Push-pull microphone preamp with UTC LS and Tamura transformers

[aazaa]

Final version, pictures and circuit: see page 7, posts #127 & 128.
Preliminary schematics are erased.

On Ebay I recently bought 2 Tamura SMG-1000 transformers and few weeks later 2x UTC LS51 at reasonable prices. It was the start of a project to build two push-pull microphone amplifiers.

On this forum hardly anyone seems to be interested in push-pull preamps. Often people convince others not to go that way.
Searching the whole site, I didn't find any circuit that was tempting.

So I started to draw a new circuit.
V1 and V2, two double triodes wired as 2 long tail pairs. (Possible to choose other tubes).

First, as reference for sound, the circuit below was built without capacitors. So no 100n cap, no 1M resistor.
Full direct connection. Simple and ... possible. Recording a few samples played on my acoustic piano proved convincing.

Of course, 100n and 1M were added to maintain a low DC between V2's anodes over the long term.
Surprisingly, the output didn't change with or without the 100n capacitor. (On scope in both cases low frequencies flat to 15Hz!). I suppose the long tail approach in V2 is correcting superbly the missing direct connection below.

Negative feedback can be easily adjusted by changing/omitting the 100k resistor.

There is a roll off in the highs. Easy to correct perfectly by adding a capacitor and a resistor.
But... I happen to prefer the sound without correction. So I'm hesitant to add correction in my final build.
On the attached circuit I have drawn a possible solution in grey: an additional switch to choose between highs flat or rolled off.

Now I am rebuilding the first preamp. Too many experiments turned the wiring into a mess.
The second preamp will be a complete UTC build. UTC LS10 and UTC LS51. I'm curious about the differences beween the two builds.

A separate experimental PSU has been built to power the two amps.
Currently also very simple:
- High voltage: AC/resistor/bridge/100µF/resistor/100µF/choke 50H/100µF.
- DC for filaments. All filaments in series to avoid overheating of the LM317HVT.
A current limiter is added to assure a soft start.

So almost all capacitors are in the PSU.
Now only 1 cap in the preamp, maybe later 2 if I choose to add correction.

The preamplifier is very quiet.

-----------------------------------------------
Edit:
+390V version. post #63
heater PSU. posts #64 & 68
hybrid version with UTC LS-30 and LSK389. post #70
corrected negative feedback - intermodulation distortion. post #90
harmonic distortion. posts #91 & 92
Improved trimming. Stabilisation of low DC and low distortion. post #100
Optimised voltages and currents. The third harmonic distortion peak. post #101
Tune the ecc83 - or not, a different approach. posts #109 & 110
Sylvania 6SN7GT and RCA 12AU7A. posts #118 & 121
Octal - UTC version (final version). posts #127 & 128
gain control. posts #130-135
Thanks. post #136

-----------------------------------------------

 

[Winston OBoogie]

Yep, you're correct that there hasn't been much talk of push-pull pre amps on here.  I've been visiting since the move from the 'other place' in 2004/5 or whenever and I'm struggling to think of a unit that gained momentum. 

I don't know why, other than there are fewer known old units that folks want to clone maybe? 
Langevin 5116B and UA1016  are two that folks might know?  There's a nice McCurdy unit (A300?) that I had a couple of at one time.  There's an example in RDH4 too.
Many other cool types that folks like Doug "EMRR" on here will know about I'm sure.

I've been playing around with one myself on and off and, as it happens, mine is DC coupled too

I can't see too clearly on this device but of what I can see, yours looks promising.  I'll open it later on a laptop so I can get details.  Sounds and looks like you don't really need help with things, but I'm absolutely interested and I'm sure others are too.

Cool beans for starting the topic 

 

[emrr]

We recently talked through a SE/PP combo using UTC iron. 

DaveP built an Altec 458A clone here.

Really, most people focus on PP  compressors here, and then go “hey, it’s also a preamp!”, but almost no one sets out to build such a thing as a dedicated preamp. 

Lots of historical choices for inspiration.  Many program (bus) amps, all of which work well as preamps.

Yours did remind me of a long ago drawing Winston posted, as he mentioned. 

Not much awareness of them in clone world, really.  Not many original designs here, between people wanting straight clones, and designers of new ideas being spooked about IP hijacking, as has happened.  Unfortunate.

I’ve a set of LS inputs and LS51 outputs waiting for a PP build, a number of ideas on paper in need of ‘feature creep reduction’. Another set of same inputs and LS50 outputs for a contrasting SE build.  I lean more into the camp of “what would one do with antique design methods in the face of modern usage?” 

I’ve mentioned elsewhere here, all the PP and SE capable UTC (and many other brands) measure better in PP than in SE.  The humbucking winding cancellation notch seen at the top of the response range in SE disappears in PP. 

Enjoy that preamp, look forward to your next version.

 

[abbey road d enfer]

Other than the curiosity factor, it would be interesting to know what are the expected advantages of such an arrangement. I reckon it's the only way for getting other members interested.

 

[Winston OBoogie]

Other than the curiosity factor, it would be interesting to know what are the expected advantages of such an arrangement. I reckon it's the only way for getting other members interested.

There need to be advantages, either perceived or real, to clone or build stuff now?



How about there are twice as many pins to play with?

Seriously, I can't speak for aazaa.  But I'm not sure the reasons I'm playing around with push-pull in my particular circuit (to reduce amplitude, harmonic, and freq. response non linearities)  would be seen as an advantage by many at all on here.  Why use tubes anyway then right?  The fact I have a lot of 3 legged silicon devices playing support roles to the tubes probably doesn't strengthen the case. either.

Or that my B+ is 520V total (2 X 260V stacked supplies) because it has synthesized inductor loads (gyrators) instead of chokes (my favourite chokes are too expensive but I wanted the signal voltage swing of chokes) so...
Maybe I'm not helping anything here?

I don't know if you were asking a rhetorical question, or being serious as I know you know what push-pull does and doesn't do. 

Help!

 

[abbey road d enfer]

I don't know if you were asking a rhetorical question

More a philosophical one. I believe it's important to know why one does things. But you have partially answered it "There need to be advantages, either perceived or real, to clone or build stuff now?" , 

or being serious as I know you know what push-pull does and doesn't do. 

Sure I do know about reducing 2nd order THD and possible noise and drive improvements, but is that all? I'm genuinely asking. It may be curiosity, which I recognize as a perfect motivation.
The worst case would be "I want to try it because (enter your favourite most hated guru) said it's better.  8)
OK, advantages is too pedestrian, let's say expectations.

 

[Winston OBoogie]

More a philosophical one. I believe it's important to know why one does things. But you have partially answered it "There need to be advantages, either perceived or real, to clone or build stuff now?" , 
Sure I do know about reducing 2nd order THD and possible noise and drive improvements, but is that all? I'm genuinely asking. It may be curiosity, which I recognize as a perfect motivation.
The worst case would be "I want to try it because (enter your favourite most hated guru) said it's better.  8)
OK, advantages is too pedestrian, let's say expectations.

OK got it.  Thanks Abbey.
For me, it was mostly curiosity.  I'd been playing with a few ideas to overcome some issues in a vari-mu comp -  Doug was right about that pathway - and the ones that worked for me and seemed applicable are the ones I'm playing with now.

I don't have an end date in mind or final goal so there's no sense of urgency, therefore things have moved slowly.

As aazaa said of his circuit, mine has also had so many experiments done and undone on it that it's barely holding itself together now.

I don't want to hijack aazaa's thread but I could gather my thoughts and post what I found out somewhere.  I highly doubt it's anything new that I've been doing, in fact I know it isn't.  Just a collection of existing stuff.

Cheers.   

 

[aazaa]

Other than the curiosity factor, it would be interesting to know what are the expected advantages of such an arrangement. I reckon it's the only way for getting other members interested.

The writing of Doug in the preceding post seems not enough to convince you:

I’ve mentioned elsewhere here, all the PP and SE capable UTC (and many other brands) measure better in PP than in SE.

Moreover I like the idea of using opposite triodes correcting each other when wired in a long tail circuit. Idem for the 'crossed feedback' connecting the opposite sides.
No decoupling capacitors needed for the cathodes.
And of course there is a pure curiosity to discover the sound of this circuit using vintage tubes and transformers.

So, a simple design. Full symmetrical wiring of tubes and transformers with a minimum use of capacitors.

But there is no need to convince. Surely no need to do the same discussion over and over.
My intention was only to share an idea with the few people interested.

Paul

 

[abbey road d enfer]

Moreover I like the idea of using opposite triodes correcting each other when wired in a long tail circuit. Idem dito for the 'crossed feedback' connecting the opposite sides.
No decoupling capacitors needed for the kathodes.
And of course there is a pure curiosity to discover the sound of this circuit using vintage tubes and transformers.

So, a simple design. Full symmetrical wiring of tubes and transformers with a minimum use of capacitors.

Pardon me for being thick, but I would think that you did not come across this topology by chance  and that you expected some specific results. What aspect of performance did you expect.
No capacitor is not a goal in itself IMO.

 

[aazaa]

Pardon me for being thick, but I would think that you did not come across this topology by chance and that you expected some specific results. What aspect of performance did you expect.
No capacitor is not a goal in itself IMO.

OK Abbey. Here is the full story. You ask for it. Warning: can be boring!
I had no intention of telling it.

First this. I have a history of building PP amps / preamps or SE amps every now and then as an amateur. It's a habit to look at existing circuits, but in the end I always draw my own circuit. Impossible for me to build a copy.
There is a strong belief in the quality of keeping things simple.

A few of months ago I had the intention to build a SE mic preamp. UTC LS10 and UTC LS 151, parafeed circuit.
As I didn't have a plate choke, I decided to do an experiment using the primary of an old Luxman PP output transformer as choke.

Experiment for fun. PP transformer, so push-pull circuit. I immediately saw the possibility of getting rid of all capacitors. Same circuit I'm using now. The primary of UTC LS151 was connected between the anodes without caps. (There is no DC between the anodes). As it was only a stupid experiment, I used the common ECC83/82 Philips tubes laying around.

Test recording. Wow! What was that....?
Best sound of my piano I've ever recorded.

A bit silly to build a PP parafeed preamp. So I patiently searched Ebay until the Tamura and later the UTC LS51 transformers appeared. Tamura was cheap as no one knew the complete data. I'm pretty sure about UTC LS51's qualities, so I decided to build the Tamura preamp first. Took the UTC LS12 from an old build with fets.
Decided to keep the ECC83 and 82.

That's it. Pure coïncidence, maybe also having the idea of trying something uncommon.

Paul

 

[abbey road d enfer]

OK Abbey. Here is the full story. You ask for it. Warning: can be boring!

Not boring. Just makes sense. Now do you have measurements that would explain why it's so good?

 

[emrr]

do you have measurements that would explain why it's so good?

I rebuilt a couple related 1932-33 preamps this month, one a repair job, one my own for comparison.  Both 2 stage SE tetrode input triode output.  Similar gain.  To my disappointment, my own unit did not measure nearly so well as the repair job unit.  I sent files to a number of friends, they universally picked my unit as the better sounding.  I’ll chalk it up to a preferential mating of the microphone, another mic might present better on the other preamp.  A discovery of a moment in time. 

 

[aazaa]

I can't see too clearly on this device but of what I can see, yours looks promising.  I'll open it later on a laptop so I can get details.  Sounds and looks like you don't really need help with things, but I'm absolutely interested and I'm sure others are too.

Cool beans for starting the topic 

I’ve mentioned elsewhere here, all the PP and SE capable UTC (and many other brands) measure better in PP than in SE.  The humbucking winding cancellation notch seen at the top of the response range in SE disappears in PP. 

Enjoy that preamp, look forward to your next version.

Winston and Doug, many thanks for the nice reactions and interesting comments!

Doug, I really like the quote of Crusty at the bottom of all your replies:
"I think this can be better. Some kind of control that's intuitive, not complicated like a single knob" - Crusty"

I'll keep that in mind when finishing my preamps.

Paul

 

[aazaa]

I rebuilt a couple related 1932-33 preamps this month, one a repair job, one my own for comparison.  Both 2 stage SE tetrode input triode output.  Similar gain.  To my disappointment, my own unit did not measure nearly so well as the repair job unit.  I sent files to a number of friends, they universally picked my unit as the better sounding.  I’ll chalk it up to a preferential mating of the microphone, another mic might present better on the other preamp.  A discovery of a moment in time.

Nice answer about the ability of measurements to prove 'why it is so good'.

After the necessary control measurements, listening should be decisive.
Very pertinent of Doug to state that the sonic result depends on the interplay of the various elements involved.

A bit unfortunate that the strange PP Luxmann/UTC parafeed device has been disassembled. At the time, I had no idea to keep it for reference.

Maybe some information about the transformers can be helpful?

First the fabulous specifications of the UTC LS151:
primary impedance: 16.000 Ohms
secundary impedance: 50,125/150,200/250,330,500/600
+- 1dB from 7 to 50.000 Hz
maximum level 400mW
relative hum -74 dB
unbal. DC in primary: max 1mA ( ...why always so much fear about it, in this forum...???)
Rdc primary 860 Ohms

Luxmann PP transformer (primary used as choke) :
pulled out of a push-pull EL84 amplifier at least 25 years ago.
Rdc: 517 Ohms
L= 350 Henry (total primary from anode to anode, measured at 50Hz-10V)

-------------------------------------------------------------------------------

Other story with the Tamura SMG-1000 transformer:
No information available. Tamura did not react on my questions.

I suppose it was intended to be used as input or as interstage.
winding 1 (primary) with center tap, Rdc:27,5 Ohms (used as low impedance output)
winding 2 (also primary?), Rdc 471 Ohms, L=1590H
winding 3 and 4 (used as PP primary): Rdc 679/677 Ohms, L=820H/820H ( only one measured)

ratios: 1/4.091/5.784/5.784
weight: 1082g

Paul

 

[aazaa]

Left channel has been rebuilt.
It is the first time I've made a turret board to mount all components.

All resistors on top of the turrets (but one, was mistake). Easy to control, change or remove them.

Measurements can be done on the turrets without fear to make shorts.
Easy adjustment of the trimpots.
Acces to the wires - hardly necessary - by unscrewing 1 nut.

Paul

 

[Winston OBoogie]

 

I do know about reducing 2nd order THD and possible noise and drive improvements, but is that all? I'm genuinely asking. It may be curiosity, which I recognize as a perfect motivation.
The worst case would be "I want to try it because (enter your favourite most hated guru) said it's better.  8)
OK, advantages is too pedestrian, let's say expectations.

Taking any typical single-ended circuit in which we’re already using a couple of dual triodes and a pair of transformers -  ask if, for the price of < $3 in passives, it’s worth running the experiment to see if that $3 is worth spending? 

Even if our expectations are only a reduction in open-loop thd., it seems a worthwhile use of time and money.

To this reduction (cancellation) in thd. we’re pleasantly surprised that we also get, for free, some or all of the following:

Noise from the power supply:  the CMRR of the circuit symmetry will reduce the effect of any ripple on the H.T. 


The input transformer:  our $3 we spent on passives also buys us extended frequency & phase linearity from the IPT. 
Not many current transformer manufacturers supply data with single-ended vs balanced secondary loading.  Anyone can run simple tests of Cinemags, UTC’s etc into an A.P. or equivalent test setup to check this, but Lundahl sometimes supply this data and, looking at their premium LL7906  (an appropriate ratio for tubes), we get almost an octave more out of it when balanced secondary loaded:  -1dB @ 45KHz vs -1dB @ 25KHz.
The LL7903, although already great at -1dB @ 70KHz single-ended terminated, is astonishing (for a high step-up) when balanced terminated.  Don’t take my word for it

After the input transformer:  for a given tube’s bias point (usually about -1.5V for a tube mic amp), we have 2 tube halves now sharing the signal.  Our overload point is higher.

I’ll skip over the rest of the amp circuit as it’s more relevant when talking about specific circuits.  Suffice to say, I’ve found things like compensation, miller, phase, lag etc.  much easier to deal with in push-pull.

Output transformer:  For a given size of transformer, in single-ended driven - we’ll have to compromise on headroom, low end response, or add additional expensive components to compete with push-pull driven.

It’s not even always necessary to specify a gap to run DC through the primary  windings and load push-pull output anodes.  A lot of quite inexpensive transformers seem more than tolerant of 1mA or so of imbalanced current at these levels.
So then we have the stored energy of the coils and we’re able to swing almost twice our H.T. voltage. 
In single-ended, we’d need a specified gap (which will compromise low end), or use a parallel fed inductor (expensive), or double our H.T. supply and then parallel feed resistively or with CCS loads our anode(s) to achieve the same result.

Seems worth $3 to me

Forgot another obvious one:  loading the output anodes with coils will shift the load line horizontally.  Granted, not as much as a pure, apparently very high z constant-current load, but the output valve will be more linear nevertheless. 

 

[aazaa]

Wow Winston, what overwhelming support! 1000x thanks.

Tomorrow I will post some reflections on the fear of reaching 1mA unbalance.

Paul

 

[Winston OBoogie]

Cool

One thing about you circuit which I'm certain you know -  Regarding the high end roll-off.
I know you're happy with it, and you also added a compensation switch to flatten it out if need be.

I'm certain the dominant pole is due to the impedance of the input transformer looking into the stray and input capacitance of the 12AX7.  This multiplied by the gain of course (miller).

An old, but easy, trick to ameliorate this somewhat is to add small pF value caps from each input anode to the opposing valve's grid.  Cross-coupled neutralizing.
This may be useful if, for instance, you wanted more feedback but were worried about having enough phase stability margin.  You would neutralize the input stage and have the 2nd stage provide the dominant pole.
Another use for this trick might be if you wanted to eek out a few dB better signal-to-noise ratio. Doubling up on 12AX7's will net you about 2dB of the theoretical 3dB improvement.  But with added input capacitance.
So neutralizing with a couple of pF caps will get you out of high-end droop trouble.

I know you know this of course

Looks good Paul.

 

[emrr]

The Altec 458A is a good example of cross-coupled neutralizing.  An interesting balancing act intended for matched condition output loading to define plate conditions and resulting specs, and an input transformer with a 150K secondary designed to land total amp gain at a particular quantity.  There's an AES paper on it, and PRR has posted about it.  An example of an amp in which one side is designed for tube failure, and the amp soldier on SE with an octave reduction in bandwidth at each extremity, until such time as the failed tube can be replaced.  7-8mA current imbalance in the normally balanced PP output. 

I would note most the increase in headroom and max output level with PP designs.  Historically you see them in bus amps for that reason, SE in preamps intended only to feed a low level bus, not stand-alone.  Once you get into the late 50's you see many more dual duty plug in amps with PP outputs which can be used for other purpose.  Those Altec stand out at a quoted headroom of +20dBm 0.5% THD 50-20K, for single stage gain without feedback beyond cathode bypass and cross-coupled neutralization. 

I believe way back in time analag had posted some PP designs with output coupling more rarely seen, like dual parafeed with the output primary floating, maybe some other things too. 

There is a trend of old tube bus amps with SE design, all of those seem to quote max outputs of +24dBm using power tubes like 6V6 or 6L6 into huge output transformers that look like power amp outputs, laid out on a lot of real estate.    Some of those are even SE parallel.  All of the tube bus amps with +30dBm/1W output capability are PP. 

 

[Winston OBoogie]

Thanks for the reference of the AES paper and the 458A Doug, I haven't read it but I'm def. interested.  Hunting for it now.

On floating the output transformer:  Yes this is what I'm doing, it works well. 
I have high gm tubes (my transformer is 3 X rp Z for load instead of the usual)  that are actively loaded (gyrators but could easily be CCS's, or chokes if you have the $ to spend) and then feeding top and bottom of the OPT windings with just one blocking capacitor in-between the windings. 
Honestly, the cap is there for my sanity right now.  The differential DC is almost negligible once things warm up, and having one less reactance there would be nice.  Servo-ing my active loads  anyone? 

Thanks again dude.


Edit P.S.  We could always just take that 12AX7, 12AU7, and two transformers and cap couple a bunch of single-ended stages together.  We'll wrap more -tve feedback around it all to even things out a bit.
We'll end up with a box that sounded and performed just like a 1950's Sears and Roebuck hi-fi.