Vacuum Tube OTL power amp!! - Page 81 - 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 31st July 2012, 10:05 PM   #801
diyAudio Member
 
atmasphere's Avatar
 
Join Date: May 2007
Oh- I forgot to mention- don't use a stainless nut- if you get it on you will have to break the bolt to get it off.
  Reply With Quote
Old 1st August 2012, 11:55 AM   #802
ilimzn is offline ilimzn  Croatia
diyAudio Member
 
Join Date: Feb 2005
Location: Zagreb
Re. Pentode OTL

Quote:
Originally Posted by atmasphere View Post
If you run a lot of feedback, yes, but I think you will find something quite curious. If you add a lot of feedback, it will not increase the power available into lower impedances. This is why I believe that increasing negative feedback does not actually decrease the output impedance.

IOW the term 'output impedance' as it is used in amplifiers today does not refer to the actual output impedance of the amplifier- it refers to the servo gain combined with the actual output impedance. Its confusing! But this is why I say that triode operation is more advantageous, because if you run the amplifier open-loop (no feedback) then you get the truth of what the actual output impedance is.
Actually, this would be confusing OLG output impedance with apparent CLG output impedance. While closed loop impedance does indeed fall, it does nothing about the intrinsic voltage drop in the device itself, as this is a physical property which no amount of gain trickery can change.

That being said, pentodes vs triodes in an OTL in fact illustrate that this is not majorly relevant to output impedance but rather maximum output power.
A pentode OTL can indeed be more powerful than the same OTL using the same pentodes connected as triodes, simply because the minimum voltage drop 'at the knee' of the pentode characteristic can be made FAR lower than in a triode connection - and the reson is that the voltage on G2 is independent from the voltage on the anode, and can be made higher - quite significantly so. In fact in this respect a PL504 is a rather drastic example - as a triode it's not at all well suited for an OTL WRT efficiency. When it comes to triodes, the number of suitable tubes that can still be bought at reasonable prices, drops to only3 - 6S33S, 6S19P (but you need lots), 6AS7G and soviet equivalents.

The ability to independently control Vg2 also makes it possible for a pentode to have lower output impedance than the same pentode connected as a triode. As with most pentode characteristic devices, it's output impedance appeoaches 1/gm, and since gm is highly (in theory, for a perfect pentode, ONLY) dependent on Vg2, we can play with that a lot.
Normally, in a triode OTL the bane of it's existance is the voltage necessary across A-K to get enough current flowing, due to Rp of the tiode (usually at Vgk=0, although A2 and AB2 drive is possible and improves things somewhat). THis leads to going way over the limits with instantaneous dissipation even though we might still be within limits for peak cathode current, and hence requires more parallel tubes to keep under control.
In a pentode OTL, assuming the load line keeps the operating point below or at most at the knee, you can get a lot of current with a low voltage drop by increasing Vg2, and with it you get increased gm. At some point you can get more than you can achieve with the same tube strapped as a triode simply because you may not run into the dissipation limit.

However, some architectures are more suitable for such an OTL than others.
Totem-pole outputs must have floating G2 supplies in order to be truly symmetrical, otherwise the top half Ig2 flows through the load, whereas the bottom half Ig2 flows to the negative supply (or ground if the whole thing is AC coupled to load). Although such designs mostly use beam power tubes rather than straight pentodes, so Ig2 is a relatively small fraction of the anode current, with many parallel tubes the difference can add up. If you go to A2 or AB2 operation, then it's a given. And, it's all worse if the load is such that the operating point gets into the knee of the pentode characteristic.
On the oposite end of the spectrum, the circlotron is the best possible architecture of a pentode OTL. Because it's actually intrinsically symmetrical Ig2 can be made to either never go through the load, or (more complex but desirable) always go through the load. If A2/AB2 operation is sued, Ig1 always goes through the load. However, with pentodes, the number of floating supplies doubles, although they stilla ctually come in floating pairs so do not need to be completely independent.

To summarise, an OTL using pentodes has the following advantages:
1) Tubes can produce high peak currents at low Vak so:
- less heat is produced
- more power can be had from the same number of tubes reliably
- higher bias (class A) is possible with less tubes as lower Vak keeps dissipation inside limits even at higher bias
- triode A2/AB2 performance can often be achieved without going positive on Vg1, i.e. staying within class A1/AB1. This results in a much simpler driver.
2) Tubes can be balanced by independently adjusting Vg2 for each tube, best done with independant Vg2 regulator (some silicon here...). This optimizes output power and distortion while still keeping the driver simple.
3) In a circlotron topology Ig2 can be recycled through the load for extra available current, increasing efficiency and output power with the same number of tubes.
4) Output tubes are operated as pentodes, so there is a small advantage regarding input capacitance of a tube. This can come in handy when driving lots of parallel tubes with a simple driver - as is the norm in an OTL.
5) With proper design, the amplifier can withstand a short circuited output indefinitely. This can be used as load protection which does not require the signal to be passed through relay contacts.

But - there are also dissadvantages, and they are all really about managing G2 dissipation:
1) Pentode OTL cannot function without a load, although the biassing requires at least a small default load to determine the center point of the output (i.e. ground). Operation would only be completely reliable with a fairly high amount of NFB. More on that below.
2) NFB may be required to get very low output impedance. Note I wrote MAY - this depends on ehat one wants to optimize the design for - it it's lowers possible loss, then fewer output tubes are required but this also results in higher output impedance.
3) Ig2 apsolutely requires special management. As with any pentode, if the load is removed from the plate, the current wants to flow through G2. This also happens if the load is lighter than a certain minimum, the result is that the operating point of the pentode is shifted below the knee, where current through plate falls, but current through G2 rises very abruptly. This is particularly important with tubes that have a top cap - if it is removed or loose, contact is lost, Vak becomes zero and a huge current will attempt to flow through G2. As an example, for a PL504, at a given Ug2, doing this will pass about half of the maximum plate peak current through G2, and this will be a quick and fiery death for the tube and possible the rest of the amplifier.
The solution is current limiting Ig2. If each G2 has it's own regulator (silicon) this is a fairly small addition which can make the amplifier unconditionally safe. There are a number of techniques which can be used to keep maximum G2 dissipation almost constant and within limits, and thus maximize the contribution of each tube in the amp. It should be mentioned, however, that doing this purely with tubes is completely impractical.
It should also be noted that having NFB does not guarantee the safety of the tubes if the load is removed from a particular tube (i.e. the disconnected top cap scenario), and may not (depending on the amount of NFB) guarantee the safety of the tubes avan if the load was removed from the output. For that, one needs a more comprehensive protection scheme, but that is really a good idea anyway.
  Reply With Quote
Old 1st August 2012, 06:06 PM   #803
diyAudio Member
 
atmasphere's Avatar
 
Join Date: May 2007
Quote:
Originally Posted by ilimzn View Post
Re. Pentode OTL

To summarise, an OTL using pentodes has the following advantages:
1) Tubes can produce high peak currents at low Vak so:
- less heat is produced
- more power can be had from the same number of tubes reliably
- higher bias (class A) is possible with less tubes as lower Vak keeps dissipation inside limits even at higher bias
- triode A2/AB2 performance can often be achieved without going positive on Vg1, i.e. staying within class A1/AB1. This results in a much simpler driver.
2) Tubes can be balanced by independently adjusting Vg2 for each tube, best done with independant Vg2 regulator (some silicon here...). This optimizes output power and distortion while still keeping the driver simple.
3) In a circlotron topology Ig2 can be recycled through the load for extra available current, increasing efficiency and output power with the same number of tubes.
4) Output tubes are operated as pentodes, so there is a small advantage regarding input capacitance of a tube. This can come in handy when driving lots of parallel tubes with a simple driver - as is the norm in an OTL.
5) With proper design, the amplifier can withstand a short circuited output indefinitely. This can be used as load protection which does not require the signal to be passed through relay contacts.
WRT point one above, in practice the 6AS7G has excellent high current capabilities as well. Simplicity of the driver demands some conversation; I use a driver that has one gain stage and then a simple cathode follower, which is able to drive multiple triode grids without distortion. So far I have yet to see a simpler amplifier based on pentodes of any sort.

2) if a 6C33 or 7241 is used, then independent bias settings are a good idea, done by also having independent driver circuits. At this point, if these tubes are used, then the driver is considerably more complex.

3) and 4) are no different if 6AS7s are used, and

5) we get the same short circuit tolerance with the 6AS7s. I have often done a short-circuit as a demonstration at shows by shorting out the amp and then running the preamp at full volume. In general OTLs seem to be tolerant of short circuits as long as there are no stability issues.
  Reply With Quote
Old 1st August 2012, 10:51 PM   #804
ilimzn is offline ilimzn  Croatia
diyAudio Member
 
Join Date: Feb 2005
Location: Zagreb
Quote:
Originally Posted by atmasphere View Post
WRT point one above, in practice the 6AS7G has excellent high current capabilities as well. Simplicity of the driver demands some conversation; I use a driver that has one gain stage and then a simple cathode follower, which is able to drive multiple triode grids without distortion. So far I have yet to see a simpler amplifier based on pentodes of any sort.
Yes, and that is exactly what I meant by a simple driver. Using pentodes can, in some cases, make it even simpler if a single differential stage offers enough gain, or )oh the horror) one can go differential pentodes but in general, decoupling the gain stage by a follower is always a good idea as it almost completely prevents blocking, offers a bit of A2/AB2 drive, and makes it possible to use less and/or smaller coupling caps.

I believe that your use of the circlotron topology is also the reason for simplicity of the driver - just compare with totem-pole drivers for fun.

Quote:
2) if a 6C33 or 7241 is used, then independent bias settings are a good idea, done by also having independent driver circuits. At this point, if these tubes are used, then the driver is considerably more complex.
Exactly - there is really no other way but separate drivers (or at least followers) for each tube. Fortunately, one can gt away with less pairs of 6S33S (and similar) but we are still talking doubling or tripling at least one driver stage. With pentodes this complication is shifted to Vg2 regulators. One drawback may be it makes automatic bias much more complex, but you don't meddle with the actual audio signal. Again, as good engineers say, there are no solutions, just different compromises... this shift is however possible due to lower Cin of pentodes, more G1s can be strapped in parallel before a redesign of the actual driver is necessary. Still, for a sane number of tubes in parallel, this is no great issue - unless one is strictly limited in available tube types (as may be the case if one wants a series produced product).

Quote:
3) and 4) are no different if 6AS7s are used, and
Hehe there is no Ig2 on triodes (I was not referring to Ig1) and normally with pentodes this current does not contribute to the output current unless we have a circlotron. Not immediately obvious but in a pentode circlotron one gets floating Vg2 with respect to tube cathode (so independant of output voltage) by cross-connecting Vg2 to Vp of the oposite bank (if a higher Vg2 is needed, it must be pegged on top of the existing Vp as a power supply 'tap'), however, the Ig2 does not go through the load then. By using local regulators for Vgs, referenced to tube cathode, but supplying the regulator from the Vp of the actual bank in question (with extra voltage added if necessary) Ig2 gous out the cathode and through the load - and unlike totem-pole outputs, it does so completely symmetrically for both halves of the circuit. This trick can really only be used in a circlotron, and it reclaims Ig2 as a useful output current, which is actually the default behaviour of a triode - Ig2 is not 'lost'.
Further, because a pentode OTL load line goes somewhat into the knee of the pentode characteristics, at this point the plate current falls and G2 current rises quite a lot so the G2 current contribution is not as small as one would think (usually 2-4% in BPTs well away from the knee). With proper Ig2 limiting (to limit G2 dissipation - no shortcuts here, G2 is tiny compared to the plate and has very low thermal inertia!), you get a good 10%, sometimes more, depending on actual tube. Simply put, this gives you 1 'virtual' extra tube for every 7-10 real ones, which is not bad at all, for just connecting the input to a Vg2 regulator differently.

4) is very similar with 6AS7 because mu of only 2 makes for negligible miller effect. Good current capability and this actually is the reason i mentioned that only 3 commonly available triodes are suitable for OTLs, 6S33S, 6S19P, 6AS7G.

Quote:
5) we get the same short circuit tolerance with the 6AS7s. I have often done a short-circuit as a demonstration at shows by shorting out the amp and then running the preamp at full volume. In general OTLs seem to be tolerant of short circuits as long as there are no stability issues.
Exactly - because the difference in Vak with voltage on the load and no voltage on the load is relatively small, it's not that much more extra dissipation driving a short. With pentodes however that is something one has to consider in a design as minimum Vak can be as much as half that of an equivalent power triode design. Because of this one can be tempted to cut down on tubes - however, output impedance considerations usually prevail (*). Even so, a pentode has a definite maximum current for maximum drive (lets assume Vg1=0 for that) so it's really a current source, hence current is intrinsically limited. In cases as above, this can help with the design.
However - open circuit is a real problem for pentodes. Any amount of current driven into an open circuit clips the amp which means Vak goes down to nearly zero, at which point there is a huge current going through G2. Recycling G2 current helps immensely with that, because the regulator is referenced to the cathode, and this will go up to the plate power supply, the G2 regulator will run out of voltage and Vg2 will drop, ditto current, but as I said, this is only possible for a circlotron. No such problem with triodes - Vak falls, there is no current and there is no alternate path for it to flow like G2 in pentodes. With proper design taking care of dissipation a triode OTL is completely safe with all loads without any extra protection.

(*) unless one relies on a LOT of NFB. It should be stressed that rather large amounts of NFB are far less of a problem with OTLs because with an OTL it's fairly simple to obtain very high full power bandwidth open loop with simple circuits. Doing the same with transformer coupled amps is nearly impossible. Assuming care is taken with stability, NFB will be FAR less atrocious with an OTL, and if one is trying to optimize for efficiency (getting the most out of available tubes), may be a good way to get good performace without compromising too much. The less compromising way would be to desing a loudspeaker specifically with a given OTL in mind, counting on higher (but constant) output impedance
  Reply With Quote
Old 2nd August 2012, 06:32 AM   #805
diyAudio Member
 
Join Date: May 2007
If you want to go low voltage Ua, donīt forget PL519 in enhanced triode mode(or whatever you want to call it), see example. Works well with something like Ua 60-70V and Ug 100V floating on top.
Attached Images
File Type: png ET.png (44.0 KB, 455 views)
  Reply With Quote
Old 5th August 2012, 10:39 AM   #806
ilimzn is offline ilimzn  Croatia
diyAudio Member
 
Join Date: Feb 2005
Location: Zagreb
Quote:
Originally Posted by revintage View Post
If you want to go low voltage Ua, donīt forget PL519 in enhanced triode mode(or whatever you want to call it), see example. Works well with something like Ua 60-70V and Ug 100V floating on top.
Yes - and this is a good example of passing Ig2 through the load, rather than having it lost, so it works like a triode. The circuit does present some challenges as to naming - one could call it 100% UL too
Full pentode would have a local regulator for each G2 referenced to the cathode of the same tube (regardless that the actual G2 current comes from the other cathode).

BTW there was a brief aside on G2 drive. A lot of swing is required, far more than the output signal - a couple 100 Vpp is not unusual. It's often dealt with by having a class B or C driver to prevent needlessly driving the G2 of the output tubes into hundreds of negative volts while the outer PP side gets full drive. G2 drive is also generally very linear, one could almost say too linear, thus best suited for class B amps, provided good control of the crossover region (biassing may be 'fiddly') - the crossover region is very small and needs to be fairly tightly controlled to avoid gm doubling-like effects (very similar to a BJT output amp).
Since we'd be passing a lot of current through the tubes, their net gain WRT Ug2 would be far below 1, so it's a rather 'bad' follower. Still, consider a build like in the pic above (except with drive to G2 and G1 at a fixed (slightly positive?) voltage - at zero output, the tubes are in the middle of the crossover area and there is no feedback to the drive voltage, so this still needs to be well controlled for bias and offset stability (note that some sweep tubes need quite a lot of time to stabilize and have quite big tolerances to begin with). However, at full swing, especially for tubes with low G2 voltages (and these would be preferred in such an application) there is considerable feedback as the cathode tries to follow Ug2. This mechanism does make the bias setting a bit more 'soft' and more manageable WRT a transformer coupled amp with G2 drive, but one still needs to provide hundreds of V, sourcing current, and at negligible distortion...

Last edited by ilimzn; 5th August 2012 at 10:52 AM.
  Reply With Quote
Old 5th August 2012, 01:12 PM   #807
dady is offline dady  Argentina
diyAudio Member
 
dady's Avatar
 
Join Date: Oct 2002
Location: Punta Prima, Menorca, Spain
Send a message via MSN to dady Send a message via Skype™ to dady
Quote:
Originally Posted by atmasphere View Post
Here's a little trick if you are using toroids:

You can stack them if you have a longer bolt to hold on to them. It works best if one is on top and maybe the other two below. You may have to play with phasing of the AC mains so that the magnetic leakage is helpful rather than harmful to the transformer cores.

The other thing is to make sure you use a stainless non-magnetic bolt! A regular steel bolt represents a magnetic short of sorts to the toroid and will heat up a lot hotter than the transformer, no matter if the manufacturer of the transformer supplied a steel bolt, don't use it.
Thanks Mr Atma for this data. Never could imagine take in mind this.
This is a good contribution.
__________________
Esteban Bikic
  Reply With Quote
Old 6th October 2012, 06:12 PM   #808
diyAudio Member
 
Join Date: Mar 2011
Location: Jarna
Got distracted from building OTL2 by (donīt laugh) OTL3.

Click the image to open in full size.

Trying to find some descent toroids having buzzing issues .
  Reply With Quote
Old 6th October 2012, 07:06 PM   #809
dady is offline dady  Argentina
diyAudio Member
 
dady's Avatar
 
Join Date: Oct 2002
Location: Punta Prima, Menorca, Spain
Send a message via MSN to dady Send a message via Skype™ to dady
Thanks god some body still is playing something.
I Guess diyaudio not work any more.....
__________________
Esteban Bikic
  Reply With Quote
Old 6th October 2012, 07:26 PM   #810
diyAudio Member
 
Join Date: Mar 2011
Location: Jarna
Itīs doing the job !

Only thing is i have to crank it up to avoid the buzzing toroid .

Click the image to open in full size.
  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
FS: Vacuum Tube Monoblock Power Amp Kit cuibono Swap Meet 13 7th May 2011 12:49 AM
Vacuum tube VCO Bitrex Tubes / Valves 1 15th March 2011 12:07 PM
HLLY MK III 3 Vacuum Tube Headphone Amplifier Pre Amp hllyelec Vendor's Bazaar 2 19th July 2009 05:54 AM
S3E Vacuum Tube angsuman Tubes / Valves 12 15th March 2008 05:12 AM
DIY Vacuum tube amp turbo_h4 Tubes / Valves 2 1st July 2004 01:48 AM


New To Site? Need Help?

All times are GMT. The time now is 07:38 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright Đ1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2