RFC: 300B driven by anode follower (6J5/6SN7) - diyAudio
Go Back   Home > Forums > Amplifiers > Tubes / Valves

Tubes / Valves All about our sweet vacuum tubes :) Threads about Musical Instrument Amps of all kinds should be in the Instruments & Amps forum

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 3rd April 2010, 02:35 AM   #1
tomchr is offline tomchr  United States
diyAudio Member
 
tomchr's Avatar
 
Join Date: Feb 2009
Location: Greater Seattle Area
Default RFC: 300B driven by anode follower (6J5/6SN7)

Folks,

I've been enjoying some amplifier design for the past couple of days. I started out with a bunch of Pinnacle 6J5 triodes and a JJ 300B. The OPT is an Edcor CXSE25-8-5k, which supposedly should have a 5 kOhm primary impedance. It actually measures 4.2 kOhm. I don't know if this is due to production tolerances or a mistake by Edcor. I don't really mind as this allows me to get a tad more power out of my 300B.

I have tried a couple of different topologies. Of course, the "standard" common cathode, cap coupled directly to the 300B. It sounded decent, but also sounded like it was out of breath at full volume. This was evident when watching the output on an o'scope as the output waveform would essentially be constant for a good 2 ms following a sharp transient. George (tubelab) describes this as "farting out" - he has an excellent write-up on the topic here. I was able to replicate this phenomenon in PSpice.

Adding a source follower, thus, arriving at the Tubelab SE circuit made a *HUGE* improvement. The amp behavior at/near clipping was greatly improved with no farting out visible on the scope (confirmed in PSpice sim also).

Even though I'm using LEDs and SS CSS'es (IXCP10M45) for bias I would prefer to avoid sand in the direct signal path. So I was looking for an alternative to the source follower. Joe Curcio pointed me to the anode follower. That works really, really well.

Attached is the schematic and prototype picture. I haven't decided if the source follower is needed or not. I keep going back and forth. The jury is still out. But it's pretty easy to swap the source follower in and out. Just let the 22 kOhm resistor float and short gate to source.

The gain of the first stage is about 20 V/V. The second stage (anode follower) is about 3 V/V. This allows me to drive the 300B to about 6.8 W at 3.6 % THD (just before clipping) with a 1 V(RMS), 1 kHz input. Adding the source follower lowers the THD at max power to 3 %.

The bias point for the 300B is 350 V on the plate, 60 mA bias current.

Do you have any suggestions for improvements to this circuit?

~Tom
Attached Images
File Type: jpg 300B_TC20100402.jpg (207.9 KB, 1497 views)
File Type: png 300B_SchematicDraft_TC20100402.PNG (28.3 KB, 1483 views)
  Reply With Quote
Old 3rd April 2010, 09:28 PM   #2
tomchr is offline tomchr  United States
diyAudio Member
 
tomchr's Avatar
 
Join Date: Feb 2009
Location: Greater Seattle Area
RFC = Request For Comments...

The verdict is in. The source follower stays. I'm using a 2SK3564 by the way. 2SK2700 or 2SK3566 would work also. Any MOSFET that can handle the voltage and has low input capacitance would be a good candidate here.

The source follower really cleans up the bass - especially at higher output power levels. The mids and highs remain largely unchanged.

By the way, the difference between the "normal way of doing it" - i.e. driving the 300B with a grounded cathode stage - and the "Tubelab SE" circuit is like night and day. I lost track of how many veils were lifted . It's awesome. The difference in sound quality between the "Tubelab SE" and my circuit above is minimal - at least to my ears. It's mostly a way for me to get half-decent gain out of the stack of 6J5's I have in my box o' tubes. I like to have full power out with 1-2 V RMS in.

I put "Tubelab SE" in quotes as I've made a few minor changes compared to the Tubelab SE. The biggest change is the LED bias. I might try resistor bias, but with the low dynamic impedance of LEDs I'm not sure resistor bias will do much to better the sound.

Also - using DC for the filaments on the 6J5's really cleans up things. Granted, I hadn't done anything to cancel the hum as I was planning to use DC all along. I just suddenly realized that I could supply the heaters from one of my lab supplies, gave it a whirl, and was quite surprised it made that much of a difference.

Now I just need to build a second channel. Or three more channels if I go for biamping.

~Tom

Last edited by tomchr; 3rd April 2010 at 09:31 PM.
  Reply With Quote
Old 3rd April 2010, 10:21 PM   #3
diyAudio Member
 
SpreadSpectrum's Avatar
 
Join Date: Nov 2007
I usually bias my fets with 1M. Might lower distortion a tiny bit?

Just out of curiosity, how much distortion do you get at 1W?
  Reply With Quote
Old 4th April 2010, 01:21 AM   #4
tomchr is offline tomchr  United States
diyAudio Member
 
tomchr's Avatar
 
Join Date: Feb 2009
Location: Greater Seattle Area
Quote:
Originally Posted by SpreadSpectrum View Post
I usually bias my fets with 1M. Might lower distortion a tiny bit?
True. It would result in lower loading of the 2nd stage. That might explain the difference I'm seeing between the anode follower and anode follower + source follower.

Quote:
Originally Posted by SpreadSpectrum View Post
Just out of curiosity, how much distortion do you get at 1W?
'bout a percent. See below.

It turns out the major difference between the two topologies is in the transient response near full power. The two scope shots show the output voltage into an 8 Ohm load. The input signal is a 10 kHz square wave. Note that the -3 dB bandwidth of the amp is only just a tad more than 20 kHz so the 10 kHz signal is fairly bandwidth limited. I'll need to extend the BW somehow. But note how the anode follower clips softly on the tops of the waveform. This is because the anode follower runs out of gas when it tries to drive the 300B into A2. The source follower, on the other hand, drives a fairly nice square wave until the FET runs out of headroom (source voltage clips at -150 V).

I'll need to tweak the circuit. I think I can improve the THD of the first and second stage if I tweak the operating point a bit.

~Tom
Attached Images
File Type: jpg 10kHz_AnodeFollowerOnly_TC20100403.jpg (62.8 KB, 1363 views)
File Type: jpg 10kHz_SourceFollower_TC20100403.JPG (82.0 KB, 1148 views)
File Type: png THD_vs_Freq.png (30.1 KB, 1079 views)

Last edited by tomchr; 4th April 2010 at 01:25 AM. Reason: Fixed attachment
  Reply With Quote
Old 4th April 2010, 02:01 AM   #5
diyAudio Member
 
Join Date: Apr 2003
Location: Southern Tier NY
Tom,
Have you tried a CCS as the source load. Before I got my test equipment, I played around between resistive and ccs as the source load in my kt88PP. I liked the sound of the ccs better so it stayed. The decision was purely qualitative though. I would be curious to see if you can measure a difference?
__________________
Living Life Doing the Waltz in 4/4 meter.
  Reply With Quote
Old 4th April 2010, 08:58 PM   #6
tomchr is offline tomchr  United States
diyAudio Member
 
tomchr's Avatar
 
Join Date: Feb 2009
Location: Greater Seattle Area
Quote:
Originally Posted by SGregory View Post
Have you tried a CCS as the source load. Before I got my test equipment, I played around between resistive and ccs as the source load in my kt88PP. I liked the sound of the ccs better so it stayed. The decision was purely qualitative though.
I played with it in PSpice. I forget the results, honestly, but I think the CCS showed marginally better THD than the resistive load. I should give that a whirl in the real world and see how it performs.

I played around with the first two stages of the amp yesterday. The THD of the two stages is minimized when their plate voltages are equal. Makes sense... You don't want any DC current flowing in the feedback network. I'll try using three LEDs in the bias for the first stage to see if I can balance things out better.

I also found one cause of the low bandwidth of the amp. The input cap of the two 6J5's in parallel along with the feedback resistance of 330 kOhm formed a pole at around 40~50 kHz. I changed the feedback network to 50k, 165k by adding 100k, 330k in parallel with the network already there. That increased the overall bandwidth of the two driver stages to almost 70 kHz.

It's yet to be determined how the changes in THD and bandwidth of the driver stages impact the overall amp once the 300B is inserted in its socket. Stay tuned.

~Tom
  Reply With Quote
Old 12th May 2011, 06:12 PM   #7
JPS64 is offline JPS64  Germany
diyAudio Member
 
Join Date: May 2011
Hello Tom,

what is the current supplied from the - supplies?
  Reply With Quote
Old 12th May 2011, 06:35 PM   #8
tomchr is offline tomchr  United States
diyAudio Member
 
tomchr's Avatar
 
Join Date: Feb 2009
Location: Greater Seattle Area
Quote:
Originally Posted by JPS64 View Post
what is the current supplied from the - supplies?
The -75 gate bias draws no DC current. It can be formed by a voltage divider from the -150 V supply.

The -150 V supply supplies about 5 mA DC but should be capable of handling at least 10 mA as it provides higher current as the grid of the 300B is driven towards 0 V.

Ya know... I should probably update this thread. I have a better schematic which includes power supply and such. In addition, I went ahead and had PCBs made professionally. I have a good handful of spares if you're interested.

~Tom
  Reply With Quote
Old 12th May 2011, 06:46 PM   #9
JPS64 is offline JPS64  Germany
diyAudio Member
 
Join Date: May 2011
I am thinking about LT1533 with DC/DC converter PICO 34163 to generate 0-300VDC (or 150VDC)@7.5W with expected 100µVpp ripple.
  Reply With Quote
Old 12th May 2011, 06:51 PM   #10
JPS64 is offline JPS64  Germany
diyAudio Member
 
Join Date: May 2011
It will be very nice if you update this thread. You know, my design is not ready. In the same time, nice to update also the thread with the national switchers: today other types are existing (with higher output voltage).

And thanks for the offer, but its not cheap to buy parts in the US.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Anode Follower yagas Tubes / Valves 19 20th November 2010 04:35 PM
Noise - 6J5 - 71a -300B amp hal3101 Tubes / Valves 14 3rd November 2008 04:42 PM
Two 6J5 versus one 6SN7 audiobliss Tubes / Valves 11 4th August 2008 10:25 PM
Question about direct coupling a anode follower into a cathode follower. G Tubes / Valves 45 29th July 2004 06:47 PM
A 76-6SN7 driven 300B arnoldc Tubes / Valves 5 18th April 2003 07:20 AM


New To Site? Need Help?

All times are GMT. The time now is 01:01 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2