Dynaco PAS becomes the ImPasse!
     
  Stuart Yaniger wrote a great article for AudioXpress in February 2009, The ImPasse Preamplifier -- a preamp to drive a pair of Nelson Pass' First Watt F4's in balanced mono.  The F4 is a unity gain amplifier requiring a preamplifier capable of driving  the full expected output swing.  As two channels of the F4 should be able to swing 40 volts in Class-A, for 100 watts of output power, the preamp must be capable of delivering 28.3 VRMS.  
     
  As yet, AX has not put article on their website, so in respect of their intellectual property I will not publish the schematic here. I would say, however, that this little circuit alone is worth the value of a subscription.  The distortion is very, very low for such a simple, yet elegant design.  
     
  A Rough Description:  
  The ImPasse uses a 6SN7 to drive a 6922 in balanced mode. There are LED's (a favorite Stu-device), NE2H neon lamps and depletion MOSFETs in the design.   
     
  In my configuration, I used a pair of Sowter input transformers (Jensen or CineMag could be used just as effectively.)  The amplifier is housed in the carcasse of an old and not-working Dynaco PAS which I had found in a dumpster. In a previous modification I had removed the old power supply circuit and installed the LastPAS power supply discussed in a 1982 issue of The Audio Amateur.  (Link1), Link2). In the current version I use Stu's Maida Regulator.  In order to provide some of the functionality of the PAS, I allow the phasing and tone controls to be "switched in" (the tone controls are made possible with an a pair of LM4562 operational amplifiers.)  For normal audiophile level listening, however, input is derived from the XLR connectors on the back.  Output is balanced mode via another XLR connector.  
     
  Since I had simulated the circuit in Multisim, I decided to transfer the layout to Ultiboard and burn some printed circuit boards.  As the preamplifier was to fit within the snug confines of the Dynaco chassis I constrained the choice of components to fit.  The completed PCB is shown below:  
     
     
     
  I used Sowter 10K:10K in put transformers as shown above.  I designed the pcb so that it could be snapped apart in case a user wanted to place this part of the PCB closer to the inputs.  Below is a close-up of the transformers:  

 
  I've always loved NEONs -- as described in the article, the two NE2H bulbs limit the grid-cathode voltage differential.  They light up when the amplifier is first turned on, then gradually dim out as current is drawn across V2.  
 

 

 
  The neons twitter when they first come on, while the tubes warm up, then go dormant.  
 
  I also like LEDs to illuminate the music:

 

  
 
  To defray the cost of the board for yours truly (with a silk-screen it's rather expensive) I offered the balance out as a "Group Buy" on DIYAUDIO and got rid of most of them.  Regrettably, I made a mistake in the orientation of the depletion MOSFETs used as a constant current source for the 6SN7's -- but this is easily remedied by making the "dance" a la:  

Looks like a pair of Irish Step Dancers if you ask me!
  The current on each half of the 6SN7's must match if output from each channel is to balance.  To adjust the current I solder in the depletion MOSFETs and the associated resistors.  Connect the drain of Q2 to the positive node of an 18V supply and the ground node to the plate end of R4.  Adjust the current flow through the two regulators by paralleling in values across R5 until you get 8mA to flow in each half.  I start with a value of 200Ω for R5, and have used parallel resistor values ranging from 2k to 2.67k a la:  
 

 
  Yaniger had provided a version of the Maida high voltage regulator, with a single TIP50 pass transistor.  If you aren't familiar with the Maida, the "Linear Brief" from National Semi's website can be found here. LB-47 .  Walt Jung wrote a 2 part series of articles for AudioXpress in 2007 which describe the use of Depletion MOSFETs as current sources.  JungCS_1 and JungCS_2 .   
     
     
     
  Initially, I ran the preamplifier with the Darlington MAIDA shown in LB-47.  I also ran it with a temperature compensated MAIDA and with a regulator using an output boosted SuperTex LR8N3 regulator.  In time, however, I found that the little Dynaco transformer got pooped out with all the demands of the ImPasse.  In addition, I wasn't satisfied with the topology I used to lift the filaments of the preamplifier tubes above ground.  Stu had mentioned a transformer available from Allied Electronics with a center tap for the 6.3V.  The Allied Transformer offered another benefit, a spare 5V winding which could be engaged to drive the filaments of the RIAA preamplifier stage of the PAS.  The photo below ilustrates the difference.  

 

     
   
     
  Not content to leave well-enough-alone, I took Stu's schematic for the power supply and transferred it to Ultiboard.  Once again, to knock down my cost I offered a bunch of the boards to the DIY community at cost.  One could certainly assemble the circuit on perf-board, but a printed circuit board, well, even here in the Dynaco it makes it look a bit more professional.  
     
 

  A little creativity was required to fit the huskier transformer into the Dynaco chassis.  I removed the screws from the trafo and replaced them with 2 inch 8-32's and hexagonal spacers and attached it to the side.  Not shown above are the rubber grommets which I used so that the wiring to the transformer and power line would not chafe.  
 

 
  When testing the power supply, I found that R103, R104 (470Ω  3W and 1KΩ 3W ran much hotter than I estimated.  While the current is only 38mA at this juncture, (0.67 and 1.44W respectively) both devices were approaching 100oC and were sure to cook when the cabinet was put back together.  I will use a pair of 3W resistors in each position so that the devices don't self-destruct.   
     
  Not shown is a heat sink on the pass transistor and LM317 voltage regulator IC.  Most of the "work" is done by R103 and R104 so the TIP50 doesn't dissipate much heat.  A few square inches of aluminum have proven satisfactory.  
     
  Well, this is the state of affairs at present.  The input selector switch is one that Joe Curcio sold for a PAS mod.  The wiring is CAT5 where the jacket makes it easy, and tefzel coated silver for the most part.  I still haven't wired up the RIAA circuit and the little tone control and buffer are still on the drawing board.  
     
  The phase/mono switch had a busted wafer when I found the preamplifier.  Perhaps it's for this reason that it was in a dumpster.  I replaced the phenolic switch with a Centralab PA-1000 which was to have become the band-switch for a ham radio receiver about 4 decades ago.  (Hint, the folks on the HP/Agilent Yahoo group were helpful in figuring out how to clean the silver contacts.)  
     
 

 

 
 

So, how does the impasse perform -- some charts to illustrate:
 

THD Spectrum for 2.83V Output
 

THD Spectrum for 20.0V Output

 

Bandpass, One Channel
 
This is the switching arrangement used to insert the Impasse preamp between the source and the pair of F4's, click on the image for a larger version:

 
 
 
 
 
 

If anyone is interested, there are still boards available for the line amplifier and the power supply  AudioXpress has granted us permission to provide you with a copy of the article. LINK