help with 1:1 transformer choice for cathode follower.

Hi.
I'm planning a tube output line stage and want it to have floating out.

The final stage is a cathode follower with current source, capacitor coupled (as i want to have a transformerless  unbalanced out too) and it should have an output impedance in the 55r range with the tube choosen.

So i need an ungapped 1:1 transformer for my floating balanced output, but i face many choice in references and don't know  which one to choose for optimal performance (excluding max output level and quality of build itself, obviously).

I suppose the dcr of primary and secondary play a role (compromising bandwitdth somewhat?) in the choice but i don't understand how and why... and like to make educated choice whan i can!

Could you explain me what to look for and why?

I thought about a Carnhill VTB 2281 first and then seen the VTB 9071 which could work too ( you can check datasheet in this link: http://www.audiomaintenance.com/downloads/carnhill_design_guide.pdf).

Thanks a lot.

While the output impedance is relevant, the question you need to ask is, what kind of load do you want to drive at what level and how much current your cathode follower can supply. That will determine what transformer to choose. More likely you will be looking at a 2:1 - 5:1, not a 1:1.

Volker ok thank you i give you the info.

It is a 6h30pi cathode follower with a ccs. It is biased around 30ma (probably some a bit more around 33ma) with 100/110v voltage for the tube (between cathode and plate of the cathode follower, this don't take into account the ccs) and bias around -6v.

From simulations i've done with Tube Cad software here is what the cathode follower should be able to do (at 32ma):

_with 600r load it should out max +22dbu (8.75v peak);
_with 10k (and higher Z load) out max is +38dbu (90.5v peak).

This is unbalanced after the coupling cap.

It won't need this high max output with higher z load but i would like to have headroom (12db or more) in this last stage.

To be clear, the whole line output is 2 stage: just before the CF there is a voltage gain stage of approximately 20db with same gain margin (around 12/14db headroom) and same family of tube (low rp, high gm).

I know this whole line stage can look total overkill but i want to try it as an experiment and see by myself if i'll gain something this way.

The whole output stage b+ is regulated and will have it's own independent psu.

The target for max out in real world use will be around +24dbu with higher z load and around +18dbu with 600r.

KrIVIUM2323 said:

I suppose the dcr of primary and secondary play a role (compromising bandwitdth somewhat?) in the choice but i don't understand how and why... and like to make educated choice whan i can!

Click to expand...

DCR is related to LF performance, not only response but also distortion; you're looking at the lowest DCR possible for lowest distortion.

Could you explain me what to look for and why?

Click to expand...

Not in a few words. You must google for transformer model and work the equations, and after that see non-linearities in transformers. Sorry, transformers are a complex subject. In addition, most of the parameters that define a transformer's real-world operation (leakage inductance, stray capacitance, magnetization curves...) are often missing from datasheets. And even if they were available, very few could make good sense of them. So manufacturers describe their products by application, input or output, line or mic level, gapped/ungapped...
You need to read the test conditions; there a lot of difference between a xfmr designed for 20Hz LF response and one designed for 40 Hz, or one designed to be driven by zero-ohm and one designed for 600r.

I thought about a Carnhill VTB 2281 first and then seen the VTB 9071

Click to expand...

The former is the logical choice, the latter is not optimized for output duties, DCR is too high.

you're looking at the lowest DCR possible for lowest distortion

Click to expand...

Ok. This is more or less what i thought at first.

Not in a few words. You must google for transformer model and work the equations, and after that see non-linearities in transformers. Sorry, transformers are a complex subject.

Click to expand...

Yes i know this is complex and not readily easy to explain, but i asked in case of someone could explain this.

Thank you very much for pointing me the track to follow anyway, i'll start reading the articles Jackob (Gyraf) uploaded in the technical document section of the forum and try to digest this (i've seen some Crowhurst articles and really like the way he explain electronic) and find other source too.

For your purposes a VTB2281 will be fine. As a rule, LF performance is determined by primary inductance and HF performance by leakage inductance and winding capacitance. Distortion is primarily determined by core material and DCR. Transformers are surprisingly complex devices. Designing them takes a lot of skill.

Interesting design concept. It will be interesting to see how it performs. Sims are a good starting point but they are rarely accurate when it comes to high levels.

Cheers

Ian

DCR  will also cause phase shift at the low end,

lets say you have 10 Henries of primary inductance, Inductive Reactance at 10 Hz will be

6.28 * 10 Hz * 10 H = 6.28 * 100 = 628

if DCR is 314 ohms (very high, just using it for easy math) then you have DCR/XL = 0.5

take the arc tangent of 0.5 and you get 45 degrees of phase shift, so the secondary will be 45 out at 10 Hz

as DCR goes down in relation to Inductive Reactance, phase shift approaches 0

you can see a phase shift graph on the Jensen Transformers website

Cj, Ian thank you for clarification.

I didn't thougt about phase issues but this make sense.

So i think the VTB 2281 with paralell primary and secondary should be ok at first.

After a bit of reading i should be looking for one with very low DCR and possibly a hi ratio nickel core for this application (for low hysteresis and lower distortion at low end).

Does this make sense or did i mixed up the reading i've done?

Interesting design concept.

Click to expand...

I don't know, it seems logical to me for an output stage: low output impedance and large headroom for clean and less coloring stage possible.
The CCS is the most interesting part of the design imho, a cascode of depletion mosfet for very high impedance and ground isolation. Will see how it perform but i have high expectations.

Sims are a good starting point but they are rarely accurate when it comes to high levels.

Click to expand...

Yes i take the sim like that, starting point and design aid. This is one of the reason i want this high headroom: to be sure that the first 6db are well below potential issues.

KrIVIUM2323 said:

So i think the VTB 2281 with paralell primary and secondary should be ok at first

Click to expand...

I think your output stage would be stressed driving those primaries in parallels at LF.

Abbey Road,

Do you think they would not have enough current drive?... Or too low total inductance load for the CF?

I could try series and parrallel anyway and keep the one i like the most.

KrIVIUM2323 said:

Abbey Road,

Do you think they would not have enough current drive?... Or too low total inductance load for the CF?

I could try series and parrallel anyway and keep the one i like the most.

Click to expand...

Too low inductance.

Jensen Transformers makes the best LA2a type output, IMHO, meant to sit on a cathode follower,

do not know if it would work in your circuit, maybe post the schematic unless you want it to be top secret,

JT-10K61-1M,  check the crazy freq response and distortion figures,  2000 Henries pri ind. 

there are also some 1:1 types listed below that>

http://www.jensen-transformers.com/transformers/line-output/

Too low inductance.

Click to expand...

Ok, i think i start to understand what could happen... i need more reading however...

Jensen Transformers makes the best LA2a type output

Click to expand...

Yes from what i've heard i agree with you about Jensen transformers in general, but for La2a output stage it is a white cathode follower if i remember correctly, not exactly the same as a cathode follower with a current source. I don't know if this one could work in the circuit but i won't try with that one at first, not easy to source in France and if you buy direct from Jensen once the shipping cost and custom fees are added they are not cheap at all... too much money for me at the moment. 

, maybe post the schematic unless you want it to be top secret,

Click to expand...

i will when free time allow, maybe in a few days. But it's a regular CF with a CCS, nothing more than that. Fairly simple.

JT-10K61-1M

Click to expand...

I've got some VTB 2291 used in a different project (9600:600/2400:150, ungapped), i will borrow them from circuit they are used in as they seem to meet the Jensen ratio spec, but i don't know inductance for them.

I must confess i would prefer a 1:1 not too loose too much db in the overall output stage.
I will draw the full out stage and upload it.

I've found 20mn to draw the schematic of the out stage.

As you'll see, i did not tell everything about the cathode follower topology and it is still simple but not as simple as i said in previous post...

All in all the bootstrapped CF won't change many things except it will theorically erase the second main source of error of classic CF (the first one being taken of using the CCS).

The CCS dynamic impedance at audio frequency should vary from 200m to 20m approximately.

In the CCS R8/R7 are grid stopper for the depletion mosfet, they could be helped by ferrite between grid and them to increase protection against oscillation without slowing down too much the mosfet behavior. Vr1 determine current.

R6 purpose is to isolate the CF cathode from the CCS  parasitic capacitance: definitive value have to be determined (somewhere around 1k should be ok). The value shown for the CF is at 32ma.

Gain stage is a classic cathode bypassed SRPP. 12.5ma and bias at -6.61v. Gain around 22 db.

Except for the bootsrapped cf it is fairly simple in fact (even the bootstrapped cf action is simple once you figured what it does to the lower CF.)

There is probably some errors in the schematics (and grid stopper resistors for tubes are not shown) as i drawn it very quickly and without checking values. But you have the general idea. 

Hello,
where comes the current for the grid of V2a?
Regards Andreas

Hello Andreas,
it seems i've forgot two resistors!  :

I've played with direct coupled voltage gain stage lately and this probably stayed in my mind when i drew the schematic...  ;D

But as i said earlier there is some errors in this schematic, it's just to give an idea of what i want to achieve in the end and help in tansformer choice.

I can understand the CF and CCS etc and the desire to minimise distortion products in this stage. However, as the CF has unity gain, the preceding SRPP stage is operating at the same level and SRPP stages are not renowned for low distortion.

Cheers

Ian

d SRPP stages are not renowned for low distortion

Click to expand...

Yes SRPP is slightly euphonic, produce higher level of distortion than let's say a beta follower or a nicely made cascode (with grid of the upper stage referenced to a regulated voltage) but i find SRPP pleasing to my ears, especially when it drive high impedance load.  And this stage have some technical advantage over other circuits that i like and want to use.

In fact i've decided to add the CF because this is the only stage that allow a wide range of load impedance to be driven equally well and with 'lesser' sonic signature (especially with bootstrapped CF). But if i had to drive only 10k and higher load i would not use a CF but an SRPP, as i like the sound of it.

As is the SRPP stage is theorically able to do more than +40dbu max out and bias at -6.61v.  As i said previously i will operate them with lot of headroom and from what i've heard from this stage behavior distortion rise proportionnaly with input level, so the voltage stage will imprint it's sonic signature but it's not very high as if i weren't using 12/14db headroom which will be the case in the end.