Look quite interesting - I'd give it a go
This is my idea of the best bang for the buck 12B4 linestage for newbies. Use tube or solid state rectification as your taste dictates. All 10M45S vary slightly, so the current set resistor values given are ballpark only. The best approach is to use a pot to set the required current then substitute with the nearest fixed value resistor. The 1k resistors are to stop the device from oscillating, so you can use anything in your parts bin from about 500R up to a couple of k (preferably carbon comp). Solder right up to the pin of the 10M45S so the connection is as short as possible.
Since you are only using the one ripple killing cap, you can splurge on a nice vintage 600V PIO (plenty available on ebay). A regular polyprop motor run is also fine. You only need a small value cap because the first CCS reduces ripple by a large amount. The voltage reg tubes and final CCS further reduce levels to that achievable only by an expensive LC combo. Use a 600V 0.01uF polyprop across the voltage reg tubes and you are away.
You have two output options.
1. Regular - approx 1000R, and fine for anything other than low impedance input SS power amps.
2. Mu stage style - only a few hundred ohms output (I haven't measured it) and suitable for driving a piece of wet string into a 5k input.
It is easy to experiment to find out which one you like best. Use the smallest value output capacitor you can get away with (they sound better and are cheaper). You can work this out from the following formula
C = 1 / 2 pi R F
where R = Input impedance of your power amp (in ohms) and F = cutoff frequency (I recommend you use 5 Hz here). C will be in Farads, so remember to multiply the result by 106 to convert to uF.
Everything fits easily in the one chassis, and it looks really cool with the lights out.
You can certainly build better 12B4 linestages, but they will probably cost you a whole lot more
Hello all,
I took a quick look around this thread and have decided to build the linestage since everyone has been so impressed with it.
Everyone seems to be using a different schematic with CCS.
Is there a reference schematic?
Can someone point me to the latest schematic with the CCS.
Thanks
Lawrence
I took a quick look around this thread and have decided to build the linestage since everyone has been so impressed with it.
Everyone seems to be using a different schematic with CCS.
Is there a reference schematic?
Can someone point me to the latest schematic with the CCS.
Thanks
Lawrence
Wow Bryan, that's EXCELLENT! 🙂
I have an old ARC LS2B MkII, which I dearly love. When I first plugged it in, replacing my old lousey preamp, it sounded like I had had blankets over my speakers, and I finally pulled them off. The difference was amazing! If I can (eventually) best my LS2 with a 12B4A... well, I better get started!
Enjoy the music!
I have an old ARC LS2B MkII, which I dearly love. When I first plugged it in, replacing my old lousey preamp, it sounded like I had had blankets over my speakers, and I finally pulled them off. The difference was amazing! If I can (eventually) best my LS2 with a 12B4A... well, I better get started!
Enjoy the music!
Thanks, mach1.
I've been trying to figure out how to start this project without spending tons of money and a lot of time mulling about, and your post was the answer. You read my mind! 🙂
I've been trying to figure out how to start this project without spending tons of money and a lot of time mulling about, and your post was the answer. You read my mind! 🙂
Mach, very cool!
Thanks alot, I will start on this after I get my other projects up and finished.
Evan
Thanks alot, I will start on this after I get my other projects up and finished.
Evan
Hi mach1,
That is an interesting design. but doesn't having the current regulator on the plate of the 12B4 really affect the overall sound of the stage? I would think it would..... so in this case are you listening to the 12B4 or the IXYS current source.... Couldn't this current source be placed in the cathode more effectively than in the plate and with less sound coloration for the solid state current source?
Thanks,
Mark
P.S. I may have to get some of these devices and try mine out both ways...... I am using the LM317 in the cathode on mine..... the addition of it gives tighter-deeper bass, better overall clarity and wider deeper soundstage than without it. It does not effect the general character of the 12B4 though. I also find that this stage sounds much better with the cathode bypass in there than without but a smaller film cap across it is an absolute must. The film cap addition across the 220uf makes an amazing difference if you listen with and without.
That is an interesting design. but doesn't having the current regulator on the plate of the 12B4 really affect the overall sound of the stage? I would think it would..... so in this case are you listening to the 12B4 or the IXYS current source.... Couldn't this current source be placed in the cathode more effectively than in the plate and with less sound coloration for the solid state current source?
Thanks,
Mark
P.S. I may have to get some of these devices and try mine out both ways...... I am using the LM317 in the cathode on mine..... the addition of it gives tighter-deeper bass, better overall clarity and wider deeper soundstage than without it. It does not effect the general character of the 12B4 though. I also find that this stage sounds much better with the cathode bypass in there than without but a smaller film cap across it is an absolute must. The film cap addition across the 220uf makes an amazing difference if you listen with and without.
It's well worth a try Mark - 10M45S chips are wonderful devices. Digikey has them at $1.76 for 1-9 and $1.59 for 10-49: they sound great, are excellent for breadboarding circuits, and cheap enough to sprinkle over your breakfast cereal. Their only real downside is the sample variation: like most mosfet based devices they are a bit all over the shop. Hence the need to use a trimpot to 'dial in' each device to the required current. 317's on the other hand are very consistent and easy to current set.
If you don't like them I'll buy any remaining stock you have.
From my experience I would place the 10M45S ahead of the 317 in terms of sonics (although I haven't tried one in the tail of a 12B4), with Gary Pimm's CCS http://home.pacifier.com/~gpimm/self_bias.htm ahead of both devices. Some people (eg Brett) prefer a simple plate resistor in this application. I don't think it makes much difference whether you place a CCS in the plate or cathode end of tube circuit; it will will impart its own signature in both locations. As you have found out, the nature of the bypass cap used in the cathode circuit can alter the sound dramatically.
In theory, use of a CCS obviates the need for a cathode bypass cap (with low currents anyway). However, as with all theory, it doesn't always work out. I would encourage anybody who builds this circuit to try both modes.
With respect to the circuit, I was a bit naughty leaving out a gridstopper resistor (although mine works fine without one) and a bleeder resistor for the psu cap (around 330k ballpark suggested). Also, my heatsinks are rated at 7 degrees C per watt (not 11 as I first thought). I just tried one rated at 12 and things started getting rather hot, so I would recommend using 7 or below.
Over the christmas break I will be constructing a frugalophile low voltage version (210V B+, only two inputs with no boutique parts in sight), partly to assess how well a cheap Hammond chassis functions as a heatsink (lower cost and much neater looking if I can get away with it).
Regards
If you don't like them I'll buy any remaining stock you have.
From my experience I would place the 10M45S ahead of the 317 in terms of sonics (although I haven't tried one in the tail of a 12B4), with Gary Pimm's CCS http://home.pacifier.com/~gpimm/self_bias.htm ahead of both devices. Some people (eg Brett) prefer a simple plate resistor in this application. I don't think it makes much difference whether you place a CCS in the plate or cathode end of tube circuit; it will will impart its own signature in both locations. As you have found out, the nature of the bypass cap used in the cathode circuit can alter the sound dramatically.
In theory, use of a CCS obviates the need for a cathode bypass cap (with low currents anyway). However, as with all theory, it doesn't always work out. I would encourage anybody who builds this circuit to try both modes.
With respect to the circuit, I was a bit naughty leaving out a gridstopper resistor (although mine works fine without one) and a bleeder resistor for the psu cap (around 330k ballpark suggested). Also, my heatsinks are rated at 7 degrees C per watt (not 11 as I first thought). I just tried one rated at 12 and things started getting rather hot, so I would recommend using 7 or below.
Over the christmas break I will be constructing a frugalophile low voltage version (210V B+, only two inputs with no boutique parts in sight), partly to assess how well a cheap Hammond chassis functions as a heatsink (lower cost and much neater looking if I can get away with it).
Regards
Mark A. Gulbrandsen said:
trying discrete current sink in cathode is worthwille (did I spelled that OK?)
during construction numerous variants of my WOT preamp,I tried also parafeed , tried WE in all iterations of stage,and also tried good old 7805 and 317 and TL431 in cathode ....
to my ears-when I use any sort of CS in cathode ,best way is discrete CS,but also most vulnerable to any sort of garbage from mains and RF polution.
in other words-most chalenging to decouple it well without maiking compromise in sound...
Which brand 12B4A is best ?
I hear guys saying that 12B4A could have noise problems as the tube ages and that they can also be microphonic enough to be a problem.
So what brands have minimised these problems ? RCA , GE ? Sylvania ? others? Someone said JAN tubes are less microphonic.
Is it correct to assume that tubes that have been slowly heated up ( 180 deg C ? ) and then cooled slowly in an oven will not have any long term noise degradation problems ?
Thanks.
I hear guys saying that 12B4A could have noise problems as the tube ages and that they can also be microphonic enough to be a problem.
So what brands have minimised these problems ? RCA , GE ? Sylvania ? others? Someone said JAN tubes are less microphonic.
Is it correct to assume that tubes that have been slowly heated up ( 180 deg C ? ) and then cooled slowly in an oven will not have any long term noise degradation problems ?
Thanks.
You are correct in that JAN toobes have less microphonics. These are the toobes that I usually go fo in any project and they usually have a longer life span. I have found that although they are very quiet they are not necessarily the best sounding toobes. Experimentation is still necessary to find the one you like and yes, ultimately you may have to settle for some microphonics.....
Mark
Mark
Thanks Mark.
Does anyone know what characteristics are responsible for best (?) sound. A few must be linearity , low noise , stability ( at RF !).
But I guess linearity is a very generalised statement as linearity must be related to operating point.
Is the typical operating point given on data sheets also the points for best linearity and lowest distortion ?
Since a lot of audio performance doesn't seem to go directly by what we see on paper specs , ....what does ? Are we at sea ( or outer space !) with regard to that ?
Cheers.
Does anyone know what characteristics are responsible for best (?) sound. A few must be linearity , low noise , stability ( at RF !).
But I guess linearity is a very generalised statement as linearity must be related to operating point.
Is the typical operating point given on data sheets also the points for best linearity and lowest distortion ?
Since a lot of audio performance doesn't seem to go directly by what we see on paper specs , ....what does ? Are we at sea ( or outer space !) with regard to that ?
Cheers.
RE: the IXYS 10M45S as a plate load for the 12B4A...
I've compared it to a traditional plate choke in parafeed circuits and prefer the choke. However, my comparisons may be apples to oranges since I've started using plate to grid feedback in combination with the choke.
I've compared it to a traditional plate choke in parafeed circuits and prefer the choke. However, my comparisons may be apples to oranges since I've started using plate to grid feedback in combination with the choke.
Hi,
Well, in reading all I could get my hands on, it seems to me the best point for the 12B4 bias was about 15MA. Still not sure to the best B+ voltage. To lower some Microphonics, lower the filament voltage to 12.2. Anyhow. that was some of the stuff I had learned. Don't know which version of this preamp is best. Mark's version at roughly 25 to 30 MA to the Etude preamp at 15MA. Some of these resistors that get hotter than the tube, slightly scare me a bit. There's also the one from Terry on Hawnthorneaudio. He has since gone to using different tubes. Maybe one of these days, I will get all of the stuff and build this preamp myself. Unfortunately, I have yet to read a concensis as to the "best" version. If someone has the "best" version, and has built them all, I am very interested.
Ray Bronk
Just to confuse the matter more, there is the real possiblilty of using a plate choke.😉
200H anode plate choke for 6sn7 6j5 5842 5687 12au7 - eBay (item 290402916907 end time Mar-16-10 18:28:46 PDT)
200H anode plate choke for 6sn7 6j5 5842 5687 12au7 - eBay (item 290402916907 end time Mar-16-10 18:28:46 PDT)
Hi Tubemack,
Yeah, that's true. I think there's one using anode loading versus cathode loading. Supposedly better. also the grid stoppers according to one view should be at least 3.3K or more between pins 2 and 7. That's supposed to kill some of the Microphonics issue.
With this 12B4 tube, the power supply is everything. The cleaner the supply the better.
Yeah, that's true. I think there's one using anode loading versus cathode loading. Supposedly better. also the grid stoppers according to one view should be at least 3.3K or more between pins 2 and 7. That's supposed to kill some of the Microphonics issue.
With this 12B4 tube, the power supply is everything. The cleaner the supply the better.
I searched, but could not find a use of the 12B4A like the following.
I'm putting it up here to see if anyone wants to comment on whether it looks like it would work well.
I know a CCS in the plate would be the 'ultimate', but since choke loading of the plate seems to have its fan club, and I have a pair of UTC CG51AX plate-to-PP grid trannies I could use as plate chokes, I'd like to try them. They have primary winding of 22H with 1000 ohm DCR. I don't know what current capacity, but I expect 20mA wouldn't kill 'em.
My questions:
1) Is the L of these trannies used as chokes sufficient for a low enough -3dB point? I suspect 12B4A's internal resistance of ~1k ohms should make for a low enough -3dB down point (<10Hz).
2) Would the 20mA of standing current cause problems with the CG51AX-as-chokes?
3) Since a bias of -12.4V would work, what about two series 6.2V zeners for the cathode load? I read about that in Morgan Jones' book, in the part about cathode bias with LED's. I've tried LED's as cathode load and liked the results.
My plan for the psu is DC-regulated filaments (using an LM317) and Bryan's full-wave bridge using a pair of UF4007 and a 6X4, feeding CLCRCRC to get the voltage down low enough (135V for each 12B4A).
Simplistic, but should work, right?
--
I'm putting it up here to see if anyone wants to comment on whether it looks like it would work well.
I know a CCS in the plate would be the 'ultimate', but since choke loading of the plate seems to have its fan club, and I have a pair of UTC CG51AX plate-to-PP grid trannies I could use as plate chokes, I'd like to try them. They have primary winding of 22H with 1000 ohm DCR. I don't know what current capacity, but I expect 20mA wouldn't kill 'em.
My questions:
1) Is the L of these trannies used as chokes sufficient for a low enough -3dB point? I suspect 12B4A's internal resistance of ~1k ohms should make for a low enough -3dB down point (<10Hz).
2) Would the 20mA of standing current cause problems with the CG51AX-as-chokes?
3) Since a bias of -12.4V would work, what about two series 6.2V zeners for the cathode load? I read about that in Morgan Jones' book, in the part about cathode bias with LED's. I've tried LED's as cathode load and liked the results.
My plan for the psu is DC-regulated filaments (using an LM317) and Bryan's full-wave bridge using a pair of UF4007 and a 6X4, feeding CLCRCRC to get the voltage down low enough (135V for each 12B4A).
Simplistic, but should work, right?
--
Did a little homework. The 12B4A has an internal resistance of about 1000 ohms, and if I plug that into the handy equation for finding the -3dB down point of an amp with a choke in the plate...
X_L = DCR + 2πfL = rp
...with the 12B4A's rp = 1200 ohms, and the choke L = 22H and DCR = 1050 ohms, then...
6.2823(1)22 + 1050 = 1188.21
So, the resulting amp would be -3dB at about 1 Hz!
Even if I'm way off, and for some reason the actual rp of the 12B4A is 1500 ohms, that would still make for a -3dB point of only ~3.5 Hz.
This needs investigating. 🙂
Unless I'm totally wrong, of course... Anybody tried this?
X_L = DCR + 2πfL = rp
...with the 12B4A's rp = 1200 ohms, and the choke L = 22H and DCR = 1050 ohms, then...
6.2823(1)22 + 1050 = 1188.21
So, the resulting amp would be -3dB at about 1 Hz!
Even if I'm way off, and for some reason the actual rp of the 12B4A is 1500 ohms, that would still make for a -3dB point of only ~3.5 Hz.
This needs investigating. 🙂
Unless I'm totally wrong, of course... Anybody tried this?
Last edited:
I cannot comment on your math, but an inductance of only 22H seems extremely low. Good plate chokes (eg. Magnequest) are usually 100-200H as far as I know.
Ray,
If you want to play with operating points, just build a CCS with a pot so you can adjust it. I'm very happy with my CCS loaded LED biased 12B4 pre (running at 17ma). My first attempt was all resistor loaded, and I had problems. Even a fancy CCS doesn't cost much more than a cheap-ish plate resistor. Ditto for LED's versus a big resistor, and for me, they made all the difference. I would recommend a good filament supply. I was restricted to 6.3V AC. Though mine are quiet enough (after swapping a tube), they do seem noise prone. I wish I had the PT capacity and room for a DC supply.
Paul
Wild Burro Audio Labs - DIY Full Range Speakers
If you want to play with operating points, just build a CCS with a pot so you can adjust it. I'm very happy with my CCS loaded LED biased 12B4 pre (running at 17ma). My first attempt was all resistor loaded, and I had problems. Even a fancy CCS doesn't cost much more than a cheap-ish plate resistor. Ditto for LED's versus a big resistor, and for me, they made all the difference. I would recommend a good filament supply. I was restricted to 6.3V AC. Though mine are quiet enough (after swapping a tube), they do seem noise prone. I wish I had the PT capacity and room for a DC supply.
Paul
Wild Burro Audio Labs - DIY Full Range Speakers
Yes, 22H is very low. I have these two nice UTC plate-to-PP-grids trannies and I'm trying to find a use for them. They're CG grade so won't work for their intended purpose if true hifi results are desired.
Plate chokes of 100H and up are absolutely necessary for commonly used triodes like 6SN7, 12AU7, 12AT7, 6DJ8, etc. That's because their rp is 5k ohms or higher, creating a too-high -3dB point in the low end if used as a plate load.
For 6SN7, if rp = ~8k ohms...
rp = 2pi(f)L+DCR
rp = 6.2823 * 50Hz * 22H + 1050ohms = 7960.5
The -3dB point would be about 50 Hz. Use a 12AT7 and you could use the resulting amplifier stage as an active high pass filter for a subwoofer!
So correct me if I'm wrong, but I think a 12B4A should make even a weenie 22H choke work as a plate load.
--
- Status
- Not open for further replies.