[Sintz]

I'm looking for instructions on how to convert my AKSA100 to a balanced XLR input

(I guess I should have checked this forum before asking Hugh directly  .
Sorry, Hugh!

I see that Hugh was able to make successful recommendations for the same question on a "55" http://(http://www.audiocircle.com/index.php?topic=21302), and would like to know if this is possible on the "100".

I've recently invested in some new equipment that allows the use of balanced XLR connections and I'd like to have a completely balanced signal path.  
Any recommendations would be appreciated.

Thanks!

[AKSA]

Hi Shayne!

Thanks for your email and post.  I'll answer here for the benefit of others as well who might want to convert to balanced input on their AKSAs.

Yes, it can be done and the solution is identical to the 55 detailed in the other thread.  Essentially, you unsolder the feedback electro (C2 - 22uF) from the ground connection (near R2), and drive this point as the inverting input.  Non-inverting input remains the same as at present.  BUT you then have uneven input impedances, 43K into the non-inverting input, and 2K5 into the inverting input.

These inputs would have to be made the same impedance, by putting in a 2K7 resistor to ground from the existing, non-inverting input, BEFORE the input cap C1.  Then you must drive these two balanced inputs from a very low source impedance, less than 100R, to achieve the necessary sound quality without distortion.  Generally, a tube preamp can't do this.

IF your preamp has an output impedance less than 100R, all is well.  Otherwise, you could have drive problems, and a buffer would need to be developed for the input to the AKSA.

Hope this is helpful, Keith tells us the sound is utterly unchanged - so there is no sonic cost to doing this, which is nice to know.

OTOH, he has not reported any improvement either, so you need to ask if it's worthwhile.  Generally, if you have a noisy environment with long interconnects, it will be quieter.

Cheers,

Hugh

[Sintz]

Hugh:

Thanks for your expedient response.

I'm only running about 0.5-1.0M ICs between my preamp and the AKSA100, so your statement regarding potential improvements in noise is noted.

I'll wait until I have the components connected before determining if this is something worth trying.

Thanks again.

[J Harris]

Hugh,

Wow, this sounds deceptively simple. Let me see if I've understood you properly:

Essentially, you unsolder the feedback electro (C2 - 22uF) from the ground connection (near R2), and drive this point as the inverting input.

Does this mean that C2 is removed entirely? Or is it it only unsoldered from the ground connection, and something more explicit is done to "drive this point from as the inverting input." I've examined the thread on converting the 55, but could not find more detail on this.

BUT you then have uneven input impedances, 43K into the non-inverting input, and 2K5 into the inverting input.

These inputs would have to be made the same impedance, by putting in a 2K7 resistor to ground from the existing, non-inverting input, BEFORE the input cap C1.

This sounds extremely straightforward. Having done this, and presuming ownership of a 100R output impedance preamp, all that's left to do is drill the rear of the chassis for XLR inputs? Btw, would it be allowable to put in a switch for single-ended and XLR and play it that way?

Best,
Patrick (J Harris)

[AKSA]

Hi Patrick,

Thank you for your post - it does not surprise me that you, an 'ol music industry man, would be interested in balanced lines!!

C2 is maintained, with the positive connection preserved, but now the negative electrode becomes the inverting input.

One thing I've not mentioned is that balanced drive doubles the gain, from 31.6dB to 37.6dB.  This can be restored by simply inserting degeneration on the emitters of the LTP.  The board has provision for this already, but I cannot tell you precisely the value of the resistors.  I would guess at around 47R;  this would halve the transconductance of the input stage according to my re calculations.

A second thing is the non-inverting input cap.  This should be VERY high quality, as it's now passing signal in series, rather than shorting signal to ground as before in shunt.  I suggest at least 22uF, which rules out film caps, and the best would be undoubtedly a Black Gate.

The impedance mismatch between the inputs is cause for some concern, but essentially they can be made equal by loading down the non-inverting (that is, the stock input) with a 2K7 resistor to ground from the HOT input point, on the non-amp side of the input cap, C1.

This amp would be hard to drive.  We've gone from a Lexus to a 50s Chevy, without power steering........  Much of the sonic character would depend on how easily your preamp could drive this very low input impedance.

Cheers,

Hugh

Yes, it could be switched.

[J Harris]

Hi Hugh,

It turns out my all-balanced tube preamp (a BAT VK-5 SE) has an output impedance of about 400 ohms per phase.

That means it won't be able to drive the AKSA as modded here, correct?

What would I need to make it work - a nice expensive transformer between the amp and preamp, correct? And I presume that this would have some bad impact on the sound.

Argh... I love my preamp and amp, but the balanced-to-single-ended converters I'm using at the amp end are picking up various radio broadcasts and sending them to the speakers.

Patrick (J Harris)

[AKSA]

HI Patrick,

Unfortunately, 400R is too much.  Even my Swift preamp has a Zout of 150R, and that's still a bit high.......

Yes, a transformer, balanced in, single ended out, 2:1 step down (equal turns on all windings).

You'd probably be best to choose 10K versions rather than the 600R types commonly found in pro-audio.

I'd suggest Jensens, from California.

Cheers,

Hugh

[keithvv]

Hi all,

Hugh is correct in that I did not report any sonic improvements in converting one of my AKSA-55 Ns to balanced input operation. I did not expect any improvements at all given my experience with balanced vs. unbalanced operations. I was only concerned that an amp designed for unbalanced input would suffer if converted to balanced operation. The extremely high performance of the AKSA was not something I wanted to sacrifice just to go to balanced inputs.

Now, you might ask, why did I want to convert an input? Simple, I had no other acceptable choice in the following situation:
1) I was remoting the amp to my study/office, while wanting to maintain a signal source in the family room. The run is over 70 feet. A tough run to drive with unbalanced output buffers, and one which is susceptible to noise pickup.
2) I wanted to have exactly the same music in the office as in the family room. The home theater processor I use as a source had only one set of unbalanced main Left / Right outputs (used in the main system) plus one set of balanced outputs. Hence, either I split the unbalanced output and risked my long run of shielded coax becoming an RF antenna, or I converted the AKSA input.

Hugh came to the rescue with a mod to the 55 that worked, worked with no perceived degradation to performance, and was MUCH cheaper (and physically smaller) than putting in a pair of Jensen 4:1 stepdown transformers.

For the record, I have performed many listening tests comparing identical sources, preamps, and amps using unbalanced vs. balanced operations. I  have found no discernable difference, let alone improvement, when going to balanced operation on runs of less than 15 feet.

I hope this helps.

Cheers,
Keith

[J Harris]

Thanks for your replies, Hugh and Keith.

Unlike Keith (what preamp are you using Keith, and with what output impedance?) it seems that I would have to buffer my preamp output with a Jensen or other brand of step-down transformer, in order to gain a better preamp match.

In my experience of using transformers they tend to add their own character to the sound (mainly step-up transformers in MC cartridge applications)... which is exactly what I was afraid of.

Although balanced operation on the AKSA would be extremely convenient for me since my BAT tube preamp has balanced outputs only, I'm scared that putting a Jensen tranny in between the two components would yield a Jensen sound, not an AKSA sound, in the final result.

So I'm thinking that probably I won't attempt the conversion... but sadly, I might look for another amp with balanced inputs that can match the impedance of my BAT... I've already invested so much in balanced cables... and I love the freedom from noise and RFI, particularly in a phono/tube setting.

Best
Patrick (J Harris)

[AKSA]

Patrick,

Stay cool for a few days - I'll do a little more work on the Zin aspects which I'm presently studying in 'The Art of Electronics', a very good text I bought recently from Amazon which discusses this very matter.

It may be possible to increase Zin on the AKSA, and who knows, I might be quite wrong!!  I'll get back to you in this forum.

Cheers,

Hugh

[J Harris]

Hey Hugh

Sure you've been busy, but any hints yet on the Zin issue?

Hope you're well - btw I noticed the top of my case was a trifle ringy. I damped the underside with coax putty from Rat Shack and the sound is clearer. For anyone who uses the case I got.

Best
Patrick (J Harris)

[AKSA]

Hi Patrick,

I've sat on a few park benches and ruminated further on this problem........

I think I was wrong.  There is in fact no reason why the Zin of the inverting input would be markedly different to the stock non-inverting input.

In fact, it would be dominated by the value of the feedback resistor (which is comprised of two) and this would be by design equal to the base bias resistor, R1, on the non-inverting input.  This is 47K.

The input impedance of the bases on the input and feedback transistors can be calculated and is high when feedback is considered.  On average, the Zin of the non-inverting input should be around 43K - an easy load - while that of the non-inverting input should be a little higher due to the feedback shunt resistor of 2.2K which should be added giving 45.5K.

In simple, practical turns this should mean that the Zin of the balanced configuration would be very high for both inputs, and there should be no problem driving it from a relatively weak source, with as high as 5K Zout.  It should be fine.......

One thing, however.  Balanced line input will double gain, so you will need to reduce gain on the amp by increasing emitter degeneration on the input diff pair.  The two resistors used here which feed the LTP emitters (T1 and T2) should be increased to 47R EACH.  This will bring gain back to around 31dB, the same as the gain in single ended mode.  This is important to sonics too, as you don't want to alter the feedback factor which is very accurately set for best sonics.

Patrick, I hope this is helpful, let us know your results!

Cheers,

Hugh

[J Harris]

Wow, thanks Hugh. I think I'll undertake this project during the long, slow weeks of January.

Patrick

[J Harris]

I've just ordered a couple of 22uF PC-mount 350VDC Black Gate caps from Angela. $16 each.

The rest just seems to be resistors... now to look for a high-quality pair of balanced input jacks.

Patrick

[Sintz]

Wow, thanks Hugh. I think I'll undertake this project during the long, slow weeks of January.

Patrick

Patrick-
Let me know how this turns out. Now that Hugh's taken the time to do the math/homework on this conversion and you've started ordering parts, I'm regaining interest in the project.  

Would you be willing to keep a parts/price list, time spent, pitfalls, etc you encounter? This could turn out to be a great sticky thread once it's been completed --- with feedback on how (if) it affects the sonics of the AKSA.

[J Harris]

Sintz -

I definitely will do so. But it could take some time and lots of questions!

Patrick (J Harris)

[J Harris]

OK, I think I've got my parts. And I've definitely got my questions

parts list:

2 X 22 uF 350 VDC Black Gate caps (I've just realized these are much too high voltage - but I don't suppose that will hurt anything, will it?)

2 X 2.7K resistors (red, violet, black, brown)
2 X 2.2K resistors (red, red, black, brown)
4 X 47 ohm resistors (yellow, violet, black, gold)
2 Neutrik balanced input jacks

Now if I've parsed this thread thorougly, the instructions should look something like this:

2.7K resistor - insert to ground from the existing non-inverting input (presumably the HOT input signal pin?)

22 uF Black Gate cap - replace non-inverting input cap (which I take to be C8?)

2.2K resistor - "feeback shunt resistor" - where should this be added? in parallel with R1? (and is R1 the resistor directly above the words "signal input on the circuit board? it becomes hard to tell what's what once you've actually installed everything on the board and covered up most of the names)

47 ohm resistor - replace the resistors feeding T1 and T2 with 47 ohm resistors. Hugh, sorry to be dense, but which resistors are these? the foil-side sand-cast resistors? or - ?

Neutrik balanced inputs - drill out rear of case and insert. These have a third input, compared to single-ended inputs - I'm presuming that a wire should run from there to star earth?

Once done, presumably I need to rebias, then remeasure everything as before, and look for the same +49V -49V on the rails?

Thanks everyone! Looking forward to this January project (and an end to my RFI/EMF with luck).

Best
Patrick

[J Harris]

Hey Hugh,

Any thoughts on my questions above? I'm eager to get started on this project and banish noise from my balanced-to-single-ended adapters!

All the best
Patrick

[AKSA]

Hi Patrick,

My apologies!  I was doing the old fart bikey thing yesterday and took a 200 miles jaunt into the country....

2 X 22 uF 350 VDC Black Gate caps (I've just realized these are much too high voltage - but I don't suppose that will hurt anything, will it?)

Yes, probably a bit high as at low voltages they don't sound good.  Could you move to a 22uF 10VW or so?  You need not use a BG (save them for your next tube project!), you could use a Sanyo OSCON, or Nichicon Muse, or Hitano EXR.

[2 X 2.7K resistors (red, violet, black, brown)
2 X 2.2K resistors (red, red, black, brown)
4 X 47 ohm resistors (yellow, violet, black, gold)
2 Neutrik balanced input jacks

Looks OK, yup!

2.7K resistor - insert to ground from the existing non-inverting input (presumably the HOT input signal pin?)

Yes, but it should go in at the input side of C1, the input cap.

[22 uF Black Gate cap - replace non-inverting input cap (which I take to be C8?)

Yes, but N-I input cap is C1, C8 is Vbe bypass cap, take care with this!!

2.2K resistor - "feeback shunt resistor" - where should this be added? in parallel with R1? (and is R1 the resistor directly above the words "signal input on the circuit board? it becomes hard to tell what's what once you've actually installed everything on the board and covered up most of the names)

Patrick there's no need to add it;  it's already on the board as R11.  All you have to do it remove C2, the feedback shunt cap, replace it with the very good quality 22uF cap above, connect the + termination the pcb and drive it at the negative terminal, which you raise from the pcb, don't solder in.

47 ohm resistor - replace the resistors feeding T1 and T2 with 47 ohm resistors. Hugh, sorry to be dense, but which resistors are these? the foil-side sand-cast resistors? or - ?

These are listed in the pcb (which should be V1.75) as R4 and R5, just above R1.

Neutrik balanced inputs - drill out rear of case and insert. These have a third input, compared to single-ended inputs - I'm presuming that a wire should run from there to star earth?

Careful here;  + input of XLR is the normal HOT input on the amp pcb.  - input of XLR is now the negative terminal of C2, which has been lifted from the pcb and now becomes the inverting input.  The ground of the XLR connects to the GND terminal pin in the pcb.  That's it!!

Once done, presumably I need to rebias, then remeasure everything as before, and look for the same +49V -49V on the rails?

You shouldn't need to rebias, it won't change the bias, but check it just to be sure.  The same +/-49V is just fine, and the amp will deliver the same power without a hitch.  It may have a little more gain if driven in differential mode;  it this is so, you can reduce the gain by increasing the value of R11 to around 2K7.  That should do it nicely!!

Good luck Patrick, and sorry for the delayed answer.

Let us know what you think of the sound.  I have high expectations!

Cheers,

Hugh

[J Harris]

Hi Hugh!

Hi Patrick,

My apologies!  I was doing the old fart bikey thing yesterday and took a 200 miles jaunt into the country....

No problem whatsoever Hugh - your plentiful help and suggestions are always appreciated! 200 miles though - you Australians get out a lot -

Yes, probably a bit high as at low voltages they don't sound good. Could you move to a 22uF 10VW or so? You need not use a BG (save them for your next tube project!), you could use a Sanyo OSCON, or Nichicon Muse, or Hitano EXR.

So a 22 uF, 25VDC Nichicon Muse FS would work fine? Probably break in faster too!

2.7K resistor - insert to ground from the existing non-inverting input (presumably the HOT input signal pin?)

Yes, but it should go in at the input side of C1, the input cap.

So just to be absolutely clear, the 2.7K resistor should go from the input side of C1 to ground, right? ground here meaning EARTH on the PCB i.e. star earth?

22 uF Black Gate cap - replace non-inverting input cap (which I take to be C8?)

Yes, but N-I input cap is C1, C8 is Vbe bypass cap, take care with this!!

Ah, now I'm confused - C1 is already a 22 uF Black Gate in my amp -- presumably because I've done the N+ upgrade. Does this mean I don't have to replace C1?

2.2K resistor - "feeback shunt resistor" - where should this be added? in parallel with R1? (and is R1 the resistor directly above the words "signal input on the circuit board? it becomes hard to tell what's what once you've actually installed everything on the board and covered up most of the names)

Patrick there's no need to add it; it's already on the board as R11. All you have to do it remove C2, the feedback shunt cap, replace it with the very good quality 22uF cap above, connect the + termination the pcb and drive it at the negative terminal, which you raise from the pcb, don't solder in.

Once again, C2 is already a 22 uF Black Gate... again part of N+... so neither C1 nor C2 need to be replaced right? I just need to lift the negative terminal of C2 off the circuit board so that it can feed the negative terminal on the XLR jack as per your instructions for wiring those?

It seems to me I already have the necessary cap upgrades as part of N+... which means I don't have to order the Nichicon Muse... and means I can do this over the long weekend coming up! Which would be wonderful!

I'll be very careful, don't worry Hugh - it's just tough to see which parts are what once they are already mounted on the PCB (which is indeed version 1.75) because the parts are covering the indications - I don't suppose you have a JPEG of the plain PCB would you? I understand if this is not something you send out to people due to intellectual property concerns....

Anyway, I think I've got it now - so long as I'm right that I've actually already upgraded the necessary caps C1 and C2 to 22 uF Black Gates... please let me know!

and have a great weekend - I have a feeling I'll be posting again here shortly  

Best
Patrick (J Harris)