• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Outside the Box- Part lll, Twin Coupled Amp

This project was originally published in the Volume 35, Issue 8 of AudioXpress magazine in August of 2004. It describes a high fidelity vacuum tube amplifier based on work done by Norman Crowhurst in the late 1950s. The circuit appears to be somewhat involved, but a closer look will reveal all. The primary object is to show what is possible with off the shelf parts & at reasonable cost.

A follow on article appeared later in the Volume 36, Issue 9 of AudioXpress magazine in September 2005.

In its present form the amplifier clips at 37 watts. The circuit is easily scaleable for more power.
 

Attachments

  • Twincoupled Amp D Nov6.jpg
    Twincoupled Amp D Nov6.jpg
    106.7 KB · Views: 270
  • Basic Circuit.doc
    89 KB · Views: 44
  • Power Supply TCW.jpg
    Power Supply TCW.jpg
    59.8 KB · Views: 254
  • Twin Coupled Amp Front.jpg
    Twin Coupled Amp Front.jpg
    324.3 KB · Views: 244
  • IM Test.jpg
    IM Test.jpg
    37.6 KB · Views: 241
  • TABLE ATWIN COUPLED AMP PERFORMANCE SUMMAR1.pdf
    6.7 KB · Views: 57
I rather like the Twin Coupled approach for making good use of the OT(s). All windings are engaged with the coupling caps in place. Effectively doubles the # of winding interleaves and uses all windings simultaneously. 4 leakage inductances paralleled, yielding 1/4 the Lleak. 2X the distr. capacitance however. If one of the OTs is flipped phase wise, then the total internal windings will be symmetrical also.

What do you think of this lower %CFB variant:
 

Attachments

  • Elliptical_Twin.bmp
    47.2 KB · Views: 74
Last edited:
I rather like the Twin Coupled approach for making good use of the OT(s). All windings are engaged with the coupling caps in place. Effectively doubles the # of winding interleaves and uses all windings simultaneously. 4 leakage inductances paralleled, yielding 1/4 the Lleak. 2X the distr. capacitance however. If one of the OTs is flipped phase wise, then the total internal windings will be symmetrical also.

What do you think of this lower %CFB variant:

As it turns out, one of the OPTs in my vers. is flipped in order to equalize the stray Cs. Otherwise the performance of the amp is badly compromised. Has anyone bread boarded the 'Ellipse'? Strange looking beast, might be interesting to try.:)

I did some things beyond what Crowhurst had published altho I'm sure he could teach all of us a lesson & much more. To begin, I tried to use all 'off the shelf' parts. Nothing boutique & otherwise exotic.
Except for function I did not select tubes at all.

Right away finding OPTs with the proper ratio is next to impossible. For those who have looked at Crowhurst's original you will find his OPT secondaries are paralleled, mine are in series. Crowhurst had purpose built OPTs. I did not. One result is a rather low DF, about the same as triodes without NFB.

The Hammond 125E OPTs are replacement type universals rated for 15W. At the time the price was right. Not so much now.

I included an optional differential front end so that if connected PP it would have good CMR & CMT. I will try to find the numbers for the curious.

Look carefully at the output section. It is running as pentodes with higher plate volts than screen. This results in more output power.

I had been curious for some time how tubes of 6BK7/6BQ7/6BZ7 family would compare as voltage amplifiers & drivers to those we normally use in audio. These are all available on the NOS market at very good prices. In order to make a comparison using this amplifier as a test set I wired up a pair of adapters so that tubes of this family could be plugged straight in. The octal plugs are made by Amphenol. Refer to the 6BQ7 Family Adaptors Photo.
One of the adapters is wired so that each element of the tube simply goes to it’s corresponding connection in the octal socket on the chassis. This adapter & tube substitutes for the 6SL7GT. Since these tubes have a somewhat higher mu than the 6SN7GTB I put a 2.2K resistor in series with each of the cathode leads on the other adapter. That reduced the gain of the other 6BQ7 in test to about the same as the 6SN7GTB that it replaced. It also provided a degree of linearization of the 6BQ7 characteristic in order to reduce distortion.

To prevent parasitic oscillation I also installed 1K, ½ watt resistors in all of the grid leads on the adapters. Refer to the test results which are better than I had anticipated.

For a comparative set of distortion results refer to Table A. From these the BQ7 family compares very well with the very expensive octals favored by the audiophools.


I like an HT switch. In order to prevent cathode stripping of the 6LU8s I put in slow ramp up of the HT in the PS.

The measurement results are by HP 334A & Picotech ADC-216 (16 bit)
And
Metrahit 29S Precision RMS DMM & Power Meter:)
The sources are HP 200CD &
GW GAG-810


I gave away my last analogue scope sometime ago. Now have four Picoscopes up to a two channel, 250 MHz, 12/8 bit vers.:)
 

Attachments

  • 6BQ7 Family Adaptors.jpg
    6BQ7 Family Adaptors.jpg
    59.9 KB · Views: 203
  • Fig 3 Alternative Front End.jpg
    Fig 3 Alternative Front End.jpg
    63.9 KB · Views: 101
Here is more test data & an IMD plot.

The correct voltage reading is by the MetraHit29S RMS DVM/Wattmeter.

The others are meters I had on the shelf as follows-
RadioShack DVM
RadioShack Analogues FET Meter
Not sure of the other.:)
 

Attachments

  • Beat Frequency Composite w Readings & IMD 100dpi 12W.jpg
    Beat Frequency Composite w Readings & IMD 100dpi 12W.jpg
    199.5 KB · Views: 59
"I prefere one good OPT."

The Twin Coupled makes 2 cheap OTs into a better one. With 1/4 the leakage L and 2X the distributed capacitance (combined) that should give a 40% improvement in bandwidth. [ Sqrt(4/2) ] And it can symmetrize the P-P sides. (but use something better than 125E OTs to start with!)

------

"The Hammond 125E OPTs are replacement type universals rated for 15W. At the time the price was right. Not so much now."

Check these out: (40 Hz to 18KHz)
EDCOR - GXPP Series Output Transformers

or for a bit more $: (20 Hz to 20 KHz)
EDCOR - CXPP Series Output Transformers

I have some double CXPP OTs here which were planned for a Twin Coupled or Elliptical Coupled experiment at one time. Haven't tried them yet. Keep getting distracted by something new. "Crazy Drive" is driving me crazy lately. Even have some of those 6LU8s here. But plenty of bigger TV Sweep tubes also. Can get a 6HJ5, 21LG6A, 6CB5 or 26DQ5 for around $4 or $3 at those Florida shops mentioned earlier (other thread), to make a 75 Watt to 100 Watt amplifier. Plenty of lab supplies too. I've never bought an "audio" tube.

I also have some long E laminations (Tempel Steel, M6 for constant V xfmrs) that would be ideal for making a twin coupled OT all on one core. (using a split bobbin design)

-----

While the 6BQ7s are pretty cheap at $3, you can get the frame grid 6JC6A on that $1 list. Even better, the frame grid 12HL7 is on the $1 list too (more suitable for a driver stage).

https://frank.pocnet.net/sheets/135/6/6JC6A.pdf
https://frank.pocnet.net/sheets/135/1/12HL7.pdf

http://www.esrcvacuumtubes.com/dollar_days.html

-
 
Last edited:
"I prefere one good OPT."

The Twin Coupled makes 2 cheap OTs into a better one. With 1/4 the leakage L and 2X the distributed capacitance (combined) that should give a 40% improvement in bandwidth. [ Sqrt(4/2) ] And it can symmetrize the P-P sides. (but use something better than 125E OTs to start with!)

------

"The Hammond 125E OPTs are replacement type universals rated for 15W. At the time the price was right. Not so much now."

Check these out: (40 Hz to 18KHz)
EDCOR - GXPP Series Output Transformers

or for a bit more $: (20 Hz to 20 KHz)
EDCOR - CXPP Series Output Transformers

I have some double CXPP OTs here which were planned for a Twin Coupled or Elliptical Coupled experiment at one time. Haven't tried them yet. Keep getting distracted by something new. "Crazy Drive" is driving me crazy lately. Even have some of those 6LU8s here. But plenty of bigger TV Sweep tubes also. Can get a 6HJ5, 21LG6A, 6CB5 or 26DQ5 for around $4 or $3 at those Florida shops mentioned earlier (other thread), to make a 75 Watt to 100 Watt amplifier. Plenty of lab supplies too. I've never bought an "audio" tube.

I also have some long E laminations (Tempel Steel, M6 for constant V xfmrs) that would be ideal for making a twin coupled OT all on one core. (using a split bobbin design)

I looked at Edcor sometime ago, prices are OK. The problem is the odd impedance ratio required, just as in McIntosh, Circlotron, Etc. Not many manufacturers of tube type transformers are interested. Specials used to be easy to get. But not now, all are driven hard by the bottom line. And the 90 day stock price report!

The Twin Coupled cct is essentially a 'Poor Mans McIntosh'. For me it was a very interesting project.

I have it all in a PDF more or less as published if anyone would like to have a copy. Just send a private email with an address.
AHsJFJhC1JVdTiwNFHjIzQRYghQooATggIBTTTRlyNVL0Z4COQLoGuCnQgAduFxBEWgoDaMibGzFWSPmU4csj+wM1PKj1okIIUjp6WEESB+BGGKhQHUrj5U6CaosGTVmjgxZfDhtogIpxgAJAgBgmZTpC5kNAxkcmaIq1B1PgSotWCgQzyIagHx4afCnFZqFdF5IKWCKgGENQxAxKdGi1yBCpWrMCCDCAC0TAdrsuGRDyAcYFIJ0QSDAAIIEiQBEYpWEh5JXTySZAQBmxYADD8KQaOQCTi8cXFZ5cGRBDYRUZd6AuiFnIIhOW1QQucCBFxJMLM7QFTgiioNZheYuAgwIADs=
 
"The problem is the odd impedance ratio required, just as in McIntosh, Circlotron, Etc."

That's why the Elliptical version is interesting. It only doubles (using 40% UL taps) the primary impedance. (versus 4X for the Twin Coupled) Edcor has some 1.6K, 1.7K, 2.2K, 2.4K, 2.5K primary OTs that could work. And the 0.4/(1+ 0.4) = 0.285 or 28.5 % CFB is easier to drive.
 
Last edited: