exeric,
Not Crazy but maybe a bit sadomasochistic...(we all have these little problems).
Problems may arise from the fact that 6V6 require a lot more drive than EL84, without some special attention, the EL84 will dominate.
The forced balance of the BH scheme means it has VERY low hum levels. A switch for the feedback would have full B+ on it, use a big switch with plenty of voltage rating. OR you could use a combination of a fixed 10K resistor + a 10K pot for the feedback resistor. Again use a beefy pot for the voltage withstand.
Duplicate the buffers and bias pots on each side so that separate bias can be applied to the EL84s and the 6V6.
Some guys are building 7591 versions which might give you the sound you want from just a single pair of tubes. The 7591 will give you a fair bit more power than the EL84 or 6V6.
If you want to scale it up significantly then my experience is that the closest sounding "big" tube to the EL84 is NOT the EL34 as you might expect but rather the KT88.
I haven't tried using different tubes on an amp like this but did build a full powerscaled guitar amp with a quad of 6V6 (Kevin O'Connors London Power "Standard"). Each pair switchable between fixed and cathode bias and between Triode and Pentode Mode. The guy I built that for often runs one pair in cathode biased triode mode and the other pair in fixed bias pentode mode. He likes that blend of tone.
Cheers,
Ian
Not Crazy but maybe a bit sadomasochistic...(we all have these little problems).
Problems may arise from the fact that 6V6 require a lot more drive than EL84, without some special attention, the EL84 will dominate.
The forced balance of the BH scheme means it has VERY low hum levels. A switch for the feedback would have full B+ on it, use a big switch with plenty of voltage rating. OR you could use a combination of a fixed 10K resistor + a 10K pot for the feedback resistor. Again use a beefy pot for the voltage withstand.
Duplicate the buffers and bias pots on each side so that separate bias can be applied to the EL84s and the 6V6.
Some guys are building 7591 versions which might give you the sound you want from just a single pair of tubes. The 7591 will give you a fair bit more power than the EL84 or 6V6.
If you want to scale it up significantly then my experience is that the closest sounding "big" tube to the EL84 is NOT the EL34 as you might expect but rather the KT88.
I haven't tried using different tubes on an amp like this but did build a full powerscaled guitar amp with a quad of 6V6 (Kevin O'Connors London Power "Standard"). Each pair switchable between fixed and cathode bias and between Triode and Pentode Mode. The guy I built that for often runs one pair in cathode biased triode mode and the other pair in fixed bias pentode mode. He likes that blend of tone.
Cheers,
Ian
Last edited:
Well, I prefer to think of it as a bit OCD. But I guess it would seem a bit SM from another perspective. As far as implementing the feedback. I definitely want to stay away from interrupting B+ while moving between resistances. So I would probably use a switch just to bypass individual resistances in series. That way there would always be continuity of resistance in between the short duration of moving the switch. I still haven't figured a way of doing the same thing with the G2 wire since it moves from the output transformer to B+. Maybe there's a way but I haven't hit on it. My old amp also had loud clicking noises while switching from pentode to triode.
There's only about a 6 volt difference in the drive requirements between the el84 and 6v6. The old amp had 4 el84s and 2 6v6s per channel so if anything there will be less of the el84 tone in this idealization. It may be that I'm obsessing on the "one that got away" but I'd really like to return to the sound of mixed 6v6 and el84 even if it seems crazy. We always want the one that just isn't good for us, if you know what I mean. I'll be trading that 6 volt region where the EL84 is being overdriven with the 6v6 coming into its own so there shouldn't an abrupt change of tone. In fact it should be better than the old one because of the source followers driving the el84. The old one just had cathode bias on both tube types and it was barely enough power for my inefficient speakers. I think without cathode biasing and your beautiful implementation of source followers driving it instead that maybe I can get away with 2 tubes instead of three. Here's hoping.
Cheers,
Eric
There's only about a 6 volt difference in the drive requirements between the el84 and 6v6. The old amp had 4 el84s and 2 6v6s per channel so if anything there will be less of the el84 tone in this idealization. It may be that I'm obsessing on the "one that got away" but I'd really like to return to the sound of mixed 6v6 and el84 even if it seems crazy. We always want the one that just isn't good for us, if you know what I mean. I'll be trading that 6 volt region where the EL84 is being overdriven with the 6v6 coming into its own so there shouldn't an abrupt change of tone. In fact it should be better than the old one because of the source followers driving the el84. The old one just had cathode bias on both tube types and it was barely enough power for my inefficient speakers. I think without cathode biasing and your beautiful implementation of source followers driving it instead that maybe I can get away with 2 tubes instead of three. Here's hoping.
Cheers,
Eric
I guess people are being quiet about it if they have?! If it was me thinking about doing that there would be one obvious choice and several not so obvious choices. The obvious choice would be to use Gingertube's 6v6 implementation that uses one source follower buffer for each channel. I think this is an obvious decision because having 4 el84 per channel shows a quest for more power which fixed bias would allow, not to mention a small ability to go to AB2.
The not so obvious decision would be whether to duplicate the source followers so each tube has its own. Transistors are cheap and the way Gingertube has designed that circuit the source followers are not big power guzzlers. Having individual source followers for each tube allows one the option of having different bias levels for each tube pair. That allows all kinds of options to blend different sounds from the 4 el84s. This is a totally different thing from on-the-fly selection of those different sounds that I suggested, which even I have to admit is a little out there.
I really am a firm believer that there is no "correct" sound in hifi. One should try to avoid "effect" type sounds in hifi. But I think in general that the hifi community is too conservative in differentiating what constitutes a sound "effect" and simply mixing tube topologies in a single amp to maximize the sound to your liking. My two cents, for what it's worth.
Yep, a koppel of page's back you can see the results. I also changes the input of the amp bud if you stick to the original input and make the el84 like i did it will work fine.
Here is the link to the page http://www.diyaudio.com/forums/tubes-valves/72536-el84-amp-baby-huey-132.html#post3305677
Last edited:
I was thinking of PPP EL84s so 4k3 UL OPTs from a Dyna ST70 could be used. That's not a good match for either 7591 or 6V6.
I guess I missed that one. Wow! And.... Thanks! A couple of questions, though...
I see you changed the LTP to 6922 (6DJ8, ECC88). It also looks like a DRV134 balanced output opamp (U11) was added to act as a preamp. How much gain comes from that opamp stage? About 6dB (X2)? That's a pretty slick trick! I think the one and only downside is its 10k input impedance. But that's really not bad.
Nice ideas there.
--
PS - What is the penalty for using a 6922? Is the only problem lower gain?
I guess I missed that one. Wow! And.... Thanks! A couple of questions, though...
I see you changed the LTP to 6922 (6DJ8, ECC88). It also looks like a DRV134 balanced output opamp (U11) was added to act as a preamp. How much gain comes from that opamp stage? About 6dB (X2)? That's a pretty slick trick! I think the one and only downside is its 10k input impedance. But that's really not bad.
Nice ideas there.
--
PS - What is the penalty for using a 6922? Is the only problem lower gain?
Last edited:
It hadn't occurred to me the reason for PPP EL84s was to match an existing OPT. I shouldn't have assumed it was for more power.
Guesto, Very nice! I recognize Menno's signature in the look. Well done.
The BH feedback scheme trades output tube gm for reduced rp.
Parallel EL84 makes a "composite" tube with:
2 x gm
1/2 rp
same mu.
So that doubling of the gm should make it work well.
You will also note that the feedback is reduced due to divider action from the diffamp triodes rp (the impedance seen looking "down" into the anodes of teh diff amp triodes). That is why I specified 12AX7 (high rp so as to not divide down the feedback too much). That 6DJ8 diffamp of Geosto's worries me a bit as 6DJ8 have VERY low rp, he will not be getting the full benefit from the BH fedback scheme. My next iteration in fact is going to have 2 x 6BR7 (EF86'ish) pentode diffamp for this very reason.
I also have a pair of Hammond 1650N (Raa = 4K3, 60W) on the shelf to try parallel EL84s or parallel 6V6.
I also have a pair of PLITRON VDV-2100-CFB/H to try a combination of cathode feedback, ultralinear and the BH shunt feedback with a quad of KT88.
These projects are on hold while I clear the decks of other projects. For that I will also "steal" the bias servo from Rob, see here: http://rmsacoustics.nl/rmsacoustics/tubeamp/tuba_intro.html
I recently celebrated my 60'th B'day by going 3 days per week in the EE day job so should shortly have more time for this stuff.
Cheers,
Ian
Parallel EL84 makes a "composite" tube with:
2 x gm
1/2 rp
same mu.
So that doubling of the gm should make it work well.
You will also note that the feedback is reduced due to divider action from the diffamp triodes rp (the impedance seen looking "down" into the anodes of teh diff amp triodes). That is why I specified 12AX7 (high rp so as to not divide down the feedback too much). That 6DJ8 diffamp of Geosto's worries me a bit as 6DJ8 have VERY low rp, he will not be getting the full benefit from the BH fedback scheme. My next iteration in fact is going to have 2 x 6BR7 (EF86'ish) pentode diffamp for this very reason.
I also have a pair of Hammond 1650N (Raa = 4K3, 60W) on the shelf to try parallel EL84s or parallel 6V6.
I also have a pair of PLITRON VDV-2100-CFB/H to try a combination of cathode feedback, ultralinear and the BH shunt feedback with a quad of KT88.
These projects are on hold while I clear the decks of other projects. For that I will also "steal" the bias servo from Rob, see here: http://rmsacoustics.nl/rmsacoustics/tubeamp/tuba_intro.html
I recently celebrated my 60'th B'day by going 3 days per week in the EE day job so should shortly have more time for this stuff.
Cheers,
Ian
Last edited:
I've been finkin...
First, happy new year to all ya mates...
I finally collected all parts for my future BH (9-pin, MOSFET follower version). I bought 8 pieces of 10GK6, matched, so that I have spare tubes when first quad one day says "goodbye and thanks for the fish".
Now, I've been finkin, is there anything bright in idea to build BH with 8 output tubes (with duplicated MOSFET followers and their CCS, coupling caps, bias adjustment trimmers), and to bias 'em around 18mA each. Them tubes should live much, much longer, I assume.
My OPTs are 8k/4-8-16R as per BH specs, and power transformer will be able to put extra amperes for added filaments.
What do you think about that idea in general? Any drawbacks or advantages that I'm not aware of?
I'll ask about eventual changes in shunt feedback network resistor values if there's any point going this way.
Thanks for reading
Regards,
Zoran
First, happy new year to all ya mates...
I finally collected all parts for my future BH (9-pin, MOSFET follower version). I bought 8 pieces of 10GK6, matched, so that I have spare tubes when first quad one day says "goodbye and thanks for the fish".
Now, I've been finkin, is there anything bright in idea to build BH with 8 output tubes (with duplicated MOSFET followers and their CCS, coupling caps, bias adjustment trimmers), and to bias 'em around 18mA each. Them tubes should live much, much longer, I assume.
My OPTs are 8k/4-8-16R as per BH specs, and power transformer will be able to put extra amperes for added filaments.
What do you think about that idea in general? Any drawbacks or advantages that I'm not aware of?
I'll ask about eventual changes in shunt feedback network resistor values if there's any point going this way.
Thanks for reading
Regards,
Zoran
Zoran,
Happy new year too you as well.
Parallel output tubes should work well - why?
The BH feedback scheme trades some output tube gm for reduced rp.
When you parallel tubes the individual tube gm add.
I have not checked what Raa is required for 10GK6 but it divides according to how many pairs you use. If a pair needs 8K then 2 pairs require 4K and 4 pairs would require 2K.
The shunt feedback resistor network should not require any changes at all.
Cheers,
Ian
Happy new year too you as well.
Parallel output tubes should work well - why?
The BH feedback scheme trades some output tube gm for reduced rp.
When you parallel tubes the individual tube gm add.
I have not checked what Raa is required for 10GK6 but it divides according to how many pairs you use. If a pair needs 8K then 2 pairs require 4K and 4 pairs would require 2K.
The shunt feedback resistor network should not require any changes at all.
Cheers,
Ian
Thanks for your answer Ian, however I think I didn't put my question clear enough (english is not my...). My "finkin" was to use existing 8k/4-8-16R OPT already made for your standard BH, but with doubled output tubes, and recommended 36mA bias shared between those tubes (18mA each). Like one would do with output transistors in SS world to share dissipation between multiple output pairs.
But, in the meantime I made a simulation in Multisim (it has 6v6 and 6sl7 models) and it looks like there's a fundamental flaw in my idea: distortion is 2x higher when compared to standard BH. So, it is like "abandon SS ideas all you enter here" (tube inferno).
BTW 10GK6 (xGK6) are identical tubes to 6bq5: specs and curves in datasheets are identical, except xGK6 having slightly increased plate dissipation and voltage. And pinout is different. And can be found for 3USD each...
Nbriles2000: where there's a will there's a way!
Regards,
Zoran
But, in the meantime I made a simulation in Multisim (it has 6v6 and 6sl7 models) and it looks like there's a fundamental flaw in my idea: distortion is 2x higher when compared to standard BH. So, it is like "abandon SS ideas all you enter here" (tube inferno).
BTW 10GK6 (xGK6) are identical tubes to 6bq5: specs and curves in datasheets are identical, except xGK6 having slightly increased plate dissipation and voltage. And pinout is different. And can be found for 3USD each...
Nbriles2000: where there's a will there's a way!
Regards,
Zoran
Hi Everyone,
I'd like to announce a new arrival.
I added three custom touches.
The first was put the whole thing in a wooden cave so that it would be easy to sit upside down while soldering everything together, and so that my top loading CD would have somewhere to sit.
The second is the micro-ammeter on the front. The momentary DPDT switch selects either the left or right channel, through a 3k3 resistor across the two cathodes in the first stage, to trim the bias from the CCS. I figured if the process involved me fiddling about with a multimeter, I'd never do it again. Besides, who doesn't love the look of an old meter.
The two knobs to the right of the meter are the two 1k pots, the next is the 50k log stereo volume pot and the last one is a 3 position rotary switch for input selector.
All the transformers are Hammond: 1650E outputs, a 370KX supply and a 193L choke. The steel box is Hammond too, 17x12x3.
My power supply is 250+250V, 5R into 30uF film, 5H choke into 440uF electrolytic, then splits into 10R into 3x30uF film for each channel. It's a little shrine to Mundorf film caps. With no load it sits on 355V, with everything connected and running it settled to 320V.
The third mod I did was that when the switch is powered off, the HV bleeds back through 36k and then through the primary of the supply transformer (which is no longer connected to mains). It drains down faster than with just 220k that I also left on the same board as the caps. If that is a BAD idea, would someone scream at me.
Things are running quite warm, but still touchable. I'm not sure my bias currents are right, but it seems to sound sweet enough so I'm assuming that it's all working, until the smoke gets out.
Thanks Gingertube for a great newbie project
I'd like to announce a new arrival.
I added three custom touches.
The first was put the whole thing in a wooden cave so that it would be easy to sit upside down while soldering everything together, and so that my top loading CD would have somewhere to sit.
The second is the micro-ammeter on the front. The momentary DPDT switch selects either the left or right channel, through a 3k3 resistor across the two cathodes in the first stage, to trim the bias from the CCS. I figured if the process involved me fiddling about with a multimeter, I'd never do it again. Besides, who doesn't love the look of an old meter.
The two knobs to the right of the meter are the two 1k pots, the next is the 50k log stereo volume pot and the last one is a 3 position rotary switch for input selector.
All the transformers are Hammond: 1650E outputs, a 370KX supply and a 193L choke. The steel box is Hammond too, 17x12x3.
My power supply is 250+250V, 5R into 30uF film, 5H choke into 440uF electrolytic, then splits into 10R into 3x30uF film for each channel. It's a little shrine to Mundorf film caps. With no load it sits on 355V, with everything connected and running it settled to 320V.
The third mod I did was that when the switch is powered off, the HV bleeds back through 36k and then through the primary of the supply transformer (which is no longer connected to mains). It drains down faster than with just 220k that I also left on the same board as the caps. If that is a BAD idea, would someone scream at me.
Things are running quite warm, but still touchable. I'm not sure my bias currents are right, but it seems to sound sweet enough so I'm assuming that it's all working, until the smoke gets out.
Thanks Gingertube for a great newbie project
