| The good, the bad, and the ugly OTL amplifiers - Click HERE for Original Thread |
| burnedfingers |
Ok, I'm looking to buy a OTL amplifier at a good price. I need information in order to formulate my decision of to buy or not to buy. I need your opinions, facts, judgements and so forth. Please convince me one way or the other here.
I have been told they are dirty little buggers that cannot be trusted. Some tell me they go thru tubes like they are going out of style and yet others tell me they can be trusted to deliver good and trouble free service. Some tell me they have no bass response and yet other people rave about the low end.
Please help seperate fact from fiction here. |
|
|
| SY |
Number one: what speakers are you using? OTLs tend to be VERY finicky about their loads. If you answer "Quad ESL," then you're a candidate.
They are horribly inefficient, tend to be unreliable, and eat output tubes. I've owned some original Futtermans (Futtermen?) and a home-made monster using 16 6528. Interesting experiences, but I'll never do that again. |
|
|
| Tweeker |
Some of them are pushing their tubes way beyond their long term limits at full power into some loads.
6C33Cs have the potential for thermal runaway if care is not taken. |
|
|
| burnedfingers |
SY,
Thank you for your reply. I have an upgraded version of the Klipsch Cornwall for speakers. Was that 16) 6528's per channel?
I'm looking at (10) 6as7GA's per channel on the amp I am looking at. Has shutdown circuitry as well for faults. |
|
|
| SY |
Yes, 16 per channel. The heat load just from the output stage heaters was nearly a kilowatt. It was a very inefficient, inelegant design, but a friend had been given 64 6528 out of a NASA shake table and wanted to do something with them...
What does the impedance curve of those Cornwalls look like? |
|
|
| tubetvr |
| quote: | | 6C33Cs have the potential for thermal runaway if care is not taken |
I have never heard or experienced that, can you tell me more? I have used 6C33C in OTL amps for many years but never had a serious problem. As a reliability test I was running one amp at full power for 24 hours without anything bad happening except that everything got very warm, I seriously doubt that there could be problems with thermal runaway and why should it happen with 6C33C?
On the subject of OTL amps I think it is important to choose a amp that is designed well, meaning that idle dissipation must be much lower than allowed maximum, that during normal operating conditions average cathode current should not exceed max allowed cathode current, that if parallell tubes are used current is derated, and that ventilation is good enough to remove the heat. However it seems that very few commercial amps fulfil these very basic requirements.
Regards Hans |
|
|
| lousymusician |
| Have you looked at these? I've never heard nor seen one, but they look promising. |
|
|
| Spasticteapot |
Anything with 16+ tubes is going to have a failuire in at least one tube on a regular basis. Make sure to get a tube tester and a few backups while you're at it!
As a side note, you might want to consider trying the SEWA solid-state amplifier found in the Pass Labs section. It's supposed to be a pretty good 7 watt amp, and if you already have an appropriate transformer/PSU components, the cost is very low. (It's also supposed to work very well with a tube preamp - it has no gain!) |
|
|
| burnedfingers |
| quote: | | What does the impedance curve of those Cornwalls look like? |
SY,
The speakers are too heavy for me to take into the shop to plot an exact curve but I did make impedance measurements with my impedance meter that will give a rough idea.
At 100HZ 5.0 ohms
330HZ 6.2 ohms
1K 9.2 ohms
10K 8.5 ohms
I tried to attach a response graph but it wouldn't take it. |
|
|
| coresta |
| Hi Burnedfingers , for sure you'll realize your dream ;) OTLs are for me the cream top of amplifiers . I did my own ones as i didn't want to use 6c33 (many failure) , 6528 (not good in OTLs) , nor 7542 (hard to find) ... king of the OTL is 6080/6as7 . With 10 per channel , you should drive 5 ohms loads without problems . Some tweaks : as previously said , make them run UNDER max rate, have a tube tester to match them , and ... :att'n: heat the NEW ones 24 or 48 hours long without HV to evaporate residual dusts : 10 years long without a failure :rolleyes: |
|
|
| ErikdeBest |
Hi Coresta
Could you post a schematic of your OTL amplifier?
Many thanks
Erik |
|
|
| burnedfingers |
| Are OTL amplifiers lacking in bass? Some people tell me yes and some say no. |
|
|
| vax9000 |
| Tubes are not for OTL. Do you want NFB? If no, the damping factor is almost always bad. If you want a damping factor close to a triode SE, you need 100 6080s, not 10 or 20. No, no tube OTL. |
|
|
| burnedfingers |
You mention the dampening factor is bad with OTL. I haven't seen a tube amplifier with high dampening factor. As a matter of fact there was a thread here a while back that discussed dampening factors in tube amplifiers and if higher dampening factors were better than lower ones. If my memory serves me correctly I believe that lower dampening factors were favored in tube amplifiers and very high dampening factors are favored with solid state amplifiers. The higher the dampening factor in solid state amplifiers the better the amplifier. I can respect your opinion here and would just like some facts and figures to go along with your opinions because I'm still on the fence here.
What is the dampening factor of a OTL amplifier, a push pull amplifier and a SE amplifier if I may ask?
With respect to feedback...Some here surely don't want any while others feel that a small amount is needed. Naturally any amplifier will sound somewhat better with a small amount of feedback.
The solid state amplifiers absolutely need it to sound listenable and for stability. Where do we draw the line here anyway? |
|
|
| Wavebourn |
vax9000, use impedance matching transformer and you will get your damping factor with less tubes. You may take only couple of them: one in cathode follower, one in current source, and you will get mach more damping factor than from one triode with OC and transformer in anode.
|
|
|
| dshortt9 |
I am presently using a Transcendent amp that uses 2 6C33C-B's per channel with quite a bit of NFB. My speakers are homebuilt with good efficiency but low impedance - around 97db and 3 ohm minimums. It is one of the best sounding amps I've had. In my top few favorites of all times.
In my opinion the NFB may not be as harmful in an OTL since it is not being used in a loop that includes a less than perfect output transformer. Just my thought as to why it does not seem to hurt the sound much.
Sometimes the best way to find out if your system will sound good with different gear is to just try it out. Not everything can be figured out with measurements, statistics, and opinions. |
|
|
| 5u4 |
| I made a Transcendent T8 several years ago. It is not a finicky amp at all, has plenty of bass, will easily drive 4 ohms, sounds really great & all that. Downside, it throws off some serious heat & the price of all those power tubes. Owning an OTL was my dream for many years & after I got one, I'd just as well listen to my little Loftin-White. |
|
|
| Giaime |
Bah. Never heard an OTL amp going too good with the Cornwalls.
I would stick with a 6L6GC push pull, or 6550/KT88. McIntosh 275 would be a good candidate ;)
McIntosh + Klipsh is something that one must hear in his life :D |
|
|
| burnedfingers |
| quote: | | Never heard an OTL amp going too good with the Cornwalls. |
Gamie,
Have you ever heard Cornwalls? Anyway they are Cornwall in size and that is about it. They are a highly developed design by DJK that totally kicks the Klipsch to the curb. They test almost ruler flat also. I refer to them as a Cornwall clone because they look almost the same. In terms of performance they are so much smoother and higher efficiency.
I'm interested in your opinion don't get me wrong. Have you heard any OTL amplifiers in which you could lend a opinion based on this information?
Quote:
McIntosh + Klipsch is something that one must hear in his life.
Well, heard them and wasn't impressed. Mac is ok but there are far better amps out there to be had.
Getting back to the OTL amps
So far we have established that they run hot. This is because of a large number of tubes producing mega heat. Note: Good for winter heat in home. There is considerable cost to replace the output tubes. They may or may not go up in a puff of smoke. They do require some feedback to sound halfway good. Note: Feedback
is 23db for the design I am looking at.
Safety: In checking into various designs it should be noted that some do not use transformers in their power design. No design like this should be considered or purchased.:hot: This notice is for SY and the diyaudio board and is meant to protect us so we don't fry our butts.
It shall be noted that some consider the OTL to sound quite good.
We should consider the type of speaker driven and the impedance curve of same. OTL's don't like to drive low impedance loads unless they have a massive amount of output tubes. Some claim long tube life if broken in correctly. I should mention that the 6C33 has a tube socket problem as it cannot withstand the massive heater current and this leads to problems. The 6C33 is a great tube however.
It would be of great interest to see some of the OTL amplifier designs that some of you have made in the past or are currently using. |
|
|
| burnedfingers |
Quote from the Transcendent audio site for the purpose of looking at the feedback aspect of the OTL amplifier.
| quote: | | Let’s look at the Transcendent T8 as an example. It uses 8, EL509 tubes in push-pull. Each 509 has a plate impedance of about 150 ohms. 150 ÷8 = 18.75 ohms. This is unacceptable. Negative feedback is employed to achieve proper performance. The amplifier uses 33 dB of negative feedback which reduces the output impedance to 0.4 ohm thereby achieving outstanding woofer control and the ability to drive 4 ohm speakers. This specification was verified by Stereophile when they reviewed the amplifier. |
Note: the 6AS7 has a plate impedance of 280 ohms so if my math is correct 280/20 sections is 14 ohms which is further reduced by the negative feedback that is applied. The feedback is 23db.
So, in your opinion is this amplifier capable of driving a 4ohm load that I would encounter with my speakers in reproducing low frequency? |
|
|
| burnedfingers |
Quote from the Transcendent audio site for the purpose of looking at the feedback aspect of the OTL amplifier.
| quote: | | Let’s look at the Transcendent T8 as an example. It uses 8, EL509 tubes in push-pull. Each 509 has a plate impedance of about 150 ohms. 150 ÷8 = 18.75 ohms. This is unacceptable. Negative feedback is employed to achieve proper performance. The amplifier uses 33 dB of negative feedback which reduces the output impedance to 0.4 ohm thereby achieving outstanding woofer control and the ability to drive 4 ohm speakers. This specification was verified by Stereophile when they reviewed the amplifier. |
| quote: | | Different tubes will provide similar results. The configuration of the OTL output stage does not matter whether it is series connected push-pull or balanced. The controlling factors here are the number of output devices and the plate impedance of the devices. A 6AS7G is a dual triode with a plate impedance of 280 ohms. An amplifier that uses 8 such tubes will have an output impedance of 280 ÷ 16 = 17.5 ohms. A fourteen tube amplifier will have an output impedance of 10 ohms. |
The amplifier I am considering has 10 6AS7GA's in it per channel thus reducing the output impedance to 14 ohms. This coupled with 23db of negative feedback should allow the amplifier to drive low frequency without a problem in my opinion. |
|
|
| Shoog |
Just an opinion,
I would speculate that most of the sonic advantages of a OTL amp can be achieved by transformer coupling a few 6AS7's into a quality toroidal output transformer. Run the 6AS7's at high current and low voltage. This type of setup gets around the inherent limitations of an output transformer, because high current low voltage has a much lower stepping ratio than a normal output tube which drastically improves the frequency response. Of course you will be running PP.
This is a very rarely tried configuration (because suitable OT are rare and 6AS7's are not liked by any but the OTL crowd) so most people cannot express an honest opinion of its sonic merits.
Running an OTL amp just doesn't make sense on so many levels, when there are good alternatives out there.
Shoog |
|
|
| Spasticteapot |
| quote: | Originally posted by Shoog
Just an opinion,
I would speculate that most of the sonic advantages of a OTL amp can be achieved by transformer coupling a few 6AS7's into a quality toroidal output transformer. Run the 6AS7's at high current and low voltage. This type of setup gets around the inherent limitations of an output transformer, because high current low voltage has a much lower stepping ratio than a normal output tube which drastically improves the frequency response. Of course you will be running PP.
This is a very rarely tried configuration (because suitable OT are rare and 6AS7's are not liked by any but the OTL crowd) so most people cannot express an honest opinion of its sonic merits.
Running an OTL amp just doesn't make sense on so many levels, when there are good alternatives out there.
Shoog |
I disagree. I'm a big fan of OTL amps, and own several.
(Of course, they're all solid-state.)
I would wonder if inexpensive tubes (availible by the truckload on eBay) could be substituted. With 32 tubes/channel, you'd need a huge power transformer, but think of the output power! |
|
|
| SY |
| quote: | | 6528 (not good in OTLs) |
What's the issue with them?
bf, from those impedance numbers, I'd guess that these speakers are not good OTL candidates. |
|
|
| coresta |
| Not linear as 6336A: The cut-off point is met very quickly. 6528 are perfect for SET . ;) |
|
|
| Giaime |
| quote: | Originally posted by burnedfingers
Gamie,
Have you ever heard Cornwalls? Anyway they are Cornwall in size and that is about it. They are a highly developed design by DJK that totally kicks the Klipsch to the curb. They test almost ruler flat also. I refer to them as a Cornwall clone because they look almost the same. In terms of performance they are so much smoother and higher efficiency.
I'm interested in your opinion don't get me wrong. Have you heard any OTL amplifiers in which you could lend a opinion based on this information?
|
My name is Giaime, and I was referring to original Klipsch Cornwall speakers. And I usually don't talk about things I don't know or heard. |
|
|
| ErikdeBest |
Hi Coresta
Thanks for the info! But I thought you did not deal with novals in your own projects? And the 5814, don't know, but being similar/substitute for the 12AU7/ECC82 they should not be that linear?
Erik |
|
|
| burnedfingers |
| quote: | | My name is Giaime, and I was referring to original Klipsch Cornwall speakers. And I usually don't talk about things I don't know or heard. |
Sorry about getting your name wrong.
I'll try not to forget |
|
|
| Tweeker |
| Id read 6C33Cs, an ultralow rp tube, can have positive temperature coefficient related issues with bias and possibly experience thermal runaway, but I guess it really isnt an issue in practice. |
|
|
| tubetvr |
| quote: | | can have positive temperature coefficient related issues with bias and possibly experience thermal runaway |
Do you remember where you read it? I am really interested in all practical information regarding the 6C33C that I can find.
What I have found so far is that there are a lot of rumours, mostly negative related to this tube, I wonder why?
Regards Hans |
|
|
| Tweeker |
Ive been trying to find the source I read about it in the past been, its been frustrating.
Some of what I read boils down to that the tube is alot less dependable at 60watts dissipation than 45watts. Also that just because you can get away with quite impressive far beyond datasheet 2A peaks, doesnt mean that youl have great MBTF doing so. The parameters may shift abit with burnin and/or hard use. |
|
|
| tubetvr |
Hi,
Yes, it seems to be more or less common knowledge that 6C33C will have very long life and be stable if used at 2/3 of max dissipation or there about. Viktor Khomenko at BAT who uses large quantities of this tube has described that parameters can start to drift if the tube is used at max dissipation continously, but he has also said that life is very long if run at or less than about 40W.
As I descríbed before I think it is important to design an OTL amp in a sensible way, I don't think this really matters on which tubes you use but due to the specs of 6C33C it has the advantage of being able to deliver useful output power from only a few tubes.
In my amp with 2 6C33C tubes I use 160V B+ and 200mA idle current so the idle dissipation is only 32W. With max output power which is 25W in 8 ohm the peak current is 2.5A but the average current that is determining the thermal load of the tube is then only 0.8A so the max dissipation will be ~115W which is still way over the max of 60W.
However you can argue that an audio amplifier will only spend a very short time on peak power and it is reasonable to assume that even if you play at clipping level the average power is at least 10dB lower than max.
10dB below max gives in my case an average current of 0.28A and an average dissipation of only 45W so the end result is that you get very long tube life which is also what I have experienced, in my own amp I am still using the original tubes after more than 6 years of use.
Regards Hans |
|
|
| Dave Cigna |
| quote: | Originally posted by burnedfingers
Note: the 6AS7 has a plate impedance of 280 ohms so if my math is correct 280/20 sections is 14 ohms ... |
It depends on the circuit. If the load is in the plate circuit of the 6AS7's (as is the case in a typical amp with an output transformer) then your math is correct. However, if the circlotron circuit is used (as is the case with Atmosphere) then a single 6AS7 will have an output impedance of something in the neighborhood of 80 ohms. That's both sections of one tube in a push-pull circlotron circuit. So 20 tubes in parallel will have an output impedance of about 4 ohms before NFB.
The White cathode follower and Futterman circuits also provide an output impedance that is lower than the plate resistance, but I don't know off hand what it is. You could probably find out with a little googling.
-- Dave |
|
|
| dshortt9 |
| I think the 6C33C-B got pad press because of the cheap sockets that were available for it a few years back. The metal was so flimsy that it made poor contact and fried. Newer sockets available now have taken care of that problem. Also, some people used them at higher than acceptable dissipation and/or voltage. Most designers prefer less than 150V. I always pre-age mine for a few hours on a seperate rig to stabilize the parameters and they work fine. |
|
|
| tubetvr |
| quote: | | The White cathode follower and Futterman circuits also provide an output impedance that is lower than the plate resistance, but I don't know off hand what it is. |
See here: http://members.aol.com/aria3/otlpaper/otlhist.htm
Lowest output impedance you will get with an inverted Futterman Rp/2*(1+u) which for 6AS7 is 280/2*(1+2) = 47ohm The Circlotron is worse and give Rp/2+u = 70ohm. Figures for 6C33C are 11 and 17ohm respectively.
Regards Hans |
|
|
| kevinkr |
I designed a disasterously unsuccessful OTL amplifier based on 6C33 about 10 yrs ago. Take a look at my website for the dire saga. ;)
The 6C33 has the potential to be an excellent tube for OTL, PP and SE amplifiers if certain simple rules are adhered to, and will be reliable if you get a good tube. (Not all tubes available are good. :xeye: )
As pointed out elsewhere in this thread a good socket is a must.
:hot: Delay B+ until cathodes are fully warmed up - this will avoid cathode stripping and more importantly the arcing that will kill your load. :hot:
:hot: Don't run them beyond 45W dissipation, they become much less reliable and drift strongly at higher dissipations. :hot:
Make sure the section to section match is good.
Make allowances in the driver design for the considerable miller capacitance of several of these in parallel, and provide for some limited grid current drive capability as well. (Cathode follower or White CF driver stage?)
My understanding is that you can run the filaments about 5% below nominal and limit inrush current as well for a substantial improvement in life.
Otherwise this might be the best choice for an otl of currently available tubes, a pair will get you about 20W into 8 ohms - try that with a small quantity of 6AS7G.. :D |
|
|
| burnedfingers |
The OTL amplifier I was looking at is on ebay and the auction closes in about 13 hours. After reading different peoples experience and the views of some experts here I'm going to draw the conclusion that I should pass on it.
Ebay item number 110050883473 |
|
|
| SY |
| Those are very pretty, but the design is pre-Futterman and not a particularly good one. Pass, unless the metalwork and tubes are worth the $$$ to you. |
|
|
| burnedfingers |
SY
I appreciate your input. Someday I might try to design and build an OTL amplifier. I'm not thrilled with the Futterman design after pulling up several schematics and patents off the web. The square wave response was another reason. It just plain doesn't look right as its got a glitch on the side indicating a possible problem in the output stage design. |
|
|
| Brian Beck |
It seems to me that an intrepid OTL fan might consider designing and building a matching speaker design that is “OTL-ready”. In other words, one could concoct an amp and a speaker that work together as a system. The main goal would be to make a high impedance speaker. If you used multiple identical drivers connected in series, you could make a high impedance speaker that would be a far better match to an OTL amp. I’m thinking of something like a line array. Suppose, just for discussion, that you made a tall three-way design. Imagine twelve tweeters in a vertical column, next to twelve midrange drivers in a vertical arrangement too. If each tweeter and midrange driver were rated at 8 ohms nominal, you’d have a roughly 100 ohm speaker. Usually, with an array like this, the drivers are connected in series-parallel arrangement to set the total impedance to close to 8 ohms (or maybe 4 ohms). But here the strictly series connection works in favor of the high-voltage, low-current “desires” of an OTL amp. For the woofer section, you probably would not want twelve 12” drivers, but imagine a pair of drivers in series in a sealed enclosure, sized appropriately to create the desired damping. If the cross-over to the midranges could set to a fairly low frequency (which is feasible since there are so many midranges to share the excursion burden), then the woofers would be operating near their system resonance frequency, and the resultant impedance in the bass range can be very much higher than 8 ohms.
I don’t think designing a whole new speaker system just to justify the use of an existing OTL amp makes much sense, but if someone were considering building a speaker and an amp anyway, then this might be an option. |
|
|
| thoriated |
My OTL's balanced driver and output stage combination appears to have an open loop impedance comparable to that of an inverted Futterman (e.g., with 8 6AS7's per channel and about 26db of ac feedback, I was getting a damping factor of about 40 into 8 ohms, before I added positive current feedback to bring that over 100) but it is a completely different topology than either a Futterman or Circlotron.
This OTL design is also DC coupled throughout (and has a ground referenced output). I have been satisfied with tube reliability, considering having had only a handful of output tubes fail in the 18+ years I have been running this amp, although I have replaced several due to falling transconductance with use over time. This is in light of the somewhat elevated output stage supply voltage of +/- 160 - 165 Vdc I am running these tubes at.
This amp behaves very well into loads as low as 2.5 ohms (the lowest I've tried), but you can tell over time that there is an increase of harmonic distortion (slight thinning of sound) at average levels of more than a few watts a side into such impedances. |
|
|
| ErikdeBest |
Hi Thoriated
Sounds really interesting, I mean, high damping factor, DC coupling, etc! Can you show us a schematic of this amp?
Many thanks
Erik |
|
|
| burnedfingers |
| Yes, a schematic would be great. |
|
|
| thoriated |
Hi -
I probably shouldn't post a schematic of the actual amplifier since I think the means by which I accomplished the DC coupling is patentable and I'm not quite ready to throw away the ability to profit from it if necessary (as I approach retirement age), but I can describe how I achieve the higher damping factor, which should be adaptable to at least some other OTLs. Basically, I have a low resistance current sense resistor (around 0.47 ohms in the case of my OTL) on the return leg of each output, and I tie the signal ground to the 'high' side of the current sense resistor. The actual output stage supply ground return is connected to the 'low' side of the current sense resistor. This allows me to sum the positive current feedback directly into the same feedback node as the negative voltage feedback by taking the current feedback off the 'low' side of the current sense resistor. With pot trimming and inverse open loop gain shaping of the current feedback circuit, I can increase the effective damping factor by up to ten times across the audio band with my OTL, so have achieved damping factors of 200 out to 20khz when trimmed (but my OTL has an open loop response down only 4db at 20khz which allows such a high damping factor to be achieved at higher frequencies).
One reason I like this approach of increasing damping factor is, besides the fact that it doesn't require any basic changes to the cicrcuit topology is that I am effectively reducing the overall negative feedback under higher current draw situations. I think this probably allows the best amplifier stability into load. |
|
|
| ErikdeBest |
Hi Thoriated
Thanks for the explanation, can't say I understand it completely from the first read after just 5 hours of sleep and one coffee, but when time passes by...
I think you are right on the uniqueness of the DC-coupling thing in tube OTL's and it's potential for patent. Good luck... (kid feeling the sweet taken away...)
Erik |
|
|
|
|
