small capacitance from heater to grid. Nonetheless, I had expected some hum to come from the heaters, but didn't hear any.
 
Hi,

please look here
HT ss stabilized

It is interesting test.
I will ptovide also a CLC results.
Thhere are two reason to use th HT ss reglulate
First is the stabiliy ( when the circuit is properly designed)
Secondary is the the ripple that is killed.
One of the test when I measure the phono stage is the s/n ratio.
Standard the test is done with a "A" weighted filter.
When you use a good , non regulated supply normally the difference on s/n with or without "A" filter is evident.
Different when you use the regulated voltage
Minor are the differences better is the quality of HT supply
Of course the start point is the perfect wiring specially on ground path.

Walter
 
Account Closed
Joined 2010
any phono preamp that's not using an op-amp forfirst stage gain makes better use of a regulated supply, but there are cases where rcrc filters are used although CLC filters are usually more efficient while with phono preamp, as @MarcelvdG already said the transformer stray field is one of the main peace killer and first tube fillament needs almost always a lift, but i found also a working alternative : Luxman CL-34...which i cloned a while ago , but i only received the iron case for it two weeks ago and complete shielding is a must with this circuit.
 

Attachments

  • luxman cl34.jpg
    luxman cl34.jpg
    133 KB · Views: 401
  • luxman_cl34_sch.pdf
    1.6 MB · Views: 178
Last edited:
Waltube - Thanks for the pointer to your PSU noise measurements. In the PDF document, is the red trace the AC ripple of a passive decoupled power supply, while the green trace is the AC ripple on the output of your shunt stabilized power supply?
--

I'm reworking the PCBs now to incorporate some of the ideas here.

First off, I'm going to add decoupling after the regulated B+.

The audio circuit is typical: 12AX7 (common cathode)>>RIAA EQ>>12AX7 (common cathode) DC-coupled to a MOSFET source follower.

There will be an RC decoupling network right when the B+ enters the board, 1k-33uF (I would have preferred 100uF, but all I have is a bunch of 33uF 400V). The second stage 12AX7 plate and MOSFET source follower drain connect to the 33uF cap.

After that, there is an RC decoupling network to the plate of the first stage 12AX7, consisting of 4.7k-33uF.

That should keep the supply impedance low into the subsonic region for the first stage 12AX7, even if I have to use a passive-filtered B+ supply.
 
Hi, I frequently use a regulator on phono preamps. I like the LR8K4. It is rated at up to 2.25 watts with the heat tab on a suitable thermal spot on the board. Without one it can handle about 1.5 watts or so. An alternative mentioned earlier is to use any LR8 to drive a FET. IRF 8XX or 7XX are quite suitable. Be aware that all the LR8s are really sensitive to reverse voltages and use a rectifier backwards from output to input. The LR8s don't tolerate short circuits at all. They have the tendency to fry immediately. If you overload one or it senses it will overload they have a funny way of protecting themselves. They reduce the output voltage to a very low level and will not restore it until they are powered down.
 
What would be the disadvantages of using a 'passive' PSU employing these monster 800uF capacitors as compared to something like a single LR8 per channel, or a Maida reg set at 310V?


What is your taste? Both approaches produce very different sound, so i cannot see how someone would enjoy both equally. Both approaches are also audibly compromised imo.

Getting a passive supply to sound good is not easy, especially without good sounding chokes. Getting a regulator to sound good is a lot easier and there are ready made solutions, all of which based on shunt regulation. A combination of CLCLC + Salas seems to work for me, but there are those who find the sound too analytic and prefer no regulation.
 
AX tech editor
Joined 2002
Paid Member
Yah…
VUNREG → REG → 100 Ω → 100 µF → VOUT

That seems a curious way to negate the hard-won clean output of a clean regulator. The 100R creates current-induced signal ripple, which the final cap has to try to get rid of again.
While at the reg output, all is clean and does not vary with signal current.

The diyaudio store has a suitable regulator kit.

Jan