Ba6a interstage

[MatthisD]

Points taken. Yes much cheaper in Euro too. Maybe an aluminium case placed over it would be good enough if external mounting it.

With an impedance loaded circuit if the reactance is raised and the impedance is maintained (reduced inductance and decreasing capacitor size) then the distortion added by the transformer will also increase in typical usage. I imagine that in this case of increased reactance/reduced impedance (as a result of the sowter's lower inductance which we're assuming) that the 4.7K resistor across each winding is aiding the response. To keep the reactance from being much higher I would increase the 40uF capacitor in steps and measure the response at each value, and if possible check the distortion at 20Hz. You could also reduce the resistors across the windings if the response starts to change for the worse. With such low inductance it is a compromise but it is worth experimenting to have it operate with minimum distortion I would say.

 

[emrr]

I keep meaning to add that there's no way the RCA interstage is specified as a 10K primary, it's simply loaded that way.  I completely believe they are loading it heavily to force improved response, then building out the series resistance to the plates to compensate for the load.  RCA did that sort of heavy loading in several preamp output transformers also, which suggests a purposeful trend in thinking.

The combo of transformer fed and resistance fed is an interesting middle ground.  It gives up some of the B+ swing advantages of pure transformer feed. 

We need someone to measure a BA-6 interstage and answer the DCR/L/open AC Z question.  I expect DCR and L more similar to the 86 limiter, just as a 1:1 rather than a step-down. 

 

[MatthisD]

I keep meaning to add that there's no way the RCA interstage is specified as a 10K primary, it's simply loaded that way.  I completely believe they are loading it heavily to force improved response, then building out the series resistance to the plates to compensate for the load.  RCA did that sort of heavy loading in several preamp output transformers also, which suggests a purposeful trend in thinking.

The combo of transformer fed and resistance fed is an interesting middle ground.  It gives up some of the B+ swing advantages of pure transformer feed. 

I agree, the AC voltage hitting the primary is lowered with those plate resistors and the two 4.7K are probably to load it for good response and also to reduce the current passing through the presumably, very thin gauge wire of the primaries. It probably wouldn't change the performance to increase the 22K to 24k, the 4.7K to 2.2k and the 40uf to 100uF to better match the Sowter's primaries...trial and error.

 

[lassoharp]

It probably wouldn't change the performance to increase the 22K to 24k, the 4.7K to 2.2k and the 40uf to 100uF to better match the Sowter's primaries...trial and error.

Good points and it might be good to look at the Drip BA6 schematic to see if any changes have been made along that line.  I noticed Drip mentioned the Sowter iron being made for that build but don't know if they meant footprint only or other.  I can't seem to find the Drip schematic.

 

[emrr]

Drip didn't used to issue their own schematics, simply link to originals.  They had part # lists and manuals showing stuffing locations in great detail. 

 

[Woot]

Yeah the Drip Ba6a manual has the complete PCB overlay with part numbers/values.

I haven't checked this against the original schematic but he states that  "This new pcb is true to the ba6a circuit with no alterations or modifications to the audio or power path" with the metering circuit removed.

 

[emrr]

I just got a new report that the Sowter interstage for the 86 limiter was fixed in circuit with heavier secondary loading, which reduced overall level by a few dB, considered acceptable trade-off by the reporter.  Hoping to get some more details soon. 

 

[DaveP]

As many of you know, I made a BA-6A a while back.  I learned the hard way about the interstage problem.

Even though the original fed the interstage from a 4.7k source, it was still too high an output resistance for the interstage I had available and bass response was poor.  I ended up using capacitive coupling instead which seemed to work ok, rather like the SA 39 in fact.

If I was to make another now, I would drive an Edcor interstage from a couple of cathode followers fed from the 6SK7's.  The output resistance could be 20 times lower and then so would the frequency response.

best
DaveP

 

[Reo]

Great info, thanks. I'm planning to build a Ba6a clone as well, but haven't picked out an interstage...I'd be up for a group buy if none of the existing models make the cut

 

[lassoharp]

if none of the existing models make the cut

The test results posted earlier in the thread show they did make the cut, easily.

 

[Reo]

Gotcha...I saw that you had talked about another round of more subjective testing a few posts back but I see now that all those tested were fine in terms of L. I think I'll go for the lundahl then.

 

[lassoharp]

I had talked about maybe doing another round just to look at distortion figures but it didn't seem to be anything critical based on listening tests - more for just documenting details of L vs distortion plots.



I'm planning on trying the Lundahl as well.  It looks like a winner.  8)

 

[bancho]

So... it's been a while and I still have my ba6a in pieces... mainly because of interstage. Still can't decide.
Did anyone tried the Lundahls ... maybe compared them to Sowters?
How about the Sowter input/output option? Or is there anything "better" to be used?

 

[bancho]

Did anyone consider Sowter 1195? It is intended for 86A interstage and it is 10k+10k/20k... in the schematics there is center tap but I don't know if Sowter's has it.  ???

edit: I saw you tested it so it should work... the question is if somebody actually used it?

 

[lassoharp]

the question is if somebody actually used it?

The Sowter BA6A was tested and used in the real circuit under typical studio conditions.  Nothing should change when it's put into a clone build . . . . . not sure what you're asking or wanting to know.  I don't see that there is anything further to be discovered that wasn't revealed in the test if that's what you're asking.

 

[bancho]

I'm sorry... I know it was debated extensively...
but are there any differences between Sowter 9830 and 1195 (this can be used connected as 20k/20k, right?) ?
sorry if this is dumb question (language barrier?)...

 

[emrr]

1195, if accurate to original, is not 1:1 and won't work properly in a ba6.  If it's 1:1 it's wrong for 86a, might work for ba6, but don't know why you wouldn't just get ba6 type. 

 

[emrr]

The problem with the Lundahl LL1635 and the Sowter BA-6 and 86 type interstages is they are 'power' rather than 'small signal' designs.  Both provide headroom levels within close range of what the 6V6 output stage provides.

Note the Lundahl LL1635 specifies source Z of 4K in PP, far lower than a GR stage would ever provide, and the DC resistance of both are very low for the job.  Tube GR stage transformer loading is a small signal job, look at the AC voltage shown on the Gates SA-39 after the GR stage for instance; 1.5VAC.  You want an interstage that can receive enough signal at a high enough Z to be sufficiently 'woken up' by the source signal, while not loading down the GR tubes.  They have enough problems as-is.  If there was a UTC A series 1:1 interstage in the 30K-50K range, it'd be far closer to the original design parameters.  There's not though. 

 

[MatthisD]

The problem with the Lundahl LL1635 and the Sowter BA-6 and 86 type interstages is they are 'power' rather than 'small signal' designs.  Both provide headroom levels within close range of what the 6V6 output stage provides.

Note the Lundahl LL1635 specifies source Z of 4K in PP, far lower than a GR stage would ever provide, and the DC resistance of both are very low for the job.  Tube GR stage transformer loading is a small signal job, look at the AC voltage shown on the Gates SA-39 after the GR stage for instance; 1.5VAC.  You want an interstage that can receive enough signal at a high enough Z to be sufficiently 'woken up' by the source signal, while not loading down the GR tubes.  They have enough problems as-is.  If there was a UTC A series 1:1 interstage in the 30K-50K range, it'd be far closer to the original design parameters.  There's not though.

Regarding the BA6A it would have to be tried with an LL1935 to know if the inductance is adequate for the frequency response spec. or acceptable to someone looking for a replacement. LL1635 inductance is given as 300H for PP.

 

[vari-mu]

  I have done for experiment  new version of the interstage transformer BA-6A  ,
ratio  1 : 1,  on a core 375 UI  ( right side ). My previous version
were 31 UI ,  8X15mm stack , ratio  2 : 1

  Turns ratio:      1 :1
  Z ratio :          30k : 30k
  Coil structure :  bifilar
LAM  : 375 UI  ,  Stack : 10 X 18 mm

  This thing is wound :
2 coils -  each coil has >
pri -    2770t  #42 ( 0.063 mm)  enamelled  wire + silk covered (blue) , DCR 730 Ohm
sec -  2770t  #44 ( 0.05 mm)    , DCR 1180 Ohm
pri & sec BIFILAR  wound

    Total :
pri -  2770t + 2770t  =  5540t ,  DCR 1460 Ohm ,  240 Hy  PP
sec -  2770t + 2770t = 5540t ,  DCR  2360 Ohm ,  240 Hy
L mesured at 100 Hz .

    Set of winding wires :