| dsavitsk |
I am looking to put together a low gain preamp. I do want a little gain so a cathode follower is not going to work. One suggestion I came across was to use a 71a in a basic grounded cathode config (I think the original suggestion was to use a choke on the plate.)
However, another post suggested that miller capacitance could be a problem in this setup. Could someone give me a 5 line primer on what this means exactly and perhaps help me determine if this really is an issue.
Also, if anyone has other suggestion for a low gain preamp, I'd be interested in those too, but the 71a has me intrigued.
Thanks,
-d |
|
|
| Eli Duttman |
The Miller Effect amplifies a triode's grid to plate capacitance by mu. The Miller capacitance is part of the load the upstream circuitry has to drive.
Search this site and other sites for information on a 12B4 based line stage. You will get everything you want, without the problem of a LARGE Miller capacitance. |
|
|
| kevinkr |
Hi Dave,
The 71A has a cgp of 7pF and a mu of slightly less than 4 which would result in a miller capacitance of worst case about 28pF - in most reasonable designs with a 100K pot or less this is not going to cause much of a problem. (Output loading will reduce the effective mu making the miller cap even smaller.)
The 71A is a great sounding tube but will probably be a little microphonic in this application.
I published an article in PFO a while back about my 26 based transformer coupled line stage. There is information you might find useful in your quest for a 71 based line stage. You can get to it easily from my site. (below)
It's definitely do-able and if you are up to the challenge why be derivative. :D |
|
|
| Eusebius |
First, all the DHTs I have used sounded better than a 12b4, so although this is a nice tube and has many fans including me, for better sound why not go with a DHT as you suggest. I tried out a whole load of DHT tubes in a line stage, and the 71a came pretty far down the list. If you want low gain, you could try the 31 (mu 3.8). The 49 in triode (mu 4.7)is intriguing - didn't really set it up properly (lots of flying leads to a 4 pin socket) , very detailed but maybe a bit sharp. Frankly I prefer some of the other DHTs anyway to all the above, but they're all about mu=9.
Filament supply will have a huge effect on the sound. There was a thread here on 300b filament supplies which was a very interesting read. Basically the best sound is a current source of some description. the RonanReg, for instance was two voltage regs followed by a current source. A current mirror also sounds good. I tried common mode chokes at the end of current sources but the sound was worse. Interestingly CMCs improved the sound from plain voltage regs. |
|
|
| kevinkr |
Filament supply is critical to both the sound and the noise floor. A ronan type regulator would be an excellent choice, there are others particularly if you are willing to experiment.
I have used foil chokes with voltage regulators (not common mode) with good results (mainly for rfi reduction)
Good tubes to consider with transformer coupling which will give you generally less than 6dB of gain are the 30, 26, 12, 01A.
With chokes the 71A, 45, 6AS7GA are good choices. |
|
|
| Eusebius |
Good tubes to consider with transformer coupling which will give you generally less than 6dB of gain are the 30, 26, 12, 01A.>>
Yes - I agree completely with that. 30 is a leaner sound and very clear, 26 is warmer, richer and more lush, 01a is clear and neutral, 12a is similar. There's a family resemblance, with the 26 rather different in sound - warmer if you like that. All good. I found the 30 less microphonic than the 31, though others report the opposite. All my 31s would ring, though it dodn't come through the speakers. The 30 seemed a bit less prone to this.
I did also try the 6B4G (Sovtek) and 2a3 (plain chinese) in my line stage, but preferred the smaller tubes. But in my opinion even the 2a3 was better than a 12b4 or other indirectly heated triodes. Going back to indirectly heated tubes seems right away to make the sound "thicker" - only way I can describe it, that last bit of clarity just goes (I tried a 30 compared with a 12b4 with a friend and we both agreed). Unless you have just switched back from a DHT you don't notice the slight cloudiness of indirectly heated tubes. And let's face it, how many people have even heard DHT small tubes in line stages. I was happy for years until I did. The first time I heard a 26 line stage it was a shock - it's quite possible my draw literally dropped. |
|
|
| dsavitsk |
Thanks for all the suggestions. I do tihink the 71 is worth trying, but some of the others may get a round as well. If microphonics are an issue, does anyone have a good strategy to reduce them?
| quote: | Originally posted by kevinkr
Good tubes to consider with transformer coupling which will give you generally less than 6dB of gain are the 30, 26, 12, 01A. |
Can you suggest a transformer? I have had trouble finding a transformer to use for a line stage.
| quote: | | Originally posted by kevinkr [b]With chokes the 71A, 45, 6AS7GA are good choices. |
Hmmm. I had a run in with the 6080/6as7 in a line stage a while back and found it to be a very noisy tube. I tried a ton of different tubes and different setups and could not get it quiet.
The 45 looks great too, but is a little too expensive for now. |
|
|
| Eusebius |
To reduce microphonics do stuff like the following:
a) Enclose the tubes, either in the chassis or some kind of tube over them
b) Try O rings
c) Make the chassis as heavy and dead as possible - I got good results with mounting the sockets straight into 4mm alu plate. The chassis should not ring when tapped.
d) If you want to be extra careful, try something like this - worked for me. Mount the two tube sockets on a rectangular 6 mm teflon sheet. You will secure this with two holes drilled across the middle, and on the two ends put rubber under the teflon, using cut pieces of a normal stationary type eraser. Put a couple of solid bolts through the bottom of the chassis and tighten with nuts on the inside of the chassis, so the bolt ends are sticking up. These go through the holes in the teflon, but cut the holes bigger so you can put a rubber grommet in each, so the bolts don't touch the teflon. Put larger rubber grommets top and bottom of the teflon and finally bolt the lot down so the teflon goes down and is held at the two ends by the rubber pieces. The teflon is non-resonant - you could use some other material - and doesn't touch the chassis because of the rubber grommets.
e) Put the chassis on some kind of soft rubber feet or other dampening material like sponge. |
|
|
| kevinkr |
Hi Dave,
Both Magnequest and James make suitable transformers. I'm using vintage HA-133 which have gotten a bit ridiculous in cost lately.
01A/12A/26/30
15K:600 ohms will give you 4dB of gain with the 26 or 01 and a source Z around 300 ohms. See the table in my article for further details. Note that I don't normally terminate into 600 ohms, but more like 10K - 100K so effectively the limiting factor on gain is the mu divided by the transformer ratio.
It's important to make sure the transformer has relatively high primary inductance to assure low distortion at low frequencies. I'd say something like 150H (xl = 18.85K ohms @ 20Hz) as an absolute minimum, this is critical with dhts with an rp of about 8K as is the case with the 26 and 01, somewhat less critical with the 12A as it's rp is about 5K..
71A/45
A transformer can be used here too, in this case something on the order of 5K ought to suffice. Here you will need about 50H for good low end performance, and you will get right around unity gain. Choke loading seems like a good choice here and will give some gain as well. 100H chokes are quite practical at the currents needed here and a lot of the problems of using transformers just go away, at the expense of considerably higher source z, (basically rp) and increased noise/microphonics due to the missing attenuation from the plate to line transformer.
Pretty much everyone around here (Northern New England Tube Group or NNETG as we are known) has built a 12B4 except me, and I have a/b'd many of these extensively and just don't need to go down that road.
I have heard one 12B4 pre-amp I really liked and it used a high transconductance pentode as the upper device in a dissimilar mu follower confiuration. Quite unusual, and the best sounding 12B4 design I have heard to date.
|
|
|
| mach1 |
Dave,
I would strongly urge you to use a separate filament transformer when using DHTs. On my 1LE3 pre, which uses a ronan reg for LT, the amount of hum, ringing and hash that came through when using a common transformer was unbelievable.
Using a separate LT transformer has dropped the amount of garbage considerably, but it still quite audible - especially an amplified version of the high pitched buzzing sound that emanates periodically from transformers.
pm |
|
|
| Eusebius |
More on microphonics -
I've had very good results with diff pairs in the line stage, which were mounted on a 4mm thick alu top plate - simple as that, just mass, no rubber damping.
I'm getting worse results with a SE setup in a normal type preamp case, 2U high with a 3mm alu bottom plate - seems to ring more, even when I put the tubes on a teflon sheet decoupled with rubber from the chassis. that did help, because it was worse before, but I'm going back to thinking you need mass. Like thick alu, granite, whatever. I'm starting to believe that the whole chassis must be quite dead - no ringing anywhere.
Otherwise the microphonics comes though the volume control, cables, all sorts of things. Anything that rings seems to set it off.
If anybody has got further than I have with this, do let me know. I was in the process of selling a line stage to a friend which we both agreed sounded great and I ended up putting it on a cushion and chucking another cushion on top of it (that silenced it!!). I took it back to the bench to solve the problem, so have to make some progress to make the sale. Andy |
|
|
| dsavitsk |
My question re microphonics was actually meant to mean wheather there was anything one should do with a DHT as opposed to other tubes. In other contexts, to reduce microphonics, I have lined the inside of a chassis with dynamat which worked pretty well.
As for transformers, thoughts on these? http://jianshin.myweb.hinet.net/PRE_OUTPUT_TRANS.htm |
|
|
| kevinkr |
Hi Dave,
You probably want to look at the local James distributor's web site if you can find it. (I had it, but with the last crash some months ago lost all of my bookmarks.)
I assume you are talking about the 71A as a candidate here, in which case 5K is a good choice, however it is not clear to me whether these are SE or PP transformers. Unless building a PP pre you don't want a PP transformer as it is not designed to cope with the quiescent dc in an SE application.
Don't expect much voltage gain with a 71A and an output transformer as opposed to a choke. You will however get much lower output Z.
Heavy mass loaded chassis, possibly shock mounting the dhts, and power supply remotely located work well. |
|
|
| Sherman |
For James transformers check out Euponia Audio, that's where I got mine! ;)
(No association, just a satisfied customer.) |
|
|
| Eusebius |
More on microphonics:
A friend just reminded me NEVER to go over the rated filament voltage, and in fact to stay 5 per cent at least under it. Reason being that DHTs can start to scream above their design filament voltage. I'll have to go back and check mine. He also thought it was possible that in diff pairs the difference in frequency of ringing might cancel out to some degree - certainly was less of a problem with diff pairs. All this is worth noting - you can have some nasty surprises with DHTs in line stages, like really crazy feedback so you have to quickly shut the amp off. |
|
|
| dsavitsk |
| quote: | Originally posted by kevinkr
Hi Dave, |
It's Doug by the way. :)
| quote: | You probably want to look at the local James distributor's web site if you can find it. (I had it, but with the last crash some months ago lost all of my bookmarks.)
I assume you are talking about the 71A as a candidate here, in which case 5K is a good choice, however it is not clear to me whether these are SE or PP transformers. Unless building a PP pre you don't want a PP transformer as it is not designed to cope with the quiescent dc in an SE application.
Don't expect much voltage gain with a 71A and an output transformer as opposed to a choke. You will however get much lower output Z. |
The plan is to try 71a with a plate load of either a ccs or a choke, or maybe just a resistor in the first go to see what happens. I am investigating transformers in case I decide I want to go that route with a different tube. And, for now, I have some edcor 10K:600 transformers that I can use to experiment. Just figured I'd do the reserach to figure out where to get thjem, how much they are, etc. so I can plan a bit.
| quote: | Originally posted by Sherman
For James transformers check out Euponia Audio, that's where I got mine! ;) |
I do have a pair of james transformers from euphonia for a headphone amp (still waiting for some parts to complete the assembly) but they don't seem to have the preamp type in stock. I have seen a few posts that James will begrudgingly send you some at a good price if you bug them enough, but otherwise I can't find any other distributors.
-d |
|
|
| EC8010 |
| quote: | Originally posted by kevinkr
The 71A has a cgp of 7pF and a mu of slightly less than 4 which would result in a Miller capacitance of worst case about 28pF |
Your mind must have been on other things; if Cag = 7pF and mu = 4, then at best the Miller capacitance will be Cag(1+mu) = 35pF. Plus Cgk, probably about 5pF, means 40pF is a likely figure. Not a problem provided that the volume control pot is 100k or less. |
|
|
| kevinkr |
EC8010 It must have been one of those days I guess, you are of course absolutely right Cag(1+mu)=Cmiller.. Ouch....
And Doug my apologies for calling you Dave, having been suitably chastened I won't do it again.. :D
You might want to take a look at some of the Magnequest wares as I suspect by the time you factor in shipping by anything other than surface a pair of James transformers shipped from the Far East is going to be in the same price range. Mike's plate to line transformers are very good. (from experience) |
|
|
| dsavitsk |
| quote: | Originally posted by kevinkr
Good tubes to consider with transformer coupling which will give you generally less than 6dB of gain are the 30, 26, 12, 01A. |
A lot of these tubes (30, 26, 01A at any rate) seem to have very similar operating points but different filament voltages (the datasheets all suggest 135V, a grid bias around 8-10V, and a plate current between 3 and 5.5mA). Is it reasonable to build an amplifier for one or another and (adjustng the filament supply) simply swap between them? Obviously I am getting ahead of myself here, but I am curious if there is some clever bit I am missing.
Doug |
|
|
| Brian Beck |
| quote: | | EC8010 It must have been one of those days I guess, you are of course absolutely right Cag(1+mu)=Cmiller.. Ouch.... |
Well, since we’re getting picky about the definition of the Miller effect, the multiplier is really based on voltage gain, not mu, and it is 1-Av, where Av is the voltage gain (note that Av is negative for common-cathode stages). Only if the plate load is >>rp, AND if the cathode resistor is bypassed, will Av approach -mu. In those cases, 1-Av = 1+mu, as you said. Otherwise the magnitude of Av will be less than mu and would have to be calculated using the usual equations. For quick and dirty estimates of Miller capacitance, I agree that 1+mu is a conservative guess. I know that EC8010 and Kevinkr know this, but just to make sure the record is straight for others… And I’ve had nothing else to post this morning yet :( |
|
|
| Eusebius |
Is it reasonable to build an amplifier for one or another and (adjustng the filament supply) simply swap between them? >>
Most definitely. This is easiest with a UX4 base. This gives you a very large choice of tubes. When I did all my initial tube testing I just used a couple of adjustable DC bench supplies, and dialled in the filament voltage. In this way I tried out the whole range of UX4 tubes. Sound is quite adequate with a good DC bench supply.
When you go over to building your proper filament supply, you can put in small pots to adjust the voltage up and down, but that's a bit trickier, and you'd need test points to check the voltage as well. Easier to do the tube rolling with an adjustable bench PSU - you have a display for the voltage and usually the current too. I have some Thurlby Thander digital ones which are nice. These are always on ebay. If you can, go for a couple of 15v at 4A. Failing that, bear in mind that a 26 is 1 amp (a common 30v 1 amp supply will work on max, since the voltage is low) and you may want to try something with higher current than that. Andy |
|
|
| kevinkr |
Hi Brian,
All true of course, and in most instances the AV will be considerably less than mu and consequently the miller capacitance will be smaller too.
Notable exceptions are any triodes loaded by a ccs where AV should very closely approximate mu at least in the case where the ccs represents the ac load impedance on the tube.
The other is with transformer coupling in the case where the load impedance is a lot higher than the intended load impedance, such would be the case for any transformer coupled line stage running into a load impedance >10x the intended load impedance. My nominally 600 ohm line stage output for example is running into 70K.. In this case the primary inductance is concern at low frequencies, otherwise the plate sees a pretty high load impedance.
Doug,
Yes that was my thought too, simple reconfiguration of the filament supplies will allow you to use the 26, 01A, 12A and 30 interchangeably. I use a small 5V switcher with an additional pi filter to power the filaments in mine, and with a minor mod to add a switch could easily accomodate the 26, 01, and 12A.. |
|
|
| arnoldc |
I think I'd join the party :D
My friend texted me yesterday and told me that I have a pair of Tungsol 26 waiting to be picked up, gratis :cool:
I have a pair of 01A already, so yeah, a cool DHT preamp will be born soon. ;) |
|
|
| dsavitsk |
Some experimentation has commenced. I picked up a pair of 01-a's and some magnequest B7 parafeed transformers. On a breadboard I loaded the plates of the tubes with a ccs at about 3.5mA, used a lm317 circuit on the filaments, and the noise and microphonics are really bad. A truck drove by on the street and the tubes picked up the vibration + the amount of background hum and noise is really high. The PS is in a separate chassis and is about 2 feet from the circuit, so I guess the place to start is with the filament supply and making the wiring neater -- it is just a bunch of clip leads right now. I also tried w/o grid stoppers, so maybe I'll add those to see if they help. I am not sure I have good feelings about this pair of tubes though.
-d |
|
|
| Eusebius |
Do you have a cathode bypass cap? That's one place to start - my SE stage sounded horrendous without it.
If all else fails, use differential pairs with a CCS underneath. I've had very little problems with this arrangements over a very wide array of DHTs. Even the humble 3A5 which I'm listening to right now sounds very good and is completely quiet. I put 3 O rings on it and it's fine. It's just mounted on a 4mm alu plate - no special suspended chassis or anything. Designs for CCS can be taken out of Morgan Jones book. For SE you can ground one of the grids on the input and take the output from one of the plates. Incidentally, I thought the 3A5 sounded better than two other common cathode double DHTs, the 3B7 and the 1J6.
If you're having a lot of trouble I really do recommend differential pairs of DHTs as a way of dealing with microphonics - it's not something you'd normally consider unless you have a balanced system (like me) but it works. |
|
|
| dsavitsk |
| quote: | Originally posted by Eusebius
Do you have a cathode bypass cap? That's one place to start - my SE stage sounded horrendous without it. |
I've attached the basic schematic I am using.
There are a bunch of things that clearly can be improved -- the 6.3V winding feeds both the ez81 and the lm317 based vreg which should be changed, I probably need some work on the cathodes of the tubes, the filament PS could be better, etc. Also, this is on breadboard and is connected with lots of clip leads. But, the question is whether these things will fix things, or whether I just have a noisy pair of tubes. I built the same circuit with 6n6p's where it worked great. Obviously there is an extra layer of work with the DHTs, but I am a little weary of plunging in too deeply. Anyone have a good way to determine if a particular tube is just inherantly noisy?
| quote: | Originally posted by Eusebius
If you're having a lot of trouble I really do recommend differential pairs of DHTs as a way of dealing with microphonics - it's not something you'd normally consider unless you have a balanced system (like me) but it works. |
The addition of grid stoppers did seem to help with microphonics a bit. One weird thing is that while there is a constant noise (60Hz, 120Hz?) there is an extra noise at about 1 second intervals that sort of sounds like a space age pinging with delay and reverb. It is very odd.
Anyhow, suggestions greatly appreciated. |
|
|
| JoshK |
| With regards to mister miller, although it requires an extra tube/channel why not try out the common-cathode circuit with DHTs? Broskie's site has some good info on it that I recently read and it seems like you get less distortion than GC and Miller C is shielded. |
|
|
| dsavitsk |
A few updates.
I swapped out the 5V linear filament supply for a switcher i had around (the one that came with my Squeezebox.) I have no idea the qualty of this PS, though I assume it is low. However, I assumed that since it is a switcher that all the noise would be HF noise and would at least be different and would show where the problems were coming from. Counter intuitively (to me anyway) the noise went way up with the switcher. I would guess that it is 120Hz noise, though I have no easy way to measure that.
The other thing that was surprising is that turning off the PS but leaving the filament supply on -- essentially letting it run from the cap bank until they discharged, all noise went away and it played pretty clear music. However, as it was about to run out of power, the noise crept back in.
I don't know what these two things mean, but I am hoping that someone else does and that they offer a clue of where to start.
Thanks,
-d |
|
|
| Eusebius |
| Have you tried the cathode bypass cap yet? |
|
|
| kevinkr |
Sounds like your plate supply is noisy (lots of ripple?) or the power transformer is close enough to the tubes to cause problems. 01A are very sensitive to both magnetic and electrostatic pickup.
You should definitely add a cap to bypass the cathode bias resistor. Note that the nominal plate resistance of 8 - 9K may be triple this amount if the cathode resistor is unbypassed depending on the value of that resistor.
I usually use a pair of small resistors (22 ohms typically) in series across the filaments and connect the bias resistor and cap to the center node. I have found that this can help a lot with noise on the filament supply - it may or may not help here. (It's just about mandatory if ac is used for heating, using ac is obviously not something you should do in a dht pre-amp.)
I actually prefer fixed bias via batteries applied to the grid, I use something like a 220K grid resistor and a 9V battery - works really well.
I have also tried nicads in the cathode circuit, they are ok, note that nimhs seem to sound really lousy. (Interestingly higher internal impedance than good nicads) Normal operation will keep them charged but they do need to be replaced every few years. |
|
|
| dsavitsk |
| quote: | Originally posted by kevinkr
Sounds like your plate supply is noisy (lots of ripple?) or the power transformer is close enough to the tubes to cause problems. 01A are very sensitive to both magnetic and electrostatic pickup. |
This is what is weird. the PS is about 2 feet away in a separate chassis, and the supply is reasonably clean, and the CCS should clean it further.
| quote: | Originally posted by Eusebius
Have you tried the cathode bypass cap yet? |
| quote: | Originally posted by kevinkr
You should definitely add a cap to bypass the cathode bias resistor. Note that the nominal plate resistance of 8 - 9K may be triple this amount if the cathode resistor is unbypassed depending on the value of that resistor. |
I'll try that tonight. I was really hoping to keep the electrolytics out of this thing, but I'll see. I'll also try adding a pair of 20R resistors on the cathodes. Would the higher plate resistance lead to more noise here?
I actually prefer fixed bias via batteries applied to the grid, I use something like a 220K grid resistor and a 9V battery - works really well.
| quote: | Originally posted by kevinkr
I have also tried nicads in the cathode circuit, they are ok, note that nimhs seem to sound really lousy. (Interestingly higher internal impedance than good nicads) Normal operation will keep them charged but they do need to be replaced every few years. |
If all else fails, I'll give this a shot too. |
|
|
| Eusebius |
I was really hoping to keep the electrolytics out of this thing>>
Sure - I use 47uF polypropylenes for cathode bypass. |
|
|
|
