Modifying S.E.X. to use 6EM7 tubes

Dr. Toobz · 21622

0 Members and 1 Guest are viewing this topic.

Offline Dr. Toobz

  • Sr. Member
  • ****
    • Posts: 432
on: November 28, 2009, 12:53:50 PM
I'm in the process of ordering a MQ iron upgrade and some other parts to allow for the use of 6EM7/6EA7 tubes in the S.E.X. as an experiment. From what I understand, the main thing is to replace R1 (620 ohms) with a 1600 ohm, 5W resistor.

Now, other than needing to change one of the caps to a higher voltage rating, couldn't I just parallel a SPST switch and 5-10W, 1k resistor with the existing 2W 620 ohm resistors already in place for each 6DN7? Using the parallel resistance formula, I would need 1012 ohms paralleled with 620 ohms to get a total resistance of 1600 ohms (the value listed in the S.E.X. manual for 6EM7's). Throwing the switch would place a parallel resistor in the circuit, thus allowing me to go back and forth from 6DN7 to 6EM7 tubes. Is this a good idea, even though the stock resistor is only rated for 2W? Or should I get rid of the 2W resistors and just replace them with the same value, rated for 5W? (Or, should I not even waste my time with the 6EM7's in the first place? :-)

Are any other changes needed to use 6EM7's? (I already have C4S boards in place of the plate resistors, which is supposed to make a big difference in terms of making the 6EM7 tubes sound like the 6DN7s). Once I get the iron upgrade and switch out the tubes, I will be sure to post very detailed observations between the two. It seems like nobody has really quantified the difference between these triodes, even on the old forum, other than to say that they only do well with CCS's instead of plate resistors. I would think that the claimed 3W output and lower plate impedance (880 ohms?) of the 6EM7's would make things a little better for speakers needing good woofer control?



Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5829
Reply #1 on: November 28, 2009, 03:56:06 PM
Except for a minor error in the calculation, you are correct. You should install a 1600 ohm 5 watt resistor for the 6EM7, and parallel it with a 1012 ohm 2 watt resistor for the 6DN7. Two resistors in parallel will always have a lower resistance than either one alone.

However the C4S current might also need to be adjusted for the 6EM7 if you are using the C4S. I can't find my notes on the C4S upgrade but I think it's running 2mA plate current. That would be around 430 ohms for the R1 resistor. I think PB (Caucasian Blackplate) has done various forms of this mod so maybe he will post. It appears to me that the plate voltage of the driver could be too high if you are running 2mA - check the voltage (I think it's pin 5?) which should not be over 250v. Post back if it is and we can work on suitable strategies.

A different variant would be to run more driver current if you are using the C4S. A good value might be 3.8mA (R1=237 ohms), the same as used by the Paramount driver. The stock SEX amp runs the driver at a low current so the plate load resistor is large, which linearizes the tube but limits its headroom. With the C4S the linearity is obtained at any current, so a greater headroom might be worth the effort. (This may be why the EM7 sounds better with a C4S.)

To clarify in case anyone is wondering, there was never a formal detailed design for the 6EM7 variant. I went far enough to be confident that such a design could be developed if there was a shortage of 6DN7s at some future date.
« Last Edit: November 28, 2009, 04:21:36 PM by Paul Joppa »

Paul Joppa


Offline Dr. Toobz

  • Sr. Member
  • ****
    • Posts: 432
Reply #2 on: November 28, 2009, 04:10:18 PM
Whoops - thought my calculation sounded funny. I guess that's why I'm a psychologist and not an engineer :-)

Thanks in advance for any info re: the C4S boards. I has wondered about if those needed to be messed with, also.

PB, care to chime in about this mod? Will I really be able to wring an extra watt or so from these tubes (and a better damping factor) without making the amp lose its "magic"?



Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5829
Reply #3 on: November 28, 2009, 04:22:10 PM
I edited my post above.

Paul Joppa


Offline Paul Birkeland

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 19732
Reply #4 on: November 29, 2009, 01:54:15 PM
Hello there,

In my experience modifying the SEX and close circuits for 6EM7's, I found that approximately 1ma of plate current (a 1k R1) on the first triode was best operationally.  With the stock cathode resistor, this will be 1ma with 1.2V of cathode voltage, so roughly 110 plate volts.  Because of the low current, the C4S transistors will not run that warmly, even though you are dropping a fair bit of voltage across them. 

You might also be interested to try an HLMP-6000 LED in place of the cathode resistor with the stock SEX C4S bias set resistors, giving you ~160v on the driver plate.  The stock SEX amp C4S with the stock cathode resistor and the 6EM7 will not leave sufficient compliance for the driver to operate properly (as PJ indicated). 

Paul "PB" Birkeland

Bottlehead Grunt & The Repro Man


Offline Dr. Toobz

  • Sr. Member
  • ****
    • Posts: 432
Reply #5 on: November 29, 2009, 03:53:40 PM
Thank you for the detailed information. I suspect I may try this mod out eventually, though I suspect that the MQ iron and chokes, along with better inter-stage and parafeed caps, may alleviate some of the speaker issues I've been having. If that turns out to be the case, little would probably be gained from going to the 6EM7 tubes (my current speakers are 96dB/1W/m, so 2W is enough power, but the output from the Specos rolls off too soon and exacerbates some potential issues already there from impedance dips in the bass region).

Any observations about the change in sound these tubes make? I seem to recall that Doc thought the smaller triode on the 6EM7 sounded less controlled than its counterpart on the 6DN7, and I also recall somebody saying the end sound was less "SET-like." Of course, I'm not sure if these observations were made with the inclusion of C4S boards on the plates as opposed to the stock resistors. I also have no idea of whether the MQ iron was used in any prior observations of the 6EM7 in the S.E.X. circuit. Those two things alone would account for a big proportion of changes in sound quality.



Offline Dr. Toobz

  • Sr. Member
  • ****
    • Posts: 432
Reply #6 on: February 19, 2010, 03:14:21 PM
Hello there,

In my experience modifying the SEX and close circuits for 6EM7's, I found that approximately 1ma of plate current (a 1k R1) on the first triode was best operationally.  With the stock cathode resistor, this will be 1ma with 1.2V of cathode voltage, so roughly 110 plate volts.  Because of the low current, the C4S transistors will not run that warmly, even though you are dropping a fair bit of voltage across them.  

You might also be interested to try an HLMP-6000 LED in place of the cathode resistor with the stock SEX C4S bias set resistors, giving you ~160v on the driver plate.  The stock SEX amp C4S with the stock cathode resistor and the 6EM7 will not leave sufficient compliance for the driver to operate properly (as PJ indicated).  

So, to use the 6EM7's, I could do the following: a) replace the 1.27k driver cathode resistors ("Rkdriver" on the schematic?) with little HLMP-6000 LED's, b) replace the 1000uF caps with something rated at a higher voltage (75V?), and c) replace the bias resistors ("R1" on the schematic) to 1600 ohm (5W). Am I completely correct?
« Last Edit: February 19, 2010, 03:23:39 PM by Dr. Toobz »



Offline Paul Birkeland

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 19732
Reply #7 on: February 20, 2010, 08:14:07 AM

So, to use the 6EM7's, I could do the following: a) replace the 1.27k driver cathode resistors ("Rkdriver" on the schematic?) with little HLMP-6000 LED's, b) replace the 1000uF caps with something rated at a higher voltage (75V?), and c) replace the bias resistors ("R1" on the schematic) to 1600 ohm (5W). Am I completely correct?

Yes on the HLMP3000, with the silver banded end towards ground.  Yes, the 1000uf caps should be 50v. 

R1 is a resistor on the C4S board (the smaller of the two physically), and that should be replaced with a 1k 1/4 watt resistor.

The 1.6k resistor replaces the 620 ohm resistor in the amplifier circuit (see the manual where it is mentioned).  So there are four parts to change. 

Paul "PB" Birkeland

Bottlehead Grunt & The Repro Man


Offline Dr. Toobz

  • Sr. Member
  • ****
    • Posts: 432
Reply #8 on: February 20, 2010, 08:45:44 AM
Thank you - this is very clear now and hopefully will serve as a reference for anybody else wanting to try this mod. If I do, I'll post my observations here! I'm not sure if I'm better off tightening up the bass with the 6EM7's (due to the lower plate impedance) or adding a teeny bit of global NFB to the circuit with the 6DN7's. I'm afraid the combination of the 6EM7's lower impedance and the addition of NFB would be overkill and would take away too much of the "SET sound," so I'm treating this as an either-or thing.



Offline Dr. Toobz

  • Sr. Member
  • ****
    • Posts: 432
Reply #9 on: June 28, 2010, 05:51:32 PM
Hello there,

In my experience modifying the SEX and close circuits for 6EM7's, I found that approximately 1ma of plate current (a 1k R1) on the first triode was best operationally.  With the stock cathode resistor, this will be 1ma with 1.2V of cathode voltage, so roughly 110 plate volts.  Because of the low current, the C4S transistors will not run that warmly, even though you are dropping a fair bit of voltage across them. 

You might also be interested to try an HLMP-6000 LED in place of the cathode resistor with the stock SEX C4S bias set resistors, giving you ~160v on the driver plate.  The stock SEX amp C4S with the stock cathode resistor and the 6EM7 will not leave sufficient compliance for the driver to operate properly (as PJ indicated). 

I ended up trying this out after all - with the stock C4S resistors and an LED in place of each cathode resistor. Is the stock resistor ideal for 160V plate voltage, or is there a more optimal value? I didn't get a chance to switch R1 to 1k and leave the cathodes as-is, but I'm wondering what would be best based on your experience and the ideal "place" on the tube curves for the little triode half.

Outcome is as expected - definitely more gain and less distortion at high volumes, and bass is nice and solid (and I'm running in no-NFB mode right now to get a baseline). Treble is more neutral than the 6DN7 - more "reference" or hi-fi, if you will, but mids seem about the same. Really, not much difference between these tubes when C4S'ed, other than the fact that my speakers seem to like the lower reflected output impedance better, and having an extra watt or two never hurts for headroom. Not all that thrilled about the hot 5W power supply dropping resistors, though - I hope things don't get too toasty under the hood.



Offline Paul Birkeland

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 19732
Reply #10 on: June 29, 2010, 10:31:06 AM
A 1k R1 and an HLMP-6000 under the cathode seems reasonable for the 6EM7 driver stage to try, but I might go down a bit on that resistor value a little bit, maybe to 749 ohms or something in that neighborhood. 

Paul "PB" Birkeland

Bottlehead Grunt & The Repro Man


Offline Dr. Toobz

  • Sr. Member
  • ****
    • Posts: 432
Reply #11 on: July 01, 2010, 06:54:37 PM
You know, the more I play around with this tube, the less I like it! You guys chose the 6DN7 wisely. The little bit of gain isn't worth it, the two sections have vastly different power requirements (and thus, a lot of power is burned off as heat in the dropping resistor), and the driver triode sounds a bit "pinched" and sterile, regardless of what bias I set. The 6DN7 is a much richer sounding tube to be sure.

No more playing with the S.E.X. amp for now - I've already started a re-build to remove all of the mods I've done and go back to stock (well, stock with MQ irons). The more I move away from the original design, the less I like the amp. The CFB is nice, but the irons can't handle DC on the secondaries. I'm not even sure I like the CCS loading, as I've found in my other projects. While my speakers seem to like a low Z-out, I've discovered that I can make up for that by increasing the distortion (2H) and making things sound "fatter." Hence, removing the 6EM7's and C4S boards. I think that's why I liked how warm the amp sounded when I first built it - speakers definitely sounded fuller, if boomier in the bass. Now the bass is just thinner. So I guess output impedance isn't the only factor here....






Offline Doc B.

  • Administrator
  • Hero Member
  • *****
    • Posts: 9648
    • Bottlehead
Reply #12 on: July 02, 2010, 07:22:38 AM
My impression of the 6EM7 was formed when we came out with the first S.E.X. amp back '94. Around the same time Ron Welborne came out with a DIY 6EM7 parafeed circuit, and I built it to see how out sounded. The impression I get whenever I have listened since is that the first triode just doesn't have the current draw to make a nice full bodied driver, rather like a 12AX7.

Dan "Doc B." Schmalle
President For Life
Bottlehead Corp.


Offline Dr. Toobz

  • Sr. Member
  • ****
    • Posts: 432
Reply #13 on: July 02, 2010, 09:50:22 AM
My impression of the 6EM7 was formed when we came out with the first S.E.X. amp back '94. Around the same time Ron Welborne came out with a DIY 6EM7 parafeed circuit, and I built it to see how out sounded. The impression I get whenever I have listened since is that the first triode just doesn't have the current draw to make a nice full bodied driver, rather like a 12AX7.

I suspect that may be why I've always thought 12AX7's sounded worse than 12AU7's, a tube I like for its warmer, fuller sound. I must not be imagining things after all! Come to think of it, high-mu tubes of any sort don't really seem to be the ones I hear cited for good sound quality....