Hi,
Yes, and that's the main source of the problem. I suggest you fix that first then we could be discussing an electronic circuit.
Horse and carriage thing, remember?
With data input from whom? Sorry, but most of the contributors here can point to at least a dozen items where this circuit can easily be improved. Noise included.
Ciao,
Yes, and that's the main source of the problem. I suggest you fix that first then we could be discussing an electronic circuit.
Horse and carriage thing, remember?
With data input from whom? Sorry, but most of the contributors here can point to at least a dozen items where this circuit can easily be improved. Noise included.
Ciao,
Frank, I will consider to buy 1:100 or 1:150 or 1:200 SUT. However in current setup two transformers make 1: (37*5) = 1:187 complex transformer. Yes its impedance is complex and it can affect frequency response. But how it can cause flicker noise in this preamp? Same did not caused any flicker noise in previous 12AX7 preamp that was mostly identical except tubes and resistors.
Yes, may be film caps are oversized, yes electros are bypassed, may be it is not necessary, but does it cause pink noise?
I can spend 170 bucks for 1:200 SUT and hours of work to replace these SUTs and replace the oversized caps with smaller ones. Will it decrease flicker noise?
What other "dozen items where this circuit can easily be improved" are? May be I could start with them?
And another thing, a single 1:200 SUT will require very short mic cable, or active load right inside the mic - means little FET preamp. That was not my initial target. I wanted "all tube" preamp.
Yes, may be film caps are oversized, yes electros are bypassed, may be it is not necessary, but does it cause pink noise?
I can spend 170 bucks for 1:200 SUT and hours of work to replace these SUTs and replace the oversized caps with smaller ones. Will it decrease flicker noise?
What other "dozen items where this circuit can easily be improved" are? May be I could start with them?
And another thing, a single 1:200 SUT will require very short mic cable, or active load right inside the mic - means little FET preamp. That was not my initial target. I wanted "all tube" preamp.
Last edited:
Hi,
Two xformers with a long capacitive cable in between are bound to wreck much more havoc on that tiny little signal than a single higher ratio one.
So, yes please do yourself a big favour and go for the best one for the job, you know your ribbon's output so ask Lundahl for the best possible one. Mount it, terminate it according to the manufacturer's recommendations and we're talking.
Then let me know the signal level, if you want to keep the balanced XLR input, need a balun 1:1 xformer or whatever and any other data needed to make a proper preamp.
There's always going to be some noise but at least then we can do something about it.
Its impedance is not that complex , it's just not correctly done. Tubes and resistors can make a big difference in noise too.
Not only are some caps oversized, the cathode decoupling of the lower bias resistors not only serve little purpose other than reducing output impedance somewhat (where you do not really need it), they also slow down slew rate and colour the sound no end.
Small value resistors contribute less noise than larger value ones, some types of resistors are noisier than others etc.
Then there's the volume pot which is going to load the next stage which in turn will alter the frequency response of the preamp. If it is meant as a volume control as I think it is and not some sort of tone control.....
And yes, too low LF poles can not only cause instability (oscillation etc.) but will also cause 1/f noise. See at least the description in Wikipedia or other work on the subject. After all we've only been talking about not much else for pages and pages in a row, right.
There's little point in it for as long as we don't have firm electrical data and design goals.
As suggested before take a look at Salas MC preamp (leave the RIAA correction mentally out) and you already have a well thought out single ended stage with relatively low noise levels.
I don't see why a xformer with a higher turns ratio will need that. Surely you can take the output from the xformer out balanced so the only signal loss is mainly copper resistance loss, then bring it to an outboard preamp and amplify the signal further. Either SE by an active balun stage or xformer or keep it balanced to wherever you want it to go.
I don't know, maybe I'm missing something....
Ciao,
Two xformers with a long capacitive cable in between are bound to wreck much more havoc on that tiny little signal than a single higher ratio one.
So, yes please do yourself a big favour and go for the best one for the job, you know your ribbon's output so ask Lundahl for the best possible one. Mount it, terminate it according to the manufacturer's recommendations and we're talking.
Then let me know the signal level, if you want to keep the balanced XLR input, need a balun 1:1 xformer or whatever and any other data needed to make a proper preamp.
There's always going to be some noise but at least then we can do something about it.
Its impedance is not that complex , it's just not correctly done. Tubes and resistors can make a big difference in noise too.
Not only are some caps oversized, the cathode decoupling of the lower bias resistors not only serve little purpose other than reducing output impedance somewhat (where you do not really need it), they also slow down slew rate and colour the sound no end.
Small value resistors contribute less noise than larger value ones, some types of resistors are noisier than others etc.
Then there's the volume pot which is going to load the next stage which in turn will alter the frequency response of the preamp. If it is meant as a volume control as I think it is and not some sort of tone control.....
And yes, too low LF poles can not only cause instability (oscillation etc.) but will also cause 1/f noise. See at least the description in Wikipedia or other work on the subject. After all we've only been talking about not much else for pages and pages in a row, right.
There's little point in it for as long as we don't have firm electrical data and design goals.
As suggested before take a look at Salas MC preamp (leave the RIAA correction mentally out) and you already have a well thought out single ended stage with relatively low noise levels.
I don't see why a xformer with a higher turns ratio will need that. Surely you can take the output from the xformer out balanced so the only signal loss is mainly copper resistance loss, then bring it to an outboard preamp and amplify the signal further. Either SE by an active balun stage or xformer or keep it balanced to wherever you want it to go.
I don't know, maybe I'm missing something....
Ciao,
No, coupling caps do not stop grid current, they just divert it - unless of course there is no grid leak resistor! To check for grid current see how much the anode voltage changes when you change the grid bias from 1M to a short and back again.mm7 said:
I fear you may have bitten off more than you can chew. I would hesitate to build a valve preamp for a ribbon mike, but I only have 45 years electronics experience.
why? what is wrong with it? it was recommended by Lundahl.
In mic there is LL2913 then in goes to LL1935.
I do not know their impedances. Not provided by Lundahl.
Tubes I suspect. Resistors are same Vishay Dale MF.
That is why they are bypassed by film caps.
250uF cathode bypass caps were advised by some guru from here. (I am lazy to search by whom).
I'll consider to decrease them to 2uF for first one (film) and 100uF for second one (electro with 1uF film bypass).
I really do not understand how it could be. I played with different "volumes" in LTSpice, yes it makes HF rolloff in "quiet" position, but the rolloff is above 20kHz.
May be, but I do not want to cut off "good signal" LF. How to remove just noise?
Not sure what do you mean. I want to build a tube preamp for a ribbon mic. It is a goal.
It will be plugged into amp with 3k3 input impedance. It is electrical data.
It would be nice to minimize noises. Another goal.
As I see, "ribbon", "tube" and "minimize noises" are not easily combineable togeather.
Tried to find. Looked backword through many pages. Is it this one?
http://www.diyaudio.com/forums/analogue-source/140635-valve-itch-phono-157.html
If it is the one, it does not have enough gain - 45dB. Ribbon mic preamp should be at least 70dB.
Do you mean to change the grid leak R to 1M and short?
How to derive Ig from Va change?
Exactly!
I do not have that much experience, thus have no fear, that is why I've started. 
Thinking this through, a lot of decisions come out.
If the Ribbon traffo OP is 30uV, even with a very average SN ratio of 60dB
(1000), then the IP stage input noise will have to be (30uV/1000)/141 in
nV/rt Hz. This comes to a bit over 0.2nV/rt Hz.
You just need more OP than 30uV for even a lowly 60dB SNR which at
0.2nV/rt Hz IP noise is virtually impossible anyway.
Even with a higher ratio transformer the only real option is either parallel
high gm jfets such as 2sk369 or same with very low noise BJTs integrated
into a gain stage of x20 to x 100 - or you can cascode them under a tube.
Choice of jfet or BJT will be determined by reflected OP Z at the Ribbon
transformer OP. The BJT's current noise will add extra noise if this is too high.
Good luck!
cheers
Terry
Hi,
That's not the point. The point is that in the case of the mu-follower (in case you did not know, that's the of the topology you're use for bot the triode stages), the bypass caps can safely be left out. I told that twice already.
Everything.
The secondary of the xformer should be loaded by the proper termination. The 1M grid leak resistor does not load the xformer neither is it need as a grid leak either since in all likelihood a) there will be nothing to leak and b) should the be a current flow then it will flow through xformer to ground. Not every xformer is happy with that but the Lundahl can take it.
c) Should there be grid current then it is quite likely that the tube is faulty. Hence it should be swapped out.
If you don't know the termination of the xformer, ask. My guess, from looking at what they prescribe for their other xformers of similar type it will be around 10K. Ask nonetheless.
With proper termination the xfromer will become more linear and the load is in // with the input resistance of the first triode.
Since this input resistance is much higher than the load of the xformer the effective input impedance is still high enough not to cause any trouble. End of story.
Net result of the above: no more bypass caps and one noisy 1M resistor less.
Both.
Nonetheless it is also in series with yet another unnecessary resistor of 100K IIRC and is preceded by a huge 5 microF coupling cap, much larger than one would need if the input impedance is 100K or larger. Not good.
The tube sees a varying input impedance which is in parallel with the next stage and may load down the previous one too much. Operative word here is may. 20KHz may seem outside the audio band to you but a drooping HF response so close to 20KHz will also have other audible effects.
Well, the good news is that by cutting off some of what no one can hear you will also lose some of that noise.
Which is likely also why you had less of it when using 12AX7s....
I see no goals, only moving goal posts.
You stepped into this thread since post #33. You're entourage consists of the most knowledge guys here on the forum (and I certainly do not include myself) , some of which deeply rooted in the recording industry.
We're at page #50 and we're still chasing our own tail. Time for some introspection maybe?
Well, the gain is not that far off but that's not the reason why I sent you there.
It's about the topology, not the gain.
Without proper impedance matching that is indeed very likely.
Which brings us back to square one.
To recap: Ask Lundahl what xformers you need for your project. From what you explained that would be a SET for the mic proper that can drive the x meters of cable to wherever the preamp needs to be, then the tube preamp proper and if it need to drive a long cable to a 3K load then the obvious choice in my book would be a SE OPT.
That, at least to me seem to be the most logical steps. I'm sure someone will step in to correct me but so be it.
Ciao,
That's not the point. The point is that in the case of the mu-follower (in case you did not know, that's the of the topology you're use for bot the triode stages), the bypass caps can safely be left out. I told that twice already.
Everything.
The secondary of the xformer should be loaded by the proper termination. The 1M grid leak resistor does not load the xformer neither is it need as a grid leak either since in all likelihood a) there will be nothing to leak and b) should the be a current flow then it will flow through xformer to ground. Not every xformer is happy with that but the Lundahl can take it.
c) Should there be grid current then it is quite likely that the tube is faulty. Hence it should be swapped out.
If you don't know the termination of the xformer, ask. My guess, from looking at what they prescribe for their other xformers of similar type it will be around 10K. Ask nonetheless.
With proper termination the xfromer will become more linear and the load is in // with the input resistance of the first triode.
Since this input resistance is much higher than the load of the xformer the effective input impedance is still high enough not to cause any trouble. End of story.
Net result of the above: no more bypass caps and one noisy 1M resistor less.
Both.
Nonetheless it is also in series with yet another unnecessary resistor of 100K IIRC and is preceded by a huge 5 microF coupling cap, much larger than one would need if the input impedance is 100K or larger. Not good.
The tube sees a varying input impedance which is in parallel with the next stage and may load down the previous one too much. Operative word here is may. 20KHz may seem outside the audio band to you but a drooping HF response so close to 20KHz will also have other audible effects.
Well, the good news is that by cutting off some of what no one can hear you will also lose some of that noise.
Which is likely also why you had less of it when using 12AX7s....
I see no goals, only moving goal posts.
You stepped into this thread since post #33. You're entourage consists of the most knowledge guys here on the forum (and I certainly do not include myself) , some of which deeply rooted in the recording industry.
We're at page #50 and we're still chasing our own tail. Time for some introspection maybe?
Well, the gain is not that far off but that's not the reason why I sent you there.
It's about the topology, not the gain.
Without proper impedance matching that is indeed very likely.
Which brings us back to square one.
To recap: Ask Lundahl what xformers you need for your project. From what you explained that would be a SET for the mic proper that can drive the x meters of cable to wherever the preamp needs to be, then the tube preamp proper and if it need to drive a long cable to a 3K load then the obvious choice in my book would be a SE OPT.
That, at least to me seem to be the most logical steps. I'm sure someone will step in to correct me but so be it.
Ciao,
One consideration.
You are planning to build this complex stage ( it seems to be) and then you will connnect it with a poor pc audio card?????
Without the use of balanced connection that in this case is vital???????
In my opinion this hard job need to have a same quality stuff where it will be connected to get a reasonable good results.
Regarding the low level voltage please consider the Fet stage with 2Sk369 (now I remember the type, thx to Terry) that is specific low noise.
Walter
You are planning to build this complex stage ( it seems to be) and then you will connnect it with a poor pc audio card?????
Without the use of balanced connection that in this case is vital???????
In my opinion this hard job need to have a same quality stuff where it will be connected to get a reasonable good results.
Regarding the low level voltage please consider the Fet stage with 2Sk369 (now I remember the type, thx to Terry) that is specific low noise.
Walter
I am tempted to say that if you don't know how to measure grid current (even with the help of a brief description) then you won't know what to do with the measurement either, so there is little point in doing it. The principle is to measure the voltage change across a 1M resistor due to grid current - so which resistor do you think might be relevant?mm7 said:
As for making a low noise vacuum tube preamp, I do not think the mu follower is the best choice for the first stage. Others may disagree but that's my gut feeling.
You're doing fine. Just ignore all the people telling you're not. Mr. Lundahl is a very smart man. Ask him about the optimum loading for the secondary of his transformer used in this configuration.
You will not have any grid leak because the transformer winding is in parallel with the 1M resistor. You may not need it at all. It may be the source of your noise as well.
Obviously, you're going for maximum gain by bypassing the cathode resistors. That's fine, but it also adds noise and distortion.
Let me take a closer look at your circuit.
Last edited:
Is R8 just to make the simulator happy? Mine won't work with balanced lines. I assume you are using a balanced input. In real life, you don't need R8, please remove it.
I doubt you need R6 either. It's probably just adding noise. Try it without it. You are not getting grid leak because of R6 is in parallel with the transformer winding, which has far lower resistance.
Your second stage is likely running too hot. You don't need that much DC current to get good performance out of this tube. Please consider reducing it.
The DC current of your first stage could be reduced as well, to lower 1/f noise.
Did you set all of the parameters for your input transformer correctly? It appears to be a good choice for this application. The ribbon mic transformer works best into a higher impedance. You do not need to "turn it around". It's already a 1:5 step up input transformer. The data sheet does not mention any resonance point or termination data, therefore you need to remove R8 and C1.
Last edited:
If a transformer secondary, with a low reflected impedance, is put in parallel with a high value resistor then the resistor cannot contribute much noise because the secondary impedance seriously attenuates it. The resistor may be unnecessary, but it is not harmful. Of course, if a coupling cap is used in order to keep DC out of the transformer then a resistor is needed.
Yeah, it does not contribute any noise at all and it is not needed. It's best to get rid of it in my opinion.
OK, for the 6S45P, the Gm is 27mA/V. Thermal noise is 2.5/Gm = 92.6 ohms, is that right? I'm not sure how to add in flicker (1/f) noise. There was a formula I found earlier but I can't find it now. If we assume that total noise is some 200 ohms for the tube, and the mic output Z is also 200 ohms, I get:
Ns/Np = SQRT (Rn/Rs)
Rn=tube noise
Rs=source noise
Ns=secondary turns
Np=primary turns
which means the turns ratio for the input transformer for lowest noise is 1:1.
So, in this case it looks like the 1:5 input transformer is adding noise by increasing the apparent source impedance. Is this really right?
Ns/Np = SQRT (Rn/Rs)
Rn=tube noise
Rs=source noise
Ns=secondary turns
Np=primary turns
which means the turns ratio for the input transformer for lowest noise is 1:1.
So, in this case it looks like the 1:5 input transformer is adding noise by increasing the apparent source impedance. Is this really right?
Hi,
That formula is only valid for applications well above the LF band. For AF it is pretty much useless.
If grid leakage would occur then the resistor + the secondary winding of the xformer will drain it to ground.
Any grid leakage will cause a current which will cause signal loss hence a channel imbalance will occur.
Under normal circumstances and assuming a non-leaky valve no grid current should occur with a correctly biased triode. If it does then you're dealing with al leaky valve and it should be replaced.
A possible workaround would be to use a grid stopper resistor (it will add noise, inevitably) which will cause a pre-distorted signal which will cancel by the amplification of the inverting first stage.
Ciao,
That formula is only valid for applications well above the LF band. For AF it is pretty much useless.
If grid leakage would occur then the resistor + the secondary winding of the xformer will drain it to ground.
Any grid leakage will cause a current which will cause signal loss hence a channel imbalance will occur.
Under normal circumstances and assuming a non-leaky valve no grid current should occur with a correctly biased triode. If it does then you're dealing with al leaky valve and it should be replaced.
A possible workaround would be to use a grid stopper resistor (it will add noise, inevitably) which will cause a pre-distorted signal which will cancel by the amplification of the inverting first stage.
Ciao,
On my website I found the formula which can be written as Ns/Np = sqrt(Req Rin)/Rs. Rin is the effective input impedance of the valve, and can consist of bias resistor, transformer losses (transformed to a parallel form), valve grid impedance etc.dirkwright said:
If we take a wild stab at Rin = 1M, then Req=200 and Rs=200 gives Ns/Np = 70.7 for optimum noise. The idea is that the source noise (multiplied by the transformer ratio), input resistance noise (loaded by the transformed source impedance) and valve noise are all equal. If the input resistance is dominated by the valve itself (unlikely at audio frequencies) then a smaller ratio is needed to get more loading, as the effective temperature of the input resistance is higher than the others.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.