I don't have a "bench supply" but I do have 2 car batteries, the charger, and various 12V "LED" supplies.
A battery powers a sine inverter which powers my soldering iron. I also have a hakko iron that runs on AA batteries for doing JFETs and the like.
Salas makes a good point about the tube shields.
Personally, my RIAA design uses a 6N1/6DJ8/6CG7 (or 6N3 with adapter) for input tube. Less gain and less chance of picking up radio in my experiences.
Also, I'm learning it's unrealistic to expect a tube phono amp to be silent at full blast. There will always be some hiss and hum but you would clip the crap out of the amp if you actually played a record at that setting anyway. If the sound of the stylus on a silent part of the record is louder than the residual noise, I'm good 🙂
A battery powers a sine inverter which powers my soldering iron. I also have a hakko iron that runs on AA batteries for doing JFETs and the like.
Salas makes a good point about the tube shields.
Personally, my RIAA design uses a 6N1/6DJ8/6CG7 (or 6N3 with adapter) for input tube. Less gain and less chance of picking up radio in my experiences.
Also, I'm learning it's unrealistic to expect a tube phono amp to be silent at full blast. There will always be some hiss and hum but you would clip the crap out of the amp if you actually played a record at that setting anyway. If the sound of the stylus on a silent part of the record is louder than the residual noise, I'm good 🙂
I should invest in some PC mount 9-pin tube sockets that are made to take shields like that. Of course I've used chassis mount sockets like that in guitar amps. This particular preamp is all on PCBs though.
That's a new one for me. I assume you're talking about the first (input) stage? In this case, the input stage is a 12AX7 with a low-current red LED (designed to operate at 0.5mA) and a 100 ohm grid stopper.
I have several cartridges I've tried with the preamp so far. They vary widely in source impedance from Denon DL110 (DCR = 160 ohms) to Shure M35X (DCR = 1000 ohms). I also tried an Audio Technica AT-VM95E (DCR = 485 ohms).
The noisiest was certainly the DL110, but I thought that was because its nominal output is only about 2.2mV, while the MM carts are about twice that.
I figure a 100R grid stopper does barely anything on a low gm triode like 12AX7. Am I wrong about that? Would I be better off simply shorting out that grid stopper?
I used tightly twisted pair hookup wire from the input jacks to the PCB audio in pads. Perhaps I should use shielded coax cable? I have plenty.
I don't have ferrite beads for this kind of application. Can someone post a link to a suitable bead for this kind of low-level signal application? I'd appreciate it.
Thanks for the input everybody.
EDIT TO ADD:
I forgot! I have a Raspberry Pi music server that runs on a 5V SMPS. I'm sure that thing is spraying RF everywhere. I'll power that off and try the preamp again, to see if that makes a difference. I'll also look around for anything else that could be spewing RFI. I'm sure there's lots.... Cable modem? Cable TV box? WiFi router? AV receiver with class D amps?
Last edited:
I had tube shields that just wrapped around and stick to the tube...
Similar to these - no special socket required:
Try 1k-10k for grid stopper. Shielded coax is better than UTP for low level stuff like this.
Ferrite beads: https://www.amazon.ca/Topnisus-Suppressor-RF-interfere-EMI-interfere-Integrity/dp/B01L6Y5CRY
Similar to these - no special socket required:
Try 1k-10k for grid stopper. Shielded coax is better than UTP for low level stuff like this.
Ferrite beads: https://www.amazon.ca/Topnisus-Suppressor-RF-interfere-EMI-interfere-Integrity/dp/B01L6Y5CRY
In this case, I wanted to get up to at least 44dB of gain because I'm trying to get the phono preamp to match the perceived output level of my DAC and CD player. It turns out that I need even more than that if I want to use my favorite Denon DL110 cart (2.2mV nominal output). More on that later...
Yes, I agree. Really, the hiss is practically non-existent in this preamp, and the hum is only evident with certain tubes and with the Denon DL110 (due to its low output I have to crank the volume up). I think the biggest issue is that the audio circuits are located about 15cm from the power transformers.
This preamp is not meant to be the be-all and end-all, not even close. It's an experiment to see if the SPICE simulations I've done come out somewhat close to reality. Simulation predicted that THD in the first stage doesn't affect the THD of the output very much. Almost all the THD in any given passive EQ two stage RIAA tube preamp (RCA-style) is in the 2nd stage. So I used a 12AX7 there with the cathode resistor unbypassed. (The reduction in PSRR that causes might be biting me with additional noise pickup. That could very well be the case.) The preamp sounds very clean, so I think SPICE's prediction is bearing fruit. But I don't know if it's meaningful. Can you really hear the difference between 0.02% and 0.1% THD playing LPs? Do the cartridge/tonearm non-linearities swamp the THD of the preamp?
I'm not going to talk about sound too much yet, suffice to say that different cartridges sound wildly different through this preamp. Huge differences. More on that later...
EDIT TO ADD:
Koda -- Thanks for the link to the ferrite beads, and for the reminder about the push-on tube shields. I might have a few of those languishing in a box somewhere. I hope I didn't purge them...
My phono stage first gain is 16, second is 50. I can't tell you how many decibels that is, but my 4.5mV cart will make +3 VU on a 12 inch 45 single. 0 VU being 1.228V RMS 1.736V peak...
Personally, I think 1% THD is a decent result. If we really wanted clinical precision, wouldn't we build power op amps? EDIT: Like this? https://www.ti.com/lit/ds/symlink/o...s%3A%2F%2Fwww.ti.com%2Fproduct%2FOPA549-HIREL
In my experience, the amp/system/tubes have to be pretty bitched to sound noticeably bad.
Personally, I think 1% THD is a decent result. If we really wanted clinical precision, wouldn't we build power op amps? EDIT: Like this? https://www.ti.com/lit/ds/symlink/o...s%3A%2F%2Fwww.ti.com%2Fproduct%2FOPA549-HIREL
In my experience, the amp/system/tubes have to be pretty bitched to sound noticeably bad.
Last edited:
Hmmm...
The conventional wisdom you read in books is that for low noise performance you want as much gain as you can get up front at the input. Then amplify less afterwards, where levels are higher. You're going the opposite direction. But hey, if it works it works!
Again, noise issues are not a problem with my Shure and Audio Technica carts that make 4.5mV or thereabouts. The only issue re: noise is with the DL110 (2.2mV). I can fix that by bypassing the 2nd 12AX7 cathode, because with this preamp the DL110 sounds waaay better than anything else I've tried. Denon DL110 and 12AX7 input stage are a very simpatico match-up. 🙂
The next surgical session will change:
- Change UTP input wiring to Mogami twin-conductor coax (Braided shield connects to chassis ground at input jack end, but left disconnected at PCB end. Black 'signal ground' wire connects from ground pin of RCA input jack to PCB ground pad.)
- Install 1k ohm grid stopper on 2nd stage 12AX7 (after the passive EQ). Perhaps a larger value? Wouldn't 10k be too much because of interaction with Miller effect capacitance of this high mu tube? Split the difference and try 4.7k?
- Install 100uF cap in parallel to 2nd stage 12AX7 cathode. That will raise the gain by about 5dB. Simulation says to 50dB total, but I find that hard to believe. (This is to increase gain to work better with Denon DL110.)
Once that's done, hopefully I can find two of those push-on tube shields and I'll order some ferrite beads.
Any other recommendations?
--
The conventional wisdom you read in books is that for low noise performance you want as much gain as you can get up front at the input. Then amplify less afterwards, where levels are higher. You're going the opposite direction. But hey, if it works it works!
Again, noise issues are not a problem with my Shure and Audio Technica carts that make 4.5mV or thereabouts. The only issue re: noise is with the DL110 (2.2mV). I can fix that by bypassing the 2nd 12AX7 cathode, because with this preamp the DL110 sounds waaay better than anything else I've tried. Denon DL110 and 12AX7 input stage are a very simpatico match-up. 🙂
The next surgical session will change:
- Change UTP input wiring to Mogami twin-conductor coax (Braided shield connects to chassis ground at input jack end, but left disconnected at PCB end. Black 'signal ground' wire connects from ground pin of RCA input jack to PCB ground pad.)
- Install 1k ohm grid stopper on 2nd stage 12AX7 (after the passive EQ). Perhaps a larger value? Wouldn't 10k be too much because of interaction with Miller effect capacitance of this high mu tube? Split the difference and try 4.7k?
- Install 100uF cap in parallel to 2nd stage 12AX7 cathode. That will raise the gain by about 5dB. Simulation says to 50dB total, but I find that hard to believe. (This is to increase gain to work better with Denon DL110.)
Once that's done, hopefully I can find two of those push-on tube shields and I'll order some ferrite beads.
Any other recommendations?
--
Yes, talking about the input stage. The Led has low dynamic resistance. Say 10-20Ω. So with the 100Ω stopper about 120Ω equivalent noise contribution but because the Denon is 160Ω its going to swamp that contribution. Being a 2.2mV Hmc cart you open the volume ~6dB more than your 5mV genuine MM carts hence more hum or hiss.
The extra gain picks up unwanted crap though, and there is less noise on a 6N1P than a 12AX7 for sure. 6N1 gain, 6N1 buffer driving passive eq, 6N2 gain, 6N1 buffer. There is no reason I can't build another board with the 6N1 and 6N2 swapped though to test it...
I use 1k in that position except I use 6N2 instead of 12AX7 🙂 How hard would it be to try 6N2 instead?
My "go-to" cap for bypassing a high gain tube is 1000uF/16V Aluminum polymer. YMMV.
Also, LED biasing can bring happy results from what I've heard - never tried it myself but you're using it now? I use unbypassed resistors in my builds (except the HP amp where the lower Zout is worth the extra THD)...
I know you won't rebuild it, but 6N1, 6N1, 6N1 makes like 60db.
I use 1k in that position except I use 6N2 instead of 12AX7 🙂 How hard would it be to try 6N2 instead?
My "go-to" cap for bypassing a high gain tube is 1000uF/16V Aluminum polymer. YMMV.
Also, LED biasing can bring happy results from what I've heard - never tried it myself but you're using it now? I use unbypassed resistors in my builds (except the HP amp where the lower Zout is worth the extra THD)...
I know you won't rebuild it, but 6N1, 6N1, 6N1 makes like 60db.
Last edited:
Borne out by my experience, exactly.
I have to turn the volume control up a couple of clicks to match the output from the genuine MM carts, and that's when I can begin to perceive just a trace of hum if I'm 30cm or so from the speakers, but only with the RCA 7025 tubes which seem to be picking up hum.
Last edited:
- I was worried the extra gain would just add extra noise. Maybe I won't bypass that resistor after all. I'll try reducing the noise instead.
OK, 1k it is. I guess carbon comp is the preferred type for grid stoppers? I think I have some in 1k 1/4W. I also have plenty of carbon film in that value.
It would be simple to swap in 6N2P. Just a change of jumper wire and ground pin 9.
I have too many 12AX7s saved up not to use them up. Vintage Telefunken, Mullard, Amperex, USA 7025s too.
I have some 1000uF 'organic' polymer. They were faddishly popular in audiophoolery for a while. Maybe still?
Yup, in the first stage 12AX7 cathode because I'm following the conventional wisdom of getting as much gain as possible from the input stage. LED bias works fine. I've used it a fair amount. It sounds a little different from capacitor-bypassed cathode R. I actually prefer R||C cathode load for line stages and in my headphone amp. I just like the way it sounds. However in this phono preamp I'm going for as clean as possible. Maybe I'll change it later.
I have an idea for a 3-stage 6DJ8 phono preamp, with the first two stages being common cathode, 1st stage bypassed (or not), 2nd stage unbypassed. 3rd stage is an anode follower, so the gain is adjustable. Finally put a source follower on the output to buffer the feedback loop around the anode follower (like we were talking about in your headphone amp thread). But that's for later. This thing is for a little box I can listen to so I don't have to listen to my Hagerman Bugle anymore. (Not that it's bad, but I like tubes.)
Last edited:
I use MF or MO types usually. yuexiang | yuexiang MO1/2W-1K+-5%-2T52 | Through Hole Resistors - LCSC.COM
CC is a gimmick IMHO. In theory, using a WW type could make it an inductor too, so reduce radio?
RE: 1mF caps, could be 🙂
How about putting the LED on the second 12AX7 and the resistor on the first?
My idea of a 3 stage 6DJ8 is my current phono with an extra stage... 12 tubes for a phono stage just seems outrageous though.
CC is a gimmick IMHO. In theory, using a WW type could make it an inductor too, so reduce radio?
RE: 1mF caps, could be 🙂
How about putting the LED on the second 12AX7 and the resistor on the first?
My idea of a 3 stage 6DJ8 is my current phono with an extra stage... 12 tubes for a phono stage just seems outrageous though.
Last edited:
I think I'll concentrate on lowering noise in this preamp. See what I can learn from the experience. If I take a few incremental steps that improve things enough, I'll have a perfectly good preamp built from parts I had languishing on the shelves. That was the primary goal. Make something from all the stuff I've been hoarding for building tyoob ampz.
According to measurements published by Merlin Blencowe in an AES article, 12AX7 is remarkably good regarding noise. Apparently it has an unusually low 1/f noise - that may also have something to do with its low anode current, as 1/f noise tends to get worse with increasing current.
If you get hum that disappears when the input is shorted, it is apparently a current that gets injected into the input by some mechanism or other, capacitive coupling or poor insulation, for example. If it were something that inductively couples into the input (like stray magnetic field from the power supply transformer), shorting should make it worse. Is there anything carrying a large AC voltage in the neighbourhood of the input or the first valve? I haven't a clue why it moves with the valves when you swap them between channels.
Regarding higher-pitched sounds: if they are really in the kHz range, microphony excited by the umpteenth harmonic of vibrations from the mains transformer maybe? I've heard sounds like that as well, this is the only explanation I could think of.
If you get hum that disappears when the input is shorted, it is apparently a current that gets injected into the input by some mechanism or other, capacitive coupling or poor insulation, for example. If it were something that inductively couples into the input (like stray magnetic field from the power supply transformer), shorting should make it worse. Is there anything carrying a large AC voltage in the neighbourhood of the input or the first valve? I haven't a clue why it moves with the valves when you swap them between channels.
Regarding higher-pitched sounds: if they are really in the kHz range, microphony excited by the umpteenth harmonic of vibrations from the mains transformer maybe? I've heard sounds like that as well, this is the only explanation I could think of.
And Klaus think's is crap 🙂 I mean the curves look good if you run it around 2W Pd though...
Tube Tester Files - 12AX7WB Sovtek
Last edited:
I don't see any noise measurements there.
I once calculated the attached list using Blencowe's data, except for the triode-connected EF86 (own measurements on a very limited number of valves). The noise densities are a kind of averages, in fact the white noise density that would result in the same total weighted noise at the output.
The ECC88 / 6DJ8 is better than the ECC83 / 12AX7, but the difference is only small despite the large difference in transconductance. The reason for that is the ECC83's unusually low 1/f noise. Biased optimally, both produce less RIAA- and A-weighted noise than the 47 kohm termination resistor at typical MM impedances (and the termination resistor far less than record surface noise).
I once calculated the attached list using Blencowe's data, except for the triode-connected EF86 (own measurements on a very limited number of valves). The noise densities are a kind of averages, in fact the white noise density that would result in the same total weighted noise at the output.
The ECC88 / 6DJ8 is better than the ECC83 / 12AX7, but the difference is only small despite the large difference in transconductance. The reason for that is the ECC83's unusually low 1/f noise. Biased optimally, both produce less RIAA- and A-weighted noise than the 47 kohm termination resistor at typical MM impedances (and the termination resistor far less than record surface noise).
Attachments
Last edited:
I've read that because carbon comps have zero inductance that makes them better as grid stoppers. Now I'm confused...
Maybe I'll split the difference and use metal film.
Why, exactly? It seems to me all you'd be doing is increasing the noise in the first stage, where you really don't want it.
I have a genuine Aikido Phono PCB in the closet, waiting for me. That thing's gigantic, like 12cm x 30cm for just the audio circuits. That PCB alone would take up the entire chassis this current phono preamp is in, connectors, audio circuits and power supply all in. There comes a point, ya know...
Last edited:
I used lots of 12AX7WB when I was repairing guitar amps around 1992, in the early days of Sovtek. They are quiet/low-noise compared to vintage American 12AX7s. They sounded 'dark' and 'less lively' in Fender guitar amps, but maybe that's just what's needed for a phono preamp or mic preamp? Too bad I chucked so many of them away after pulling them from clients' guitar amps. Everybody thought they weren't worth having around, and coveted the vintage American, German and British tubes. I don't think I have any Sovtek 12AX7s now.
Quite. There aren't any. Just his opinion. I'd be interested in if Merlin ever tested 6N2 like the 12AX7...
You can also bypass an LED. Only that its bypass capacitor must be 1000uF or bigger to be effective down to low frequencies because the LED has low impedance. They are not huge such caps when of low voltage rating. Its only an LED's VF drop they have to withstand in a cathode circuit. 1000uF has 8Ω impedance at 20Hz and only 1.6Ω at 100Hz so if the LED has 20Ω the signal will prefer the capacitor's route in its most significant portion.