I'm building Tube Amps, and I can't stop!
 
In addition to building a small amplifier based around Russian military tubes and a Chinese SMPS, I have been accumulating parts to build a EH 7591a based amplifier.  The Output transformers will come from a Rowe/AMI R-4359 amp chassis that came from a jukebox.  The original amp was based on 7868 tubes.  7591's and 7868's are electrically very similar, just in a different basing and envelope, so I also plan to stick with the R-4359 schematic fairly closely.
 
Output Transformers, and the basic circuit design of each channel is where the similarities to the Jukebox Amp will end.
 
The amplifier will be built in a rack-mount chassis (which is also surplus).  I am not using the Power Transformer, 5AUG Rectifier tube or any of the original power supply circuitry.  Instead, the needed voltages for the tubes will come from a combination of Switch Mode Power Supplies, as I have done for my other projects.  A 20Amp rated SMPS will provide 12VDC from a 100-240V input.  The 12VDC (at lots of Amps) will power a significantly modified Chinese Inverter module along with a rectifier, filter, and feedback system of my own design.  This will provide approximately 380VDC to the plates of the EH 7591a's.  Other modular SMPS's will provide the ~6.3V needed for the 7591's heaters, and a +/- output for the tube biasing circuit.
 
It will also contain a high performance DAC to allow USB Audio, as well as Optical and Coaxial SPDIF inputs.  Of course there will be straight analog inputs, too.  A digital, audio grade relay board will allow push-button switching between the digital inputs and 3 analog inputs.
 
It will also have a VFD Spectrum display on the front panel.
 
Yes, it's an ambitious project, and it has to wait for other projects ahead of it, but in the meantime, here is a rendering so that you can at least see some artwork:
 

Just a minor status update.
 
I have cut the top panel (but not started drilling as of yet).
 
I have also decided to use wood side panels.  I asked my wife to bring me some thin finishing wood from Lowe's, and she brought home some plywood where the inside layers are wood paste (particle board).  It turns out that the piece of plywood that was used for shipping me the original Rowe/AMI chassis was actually better quality plywood.  It is also the same thickness.
 
So I managed to cut 2 decent pieces from the shipping plywood.  I have sanded, stained, and polyurethane'd them, and they came out pretty nice looking.  There should be no problem fitting the unit in a standard rack with the thin plywood sides.  Additional wood work may be necessary to make the rack mount ears fit properly.
 
The next step is to buy a threaded insert rivet tool and do a little structural engineering so that the chassis is structurally sound enough to be supported by the ears in the event that it is actually installed in a rack.
 

It's another year, and little additional work has been done on this project;  However there's been a significant design change.
 
As mentioned in the previous post, I was attempting to work on the final layout (and drill plan) for the top panel.  During this effort, I became very concerned about both physical interference and electrical interfernce due to the proximity of the input circuitry to the Vacuum Flourescent Display (VFD) Spectrum Analyser module.  The VFD employs high voltages, and display scanning in the AF range.
 
After some addtional thought, I came up with a new layout for the amplifier.  I have moved the 2 OPT's together at the back of the chassis, and placed the Output Tubes on the sides.  This allows space to move the 12AX7's back signifcantly away from the VFD's.
 
While this has the disadvantage of moving the Power Tubes closer to both the 12V and HV SMPS's, the signals around the Power Tubes are much higher level, and therefore less subject to interference.
 
It also tends to "Hide" the power tubes behind one another, but this is not expected to visually detract from the apperance of the amplifier from any real viewing angle.
 
I will upload the latest render when I get a chance.  Latest Render added below, showing the new tube layout.
 
After a little time to review and think about this significant design change, I should be able to start drilling the top panel.
 
As of right now, the plan is to paint the entire chassis in-house to save time and money.
 

As we get closer to the actual assembly/wiring part of this build, I realized that I had not shared the following picture here:
 

• Not Shown: Output transformers. They are on-hand, but were still attached to the Rowe R-4359a chassis when this picture was taken.
  
• Not Shown: (Most) fasteners, discrete components, and quantity components such as wire, heat shrink etc.
  
• Not Shown: 6VDC Buck converter. I do have one of sufficient current rating on hand, but I'm not sure if I'll use it. The alternative is to just series wire the 7591a heaters for each channel.
  
• Yes, there really are tubes in the boxes. The 12AX7's are also branded Electro-Harmonix. All are new and new build.
  
• As you can see, most of the metal work on the chassis is complete.
  
• Inside the Chassis left to right: 90-250VAC to 12VDC SMPS (position approximate) -- Brackets for the DAC OLED display, and the VFD -- Input switch board -- DAC. Closer to the opening is the VFD Spectrum Analyzer board (partially viable) and the Lexan piece (with protective blue tape). The Speaker binding posts are installed.
  
• On the right is the High Voltage SMPS with temporary heat sinks. The actual heatsink will be cut from the part shown at the right edge.
  
• Front center are the Bias supply inverter boards, and the bias fail protection relay.
  
• The 2 switches just to the left of the Bias inverters are for the Digital and Analog input select buttons.
  
• Blue LED (appears clear) = Power, and the Red LED = Bias Fail warning light. I was going to use an Amber LED for power, but the family insisted that power LED's must be blue these days.  Command override.  I have decided to go back to using Amber for the power light.
  
• Note that the OLED DAC control board on the very left will have to be disassembled, and wired separately. IR (remote) will not be supported. The volume encoder will be on the back panel.
  
• Note Little Rubber Feet "LRF Technology" is a very important factor. Amps are like puppy dogs. If they have big feet, you know they will grow up big and strong.
  
• The area around unused holes in the back panel will be cut-out, and covered with perforated metal for ventilation. The top will also be vented to allow convection cooling airflow around both SMPS's. (Perforated metal for both ventilation, and protection around the HV SMPS is not shown, and has not been ordered yet.) Note that as shown below, the HV SMPS will actually be on the left side.
  
• The Chassis top (extreme top of the picture) has been cut apart.  Only the sides are retained.  The top chassis is the aluminum part shown in the previous post.
  

 
I want to give that large power switch on the left a separate paragraph. That is a Soviet Military Surplus NOS toggle switch as was actually used in Soviet Aircraft such as the MIG-21, Transports, and Helicopters. The spot on the end of the toggle glows in the dark. It gives a very satisfying clunk when switched.  It is rated at 20A/27B (Note that the "B" is the cyrillic letter "V" = Volts). The eBay vendor claims that it is OK for use at up to at least 250VAC.

Project Update:
 
Note:  This entry is also posted to the Build Thread for this project on AudioKarma.org
 
This thread is very nearly 2 years old, but the project is not dead. In fact I've gotten back into it and made some progress.
 
The High Voltage Switch Mode Power Supply (Inverter) heatsink has been cut/drilled/installed, and the PS board and heatsink have been mounted to the chassis. The Heatsink is isolated from the Inverter Driver MOSFETs, and is bolted to the chassis for noise reduction and heat dissapation.
 
I think I have reached an engineering milestone -- I have made something so overly complicated that design changes can no longer add complexity, only simplifiy it -- or perhaps that's just wishfull thinking.
 
I have decided to eliminate the two 12VDC Negative boost converters that were going to generate a single (-) bias voltage for each channel.
 
In addition to the 380V winding, the HV PS has an 18V winding on the secondary of the transformer. Since it is a separate winding, it should be possible to use this winding to generate a negative DC supply in the same manner as is typically done in a classic grid biased or fixed bias amp.
 
Then I came across these on a "certain popular auction site":
 

 
These boards contain a bridge rectifier wired for a negative voltage, filter capacitors, and (4) voltage dividers to provide a bias for each of 4 tubes.
 
Note that because of the higher frequency, I should be able to use much smaller capacitors. I intend to leave the ones it comes with, and just bypass them with some smaller value non-electrolytic caps to help with the higher frequencies.
 
This will give me individual bias adjustment pots for each tube.
 
It is also much safer, as it contains only passive components. Failure of the HV PS will be fail-safe.
 
This means that I can also get rid of the "Bias Fail" relay arrangement. This will give me back much needed space inside the chassis, and greatly simplify wiring. It also eliminates the load of the 2 automotive relays
 
It also gives me more negative voltage if needed. The previous boost supplies maxed out (minimumed out?) at ~-15VDC when driven from a +12VDC source. I'd be in trouble if I needed more (absolute value) than that. Running from ~18VAC, the pictured board should be capable of well over (again absolute value) -20VDC if needed.
 
Ideally, the 4 bias pots should be accessible through holes or an opening in the chassis. Due to how far along I am, I don't think that will be possible in this design. Adjusting the bias will require opening the chassis.
 

Here's a Minor Update, Cross Posted from AudioKarma.org, with edits.
----
 
 
Engineering is always a compromise.
 
I don't like compromise.
 
...So I compromised.
 
Many articles debate the value for the Grid Bias Resistors for 7591 based amplifiers -- Particularly Fishers, but this applies to other grid biased 7591 amps, too.  Many of these articles suggest installing 220K, or even 200K, with the lower values helping to prevent bias run away. So I had previously installed 200K Grid resistors.
 
However, I felt that the Amp (only a single channel at this time), was not putting out the power that P-P 7591's should produce -- probably because the Phase Inverter was being loaded down by the 200K Grid Bias Resistors.
 
I didn't think that the change from 200K to 220K would make a lot of difference on the load on the PI; But going to 270K worried me, especially with my experience with (used) 7591 family tubes.
 
240K is also a value for Metal Film Resistors, and an assortment pack I bought from MPJA contained this value. <-- Yes, that's an unsolicited plug. 2W is overkill for a grid resistor, but 2W resistors in the values included in that assortment have lots of applications in Tube Amps.
 
So based on totally empirical "guess the number", I installed the 240K resistors.
 
The amp is definitely making more horsepower with theses resistors installed.  For all you Fisher owners, and other Owners of 7591 based amps that may be worried about run-away tubes, but don't want to hurt your sound, 240K may be the value for you.
 
I also changed the diodes on the Bias Board to High Speed Glass Diodes that are more appropriate for the High Speed Switching PS that is providing the 18VAC for the (-) bias.
 

As Promised, a Couple of Pictures:
 

 
 

 
 
Notes:

• No High Voltage Cage.  Home Depot was out of the perforated metal sheeting I was going to use.
  
• There is still protective masking tape on the top.
  
• Front Lens/Bezel not installed -- Again, this is to protect it from scratches, as I continue building.
  
• Only the 2 Tubes on the left are the new ones purchased for this project.
  
• Not Visible, but the little metal "L" brackets that allow the top to remain stable when the unit is opened are still installed.  There is still a non-trivial amount of work to be done inside.
  
• Little Rubber Feet (LRF) installed.  That's kinda like when an airplane under construction has "weight on wheels" for the first time.
  

 
 
UPDATE:
 

 
Added picture of back panel.
 

• Note USB/SPDIF Optical/SPDIF Coax inputs on the left.
  
• DAC Volume Knob installed.  Note that this is a digital encoder, there are no wipers, and no audio flows through this device.
  
• Only the 8 ohm terminals are currently connected internally.
  
• Hole plugs installed.  At this point, it is hard to tell that this chassis was re-used from another device.
  
• Hinge brackets to allow easy servicing are visible in this view.  These will be removed.
  

For now, I am considering this project essentially complete.
 
Here are the official "Glamor Shots" of this project:
 



 
It generally sounds great and has plenty of horsepower and bass. It also has the clear, crisp highs one would expect from a tube amp.
 
I still may do some more work on the Driver/PI circuit in the future if I learn more.
 
I also may still do some chassis modifications, including:
    

•  2 holes just under the Spectrum display for access to the clock/mode set buttons. Right now you basically can't set the clock, but I don't think the clock display is important.
  
•  I may install a toggle switch to the right of the display to allow the Spectrum analyzer to be turned off.
  
•  I need to install some reflective tape inside the HV enclosure to keep the Green LED light from leaking through the paint at the front.