| corbato |
I am not getting any younger and slowly it is dawning on me that I should discard the assorted riff-raffs of the past and build a gear that will see me through rest of my life. So here we are. The aim is to build a RIAA and Line stage that will not make me want to change it ever. Can we discuss possibilities with following broad parameters?
- No feedback. As far as possible.
- No interstage/output transformers
- Minimum stages
- Only use any or a combination of following tubes
- 6SN7
- 6SL7
- 6DJ8
- 12AX7
- 6267
My options are limited to the above set of tubes. So esoteric types will not be of much help. In past I’ve built several ‘generic’ types 12AX7’s and 6DJ8 topologies and somehow felt all of them lacking in transparency, detail and LF response.
Thanks in advance. |
|
|
| SY |
| Would the presence of a FET in there wound your pride? |
|
|
| corbato |
| Not at all..honourable Sir..not at all. |
|
|
| fdegrove |
Hi,
What types of cartridge do you need to cater for?
Fetsssss....I'll start throwing mayonaise your way soon...:smash:
Cheers,;) |
|
|
| corbato |
Oops.Should have specified that in the begining.
MM. I have Shure V15 and Shure M97. Nothing fancy. I'm assured of its continued availability here for a reasonable outlay. |
|
|
| EC8010 |
So you're looking at passive RIAA equalisation with valves biased for optimum linearity, and a cathode follower line stage to drive cable capacitance. 6DJ8 can make a quiet disc input stage, and 6SN7 would make good succeeding stages. Are moving coil input transformers forbidden, or is your cartridge moving magnet? What sensitivity is the power amplifier?
Edit: Ah, so you actually require 400-500pF of input capacitance. What is the capacitance of your arm and cable? |
|
|
| corbato |
I think I'll actually require 250 to 300pF as a compromise load.
My current and future amp will have an Input sensitivity of greater than 500mV. |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by corbato
The aim is to build a RIAA and Line stage that will not make me want to change it ever. Can we discuss possibilities with following broad parameters?
- No feedback. As far as possible.
- No interstage/output transformers
- Minimum stages
- Only use any or a combination of following tubes
- 6SN7
- 6SL7
- 6DJ8
- 12AX7
- 6267
|
I would say that the above is limiting enough that no matter what, you will be wanting to change.
You have eliminated input/output transformers, that presumably also includes Transformer Volume controls?
For a "limited" Phonostage I'd recommend this one:
http://diyaudio.com/forums/showthre...5624#post335624
Linestages are more difficult. How about a parallel 6SN7 or ECC88/6DJ8 with sone output to grid feedback to get the gain low, the output impedance downs and all? Take a 22k Anode load and individual 1k Cathode resistors plus bypass Cap, 1uF ouput cap and a 220k .... 470k Resistor from output to grid. Then a 100k Resistor from the grid to the 100k Log pot or directly attach the wiper of 470/500k or even better 1M Carbon Track linear Pot.
I have drawn this out in the attached piccie, this stage has around 6db gain << 1K Output impedance. You can pretty much do the same thing also with a ECC88/6DJ8 in the linestage of course, simply reduce the cathode resistors to 750R each.
I think if you slightly scale the Powersupply I suggested for the Phonostage you could use it for the linestage too. Not really my kettle of fish, but you set the groundrules to preclude almost all with any seriously good sonic potential, so what can I do? The results should be okay enough I guess.
Sayonara |
|
|
| fdegrove |
Hi,
Given the restrictions of the tubelist I personally favour the 6DJ8, both for phono as linestage.
You don't need the gain in a linestage but its high transconductance is always a good thing to have.
While we're talking linestages, I always wondered why no one bothers to design a linestage and headphone amp all in one box?
Instead of using a local NFB loop why not use a good CF and be done with it?
Just my twopence,;) |
|
|
| thoriated |
Hi, Ashok -
I would ditch the 6DJ8 and definitely not use any silicon anywhere near the signal path. Even one transistor and you've lost a lot of potential on your very best analog recordings, IMO. Maybe 6SN7's 6SL7's and 6BX7's or 6BL7's if you want low impedance/high transconductance perhaps with an initial gain stage followed by passive RIAA eq, then a second gain stage/buffer driving a passive gain control with an otherwise passive architecture would seem most satisfactory. I did something somewhat similar about ten years ago and there's no comparison with anything commercial I've heard when playing LPs. |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by fdegrove
Given the restrictions of the tubelist I personally favour the 6DJ8, both for phono as linestage.
|
I find all else being equal actually mixing different valves is a good idea. The ECC83/ECC88 Phonostage (BTW, for hopeless Octalists - yes, you can use a 6SL7 in the first stage) tends to work quite good because the two valves have pretty much "opposite sonic personalities". Combine that with the moderatly clean, but normally rather boring sounding 6SN7 in the linestage with a little feedback to get the measured and audible performance up and you have a "nice combo" for a full function preamp.
| quote: | Originally posted by fdegrove
Instead of using a local NFB loop why not use a good CF and be done with it? |
Because even the Allan Wright / Joe Rasmussen "SLCF" sounds less that ideal. If an active linestage at all I'd avoid the follower. Quite an easy job to have a listen. Just my sixpence of course and views/tastes differ.
Sayonara |
|
|
| fdegrove |
Hi,
| quote: | | I find all else being equal actually mixing different valves is a good idea. |
In principle I'm in favour of that philosophy as well provided you know your valves and their sonic fingerprint as well as you seem to do.
For a MM phono stage the ECC83-88 combo looks a good idea.
If it were a MM phono stage I'd approach it in an entirely different way.
| quote: | | Because even the Allan Wright / Joe Rasmussen "SLCF" sounds less that ideal. |
Never had the chance to listen to one so far which is why I'm still very much in favour of the WCF which comes as close to neutral as its passive components allow for to my ears.
| quote: | | Just my sixpence of course and views/tastes differ. |
No arguing there...
Maybe time for us to open a thread revealing what we expect our gear to do sonically.
Cheers,;) |
|
|
| EC8010 |
I can't help feel that there's been a lot of fuss and bother before full requirements have been determined.
Does the pre-amplifier need to drive significant capacitance?
Is the arm capacitance confirmed as 175pF?
Is only LP required?
(The input requirements might allow a 6SL7.)
(The output requirements might allow a common cathode 6SN7.)
Let's find out what it needs to do. Then we can argue about the best way of doing it. |
|
|
| fdegrove |
Hi,
Drrrrrrrr....scrolling back to post # 1 I read:
| quote: | | I am not getting any younger and slowly it is dawning on me |
that I should definitely try out that latest HGH hormone that promises me eternal youth...Ooops, wrong forum.
Seriously, given the Shure as an healthy MM with good track record (pun intended)...
Yup, a 6SL7 and a 6SN7 would make a nice preamp capable of phono and line.
Naturally, I'd stick more valves in the PSU than in the circuit proper...That's just me though...
Errrrr....why is it so important to know cable capacitance before we even put pencil to paper?
Cheers, ;) |
|
|
| EC8010 |
| quote: | Originally posted by fdegrove
Errrrr....why is it so important to know cable capacitance before we even put pencil to paper? |
Early Shures are critical on load capacitance, and need 400-500pF to get their electrical equalisation of the transducer correct. Later cartidges (post CD-4) tend towards 250pF. Oddly, the significance is that an early Shure allows you to get away with something that a later cartridge wouldn't allow. I suspect that Corbato's Shures could be driven very nicely into a 6SL7, but we need to know the cable capacitance precisely to see if the cartridge will tolerate the Miller capacitance plus that cable capacitance. |
|
|
| fdegrove |
Hi,
| quote: | | I suspect that Corbato's Shures could be driven very nicely into a 6SL7, but we need to know the cable capacitance precisely to see if the cartridge will tolerate the Miller capacitance plus that cable capacitance. |
Okidoki...
I usually consider it good practice to keep phono leads a short as feasable and use as low capacitance cable as I can get my hands on even if that means a little added inductance.
Some SL7s can be quite extraordinary as you may know already...
Thanks for the 'splainin',;) |
|
|
| tubetvr |
| quote: | | but we need to know the cable capacitance precisely to see if the cartridge will tolerate the Miller capacitance plus that cable capacitance. |
An alternative is to use feedback around the input stage, feedback connected to the cathode of a cathode grounded stage lowers the miller input capacitance with the same amount as the feedback ratio, this fact is often overlooked when discussing RIAA phono stages.
Regards Hans |
|
|
| EC8010 |
| You're welcome, Frank. The thing that surprised me was that insufficient capacitance was actually worse than too much (hence the Ortofon add-on capacitor). |
|
|
| fdegrove |
Hi,
| quote: | | The thing that surprised me was that insufficient capacitance was actually worse than too much (hence the Ortofon add-on capacitor). |
Oh, yes...Absolutely. Not enough C and you'll end up with a rising top end wondering who stole the bass.
Hans,
You're absolutely correct but Corbato expressed a wish for as little FB as need be.
Cheers,;) |
|
|
| tubetvr |
| quote: | | You're absolutely correct but Corbato expressed a wish for as little FB as need be. |
Ouch! I didn't see that, I just wanted to point out the fact that also input capacitance is reduced by feedback as many people doesn't seem to remember that, fore instance even Morgan Jones in his book "Valve amplifiers" doesn't mention this fact and is discarding 12AX7's due to high Miller capacitance.
Regards Hans |
|
|
| Positron |
Hi Corbato,
Using a 100k with 100k resistor will severly lessen the high frequency response. The feedback circuit won't work on this problem. Look below at some figures of high frequency loss.
Consider a stage of gain of only 10, with 4pf grid to plate capacitance (two tubes in parallel so twice the gp capacitance) and no stray capacitance (very optimistic). I have computed (computer program) the -1db and -3db responses.
1) with 100k volume control set to midpoint
-1db = approx 6khz
-3db = approx 14khz
2) 100k pot set with wiper 30k from ground to wiper arm (70k to source) and 100k resistor to grid
-1db = approx 6khz
-3db = approx 14.5khz
3) 100k pot set with wiper 30k from ground and 70k to source, 1k to grid
-1db = approx 43.5khz
-3db = approx 86khz
Test "2 and 3" seems ok but as the volume pot is varied, the HF response will change dramatically as indicated in the tests. Since you don't know where the volume control will be set, the HF response can vary wildly. I would definitely not use a 100k resistor to the grid.
Also remember that all other sources, stages also degrade the HF performance and -db are additionally cumulitive. The figures I present don't even include the HF response of the 6sn7 to the following stage (amp input including the IC capacitances).
Hope this helps. |
|
|
| tubetvr |
| quote: |
Using a 100k with 100k resistor will severly lessen the high frequency response. |
Yes it will, it is not a good solution in this case.
| quote: | | The feedback circuit won't work on this problem. |
It will to the same degree as the gain is lowered, the Miller capacitance is determined by the stage gain, not by u.
A better solution for a wideband line stage is to use for instance 2 gain stages with heavy cathode coupled feedback, in that case the Miller effect can be in practise eliminated, another solution is to use a cathode follower as an input stage. The idea of a large resistor in series with the grid is a bad idea anyway as it is difficult to get below ~5-10pF including stray capacitance, even without a resistor a 100kohm pot is not ideal if you want to have high constant bandwidth.
My own line gain stage has an open loop gain of ~2000 but I use heavy feedback to reduce this to as low as 4 when I use it to amplify signals from a CD player, the input capacitance is of course neglible in this case and approaches Cgp only. I use a 100kohm pot and I can achieve a constant bandwidth of ~200 kHz which is more then enough for me.
Regards Hans |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by EC8010
I can't help feel that there's been a lot of fuss and bother before full requirements have been determined.
|
I don't think so. The "requirements where set out clearly:
1) Phono & Line (eg. generic "full function" preamp)
2) Use certain Valves
3) Don't use Interstage etc. Transformers
| quote: | Originally posted by EC8010
Does the pre-amplifier need to drive significant capacitance?
|
Invariably the answer is that at some point in time it MIGHT BE required to do so.
| quote: | Originally posted by EC8010
Is the arm capacitance confirmed as 175pF?
|
Most Moving Magnet Cartridges require a lot of capacitance as load (even modern ones) to give a balanced response. This is becaus ethey all use the resonance between their Coils inductance and the external capacitance as resonant circuit to bring the High Frequencies up.
Most MM are in the region of 0.5 - 0.8H and require a resonance at around 15 - 18KHz. This implies quite high capacitive loads, old SME Arms came with a PAIR of 250pF Capacitrs conneted in parallel to the cable, if there was not enough treble you simply clipped one out.
Anyway, in my suggested ECC83/ECC88 Phono Circuit the ECC83 operates at a stage-gain of around 37 (due to local feedback and lowish anode load), resulting in around 60pF Miller amplified input capacitance plus a few puff (pF) extra for the other capacitances. Even with 175pF Arm/Cable capaictance this is by far too low to work with most MM's, requiring extra capacitive load. So the Miller Capacitance is not an issue.
I still feel that the choice of Valves and refusal to allow transformers will severely limit the possible performance, but I think it is not difficult to fulfill the requirements.
Sayonara |
|
|
| corbato |
| quote: | Originally posted by Kuei Yang Wang
How about a parallel 6SN7 or ECC88/6DJ8 |
Thanks Thorsten Sahib for the schematic. I think I'd avoid teh 100K input grid resistor and look out for 500K pot. Your suggestions pls?
| quote: | Originally posted by thoriated
Maybe 6SN7's 6SL7's |
Exactly. Particularly when I realise that I've an almost unlimited source of NOS 6SN7/6SL7's.
| quote: | Originally posted by fdegrove
that I should definitely try out that latest HGH hormone that promises me eternal youth...Ooops, wrong forum. |
I have an excellent source of pure audiophile grade viagra. Want some :)| quote: |
Naturally, I'd stick more valves in the PSU than in the circuit proper...That's just me though... |
Tube regulated supplies? With OA2's and such. Why not.
The pre amp will be placed just next to the TT and will have outboard PS.
Ok. While its emerging that mix and match tubes are in thing, how about this for a Phono? |
|
|
| SY |
Cascodes in the first hole do have an advantage in that the Millering problem is much reduced. Getting low-noise performance from 6SL7s can be quite an exercise and one I might hesitate to undertake. The reason I asked about FETs before is that the bottom of an input cascode is the perfect place to use a low-noise FET. Let the tube do what it does best, swing a lot of voltage, and the FET do what it does best, drive current out of its drain with a high gm. Low Millering means great flexibility in choosing input capacitance.
For mm use, a VERY inexpensive FET which is easy to get and very quiet is the NTE458. In my preamp, I cascode this with a 6DJ8-type tube on the top. |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by corbato
Thanks Thorsten Sahib for the schematic. I think I'd avoid teh 100K input grid resistor and look out for 500K pot. Your suggestions pls? |
Vishay/Sfernice P11 Cermet track.
| quote: | Originally posted by corbato
Tube regulated supplies? With OA2's and such. Why not.
|
I prefer passive filtering with enough capacitance and high quality cap's, especially in the final position.
| quote: | Originally posted by corbato
Ok. While its emerging that mix and match tubes are in thing, how about this for a Phono |
Hmmm. the 6SL7 in the first stage is not going to bias at much voltage and the current will be low. That will severely prejudice any FET you could possibly use there. You could lift up the 6SL7 Grid of course, but that will add complexity and will still not solve the low current issue. In the interest of overload margins and noise I'd swap the positions of the two RIAA.sections as well
Secondly, most low noise FET's have a rather stratospheric transconductance which would make your circuit have by far too much gain, even for MC Pickups, let's say a FET with a gm (Yfs) of 10mA/V is used, the gain of the first cascode stage will be 1000. Even if we follow this with the 3180/318uS RIAA section the signal send into the forst 6SN7 would be at around 100 times the input voltage, then amplified by at least aanother factor, so for a MM Input you would have around 60db gain after the second stage and 80db after the third stage. Even or MC pickups this is BY FAR excessive.
Further, the final stage will have a rather high output Impedance.
If you insist on having a J-Fet in there (which is not at all needed for MM BTW) drop at least one stage, use a compound RIAA similar to that shown in my Phono and use 6SN7 in both stages, bypass the second stages cathode resistor, 47k Load, 1K cathode resistor & 250V +B.
The J-Fet should be 2SK170 with around 5mA Idss (BL Grade), no source resistor and the Cascode 6SN7 should have around 10k Anode Load. The 6SN7 should bias up suitably to have enough operating Voltage for the FET. The resultant Phono will have minimal input capaitance (a bad thing for a MM phono BTW as you will need to add load capacitors) and around 54db gain.
Sayonara |
|
|
| SY |
Biasing up the grid of the first stage tube is pretty easy. In my preamp, there's a 12V rail used to power the heaters. I bring a 680K resistor from that rail to the first stage grid, then bypass the grid with a 0.33u cap. Two extra parts, no big deal.
The transconductance issue is similarly easy to dispose of. First, recognize that the gm is nowhere near the spec if you're not running at idss. Second, because of the levels here, you can sacrifice a miniscule degradation in noise to have some degeneration in the source lead. For the FET I specified (NTE458), a 6DJ8-family tube, a 620 ohm source resistor, a 100K plate resistor, and a 230V B+, I get a gain of 110, almost exactly what I targeted. The stage has no trouble swinging +/-50V, so overload is not likely to be an issue. Noise is low enough that at normal volume settings, I need to get my ear within a few inches of the speaker to hear any noise. |
|
|
|
