Rouge Un - my attempt at a red Lundahl + 5534 preamp

[thomasdf]

Hello

A long time ago I was given a set of spare parts, transformers and such among which was a pair of Lundahl LL1538 and LL1539.
Knowing a famous console brand used the 1538 as mic input in a legendary preamp / channel design I thought I'd give this design a shot, in 51x format.
I pretty much went 1:1 on the schematic but this experience is a lot of first times for me : Kicad (used to be a Eagle user... but it's way too expensive) and 500/51x format.

Here is where I am so far :

I have some questions regarding the design :
- what's the deal with R113 / R114 and the trimmer there ? On some schematics the value is "L5240" ?
- what are the most appropriate values for the 5534 capacitors ?
- output transformer : orig transformer is the infamous "pumpkin" is 1:2+1 so am I right to think my option B wiring is better ? LL1539 is 2:1+1 so option B here is reversing the transformer and using it 1:2 and using 2nd tap of primary as 1:1 output for feedback.

I am curious to try to swap U1 for a TDA1034, if I can source some. Maybe I'll add a parallel footprint.

I am open to any input, advice etc. I plan to share the PCB design / gerber when I am done.

Cheers !

Thomas

 

[Dylan W]

R113/114 look wrong to me. The schematic I have doesn’t have them. Can you post that section of the schematic you’re working from?

1:2 is the most true to the original (option B). You may want to ask Lundahl which windings to use for primary, secondary, and feedback in this arrangement because they’re not symmetrical on the 1539.

There was some talk previously that a quadfilar output tx would probably work well here. In that case it wouldn’t matter since all the windings are symmetrical.

This has been on my list to try, will follow with interest to see how you make out.

 

[thomasdf]

Thanks for chiming in ! Here are excerpt of the schematics. R113 / 114 or not present in the console version (2nd file), they are in the standalone preamp (1). I wonder what's their purpose.

Good call about contacting Lundahl, I'll do that.

 

[ruffrecords]

I suspect this is the means of adjusting the positive feedback to null out the primary resistance of the output transformer. There is a Lundhal app note for this I believe.

Cheers

Ian

 

[Ike Zimbel]

Hello

A long time ago I was given a set of spare parts, transformers and such among which was a pair of Lundahl LL1538 and LL1539.
Knowing a famous console brand used the 1538 as mic input in a legendary preamp / channel design I thought I'd give this design a shot, in 51x format.
I pretty much went 1:1 on the schematic but this experience is a lot of first times for me : Kicad (used to be a Eagle user... but it's way too expensive) and 500/51x format.

Here is where I am so far :

View attachment 133706I have some questions regarding the design :
- what's the deal with R113 / R114 and the trimmer there ? On some schematics the value is "L5240" ?
- what are the most appropriate values for the 5534 capacitors ?
- output transformer : orig transformer is the infamous "pumpkin" is 1:2+1 so am I right to think my option B wiring is better ? LL1539 is 2:1+1 so option B here is reversing the transformer and using it 1:2 and using 2nd tap of primary as 1:1 output for feedback.

I am curious to try to swap U1 for a TDA1034, if I can source some. Maybe I'll add a parallel footprint.

I am open to any input, advice etc. I plan to share the PCB design / gerber when I am done.

Cheers !

Thomas

Are the phantom power resistors, R121,122 really 3K6? I'm used to seeing 6K81 in that application.

 

[Dylan W]

Did a quick sim of the first stage, if I’m looking at it correctly minimum gain is about 1.5, so I would try 15pf for compensation. 5534 doesn’t need compensation above gains of 5. Unity gain comp is 22pf. A good rule of thumb for gains between 1 and 5 is 22pf/gain.

Second stage is unity so 22pf seems logical.

Btw R15/16 should be 6K2, R8 should be 1K8. There are a few other omissions, for example should be a 47uf cap between bases of the output transistors. I believe C4 should be 22uf “N/P” (bipolar), not 22nf

 

[thomasdf]

Thanks for the info and corrections. Crazy how many mistakes I can let slide ! Dylan W, thanks for your sharp eye

Here is an updated version :


The values of the phantom resistors are uncommon indeed, but my guess is it has to do with the transformer specs and hookup ? Maybe there are other ways of implementing it ?

I sent an email to Lundahl, to be continued !

 

[vintagelove]

Hello, it's regard to unity gain and compensation caps... if you can find a genuine (be careful of fakes from China) ma332 opamp, you might want to try that in the circuit. They should drop right in and you could likely use it without the compensation cap, which will speed up the slew rate.

 

[Mixermend]

As well as the Phantom resistors surely being incorrect, as mentioned previously, they should be 6K8 - it looks to me that the input transformer primary windings are connected out of phase - is this just a drawing error?

 

[thomasdf]

The value of these phantom resistors are 100% as in the 3 schematics I have referenced... as weird as it may seem. But I can't see why they couldn't be 6K8, since the 48V is not schematic dependent.
Re the transformer, I was dead certain of the wiring but on the console schematic which specifies the use of the LL1538 (the one I have) you are right, the 2nd primary is wired in reverse polarity.
I'll update the schematics

 

[My3gger]

Lower phantom voltage would require lower resistor values, but i doubt they would use less than is standard phantom voltage.

 

[thomasdf]

Here is the latest revision of my drawing :

I am starting to play around with the PCB design, now. Kicad is new to me, and quite different from Eagle so I am still managing things out a little. Especially for the physical placement of front panel elements, rotary switch and polarity and 48V switches.
Still need to figure out what 2PDT / SPDT switches to use , and find the correct footprint.
Rotary will be Grayhill, I think ... Quite expensive but there's a reason for that I suppose

 

[thomasdf]

I got an answer from Lundahl that got me confused :

Du to the internal structure of the windings in the LL1539, it can not be used in this application.
I will suggest you take a look at the LL1555. This model has all winding ends available at the pins and can be configured for your application. Please keep in mind that Lundahl transformers consist of two coils, and it is important that windings on both coils must be used to function correctly.

Here are the datasheet of the LL1539 which I have a would like to use :
http://www.lundahl.se/wp-content/uploads/datasheets/1539.pdf
and the LL1555 :
https://www.lundahltransformers.com/wp-content/uploads/datasheets/1555.pdf

If I stick to option B, with the output transformer reversed here is what I get :
DCR reading of 20 ohms on tap 7-8 and tap 11-12 which I planned to use as primary and tertiary feedback, and a DCR reading of 87 ohms on 1-2, with 3-4 linked, planned to use as my output tap.

If I use it in the correct orientation (that was option A) I can use it as 1+1 : 1, confirmed by my DCR reading of approx 50 ohms on taps 1-4 (primary), 3-5 (tertiary feedback) and 8-12 with 7&11 linked (output tap). This would drop the output level by half, which is too bad.

I can make additional measurement if needed, but I am confused because from what I see on the datasheet and my measurements the 1539 has 4 distinctive taps that can be arranged in various ways. I am pretty sure using it in reverse is possible, but maybe it's time for some transformer gurus to chime in.

 

[beatnik]

Is this the original output transformer ?

https://www.audiotestserviciotecnic...arnhill-vt30408-output-transformer-focusrite/

It would be sensible to make one prototype with the original transformer configuration, and then use it as a benchmark to experiment with alternatives.

The original transformer was custom made for this circuit so it maybe not so easy to sub with an off the shelf part.

 

[thomasdf]

the original transformer is 1:2+1, into 600 ohms. Nothing extraordinary except for its build quality and tonal quality.
I am not looking to build a 1:1 replica, my initial goal is to use these transformers that are just sleeping in a drawer, and I realized the 1538 is the iconic input transformer for this design, and the output might work.. but you are right I need to experiment !

 

[delatsch]

I am not sure if this is the original, but as for the winding, it seems to be similar to UA5002 minus the other windings so either two pairs of bifilar windings where primary and secondaries are mixed or like 2520 output from API 2503 with quadfillar winding with four equal windings, one being a primary, two tied together for secondary and one left for aux or tertiary in this case. I am sure any of those would work fine.

 

[delatsch]

I am interested in driving similar simple OT directly with 5534 since it was made to drive 600 ohm load, have everything ready to test it will see if it can work, I don't know if anyone tried it, not sure what the output stage of Portico 5012 is, it also uses a lot of 5534s.

 

[beatnik]

I think the original transformer would spell 2 : 1 : 1 rather than 2 : 1 + 1 and these two are not exactly the same thing.

From what I can understand from Lundahl response it means that both winding on one side of the transformer must go to the same connection, you cannot take one half of the winding and connect it somewhere else.

With this in mind the suggestion for LL1555 makes sense because you can configure it as having three separate coils.

 

[Dylan W]

The LL1539 has four separate windings, so I don’t see any reason why it won’t “work.” It may or may not be optimized in terms of capacitance, freq response in that configuration, but it should pass signal. Really no way to know if it’ll sound good without experimenting, you may need to try multiple configurations of windings to see which works the best as primary, etc. I also like @beatnik ’s idea of making a benchmark with the original tx. You could also do the same with a quadfilar type, which is almost guaranteed to perform well in this application.

 

[beatnik]

I think it need more than just passing a signal.

I am inclined to believe what Lundahl says but since you have the 1539 already there is no harm trying out.

The original transformer is not that expensive, I am wondering if Carnhill would sell directly or it's a proprietary design.

If they don't you should just buy a spare one from the link above and reverse engineer it, I know several toroidal transformer makers that could make a small batch if we can organise a group buy.